Ju 

Hyun Lee
Michael J. Ostwald
Ning Gu

Design Thinking:
Creativity,
Collaboration
and Culture
Design Thinking: Creativity, Collaboration
and Culture
Ju Hyun Lee Michael J. Ostwald
• •

Ning Gu

Design Thinking: Creativity,

Collaboration and Culture

123
Ju Hyun Lee Michael J. Ostwald
UNSW Built Environment UNSW Built Environment
University of New South Wales University of New South Wales
Sydney, NSW, Australia Sydney, NSW, Australia

Ning Gu
UniSA Creative
University of South Australia
Adelaide, SA, Australia

ISBN 978-3-030-56557-2 ISBN 978-3-030-56558-9 (eBook)

https://doi.org/10.1007/978-3-030-56558-9
© Springer Nature Switzerland AG 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

In its simplest form, “design thinking” refers to the reasoning processes that occur
during the act of creating a product. It encapsulates the cognitive strategies and
behaviours of people who are engaged in developing innovative solutions to
problems, or identifying new opportunities in a complex marketplace or ecosystem.
Importantly, design thinking offers a valuable counterpoint to “scientific thinking”.
The former is typically characterised as user-centred, inventive and even produc-
tively disruptive, while the latter is regarded as methodical, logical and reductive.
Despite the simplicity of this characterisation, design thinking is fundamentally
concerned with developing creative or original responses. At a time when the
world’s headlines are dominated by grand challenges such as ecological dilemmas,
economic crises and resource shortages, the capacity to develop innovative solu-
tions to problems and sometimes even re-define the problems has never been so
critical.
Because design thinking is primarily used to solve “real-world”, “ill-defined” or
“wicked” problems, it rarely follows a linear path to its destination. Instead, the
cognitive paths taken by designers are more likely to include spiralling off-ramps,
intricate intersections, elaborate cul-de-sacs and opportunistic shortcuts. Novice
designers, and those from disciplines more accustomed to linear processes, typically
struggle with complex, real-world problems. Not only are there a myriad of factors
to consider, but the designer may also face the prospect that an optimal solution
does not exist. In a sense, the labyrinthine pathways that confront the novice
designer not only do not contain clear signposts to freeways; they sometimes don’t
even lead to the desired destination. In contrast, expert designers develop cognitive
strategies to assist them to navigate between different starting points and alternative
destinations. They critically review and repeatedly question both the purpose
of their travels and possible routes to its conclusion. Expert designers expect to
navigate winding paths and discover dead ends while looking for hidden tunnels or
fortuitous bridges to their journey’s end. They combine convergent and divergent
thinking along with deductive and inductive reasoning to propose creative solutions
which satisfy their goals.

v
vi Preface

Designers, whether novice or expert, are not without resources to assist them to
navigate the maze of possibilities that is the design process. In the context of our
book, four of the most powerful resources for design thinking are computational
platforms, creative thinking, collaborative processes and diverse teams. The first
of these encompasses advanced computational systems and digital platforms, which
is a new addition supporting design thinking with the recent advancements of
Information and Communication Technology (ICT). These digital design envi-
ronments support rapid modelling, testing and evaluation of alternative options.
They also enable the development of enhanced cognitive strategies and skills. The
second resource is arguably the reason design thinking is so sought after. Certain
cognitive strategies in design have been linked to creative responses or solutions.
Creativity is the capacity to find alternative, novel or surprising paths to a desti-
nation. The third resource arises from the realisation that designers rarely work
alone and is currently also associated with the ways computational platforms can be
used to support collaboration and to leverage collective intelligence in the design
process. The cognitive operations of the crowd are increasingly being used to
develop collaborative solutions to problems. The final resource reflects on the fact
that most designers operate in teams, and some of the most effective teams are
diverse ones. The utilisation of digital tools and networked technologies has made
teamwork across geographical and cultural boundaries occurred more frequently
and readily. Whereas a cognitively homogeneous team will tend to blaze a
pre-determined trail to a finite destination, a multi-cultural team is more likely to
assess unconventional paths and question collective assumptions. It is not sur-
prising then that diverse teams are often linked to more creative solutions. There
are, however, challenges and opportunities for design thinking when different
cultures and languages are combined in the design process. Design is not a uni-
versal language and design thinking both shapes, and is shaped by, the language
used to communicate it.
Collectively, these themes in design thinking—computational, creative, collab-
orative and cultural—define the scope of this book. The first theme sets the context
for the book, and the latter three shape its content. The focus of this book is on
design thinking in digital or computational environments. Within this general
context, the book examines three themes: creativity, collaboration and culture.
The catalyst for Part I in this book is the claim that certain cognitive strategies or
approaches to design thinking are more likely to result in creative outcomes. Part I
uses evidence derived from empirical studies to develop an understanding of the
way computational environments shape creative design thinking and may lead to
more inventive outcomes. Part II considers the cognitive dimensions of the oper-
ations of design teams, crowds and collectives. Collaborative design thinking has
received relatively limited attention in past research, and there are even fewer
frameworks available to understand the way collective intelligence might operate in
design. Drawing on a combination of a critical literature review, conceptual
frameworks and empirical evidence, Part II expands our knowledge of design
cognition in a collaborative context. Finally, Part III delves into territory that has
only rarely been considered by design researchers, the impact of culture on design
Preface vii

thinking. While, in a pragmatic sense, linguistic and cultural differences may dis-
rupt communication in a design team, at a more fundamental level they may also
shape the way people think. Part III contains some of the most advanced research
undertaken into this last dimension of design thinking.
This book has been written for designers, students and scholars who are inter-
ested in understanding the cognitive operations that occur in design thinking. The
primary methods used for this purpose are explained in Chap. 1, and experimental
results are presented throughout the book to support discussions about and reviews
of past research.
The authors of this book have backgrounds in architecture, interior design and
urban design, and all three have expertise in computational design. We have sep-
arately completed major design projects in different sectors of the economy and in
multiple countries. Significantly, we have different native languages (Korean,
English and Chinese, respectively) and have also worked throughout our careers in
multi-lingual and multi-cultural teams. These factors have shaped our approaches to
several themes in this book and our interests in the role of collaboration and culture
in design thinking.
An additional factor shaping the content of this book is that we approach design
thinking from a dual perspective: First, as qualified, experienced designers and
second, as researchers who have conducted extensive formal studies in the field.
This combination is not as common as readers might think. Many designers resist
the idea that their cognitive processes can be studied, insights revealed and patterns
identified in experimental settings. Conversely, a surprising number of researchers
in the field of design thinking have only limited experience designing. This dual
perspective—designers and researchers—shapes the way we interpret several
themes in this book. It helps us to “reality test” our own experimental results, and to
be productively sceptical about claims and theories. It means, for example, that we
provide several alternative definitions of key concepts, rather than advocating for a
singular or emphatic one. We also draw from and acknowledge past research in the
field, and in other pertinent fields, even though we will not always agree with it.
Finally, the new empirical research presented in this book, along with much
of the literature reviewed for it, was developed from studies in the traditional design
fields of architecture, industrial design and interior design. These are core disci-
plinary backgrounds in design cognition research. Nevertheless, the lessons and
findings of the research in this book are applicable to design thinking across other
fields including engineering, business, management, science and the arts. This is a
key principle of the new discipline of design thinking that the lessons learnt in the
traditional design domains offer significant opportunities for the world. Or, to return
to the analogy used earlier in this preface, design thinking can open up new
pathways to innovation, regardless of the disciplines or fields involved.

Sydney, Australia Ju Hyun Lee

Sydney, Australia Michael J. Ostwald
Adelaide, Australia Ning Gu
Acknowledgements

This book has evolved out of research undertaken over the last decade by the
authors, and it has been supported by many people whose contributions we wish to
acknowledge. We would especially like to thank our colleagues and research
assistants, Ji Suk Lee, Maria Roberts, Chris Burns, Katie Cadman, Dr. Sue Sherratt,
A/Prof. Julie Jupp, Prof. Richard Tucker, Prof. Anthony P. Williams, Prof. Jane
Burry, Prof. Mary L. Maher, Prof. Marc A. Schnabel, Prof. Mark Taylor, Prof. Mi
Jeong Kim and Prof. Robin Drogemuller.
The ideas contained in this volume were also shaped by the generous responses
of the editors and anonymous referees of the following journals, conferences and
book projects: International Journal of Technology and Design Education,
Architectural Science Review, International Journal of Design Creativity and
Innovation, International Journal of Architectural Computing, Collaboration and
Student Engagement in Design Education, Partners for Preservation, Design
Computing and Cognition (DCC), Computer-Aided Architectural Design Futures
(CAADFutures), Design Research Society (DRS), ACM Conference on Creativity
and Cognition, Architectural Science Association (ASA/ANZAScA) and
Computer-Aided Architectural Design Research in Asia (CAADRIA).
Some sections of this book are derived from data and manuscripts that were
previously published and have been substantially revised, expanded or updated for
the present work. Specifically, Chap. 2 includes a revised section from two pub-
lications: Lee, Ju Hyun, Ning Gu, and Anthony P. Williams. 2014. Parametric
design strategies for the generation of creative designs. International Journal of
Architectural Computing 12 (3); Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald.
2015. Creativity and parametric design? Comparing designer’s cognitive approa-
ches with assessed levels of creativity. International Journal of Design Creativity
and Innovation 3 (2). Chapter 4 draws on material published in Lee, Ju Hyun, and
Michael J. Ostwald. 2019. Measuring cognitive complexity in parametric design.
International Journal of Design Creativity and Innovation 7 (3). Chapter 6 revises a
model and builds on materials published in Lee, Ju Hyun, and Ning Gu. 2019.
Historical Building Information Model (BIM)+: sharing, preserving and reusing
architectural design data. In Partners for Preservation: Advancing Digital

ix
x Acknowledgements

Preservation through Cross-community Collaboration, ed. Jeanne Kramer-Smyth.

London: Facet Publishing. Chapter 8 includes revised sections and extended ideas
originally presented in Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2016. The
language of design: Spatial cognition and spatial language in parametric design.
International Journal of Architectural Computing 14 (3). Chapter 9 revises and
expands material published in two publications: Lee, Ju Hyun, Ning Gu, and
Michael J. Ostwald. 2019. Cognitive and linguistic differences in architectural
design. Architectural Science Review 62 (3); Gu, Ning, Michael J. Ostwald, Ju
Hyun Lee, and Maria Roberts. 2019. Developing Pedagogical Solutions to
Linguistic and Cultural Barriers in Design Education Supporting Asian
Architecture Students. Canberra, Australia: Australian Government Department of
Education and Training, Skills and Employment.
We gratefully acknowledge the support of the Australian Government
Department of Education and Training, Skills and Employment, UNSW Scientia
Fellowship Program, University of New South Wales, Sydney, and University of
South Australia. All of the design experiments reported in this book were the
subject of formal Human Research Ethics Approvals: No. H-20110313 and
No. H-20160219 (University of Newcastle).
Contents

1 Introduction: Exploring Design Thinking . . . . . . . . . . . . . . . . . . . . 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Defining Design Thinking . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Research into Design Thinking . . . . . . . . . . . . . . . . . . . 5
1.1.3 The Digital Anthropocene . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Creativity, Collaboration and Culture . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Creativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Collaboration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.3 Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Research Method I: Protocol Analysis . . . . . . . . . . . . . . . . . . . 14
1.3.1 Studying the Design Process . . . . . . . . . . . . . . . . . . . . . 14
1.3.2 Protocol Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.3 Coding Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.4 Research Method II: Expert Panel Assessment . . . . . . . . . . . . . 21
1.4.1 Design Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.4.2 Evaluation Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Part I Creativity
2 Design Strategies and Creativity . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2 Study I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.2.1 Research Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.2.2 Coding Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.3 Sketch-Based Design Strategies . . . . . . . . . . . . . . . . . . 43
2.3 Study II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3.1 Parameter and Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.2 Research Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

xi
xii Contents

2.3.3 Results . . .............. . . . . . . . . . . . . . . . . . . . . 53

2.3.4 Parametric Design Strategies . . . . . . . . . . . . . . . . . . . . . 56
2.4 Design Strategies .............. . . . . . . . . . . . . . . . . . . . . 58
2.5 Conclusion . . . . . .............. . . . . . . . . . . . . . . . . . . . . 60
References . . . . . . . . . . .............. . . . . . . . . . . . . . . . . . . . . 61
3 Creative Micro-processes in Parametric Design . . . . . . . . . . . . . . . 65
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2 Four Parametric Design Activities . . . . . . . . . . . . . . . . . . . . . . 67
3.2.1 Changing Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.2 Perceiving Geometries . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.2.3 Introducing Algorithmic Ideas . . . . . . . . . . . . . . . . . . . . 74
3.2.4 Evaluating Geometries . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.3 Creative Micro-processes in Parametric Design . . . . . . . . . . . . . 76
3.3.1 Patterns of Parametric Design Activities . . . . . . . . . . . . 76
3.3.2 Creative Micro-processes . . . . . . . . . . . . . . . . . . . . . . . 79
3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4 Measuring Cognitive Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.3 Measuring Cognitive Complexity . . . . . . . . . . . . . . . . . . . . . . . 89
4.4 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.4.1 Entropy Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.4.2 A Coding System for Content Complexity . . . . . . . . . . 93
4.4.3 Linkography Measures for Structural Complexity . . . . . . 95
4.5 Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.5.1 Content Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.5.2 Structural Complexity . . . . . . . . . . . . . . . . . . . . . . . . . 102
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

Part II Collaboration
5 Collaborative Design: Team Cognition and
Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.2 Cognition and Representation for Teamwork . . . . . . . . . . . . . . 116
5.2.1 Team Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.2.2 Cognitive Representation . . . . . . . . . . . . . . . . . . . . . . . 121
5.2.3 Design Team Cognition (DTC) Model . . . . . . . . . . . . . 125
5.3 Design Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.3.1 Problem and Solution Spaces . . . . . . . . . . . . . . . . . . . . 129
5.3.2 Geometric and Algorithmic Modes . . . . . . . . . . . . . . . . 132
Contents xiii

5.4 Design Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134

5.4.1 Design Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.4.2 Spatial Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6 Design Thinking and Building Information Modelling . . . . . . . . . . 147
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.2 Digital Design Collaboration and BIM . . . . . . . . . . . . . . . . . . . 148
6.2.1 Design Collaboration Through Digital Platforms . . . . . . 149
6.2.2 BIM Collaboration Essentials . . . . . . . . . . . . . . . . . . . . 150
6.3 An Advanced BIM Knowledge Framework . . . . . . . . . . . . . . . 152
6.3.1 A BIM Knowledge Framework . . . . . . . . . . . . . . . . . . . 153
6.3.2 Six Phases of the BIM Knowledge Framework . . . . . . . 155
6.3.3 Five Modules of the BIM Platform . . . . . . . . . . . . . . . . 157
6.4 BIM Futures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
7 Design Thinking and the Digital Ecosystem . . . . . . . . . . . . . . . . . . 165
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.2 A Digital Design Thinking (DDT) Framework . . . . . . . . . . . . . 167
7.2.1 Types of DDT Processes . . . . . . . . . . . . . . . . . . . . . . . 167
7.2.2 Key Functionalities of Interactive and Collective
Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
7.3 Detailed Examples and Analysis . . . . . . . . . . . . . . . . . . . . . . . 175
7.3.1 Mobile Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.3.2 Situated Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
7.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
7.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

Part III Culture

8 Design Thinking Across Borders . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
8.2 Research Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
8.3 Design Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
8.3.1 Coding Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
8.3.2 Linkographic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.4 Cognitive and Syntactical Complexities . . . . . . . . . . . . . . . . . . 199
8.4.1 Cognitive Complexity . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.4.2 Syntactical Complexity . . . . . . . . . . . . . . . . . . . . . . . . . 203
xiv Contents

8.5 Spatial Language . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

8.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
9 The Language of Design Thinking . ....... . . . . . . . . . . . . . . . . . 211
9.1 Introduction . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 211
9.2 Design Thinking and Language Model . . . . . . . . . . . . . . . . . . . 213
9.3 Methodological Procedure . . . . ....... . . . . . . . . . . . . . . . . . 214
9.3.1 Design Experiment . . . . ....... . . . . . . . . . . . . . . . . . 214
9.3.2 Analysis Procedure . . . ....... . . . . . . . . . . . . . . . . . 215
9.4 Results . . . . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 215
9.4.1 Coding Results . . . . . . ....... . . . . . . . . . . . . . . . . . 215
9.4.2 Correlation . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 227
9.4.3 Comparison . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 229
9.5 Discussion . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 230
9.6 Conclusion . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 232
References . . . . . . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . 232

Part IV Conclusion
10 Conclusion: Three C’s of Design Thinking . . . . . . . . . . . . . . . . . . . 237
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
10.2 Creative Design Thinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
10.3 Collaborative Design Thinking . . . . . . . . . . . . . . . . . . . . . . . . 241
10.4 Cultural Design Thinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
10.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
About the Authors

Dr. Ju Hyun Lee is a Scientia Fellow and Senior Lecturer in the Faculty of the
Built Environment, UNSW Sydney. He has made significant contributions to the
fields of architectural and computational design, since his first lecturing position
(Digital Design) in 2003, South Korea. He was invited to become a visiting aca-
demic at the University of Newcastle in 2011, where he as a senior lecturer com-
pleted 5 years of post-doctoral studies in design computing and cognition. He was a
Senior Research Fellow at the University of South Australia in 2018. He is
co-author with Michael J. Ostwald of Grammatical and Syntactical Approaches in
Architecture (IGI Global 2020). He led two externally funded research projects,
supported by OLT and DFAT, in Australia.

Prof. Michael J. Ostwald is Associate Dean of Research and Professor of

Architecture at UNSW, Sydney. He has a Ph.D. in architectural history and theory
and a D.Sc. in design mathematics and computing. He is Co-Editor-in-Chief of the
Nexus Network Journal: Architecture and Mathematics (Springer) and on the
editorial boards of ARQ (Cambridge) and Architectural Theory Review (Taylor and
Francis). He is Co-editor with Kim Williams of Architecture and Mathematics from
Antiquity to the Future (Springer 2015), Co-author with Josephine Vaughan of The
Fractal Dimension of Architecture (Birkhäuser 2016), Co-author with
Michael J. Dawes of The Mathematics of the Modernist Villa (Birkhäuser 2018) and
co-author with Ju Hyun Lee of Grammatical and Syntactical Approaches in
Architecture (IGI Global 2020).

Dr. Ning Gu is Professor in Architecture at UniSA Creative, University of South

Australia (UniSA). He is Deputy Director of Australian Research Centre for
Interactive and Virtual Environments (IVE). His most significant contributions have
been made towards research in design computing and cognition. The outcomes of
his research have been documented in over 180 peer-reviewed publications. His

xv
xvi About the Authors

research has been supported by prestigious Australian research funding schemes

from Australian Research Council, Office for Learning and Teaching, and
Cooperative Research Centres. He is an Associate Editor of Architectural Science
Review (Taylor & Francis) and has guest-edited/chaired major international
journals/conferences in the field.
Abbreviations

AEC Architecture, Engineering and Construction

AI Artificial Intelligence
ANOVA Analysis of variance
AR Augmented Reality
ASE Analysis–Synthesis–Evaluation
BIM Building Information Modelling
C Content-based functionality or Clause
CAD Computer-Aided Design
CAT Consensual Assessment Technique
CD Collective Design
CI Collective Intelligence
CM Critical Move
CPSS Creative Product Semantic Scale
CSCW Computer-Supported Cooperative Work
DDT Digital Design Thinking
DMM Distributed Mental Model
DMS Document Management System
DT Divergent Thinking
DTC Design Team Cognition
F Functionality or Frequency
FBS Function–Behaviour–Structure
FM Facility Management
GIS Geographic Information System
GPS Global Positioning System
H Entropy value as a level of complexity
HC Cognitive complexity
HS Structural complexity
I-C-L Interacting–Collecting–Leveraging
ICT Information and Communication Technology
IFC Industry Foundation Classes

xvii
xviii Abbreviations

IM Interactive Media
IoT Internet of Things
L Location- and time-based functionality
LBS Location-Based Service
LOD Level of Development
LTM Long-Term Memory
MAR Mobile Augmented Reality
MM Metal Model
MR Mixed Reality
P Probability
PCP Personal Construct Psychology
PDA Personal Digital Assistant
SD Standard Deviation
Sig. Statistical Significance
SMM Shared Metal Model
SNS Social Network Service
STM Short-Term Memory
T T-unit
TM Transactive Memory
TMM Team Mental Model
TTCT Torrance’s Tests of Creative Thinking
TUI Tangible User Interface
U User-based functionality
UMPC Ultra-mobile PC
VPE Visual Programming (language) Environment
WPS Wi-Fi Positioning System
Chapter 1
Introduction: Exploring Design Thinking

Abstract This chapter provides a background to the concept of “design thinking”,

as it is defined and used in the field of design research. Thereafter, the chapter
introduces three themes in design thinking—creativity, collaboration and culture—
which have become increasingly important in the last decade. It briefly describes
the content and structure of the present book, which includes detailed explanations,
explorations and developments of design thinking in terms of creativity, collaboration
and culture. In addition, throughout this book, empirical studies of these themes
are typically presented using data derived from protocol analyses of experiments,
sometimes coupled with expert assessment. Thus, this chapter also outlines these
methods and their capacity to support exploration of design cognition, processes and
activities.

1.1 Introduction

In the last two decades, “design thinking” has grown in global significance, placing
the design process and its products at the forefront of strategic thinking and planning
around the developed world. Design thinking has been praised for promoting inno-
vative solutions to complex, multi-variable problems in society and industry (Brown
2009; Martin 2009; Neumeier 2009). It has also gained considerable popularity and
credibility in business and management because it supports the creation of new ideas
that can present a unique competitive advantage in a market, or even potentially
create a new market. Design thinking has not only been incorporated into business
strategy (Brassett and O’Reilly 2015) but into organisational and social structures. In
a complex world, under increasing economic, cultural and political pressure, design
thinking appears to offer “a fruitful balance between intuitive thinking and analytical
thinking” (Martin 2009, p. 137).
This book has a particular focus on design thinking in a world where digital
and computational processes are omnipresent. The origins of design thinking are
often traced to research conducted from the 1960s to 1990s, and much has changed
in the world since then. Moreover, many of the major theories of design thinking
were developed from interviews with designers who were educated in the 1930s and

© Springer Nature Switzerland AG 2020 1

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_1
2 1 Introduction: Exploring Design Thinking

1940s. They used hand-sketching, tracing paper, materials almanacs, slide rules and
physical models to develop and test their ideas. Over time, however, such tools and
methods have diminished in significance in the design professions, being replaced
with Computer-Aided Design (CAD) drawings and Building Information Modelling
(BIM) models, rapidly fabricated prototypes, Virtual Reality (VR) and Augmented
Reality (AR) simulations (Gu and Ostwald 2012; Ostwald 2012). The contrac-
tual conditions governing design as a process have also changed in the last few
decades and so too have liabilities and responsibilities. Even design pedagogy has
evolved throughout this period, rejecting Beaux Arts and master-apprentice models
in favour of structured-didactic programs, student-centred learning, inquiry-based
approaches and flipped classrooms. Contemporary education uses recorded lectures,
online discussion boards and “live” portfolios. First-year design students are imme-
diately introduced to the software and processes they will need when they graduate
and commence work. Today’s students are also “digital natives”, more at home with
a laptop than pencil and paper. They are less inclined to use sketchbooks, tracing
paper or cardboard models, and have never heard of materials almanacs or scale
rulers. Considering all of these developments in design practice, is it reasonable to
assume that design thinking is still the same?
There are multiple reasons why the design process may be different today because
of the advent of digital tools and systems. For example, classic models of the design
process always set aside time for reflection and evaluation (Schön 1983), which
is now potentially diminished because of the efficiency of new technology. The
modern three-dimensional (3D) computer model is not just an advanced form of hand
drawing. In a narrow representational sense, a hand-drawn elevation of a design could
look almost identical to one produced in a computer; however, the similarities end
there. The computer model can be tested for structural, environmental and economic
performance and optimisation. A change in the computer model flows through all of
these software simulations, updating the environmental or manufacturing properties
and showing the designer the cost implications of a decision. The hand drawing and
the BIM model are definitely not the same in their capacity to model, test or adapt a
design. More significantly, contemporary CAD tools like parametric software have
automated multiple stages of the design ideation, visualisation and testing processes
(Woodbury 2010). The designer’s role in a parametric process is to define the limits,
controls and functions that are required of an object or building, along with identifying
the evaluation criteria for it to be successful (Jabi 2013). Thereafter, the designer’s
role is to let the software generate compliant solutions, before taking responsibility
for the process once more (Ostwald 2010). Regardless of whether we call the current
era the “digital Anthropocene” or the “post-digital era” (because the ubiquity of
digital processes effectively renders them invisible), the modern world’s operations
and systems are shaping design thinking in ways that were not anticipated a decade
ago.
This book considers three themes in design thinking, all of which relate to the
impacts of digital technology. The first of these is creativity, which continues to
be a major motivation for many disciplines to adopt design thinking. Despite the
ongoing fascination with creativity, parametric design, Artificial Intelligence (AI)
1.1 Introduction 3

and machine learning are all now being linked to increased creativity, raising ques-
tions about the design process and its outcomes. The second theme is collaboration. In
the past, design thinking has tended to be presented or discussed as if it is an individual
process. While some design process models acknowledge co-design or consultation,
the majority of the well-known theories rarely mention that design has always been a
team process. This is especially the case today, where digital tools in the design fields
are innately collaborative, allowing or even requiring multiple people from different
backgrounds to simultaneously participate in the design process. Design process
models based on the “crowd” and the “swarm” also raise significant questions for
the field of design thinking (Phare et al. 2016, 2018). Finally, in an increasingly glob-
alised world, design teams have members drawn from different nations with different
languages and cultures. This poses a different set of problems and opportunities for
design thinking and cognition. Furthermore, design is not, as later chapters in this
book reveal, a universal language, and this signals the need for a better understanding
of culture and language in design.
This book examines these three themes—creativity, collaboration and culture—in
design thinking. The present chapter introduces design thinking as a concept and a
field of research. It also describes the context of the book, which is motivated by
gaining a better understanding of the impacts of advances in technology on design
thinking. Thereafter, it outlines the three themes which structure the book and the
content of its chapters. Finally, because the results of design experiments by the
authors are used to ground this book, the two major research methods used are
also introduced: protocol analysis (Cross and Cross 1995; Goldschmidt 1995) and
expert panel assessment (Amabile 1983a; Besemer and O’Quin 1987). Subsequent
chapters introduce specific methodological factors including the different coding
schemes used to explore the three research themes.

1.1.1 Defining Design Thinking

Despite being widely used today, the phrase “design thinking” doesn’t have an agreed
definition or use. Further complicating this matter, there is a growing field of research
and pedagogy called Design Thinking that, surprisingly for some, doesn’t necessarily
have a close connection to the classic design disciplines of architecture, industrial
design and graphic design. Attempts to explain what is meant by design thinking
often resort to discussions about the difference between a process and a product.
Such explanations typically explain that a shift has occurred over time, from “design”
being a discipline-specific product to being a trans-disciplinary cognitive process.
The reality, however, is more complex and it helps to understand that these two
positions are not mutually exclusive. De-coupling design as a process from design
as a product is not a simple thing to do, even though it is occasionally useful for
researchers to consider one or other in isolation.
Part of the difficulty with defining what is meant by “design thinking” is that the
word design is both a verb and a noun, with its context in a sentence assisting the
4 1 Introduction: Exploring Design Thinking

reader to understand its role (Glanville 1999). In the former usage as a verb, “design”
refers to the process of conceptualising, refining, testing and documenting an object
or outcome. In the latter usage as a noun, it refers to the product of this process,
being a detailed representation or plan of an as yet unrealised object or outcome.
Design is therefore, both a product and a process, and the difference between the
two is contingent on its context or use. Importantly, the process of design necessarily
includes stages where the products of design are created and reviewed. Of equal
significance, the designed product is never static or complete; it is always part of
a larger process that includes approvals, revisions, manufacturing, post-occupancy
or post-user variations and repurposing. The classical recursive model of the design
process features a series of stages, which cycle through ideation, problem-solving,
representation and testing, before returning to ideation once more to start a new cycle.
This sequence is repeated multiple times throughout the design process, each loop
resulting in a more refined or compliant product. As such, the grammatical meaning
of “design” shifts restlessly between noun and verb, spiralling between process and
product and back again. It is certainly possible to talk about the “design process”, or
the “designed product” in isolation, but from a larger perspective the two states are
innately temporary.
The phrase “design thinking” also has multiple potential meanings, even before
specific definitions are considered. It could, for example, refer to the process of
thinking creatively and how this occurs. It could also describe the product of creative
thinking and its characteristics. The context in which the phrase is used is simi-
larly significant as, for example, Rowe (1987) uses “design thinking” to describe
the procedure used by architects when responding to a client’s brief or program. For
Rowe, “design” is a noun and “thinking” is an adverb, and the phrase “design think-
ing” refers to the thought process underpinning a proposition. In contrast, Buchanan
(1992) defines “design thinking” as a cognitive process which can be used to solve
a particular type of complex or “wicked” problem. For Buchanan, “design” is a
verb and “thinking” is a noun, and the phrase “design thinking” describes the gener-
alised thought process of the designer. Such critical differences of opinion about
the meaning of “design thinking” are often forgotten in books and courses about
the topic, although the field of design thinking more commonly defines its scope as
being focussed on the cognitive process.
This book does not propose a new definition of “design thinking” or attempt to
question the validity of its use to delineate a field. Instead, it takes a more inclu-
sive approach, accepting that new knowledge about design thinking is developed
by closely and critically observing the interactions between design as a process and
as a product. Indeed, despite design thinking’s notional focus on process, it often
positions the resultant product as evidence of the validity of the underlying cogni-
tive operation. Furthermore, the awards and recognition to a designer or product are
commonly used to argue for the inherent quality or innovation of the underlying
process (Schön 1983). As such, separating process and product in design is not only
difficult; it might even be counterproductive.
A further factor to consider when approaching the topic of design thinking is
that all past empirical research into the topic relies on evidence that is at least one
1.1 Introduction 5

step removed from the process of design. As the following section discusses, such
evidence is typically presented in the form of interview transcripts, survey data and
activity logs developed from observational studies. This evidence is invaluable to
the field, but it must be generalised into models of design cognition in order for it
to be applied. The effectiveness of these models also requires a level of validation,
which inevitably returns to an assessment of the designed product and its functional
or innovative properties. Thus, both the process and the product may need to be
considered to develop robust or significant findings of design thinking.
All of these factors have shaped the way design thinking is positioned and
described in this book. The decision to avoid framing a finite definition is a practical
and logical one. The purpose of this book is to develop and test new knowledge about
design thinking in its broader sense, to allow readers to make their own decisions
about its relevance to other fields and practices.

1.1.2 Research into Design Thinking

The origins of design thinking are typically traced to the works of the early twentieth-
century philosophers of science, who formulated various logic models to explain the
world and our understanding of it. They created frameworks or systems to study a
wide range of processes and also to question how we construct knowledge about
them. Following the footsteps of Karl Popper and Thomas Kuhn, and shaped by
structuralist and post-positivist thinking, designers and architects soon became inter-
ested in developing what Wallas (1926) calls a “scientific explanation” of thinking.
By the 1960s, an increasing number of designers had proposed scientific, mathe-
matical and philosophical theories to explain design as a process (Alexander 1964;
Asimow 1962; Gordon 1961; Osborn 1963). Soon thereafter, cognitive scientists led
by Simon (1969), and engineers like McKim (1973), began to treat design as a type of
science. In the following decade, Cross (1982) and Lawson (1980) formulated many
key concepts in design thinking, noting that designers use different problem-solving
techniques to non-designers. Various behaviours like “satisficing” and “solution-
centred” thinking soon began to be used to characterise what was becoming known
as “designerly ways of knowing” (Cross 2011, 2018).
In the 1990s, the volume of research produced about design thinking increased
rapidly and multiple models were soon developed for explaining the design process.
Most of these models conceptualise the “creative design process” as a series of actions
or steps that take place in the mind of the designer. Thus, the design process is innately
cognitive and is also described as “design cognition”, even though the actual mental
activities that take place during design are still largely unexplored (Chan 2015). As
an example of a design model, Brown and Wyatt (2010) propose a systems approach
with three overlapping spaces: inspiration, ideation and implementation. A five-stage
design thinking model developed by the Hasso Plattner Institute of Design at Stanford
(2010) is also widely applied to teach design in various domains. Its stages, which
6 1 Introduction: Exploring Design Thinking

partially mirror those in the classic architectural process model of the mid-twentieth
century, are as follows: empathise, define, ideate, prototype and test.
By the twenty-first century, design thinking had become a field of research in its
own right. Its focus also shifted during this time, away from applications in the tradi-
tional creative design disciplines to those in business, management or engineering.
Design thinking, in this new field, is no longer the sole domain of design profes-
sionals, but all who seek solutions to complex problems. As such, design thinking
is often viewed today as a type of problem-solving process that aims to produce
innovative outcomes and to address ill-defined or “wicked” problems. During this
period, research into design thinking, intelligence and behaviour soon expanded to
include six disciplines—psychology, linguistics, neuroscience, computer science,
anthropology and philosophy (Miller 2003). This breadth of application is evidence
of the growing importance of the field but is also a reflection of the origins of the
methods that have been used to try to understand design as a cognitive process.
Research into design thinking and cognition is typically undertaken using one
of five methods (Cross 2011). The first, and also one of the earliest, is interviews
with designers. The second is observation of designers and case studies of their
work. The third is experimental studies in controlled contexts. The fourth is simu-
lation and finally, reflection and theorising. Furthermore, some of the most effective
research studies both design cognition and design activity, identifying key character-
istics, patterns and strategies in both the process and product of design (Atman et al.
2005; D’Souza and Dastmalchi 2016; Georgiev and Georgiev 2018; Ho 2001; Kim
and Maher 2008; Kruger and Cross 2006; Lee et al. 2015, 2019; Lee et al. 2014a;
Lee and Ostwald 2019; Yu et al. 2013). Through research published in the journal
Design Studies and the Design Thinking Research Symposium (DTRS) series (Cross
2018; Dorst 2018), design cognition has become the focus of sustained attempts to
understand its operations and activities.

1.1.3 The Digital Anthropocene

One of the most famous books about the challenges associated with the rise of the
modern, technological world was Marshall Berman’s (1982) All That is Solid Melts
into Air. In this book, Berman observes that, paradoxically, progress is also a destruc-
tive force, as every technological advance effectively erases the ones that preceded
it. This was certainly the case for early twentieth-century industrial designers and
architects. The early Modern movement promoted a new, machine-driven process of
production, allowing designers to create technically superior products that surpassed
the performance and durability of the handmade or crafted equivalents of the previous
era. By the 1960s, however, these machine-made products were being replaced by
those of the next generation of designers, using electronic components and synthetic
materials. In the 1990s, bespoke production systems allowed designers to create
unique, digitally enhanced objects which soon surpassed the performance or desir-
ability of those of the mid-century modern movement. In this sense, the trajectory of
1.1 Introduction 7

the world is innately forward or upward, forever seeking new ideas or opportunities,
while simultaneously leaving the past behind. The past, viewed in this way, is that
which “melts into air”, as the ideas, techniques and processes used by designers, and
which shape the world we live in, are all slaves to progress. Even the most tangible
or “solid” concepts and theories from the past are always being challenged.
Design thinking is a progressive field that acknowledges the impacts of new tech-
nology and ideas. For example, as noted previously, parametric design appears to offer
an innovative way of generating new design solutions based on algorithmic thinking
(Lee et al. 2015). Mobile and pervasive computing enables interactive and collective
design thinking, continuously constructing and sharing creative processes and prod-
ucts (Lee et al. 2013b). The design process also adopts, as needed, innovative design
tools and operates in a globalised design ecosystem (Singh and Gu 2012). These new
paradigms of design for a digital Anthropocene are encapsulated in this book under
three headings: Creativity, Collaboration and Culture of the “three C’s” (Fig. 1.1).
Creativity as a theme in design thinking highlights computational design employing
various strategies and complex cognitive activities. Collaboration in design thinking
refers to interactive, collective thinking, teamwork and networked design collabora-
tion. Culture in design thinking is concerned with designing and thinking in multi-
cultural contexts, encompassing the language of design and cognitive and linguistic
differences in design thinking.
These themes are explored in the present book through a series of empirical
studies of design cognition. The purpose is to develop new knowledge and ideas,
not necessarily to erase old ones, although some existing theories in design thinking
need to be questioned before they too “melt into air”, as progress is made in the field.

CreaƟvity Computational design thinking

Design strategies
Cognitive complexity

DESIGN
THINKING
in the digital era

CollaboraƟon Culture
Interactive and collective thinking Multi-cultural design thinking
Cognition and teamwork The language of design
Networked design collaboration Cognitive and linguistic differences

Fig. 1.1 Themes in design thinking for the digital era

8 1 Introduction: Exploring Design Thinking

1.2 Creativity, Collaboration and Culture

1.2.1 Creativity

Creativity and novelty are often understood as closely related concepts in design,
much like the reciprocal relationship between aesthetic and technical values. Each
pairing requires both capacities and properties to be completely successful. For
example, Kryssanov et al. (2001) define creativity as the combination of “novelty”
and “appropriateness”. Creative ideas are not just new or unexpected, but provide
solutions that are useful, efficient and valuable (Georgiev and Georgiev 2018). In
addition, complementary models of creativity also define psychological creativity
and historical creativity (Boden 2004), as well as human creativity and compu-
tational creativity (Maher 2010). For example, psychological creativity refers to “a
surprising, valuable idea that’s new to the person who comes up with it” (Boden 2004,
p. 2), whereas computational creativity is “expressed in the formal language of search
spaces and algorithms” (Maher 2010, p. 22), addressing the joint products of humans
and computers. Iordanova et al. (2009) also argue that as generative modelling tools
replace those of previous eras (for example, as parametric modelling replaces hand
drawing and traditional CAD modelling), new types of creativity are enabled. Simi-
larly, generative design tools are often presented as supporting creativity, because
they allow for rapid exploration of design alternatives at an early stage of the process
(Blosiu 1999; SHoP/Sharples Holden Pasquarelli 2002; Spiller 2008). In such exam-
ples, creative potential is linked to increased capacity to generate options that fulfil
functional parameters.
Multiple theories exist about the role of creativity in design thinking. For example,
Gordon’s (1961) theory of synectics proposes that creative processes can be system-
atised and learnt, and that people are more effective if they understand the creative
process. Lawson (1980) postulates that creativity emerges from the overlap between
convergent and divergent thinking. Convergent thinking leads to a singular, appro-
priate solution, while divergent thinking produces multiple ideas and alternatives.
Blosiu (1999) suggests the use of a laterally integrated design model, which not only
generates design alternatives but also develops creative capacity Bucciarelli (2001)
also highlights analytic and synthetic design thinking in engineering education,
drawing on Gordon’s synectics.
A creative design can also potentially be generated using specific computational
processes. For instance, Rosenman and Gero (1993) identify the processes of combi-
nation, mutation, analogy and first principles, and Gero (2000) later revises this
to include their computational equivalents. Such computational models of design
thinking, and their support for creativity through design exploration and evaluation,
are considered in more detail in the present book. Specifically, design thinking with
parameters and rules in parametric design is examined for its capacity to generate
and evolve innovative or original ideas.
Sternberg and Lubart (1999) categorise six scientific approaches to investigating
creativity: (i) pragmatic, (ii) psychodynamic, (iii) psychometric, (iv) social–personal,
1.2 Creativity, Collaboration and Culture 9

(v) cognitive and (vi) confluence approaches. Their first, the pragmatic, addresses
commercial and social benefit, while their second, the psychodynamic approach,
is concerned with the role of psychology. Since Guilford’s (Guilford 1950) Diver-
gent Thinking (DT) test first measured individual fluency, originality, flexibility and
elaboration, psychometric approaches have been widely used to measure creativity.
Several versions of the DT test also exist, such as the Structure of the Intellect (SOI)
and Torrance’s Tests of Creative Thinking (TTCT), and have dominated the field of
creativity assessment for several decades (Runco and Acar 2012). The TTCT consists
of four measures, two figural and two verbal, and is highly respected in education as
well as in industry (Kim 2006). Recently, Jung and Chang (2017) show how the Some-
thing About Myself (SAM) test can be used to assess creativity, along with the Latent
Ability Detection (LAD) test to examine multiple intelligences. They also examine
creative-convergence design and reveal that spatial and linguistic intelligences are
positively related to creativity. However, such paper-and-pencil tests of the psycho-
metric approach are unable to investigate design exploration using richer visual
representations and the creative design processes it supports. This is, in part, because
sketches and drawings often play a central role in models of the design process (Koko-
tovich and Purcell 2000; Goldschmidt and Smolkov 2006; Schön and Wiggins 1992;
Suwa and Tversky 1997). Thus, the psychometric approach to assessing creativity
is limited to divergent thinking inside an individual mind. Sternberg and Lubart’s
(1999) fourth approach, the social–personal, emphasises the significance of social
contexts or personalities in creativity, whereas the fifth, the cognitive, is the most
important for understanding creative design processes (Chan 1990; Kim et al. 2007;
Kruger and Cross 2006; Lee et al. 2013a; Lee et al. 2017; Suwa et al. 1998). Stern-
berg and Lubart’s (1999) last approach to creativity, the confluence approach, has
growing acceptance across many disciplines (Katz 2002). This book employs both
the cognitive and confluence approaches to investigate creative design thinking.
One established model of creativity is Rhodes’ (1961) four P’s—person, process,
press and product—which has been used as a multifaceted framework to inves-
tigate creativity (Askland et al. 2012; Williams et al. 2010). The person is the
designer, the process is the cognitive operation and the product is the outcome of
the design. As such, it replicates three components of design thinking discussed
previously. “Press” in Rhodes’ model refers to environmental or contextual factors
which shape creativity. This factor is probably the least researched of the four P’s in
design thinking. For example, multiple studies address person, process and product
in creativity (Hasirci and Demirkan 2007; Simonton 2003), while investigating both
design processes and designed products has become more common over time (Gold-
schmidt and Smolkov 2006; Lee et al. 2015). Furthermore, exploring the design
process can also involve the identification of individual design strategies and prefer-
ences, which indicate a person’s approach to creativity. The creative design process
is often explored cognitively using protocol analysis (employing various coding
schemes), while creativity in the design product is measured using a confluence
approach, such as expert panel assessment (e.g. Consensual Assessment Technique
or CAT). Press, however, is only rarely investigated in design thinking experiments,
although the present book’s interest in culture and language does begin to address
10 1 Introduction: Exploring Design Thinking

some environmental or contextual factors. Part I of this book is focussed on creativity

and design thinking in the digital Anthropocene.

1.2.2 Collaboration

The oldest models of creativity are typically traced to Plato and the belief that an
artist’s vision and inventiveness are dependent on the existence of a transcendent
muse or direct connection to the gods. Philosopher Immanuel Kant’s views have some
similarities as he maintains that creativity is an innate and valuable gift bestowed
on an individual (Sawyer 2006). These, so-called “romantic” views of creativity, all
emphasise that creativity is an inborn or natural characteristic of some people and
not of others (Williams et al. 2010). The alternative perspective on creativity, the
“rationalist” view, holds that creativity can be taught or learnt by anyone. Drawing
on the work of Aristotle, rationalist attitudes to creativity provided the foundations
for the first attempts to understand design thinking in the twentieth century (Guilford
1950; Sawyer 2006). One of the few commonalities between the romantic and the
rationalist models of creativity is that both focus on the role of the individual. This
is not surprising, given the way society valorises individuals in creative processes,
even though they rarely, if ever, work alone. Standard twentieth-century management
theory even emphasises that creative problem-solving requires brainstorming rather
than group thinking. Despite this, collaboration in the design process, and especially
as enabled using digital tools, is an area that requires further research. In order to
understand this need, a brief review of past research into collaboration in design
thinking and cognition is useful. Furthermore, this review also touches on the topic
of communication, which is relevant to the third theme in this book, culture.
Collaborative design processes and communication in design teams have been
examined in past experimental research using protocol analysis (Cross and Cross
1995; Dong 2005; Goldschmidt 1995; Stempfle and Badke-Schaub 2002; Valken-
burg and Dorst 1998). The products of team design processes have also been evaluated
or analysed using expert panels (Mulet et al. 2016; Karlusch et al. 2018). Unlike the
singular “person” aspect of creativity in Rhodes’ model, team members’ thinking
styles and preferences can be measured using several tests. Three examples of these
tests are the State Action Part Phenomenon Input oRgan Effect (SAPPhIRE) model
of causality (Mulet et al. 2016; Sarkar and Chakrabarti 2011), the Herrmann brain
dominance test (Herrmann 1991) and Kirton’s adaptor–innovator test (Kirton 1994;
López-Mesa and Thompson 2006). Dong (2005) also employs a Latent Semantic
Analysis (LSA) method to investigate design team communication. In one of the first
practical protocol studies in the field of design, Cross and Cross (1995) examine a
social process of design with a focus on six points of view: “roles and relationships”,
“planning and acting”, “information gathering and sharing”, “problem analysing
and understanding”, “concept generating and adopting” and “conflict avoiding and
resolving” through an in-depth analysis of a teamwork experiment. Valkenburg and
Dorst (1998) expand Schön’s theory of reflective practice (Newman 1999; Schön
1.2 Creativity, Collaboration and Culture 11

1983, 1984) in design teams to address four activities: naming, framing, moving
and reflecting. Stempfle and Badke-Schaub (2002) also suggest the use of four
basic cognitive operations—generation, exploration, comparison and selection—to
understand the collective design process. D’Souza and Dastmalchi (2016) identify
smaller creative events potentially connected to big creative leaps in a collabora-
tive design process. They further suggest that a balance between individual perfor-
mance and team dynamics is essential to a successful multi-disciplinary team process.
Research also proposes that individual thinking and reasoning styles can impact on
knowledge-sharing capabilities (Mulet et al. 2016).
Previous experimental and theoretical studies of cognition and team design
emphasise four main approaches: (i) design space, (ii) design strategy, (ii) design
productivity and (iv) spatial representation (Lee et al. 2017). These approaches are
significant for the way teams operate as part of a creative design process. In the
first of these approaches, design is regarded as a “co-evolutionary” process, which
needs to shift backwards and forwards between problem and solution spaces for
a creative product to be produced (Lee et al. 2015; Maher and Poon 1996). The
way in which members of teams move between these two spaces would appear to
be key to understanding and improving their collective, creative performance. To
explore team communication in terms of problem and solution spaces, Stempfle
and Badke-Schaub (2002) present four basic cognitive operations, “generation” and
“exploration” to widen a problem space and “comparison” and “selection” to narrow
a problem space. The second approach, design strategies, deals with the definition
of sub-goals which limit or enable certain operations, like working-forward and
working-backward strategies. These cognitive strategies are linked to personal pref-
erences and habits and contribute to defining designers’ thought processes (Lee et al.
2014a). The significance of this is that it confirms that there are different teamwork
strategies involved in the average design team. The third approach, design produc-
tivity, is associated with the effectiveness and efficiency of a cognitive process that
is either individually or collectively undertaken (Goldschmidt 1995). Such is the
importance of this way of conceptualising the effectiveness of design cognition that
mathematical methods have been developed for interpreting linkograph data (Kan
and Gero 2008; Lee and Ostwald 2019). Linkograph is a graph that illustrates how
one idea, action or event interlinks with another (Goldschmidt 1995). These cognitive
themes are related to how creativity is accommodated and supported in collaborative
design processes. The last cognitive approach, spatial representation, is related to
the capacity to conceptualise and communicate spatial information. It is associated
with linguistic complexity and fluency, and is also a factor in multi-cultural design
communication.
The “collaboration” theme in the present book also considers interactive and
collective design thinking, which are linked to “social creativity” in the “digital
ecology”. For example, mobile devices like smartphones produce active relation-
ships between people, objects and locations, supporting mobile social interaction
beyond Greif and Cashman’s “Computer-Supported Cooperative Work” (CSCW)
(Grudin 1994). Recent information technologies embed microprocessors in everyday
objects, promoted the spread of wireless technologies and advanced the use of sensor
12 1 Introduction: Exploring Design Thinking

networks and the internet (Lee et al. 2014b). Beyond human collaboration and
office automation, the “internet of things”—“ubiquitous computing” or “pervasive
computing”—allow for a diverse range of information communication processes.
This level of mass interaction is potentially the catalyst for a new type of creativity
(Phare et al. 2016, 2018). This “mobile creativity” involves interaction and collab-
oration with other individuals and is more strongly related to social creativity than
individual creativity (Lee et al. 2013b). Thus, new collaborative, creative design
processes evolve with these interactive and collective surroundings beyond phys-
ical environments. In this context, virtual workplaces, including BIM platforms,
have attracted the attention of many researchers. BIM has been used as an innova-
tive collaboration tool in building design, construction and management. Gu et al.
(2015) highlight the significance of BIM-related products, processes and people in
collaboration, and Mulet et al. (2016) compare the thinking and reasoning styles of
design teams in a face-to-face environment and a virtual working environment. Mulet
et al.’s (2016) results suggest that design teams in virtual environments produce a
slightly higher degree of novelty in their design outcomes. Tang et al. (2011) also
compare a traditional design environment (face-to-face, pen-and-paper based) with
a digital sketching environment (emulating the traditional sketching environment).
They examine the design processes through protocol analysis using a Function–
Behaviour–Structure (FBS) coding scheme. Their results show no significant differ-
ences between design environments in terms of design process, suggesting that the
higher level cognitive activities can be effectively supported in digital environments
and may not be influenced by specific tools. Part II of this book examines these
new collaborative design environments in terms of interactivity and collectiveness
in designing.

1.2.3 Culture

The third theme in this book, culture in design thinking, is probably the least well
developed in past research. A few pioneering projects have identified the impor-
tance of multi-cultural education and practice in an increasingly globalised world
(Lee et al. 2019; Santandreu Calonge and Safiullin 2015; Karlusch et al. 2018).
Santandreu Calonge and Safiullin (2015), for example, argue that multi-national or
multi-cultural teams are more creative than intra-national teams, because individuals
from different cultural and linguistic backgrounds in a team can develop a wider
variety of perspectives and points of views. That is, diversity is a “synergy enabler”
to produce positive outcomes. Karlusch et al. (2018) also identify the importance of
lateral thinking and incorporated intelligence in fostering creativity in diverse teams.
Design naturally requires engagement with multiple complex processes and
conceptual configurations that arise from visual and verbal communication. Design
expertise is often spread across project teams and different locations and uses collab-
orative design management processes, where design communication is supported by
1.2 Creativity, Collaboration and Culture 13

representations such as sketches and drawings (London and Singh 2013). Nonethe-
less, the experimental results of past research indicate that cognitive and linguistic
design abstraction is related to individual linguistic and cultural experiences and pref-
erences (Lee et al. 2019). That is, for international architects and designers, language
is not just a spoken or written system, it also shapes how they use and understand
design representations.
The relationship between thought and language, or cognition and culture, is a
complex one. For example, psychological, social and semiotic researchers repeat-
edly note that ideas are intrinsically tied to the language in which they are both
constructed and expressed (Bonvillain 2010; Lewis 2012). Structural linguists and
post-structuralist philosophers agree on this one point, although for different reasons.
For example, de Saussure (1959) argues that language is the fundamental basis on
which we understand and represent the world, and therefore the structure of our
language is the structure of our world. Derrida (1976) famously questions the “logo-
centric” hierarchy in de Saussure’s philosophy and semiotics wherein thought is
argued to be superior to speech, which is in turn superior to writing. However, a
central message in Derrida’s work is that language is never a transparent, stable
means of communication.
In contrast, Chomsky (1965), extending and also questioning de Saussure’s ideas,
identifies common structural threads between languages across cultures. His research
proposes that human language is universal or innate. Pinker (1994) also presents a
compelling argument that there are thought processes at work long before people
develop language skills. That is, thought comes before language. For example, as
humans can acquire complicated linguistic skills from the earliest age without explicit
lessons, people must be born with the ability to learn and reason. Thus, thought
could be, to a certain extent, independent of language, or precede it developmen-
tally. Furthermore, Arnheim’s “visual thinking” argues that “the cognitive operations
called thinking are not the privilege of mental processes above and beyond percep-
tion but the essential ingredients of perception itself” (Arnheim 1997, p. 13). He
also highlights the cooperation of verbal language and imagery and proposes that
“language tends to suggest functional rather than formal categories and thereby to go
beyond more appearance” (p. 239). Artistic expression is potentially another form of
communicative reasoning, often related to “non-verbal thought” although it has been
claimed that visual perception is an intelligent act with no relation to language (Ware
2008). These opposing arguments illuminate a core disagreement about culture and
cognition that is pertinent to design thinking.
In the discipline of linguistics, it is accepted that language shapes both the way
ideas are presented or explained and the thought processes used to develop them.
Wittgenstein summarises this idea as, “the limits of my language mean the limits
of my world” (1922, p. 74). Significantly, linguistic researchers have observed that
this phenomenon includes the construction and communication of spatial and formal
relationships and reasoning (Herskovits 1986; Tenbrink and Ragni 2012; van der
Zee and Slack 2003). For example, Chinese speakers do not typically follow a coun-
terfactual story structure as they use different spatiotemporal metaphors (Casasanto
2008), while Korean speakers lack a system of particles and prepositions comparable
14 1 Introduction: Exploring Design Thinking

to words like in, on, up and down in English (Choi et al. 1999). Nonetheless, the real-
isation that design language might shape design itself, and vice versa, has been less
commonly observed (Dong 2009; Lee et al. 2016; Lee et al. 2019). The problem is
not only that the globalised design environment is so reliant on linguistic skills, but
also that language as a system is a reflection of the way people think and of their
sociocultural values, both of which are central to the process of design (Gleitman
and Papafragou 2005). Part III of this book explores multi-cultural design commu-
nication through protocol studies employing both cognitive and linguistic coding
schemes.

1.3 Research Method I: Protocol Analysis

1.3.1 Studying the Design Process

Empirical research into design thinking requires a method that can capture and
compare aspects of cognitive processes. There are multiple models of the creative
process that can be experimentally tested. One of the oldest examples, from Wallas
(1926), proposes the use of a four-stage model of the creative process—prepara-
tion, incubation, illumination, and verification—to formally investigate creativity in
human thought processes. Guilford’s (1967) Structure of Intellect (SI) theory identi-
fies five types of mental operations—cognition, memory, convergent thinking, diver-
gent thinking and evaluation—that can be examined in cognitive experiments. More
recently, Hayes (1989) also categorises creative cognitive processes into five acts—
preparation, goal setting, representation, searching for solution and revision—which
facilitate creativity in the design process to be studied.
Dacey and Lennon (1998) argue that cognitive processes can be researched and
understood using two sets of theories. The first set comprises the combination and
expansion of three early creativity models, associationism, Gestalt and cognitive-
developmental approaches. These theories develop various aspects of the conceptual
combination of ideas and suggest that creative cognition is more than problem-
solving, highlighting selective and evaluative stages in the creative process. The
second set of theories involves metaphors, analogies and mental models. Metaphors
shift the interpretation from one conceptual understanding to a new point of view, a
so-called “ontological shift”, while analogy is a cognitive process referring to concep-
tual parallels. Choi and Kim (2017) also suggest that analogical and metaphorical
reasoning can be used to develop creative design thinking in tertiary education.
This set of theories and the other cognitive models mentioned in this section have
facilitated formal studies on creativity across multiple disciplines. In many cases,
the methodological approach used to develop the empirical evidence from design
experiments to support the development of these models and theories is protocol
analysis.
1.3 Research Method I: Protocol Analysis 15

1.3.2 Protocol Analysis

Protocol analysis is a method for analysing recorded actions, behaviours, verbalised

responses or written information in a rigorous and repeatable way. A typical use of
this method would be to analyse a recording of an experiment in accordance with an
agreed or accepted set of rules (the “protocol”). Each time a particular action occurs,
or a word is used in the recording, it is coded, counted and its timing noted. This
process extracts rich data from the experiment that can be directly compared with the
results of other participants in the experiment to identify similarities or differences
and test theories about human cognitive behaviours.
Protocol analysis became popular in cognitive psychology and design research
in the 1980s and 1990s because it offered a way of methodically counting and
comparing human processes. Since then protocol analysis has become the most
widely accepted cognitive research technique for analysing the design process (Chai
and Xiao 2012; Coley et al. 2007). It provides a previously unavailable level of
scientific or mathematical rigour to the assessment of complex human operations.
The strengths and weaknesses of protocol analysis can be readily understood
by comparing it with a standard Likert-scale survey method. Surveys can engage
a large number of participants for a short amount of time, generating data that is
straightforward to extract. If the response rate is high enough, the results may be
statistically generalisable to a larger population even though they are often grossly
simplified. In contrast, protocol analysis engages a small number of participants
for an extended period of time. The results are analysed in great detail and depth,
often involving multiple simultaneous data collection and assessment processes.
The results are time-consuming to develop but they are rigorous and rich. They
cannot, however, be easily generalised to a larger population. Depending on the
topic being investigated, both surveys and protocol analysis are valuable methods
for researchers. However, for studying design cognition, depth, subtlety and richness
are of paramount importance and thus protocol analysis has become the standard
approach for many studies.
Protocol analysis necessarily requires recruitment of participants to take part in
experiments. Sometimes volunteers can be sought, whereas often incentives (typi-
cally gift vouchers) are used to compensate people for the time and effort. In most
countries, formal human ethics clearances are required before the experiments may
commence. These clearances cover pre-experiment disclosures, participant consent
paperwork, briefs for participants and their rights, and information about data use
and storage. Participants are generally anonymous and agree to being identified only
by number or coding.
In its typical use in design cognition or design thinking research, protocol analysis
has five components which are needed to meet a specific aim or objective. They
are (i) experimental settings, (ii) verbalisation, (iii) transcription and segmentation,
(iv) encoding with coding schemes and (v) arbitration and intercoder reliability.
Most protocol studies possess these same five elements, each of which are explained
hereafter, even if they may vary for particular applications.
16 1 Introduction: Exploring Design Thinking

(i) Experimental settings are the parameters or limits for a study. These might
include the location where the experiment takes place, the scenario or test
participants will undertake, the hardware or tools provided and a time limit.
For example, in a typical experiment in design thinking a participant is given
a design brief and a selection of design media (for example, pen-and-paper or
CAD tools), a specific outcome or goal and a time limit. The type of partic-
ipants required (for example, novice or expert) and experiment venues (for
example, simulated or real workspace) must be carefully considered to meet
the scope and feasibility of the research. Figure 1.2 shows an example of a
design experiment undertaken by the authors in a university computer labo-
ratory simulating a parametric workstation in a design office. Each session of
the experiment is video recorded with one camera giving a clear view of the
designer’s overall activities and the other focusing on the computer screen to
capture representation activities. For this experiment, participants were given
a design brief (for a high-rise building), software (Rhino and Grasshopper) and
one hour to complete the task. Only participants with a sufficient fluency or
expertise in operating on the software were allowed to take part.
(ii) Verbalisation describes the methods used to develop audio protocol data.
There are two common techniques, concurrent and retrospective verbalisa-
tion. For the former, participants are asked to verbalise their thoughts and
actions while designing (concurrent) and for the latter, immediately afterwards
(retrospective). There are benefits to both concurrent and retrospective verbal-
isations (Coley et al. 2007). The former approach, often called “think-aloud”,
could interfere with the design process, distracting the participant from acting
normally. Participants in the latter, retrospective approach, often forget details
or have incorrect recall. A mixed technique with both a think-aloud verbalisa-
tion during designing and a post-experiment interview can be the most effective
combination (Lee et al. 2015). Most commonly, the verbalisation is recorded
using the same video recording equipment used for capturing actions and
behaviours. This allows the researcher to match actions to words, providing
an added level of assurance when coding the data.
The experiments described in the present book were video recorded, with a
parallel “think-aloud” audio recording, and a post-experiment interview, to

Fig. 1.2 An example of experimental settings

1.3 Research Method I: Protocol Analysis 17

provide retrospective data to explain the participant’s thoughts and activities.

Before each design experiment, a researcher also explained a design brief and
undertook a short (for example, five-minute) “practice-run” of think-aloud
verbalisations with each participant.
(iii) Transcription and segmentation are the processes of extracting data from the
recordings and breaking them down into smaller fragments. The audio data
are typically transcribed in full, although this may not be necessary for some
experiments or research questions. After transcribing the audio, each transcript
is divided into smaller segments, sometimes called “episodes”. The episode
timing can be determined using either set times—every 30 s or every minute—
or using content. In the latter case, segmentation by content, there are two
methods (Suwa et al. 1998). The first uses pauses or syntactic markers as
breaks, and the second reviews the participant’s spoken content to determine
where an idea has started and ended. It is also possible to use both time-based
and content-based segmentation processes simultaneously, examining the rela-
tionship between time and content. Professional services often transcribe and
segment (most often time-based) audio data for cognitive research projects.
Such formal and rigorous transcriptions are valuable for analysing results as
well as for supporting open research with peers and further dissemination.
(iv) Encoding with coding schemes is arguably the most important procedure in
protocol analysis. There are two parts of this component, the coding scheme and
the coding process. The “coding scheme” is a guideline, schema or reference
identifying precisely which activities are being captured. Each category in the
coding scheme must be clearly defined, preferably with examples, of what
constitutes evidence of an observed action being classified in a particular way.
For many uses, it is better to adopt an existing “tried and tested” coding scheme.
This also allows for comparisons to be made between experiments by different
teams, for benchmarking.
Previous coding schemes in design research fall into two types, targeting
the “content-oriented” aspects of designing and its “process-oriented” aspects
(Coley et al. 2007). For example, Suwa et al.’s coding scheme (1998) for coding
designers’ cognitive actions is content-oriented, while Gero’s FBS coding
scheme (Gero 1990; Mc Neill et al. 1998) is process-oriented. Most cogni-
tive studies use only a single coding scheme, although a few employ multiple
coding schemes simultaneously, to better compare different aspects of the data.
For example, Mc Neill et al. (1998) use a four-dimensional coding scheme: level
of abstraction, FBS, micro- and macro-strategies. Lee et al. (2019) present a
triple-coding system consisting of design cognition, design information and
spatial language coding schemes.
The coding process is where a researcher views (and listens) to a recording, and
identifies every action, behaviour or verbalisation listed in the coding scheme as
it occurs to allow quantification of the design data. This is an exacting and time-
consuming process, where the increased level of scientific and mathematical
rigour is provided to the analysis.
18 1 Introduction: Exploring Design Thinking

(v) Arbitration and intercoder reliability are processes used to ensure the reliability
of the coding. There are two approaches to this, arbitrated and self-arbitrated.
In the arbitrated version, two or more coders undertake the encoding process
(and sometimes, transcription as well) in parallel. Then the multiple sets of
coded data are compared, and any disagreements are reviewed and resolved. In
the self-arbitrated version, the same coder takes two or more “runs” coding the
data, separated by a period of time (typically 2–3 months). When completed
the different versions are compared and any disagreements resolved to ensure
the reliability of the encoding.
Methods also exist for assessing Intercoder Reliability (ICR), percentage agree-
ment being the simplest one. Holsti’s coefficient, Krippendorff’s alpha and
Cohen’s kappa (κ) indexes are also used for this purpose although there is no
consensus on the single best measure (Cho 2011; Krippendorff 2018; Mouter
and Vonk Noordegraaf 2012). “The selection of proper index will depend on
the levels of coding, number of coded categories if coded nominal, number of
coders, and number of coded units” (Cho 2011, p. 344).
Research using protocol analysis should always address and describe these five
aspects of the experiment. Relevant chapters in this book record this information for
specific experiments accordingly.

1.3.3 Coding Scheme

To conduct protocol analysis, a coding scheme must be selected, modified or designed

to suit the specific research aims and objectives. If modified or designed, the scheme
needs a clear underlying logic, preferably linked to established theories or models.
Table 1.1 is an example of a coding scheme which supports the description and iden-
tification of cognitive thinking and activities for evaluating creativity in parametric
design processes.
This example coding scheme starts with four “levels” of creativity, representa-
tion, perception, goal setting and searching for solution, selectively developed from
Hayes’ (1989) creative acts model. The representation level encapsulates modelling
activities in parametric design. This level is similar to Suwa et al.’s physical actions
(1998) and Kim and Maher’s three-dimensional modelling actions (2008). The
perception level addresses the process of seeing relationships between elements
(Flowers and Garbin 1989) or components, referring to the activities of visual imagery
in the creative process. Perception is also related to the “incubation” stage of Wallas’
creative model (1926). The goal-setting level describes the activities of “problem-
posing” or “problem-formulation”, while the searching for solution level highlights
divergent and convergent thinking processes.
1.3 Research Method I: Protocol Analysis 19

Table 1.1 A coding scheme to investigate creativity in parametric design process

Level Category Subclasses Description
Representation Geometry RG-Primitive Make primitive
geometries
RG-Change Change existing
geometries
RG-Transformation Perform deformations
(or morphing of forms)
RG-Variation Make variations (or
generate forms)
Algorithm RA-Parameter Make initial parameters
RA-ChangeParameter Change existing
parameters
RA-Rule Make initial rules
RA-ChangeRule Change existing rules
RA-Reference Retrieve or get an
internal/external
reference
Perception Geometry PG-Geometry Attend to existing
primitive or changed
geometries
PG-Transformation Attend to transformed
geometries
PG-Variation Attend to variations (or
generated forms)
Algorithm PA-Parameter Attend to existing
parameters
PA-Algorithm Attend to existing
algorithms
PA-Reference Attend to existing
reference data
Goal setting Formulating Goal SF-Initial Goal Introduce an initial goal
SF-Previous Goal Introduce new ideas
extended from a
previous idea (or goal)
SF-Knowledge Introduce new ideas
derived from knowledge
(or experience)
Searching for solution Evaluation SE-Geometry Evaluate primitive or
(Geometry) changed geometries
SE-Transformation Evaluate transformation
SE-Variation Evaluate variations (or
generated forms)
(continued)
20 1 Introduction: Exploring Design Thinking

Table 1.1 (continued)

Level Category Subclasses Description
(Algorithm) SE-Parameter Evaluate existing
parameters
SE-Algorithm Evaluate existing
algorithms
SE-Reference Evaluate existing
reference data
Adopting Solution SA-Geometry Adopt new solutions to
(Geometry) primitive or changed
geometries
SA-Transformation Adopt new solutions to
transformation
SA-Variation Adopt new solutions to
variations (or generated
forms)
(Algorithm) SA-Parameter Adopt new solutions to
SA-Parameter parameters
SA-Algorithm Adopt new solutions to
algorithms
SA-Reference Adopt new solutions to
retrieve or get
internal/external
reference

Within each level, there are two categories (except for goal setting), in this example
one for standard CAD environments (geometry) and the other for parametric environ-
ments (algorithm), and within each category there are further subclasses. The cate-
gories and subclasses for detailed actions in this example are adapted from established
coding schemes for design actions (Suwa and Tversky 1997) and cognitive actions
(Kim and Maher 2008). Because this example coding scheme is for comparing para-
metric and standard design processes, geometry and algorithm categories are used to
capture the cognitive activity in each. For example, “Geometry category” includes
general modelling activities (line or shape drawing) in software. In contrast, “algo-
rithm category” represents generative algorithms that involve “parameter” and “rule”
as well as “reference”. This coding scheme is an example of an adapted scheme,
drawing on aspects of three common and accepted models, and then modifying them
to suit the focus of the experiment. Chapters in this book record and briefly explain
the coding schemes used for specific experiments, all of which follow this pattern.
1.4 Research Method II: Expert Panel Assessment 21

1.4 Research Method II: Expert Panel Assessment

1.4.1 Design Product

Thus far in this chapter, most of the content has focussed on design as a verb or
process, denoting an activity that might lead to creativity (Glanville 1999). It is
difficult, however, to talk about creativity without considering design as a noun or
product. This is because the creativity implicit in a product is often more apparent
than in a process, and a creative product is also a common form of evidence for a
creative process.
From the earliest architectural treatises, the completed building has been viewed
as encapsulating the structural, functional and innovative dimensions of the work.
For example, in approximately 20–30 BC, Vitruvius, a Roman architect and military
engineer, argued that a successful building should have three properties—firmitas,
utilitas and venustas—which are translated as either firmness, commodity and delight
or soundness, utility and attractiveness, respectively (Kuiper 2010; Rowland and
Howe 1999). These three interrelated components have since been integrated into
many research frameworks in architecture where the product’s success is assessed
in both utilitarian and aesthetic terms. Rowe (1987), for example, treats design as
both a form of fine art and a technical science, and Chan (2015) argues that design
is characterised as the creation of satisfactory solutions or beautiful artefacts that
fulfil certain functions. In addition, Vitruvius’s delight (venustas) can refer to an
observer’s intellectual, emotional or psychological response. That is, design is not
only about the function of a building, garment or other object, but also satisfying
people’s emotions and desire for meaning or value (delight).
Whereas protocol analysis can be used to interrogate the cognitive process of
design, it cannot be used to directly assess the creativity that occurs in this process
that often requires a review of the product. The creativity implicit in a product is best
assessed by experts or judges in accordance with a set of evaluation criteria. Teresa
Amabile (1983a), for example, identifies four components for any creative response:
domain-relevant knowledge and abilities, creativity-relevant skills, intrinsic task
motivation, and the social environment (Amabile 1983a; Hennessey and Amabile
1999). Amabile’s Consensual Assessment Technique (CAT) has been used for
assessing both artistic and verbal creativities and been widely applied in research in
education, arts, business and advertising. In particular, CAT has been used to assess
the creativity evident in a designed product (Amabile 1983a; Christiaans 2002; Thang
et al. 2008). A more general version of this approach is the expert panel assessment,
where the creativity of an artefact is measured using the combined assessment of
experts in that field.
The following is a simple example of how expert panel assessment works in design
thinking research. First, let us assume that a design experiment has occurred, with
20 participants each producing a design for a simple house. A group of ten experi-
enced architects (the “expert panel”) are then presented with the 20 house designs
to assess. They are also given a detailed set of evaluation criteria, covering various
22 1 Introduction: Exploring Design Thinking

functional and aesthetic requirements. The expert panellists then independently view
and assess each of the house designs against the provided criteria, scoring them on
a five-point scale. At the completion of this, the scores provided by experts are
averaged (representing a consensus view) and analysed mathematically (to identify
outliers and trends). The resultant data provides an effective measure of creativity,
innovation or other criteria for each house design, which are also able to be linked
to the process that developed the design. Thus, expert panel assessment provides a
level of information about the product, whereas protocol analyses are used to capture
and encode the process. In combination, they allow researchers to connect creative
outputs to cognitive operations.

1.4.2 Evaluation Criteria

The primary evaluation criteria for an expert panel to assess creativity typically
comprise three main categories: novelty, usefulness and aesthetics, or variations of
these. The subscales within these categories are often selectively adopted from either
CAT (Amabile 1983b) or the Creative Product Semantic Scale (CPSS) (Besemer and
O’Quin 1987) or a combination of both that is most useful for the purposes of the
specific research question. CAT requires an analysis of inter-judge reliability in the
ratings on each dimension (Amabile 1983a). Factor analysis on the various dimen-
sions of judgement, including several subjective dimensions, should be conducted to
determine the degree of independence between one and the other dimensions. The
various versions of the CAT have several dimensions of criteria such as the creativity,
technical and aesthetic dimensions. The CPSS also uses three dimensions—novelty,
resolution and elaboration and synthesis—to rate creative design products and 70
bipolar subscales. Iordanova et al. (2009) add indicators such as abundance, flexi-
bility, evolution and originality of ideas and dynamic factors can also be included in
the criteria (Liu and Lim 2006).
The total number of subscales of the evaluation criteria should also be care-
fully considered. A large set of subscales might be important for comparing design
outcomes in detail, while the relatively small set is more useful for correlating results
with other data, like that derived from protocol analysis. Although it is acknowledged
that the large set can also be used for the correlation through the mean values of factors
using factor analysis, it would be too time-consuming for measuring and analysing
it.
As an example, the evaluation criteria for measuring the creativity of a parametric
design outcome are developed in three categories: Novelty, Value and Aesthetics
(Table 1.2). The subscale criterion combines CAT (Hennessey and Amabile 1999) and
CPSS (Besemer and O’Quin 1987), while subscales for parametric-design-specific
elements, like “well transformation”, are added after a review of literature and models
for parametric design. The subscales in Table 1.2 are also developed from the results
of correlation and factor analysis of related works (Besemer and O’Quin 1987; Chris-
tiaans 2002; Hennessey and Amabile 1999). The validity of evaluation criteria can
1.4 Research Method II: Expert Panel Assessment 23

Table 1.2 An example of

Novelty Value Aesthetics
evaluation criteria for
creativity in parametric design Originality (Idea) Function Aesthetic form
Complexity Usefulness Elegance
Evolutiona Understandable form Well transformationa
a Specific criteria referring to parametric design evaluation

be constructed by the analysis of inter-judge reliability and factor analysis (Amabile

1983a).
In this example set of evaluation criteria, Novelty consists of “originality (idea)”,
“complexity” and “evolution”. Originality refers to the degree to which the design
itself displays a novel idea (Amabile 1983a). Complexity not only refers to the
degree to which the design shows a level of intricacy (Amabile 1983a), but also
relates to the features of a parametric design. Evolution is a specific subscale for
parametric design that refers to the degree to which the design shows evolutionary
and progressive features. The category Value is measured using three subscales:
“function”, “usefulness” and “understandable form”. Function refers to the degree
to which the outcome fulfils or demonstrates the function of a given design task.
Usefulness represents the degree to which the design shows the quality of being of
practical use (Besemer and O’Quin 1987). Understandable form refers to the degree
to which the model makes sense and is sufficiently coherent or lucid in terms of its
stated purpose. Aesthetics also includes three subscales: “aesthetic form”, “elegance”
and “well transformation”. Aesthetic form refers to the degree to which the design is
tasteful or appealing (Amabile 1983a). Elegance represents the degree to which the
design is attractive or stylish (Besemer and O’Quin 1987). Well transformation is
derived from the CPSS’s “well crafted” (Christiaans 2002) and refers to the degree to
which the model has been technically transformed and is more suited in the parametric
design environment.
To employ these evaluation criteria, experts should be familiar with the specific
design domain and the techniques for producing the design. A series of assessment
tasks are also suggested to evaluate the design product. For example, in addition to
the criterion-based assessment using the evaluation criteria, an “independent non-
criteria-based assessment” and a “comparative non-criteria-based assessment” can
be considered. The latter one, evaluating design models relative to one another,
is frequently used in design research. Most of the experimental design protocols
described in this book employ protocol analysis and expert panel assessment in
parallel to explore design cognition and creativity. However, each study uses its own
coding schemes and evaluation criteria to investigate particular research aims and
objectives.
24 1 Introduction: Exploring Design Thinking

1.5 Conclusion

This book is intended to explore new ways of investigating and promoting design
thinking in the digital Anthropocene. A triangular model comprising the themes of
“creativity”, “collaboration” and “culture” is used to improve our integrated under-
standing of new ways of design thinking in the globalised design ecosystem. Each
theme is explored through empirical studies that have been conducted by the authors
over the last 10 years. The volume of protocol data and experimental results reported
in this book are also significant, as too are the advanced and innovative methodolog-
ical approaches presented, which are especially useful for the future generation of
researchers in the field. Finally, this chapter commenced by proposing that design as
verb and noun cannot or should not be easily separated, and the closing sections of
the chapter have demonstrated why, in a methodological sense, this is also the case.

References

Alexander, Christopher. 1964. Notes on the Synthesis of Form. Cambridge, MA: Harvard University
Press.
Amabile, Teresa M. 1983a. The Social Psychology of Creativity. Springer series in social psychology.
New York: Springer.
Amabile, Teresa M. 1983b. The social psychology of creativity: A componential conceptualization.
Journal of Personality and Social Psychology 45 (2): 357–376. https://doi.org/10.1037/0022-
3514.45.2.357.
Arnheim, R. 1997. Visual Thinking. Los Angeles: University of California Press.
Asimow, Morris. 1962. Introduction to Design. Englewood Cliffs, NJ: Prentice-Hall.
Askland, Hedda Haugen, Michael J. Ostwald, and Anthony Williams. 2012. Assessing Creativity:
Supporting Learning in Architecture and Design. Sydney: Office for Learning and Teaching.
Atman, C.J., M.E. Cardella, J. Turns, and R. Adams. 2005. Comparing freshman and senior
engineering design processes: An in-depth follow-up study. Design Studies 26 (4): 325–357.
Besemer, S.P., and K. O’Quin. 1987. Creative product analysis: Testing a model by developing a
judging instrument. In Frontiers of Creativity Research: Beyond the Basics, ed. S.G. Isaksen,
341–357. Buffalo, NY: Bearly Limited.
Blosiu, J.O. 1999. Use of synectics as an idea seeding technique to enhance design creativity.
In1999 IEEE International Conference on Systems, Man, and Cybernetics, 1999. IEEE SMC ‘99
Conference Proceedings. Tokyo.
Boden, Margaret A. 2004. The Creative Mind: Myths and Mechanisms. London, New York:
Routledge.
Bonvillain, Nancy. 2010. Language, Culture and Communication: The Meaning of Messages. New
York: Pearson.
Brassett, Jamie, and John O’Reilly. 2015. Styling the future. A philosophical approach to design
and scenarios. Futures 74: 37–48. https://doi.org/10.1016/j.futures.2015.07.001.
Brown, Tim. 2009. Change by Design, How Design Thinking Transforms Organizations and Inspires
Innovation. New York: Harper Collins Publishers.
Brown, Tim, and Jocelyn Wyatt. 2010. Design thinking for social innovation. Stanford Social
Innovation Review (Winter).
Bucciarelli, Louis L. 2001. Design knowing & learning: A socially mediated activity. In Design
Knowing and Learning: Cognition in Design Education, ed. Charles M. Eastman, W. Michael
McCracken, and Wendy C. Newstetter, 297–314. Oxford: Elsevier Science.
References 25

Buchanan, Richard. 1992. Wicked problems in design thinking. Design Issues 8 (2): 5–21.
Casasanto, Daniel. 2008. Who’s afraid of the big bad Whorf? Crosslinguistic differences in temporal
language and thought. Language Learning 58 (1): 63–79.
Chai, Kah-Hin., and Xin Xiao. 2012. Understanding design research: A bibliometric analysis of
design studies (1996–2010). Design Studies 33 (1): 24–43.
Chan, C.S. 1990. Cognitive processes in architectural design problem solving. Design Studies 11
(2): 60–80.
Chan, Chiu-Shui. 2015. Style and Creativity in Design. Cham: Springer International Publishing.
Cho, Young Ik. 2011. Intercoder reliability. In Encyclopedia of Survey Research Methods, ed. Paul
J. Lavrakas. Thousand Oaks, CA: Sage Publications.
Choi, Han Hee, and Mi Jeong Kim. 2017. The effects of analogical and metaphorical reasoning
on design thinking. Thinking Skills and Creativity 23: 29–41. https://doi.org/10.1016/j.tsc.2016.
11.004.
Choi, Soonja, Laraine McDonough, Melissa Bowerman, and Jean M. Mandler. 1999. Early sensi-
tivity to language-specific spatial categories in English and Korean. Cognitive Development 14
(2): 241–268.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Christiaans, Henri H. C. M. 2002. Creativity as a design criterion. Creativity Research Journal 14
(1): 41–54.
Coley, F., O. Houseman, and R. Roy. 2007. An introduction to capturing and understanding the
cognitive behaviour of design engineers. Journal of Engineering Design 18 (4): 311–325.
Cross, Nigel. 1982. Designerly ways of knowing. Design Studies 3 (4): 221–227.
Cross, Nigel. 2011. Design Thinking: Understanding How Designers Think and Work. Oxford, New
York: Bloomsbury.
Cross, Nigel. 2018. A brief history of the design thinking research symposium series. Design Studies
57: 160–164. https://doi.org/10.1016/j.destud.2018.03.007.
Cross, Nigel, and Anita Clayburn Cross. 1995. Observations of teamwork and social processes in
design. Design Studies 16 (2): 143–170. https://doi.org/10.1016/0142-694X(94)00007-Z.
Dacey, John S., and Kathleen H. Lennon. 1998. Understanding Creativity: The Interplay of
Biological, Psychological, and Social Factors. San Francisco: Jossey-Bass.
Derrida, Jacques. 1976. Of Grammatology. Baltimore: Johns Hopkins University Press.
de Saussure, Ferdinand. 1959. Course in General Linguistics. New York: The Philosophical Society.
Dong, Andy. 2005. The latent semantic approach to studying design team communication. Design
Studies 26 (5): 445–461.
Dong, Andy. 2009. The Language of Design: Theory and Computation. London: Springer.
Dorst, Kees. 2018. DTRS: A catalyst for research in design thinking. Design Studies 57: 156–159.
https://doi.org/10.1016/j.destud.2018.03.005.
D’Souza, Newton, and Mohammad Reza Dastmalchi. 2016. Creativity on the move: Exploring
little-c (p) and big-C (p) creative events within a multidisciplinary design team process. Design
Studies 46: 6–37. https://doi.org/10.1016/j.destud.2016.07.003.
Flowers, John H., and Calvin P. Garbin. 1989. Creativity and Perception. In Handbook of creativity.
Perspectives on individual differences, ed. John A. Glover, Royce R. Ronning, and Cecil R.
Reynolds, 135–145. New York, NY, US: Plenum Press.
Georgiev, Georgi V., and Danko D. Georgiev. 2018. Enhancing user creativity: Semantic measures
for idea generation. Knowledge-Based Systems 151: 1–15. https://doi.org/10.1016/j.knosys.2018.
03.016.
Gero, John S. 1990. Design prototypes: a knowledge representation schema for design. AI Magazine
11 (4). https://doi.org/10.1609/aimag.v11i4.854.
Gero, John S. 2000. Computational models of innovative and creative design processes. Technolog-
ical Forecasting and Social Change 64 (2): 183–196. https://doi.org/10.1016/S0040-1625(99)001
05-5.
Glanville, Ranulph. 1999. Researching design and designing research. Design Issues 15 (2): 80–91.
26 1 Introduction: Exploring Design Thinking

Gleitman, Lila, and Anna Papafragou. 2005. Language and thought. In Cambridge Handbook of
Thinking and Reasoning, 633–661.
Goldschmidt, Gabriela. 1995. The designer as a team of one. Design Studies 16 (2): 189–209.
https://doi.org/10.1016/0142-694X(94)00009-3.
Goldschmidt, Gabriela, and Maria Smolkov. 2006. Variances in the impact of visual stimuli on
design problem solving performance. Design Studies 27 (5): 549–569. https://doi.org/10.1016/j.
destud.2006.01.002.
Gordon, W.J.J. 1961. Synectics: The development of creative capacity. New York: Harper.
Grudin, J. 1994. Computer-supported cooperative work: History and focus. Computer 27 (5): 19–26.
https://doi.org/10.1109/2.291294.
Guilford, Joy P. 1950. Creativity. American Psychologist 5: 444–434.
Guilford, Joy P. 1967. The Nature of Human Intelligence. New York: McGraw-Hill.
Gu, Ning, and Michael J. Ostwald. 2012. Computational methods and technologies: Reflections
on their impact on design and education. In Computational Design Methods and Technologies:
Applications in CAD, CAM and CAE Education, ed. Gu. Ning and Xiangyu Wang, 412–419.
Hershey, PA: IGI Global.
Gu, Ning, Vishal Singh, and Kerry London. 2015. BIM ecosystem: The coevolution of products,
processes, and people. In Building Information Modeling, 197–210. Wiley.
Hasirci, D., and H. Demirkan. 2007. Understanding the effects of cognition in creative decision
making: A creativity model for enhancing the design studio process. Creativity Research Journal
19 (2–3): 259–271.
Hasso Plattner Institute of Design at Stanford. 2010. An Introduction to Design
Thinking PROCESS GUIDE. https://dschool-old.stanford.edu/sandbox/groups/designresources/
wiki/36873/attachments/74b3d/ModeGuideBOOTCAMP2010L.pdf. Accessed 4 May 2020.
Hayes, John R. 1989. Cognitive processes in Creativity. In Handbook of Creativity. Perspectives on
Individual Differences, eds. John A. Glover, Royce R. Ronning, and Cecil R. Reynolds, 135–145.
New York: Plenum Press.
Hennessey, B.A., and T.M. Amabile. 1999. Consensual assessment. In Encyclopedia of Creativity,
ed. M.A. Runco and S.R. Pritzker, 346–359. San Diego, CA: Academic Press.
Herrmann, Ned. 1991. The creative brain*. The Journal of Creative Behavior 25 (4): 275–295.
https://doi.org/10.1002/j.2162-6057.1991.tb01140.x.
Herskovits, A. 1986. Language and Spatial Cognition: A Interdisciplinary Study of the Prepositions
in English. Cambridge, England: Cambridge University Press.
Ho, Chun-Heng. 2001. Some phenomena of problem decomposition strategy for design thinking:
Differences between novices and experts. Design Studies 22 (1): 27–45.
Iordanova, I., T. Tidafi, M. Guité, G. De Paoli, and J. Lachapelle. 2009. Parametric methods of
exploration and creativity during architectural design: a case study in the design studio. In Joining
Languages, Cultures and Visions: CAADFutures 2009, ed. T. Tidafi, and T. Dorta, 423–439.
Jabi, Wassim. 2013. Parametric Design for Architecture. London: Laurence King.
Jung, Jung-Ho, and Don-Ryun. Chang. 2017. Types of creativity—Fostering multiple intelligences
in design convergence talents. Thinking Skills and Creativity 23: 101–111. https://doi.org/10.
1016/j.tsc.2016.12.001.
Kan, Jeff W. T., and John S. Gero. 2008. Acquiring information from linkography in protocol studies
of designing. Design Studies 29 (4): 315–337. https://doi.org/10.1016/j.destud.2008.03.001.
Karlusch, Albrecht, Wolfgang Sachsenhofer, and Kathrin Reinsberger. 2018. Educating for the
development of sustainable business models: Designing and delivering a course to foster creativity.
Journal of Cleaner Production 179: 169–179. https://doi.org/10.1016/j.jclepro.2017.12.199.
Katz, Helene. 2002. The relationship of intrinsic motivation, cognitive style, and tolerance of
ambiguity, and creativity in scientists. College of education and human services: Seton Hall
University.
Kim, Kyung Hee. 2006. Can we trust creativity tests? A Review of the Torrance Tests of Creative
Thinking (TTCT). Creativity Research Journal 18 (1): 3–14. https://doi.org/10.1207/s15326934
crj1801_2.
References 27

Kim, Mi Jeong, and Mary Lou Maher. 2008. The impact of tangible user interfaces on designers’
spatial cognition. Human Computer Interaction 23 (2): 101–137.
Kim, M.H., Y.S. Kim, H.S. Lee, and J.A. Park. 2007. An underlying cognitive aspect of design
creativity: Limited Commitment Mode control strategy. Design Studies 28 (6): 585–604.
Kirton, M.J. 1994. Adaptors and Innovators. Styles of Creativity and Problem Solving. London:
Routledge.
Kokotovich, Vasilije, and Terry Purcell. 2000. Mental synthesis and creativity in design: An
experimental examination. Design Studies 21 (5): 437–449. https://doi.org/10.1016/S0142-694
X(00)00017-X.
Krippendorff, Klaus. 2018. Content analysis: An introduction to its methodology. Thousand Oaks,
CA: Sage publications.
Kruger, Corinne, and Nigel Cross. 2006. Solution driven versus problem driven design: Strategies
and outcomes. Design Studies 27 (5): 527–548.
Kryssanov, V.V., H. Tamaki, and S. Kitamura. 2001. Understanding design fundamentals:
How synthesis and analysis drive creativity, resulting in emergence. Artificial Intelligence in
Engineering 15 (4): 329–342.
Kuiper, Kathleen. 2010. The Britannica guide to theories and ideas that changed the modern world.
New York, NY: Britannica Educational Publishing, Association with Rosen Educational Service.
Lawson, Bryan. 1980. How Designers Think. Architectural Press; Westfield, N.J.: Eastview Editions.
Lee, Ju Hyun, and Michael J. Ostwald. 2019. Measuring cognitive complexity in parametric design.
International Journal of Design Creativity and Innovation 7 (3): 158–178. https://doi.org/10.
1080/21650349.2018.1555492.
Lee, Ju Hyun, Michael J. Ostwald, and Julie Jupp. 2013a. Understanding cognitive activities in
parametric design. In Global Design and Local Materialization, 38–49. Berlin, Heidelberg:
Springer.
Lee, Ju Hyun, Mi Jeong Kim, and Mary Lou Maher. 2013b. Designing for interactive and collective
mobile creativity. In Paper presented at the Proceedings of the 9th ACM Conference on Creativity
and Cognition, Sydney, Australia,
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014a. Parametric design strategies for the gener-
ation of creative designs. International Journal of Architectural Computing 12 (3): 263–282.
https://doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, Hyunsoo Lee, Mi Jeong Kim, Xiangyu Wang, and Peter E.D. Love. 2014b. Context-
aware inference in ubiquitous residential environments. Computers in Industry 65 (1): 148–157.
https://doi.org/10.1016/j.compind.2013.08.005.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2015. Creativity and parametric design? Comparing
designer’s cognitive approaches with assessed levels of creativity. International Journal of Design
Creativity and Innovation 3 (2): 78–94. https://doi.org/10.1080/21650349.2014.931826.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2016. The language of design: Spatial cognition
and spatial language in parametric design. International Journal of Architectural Computing 14
(3): 277–288. https://doi.org/10.1177/1478077116663350.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2017. Cognitive Challenges for Teamwork in
Design. In Collaboration and Student Engagement in Design Education, 55–75. Hershey, PA,
USA: IGI Global.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2019. Cognitive and linguistic differences in
architectural design. Architectural Science Review 62 (3): 248–260. https://doi.org/10.1080/000
38628.2019.1606777.
Lewis, Rhodri. 2012. Language, Mind and Nature. Cambridge: Cambridge University Press.
Liu, Yu-Tung., and Chor-Kheng Lim. 2006. New tectonics: A preliminary framework involving
classic and digital thinking. Design Studies 27 (3): 267–307.
London, K., and V. Singh. 2013. Integrated construction supply chain design and delivery solutions.
Architectural Engineering and Design Management 9 (3): 135–157. https://doi.org/10.1080/174
52007.2012.684451.
28 1 Introduction: Exploring Design Thinking

López-Mesa, B., and G. Thompson. 2006. On the significance of cognitive style and the selection
of appropriate design methods. Journal of Engineering Design 17 (4): 371–386. https://doi.org/
10.1080/09544820500274100.
Maher, M.L. 2010. Evaluating creativity in humans, computers, and collectively intelligent systems.
In DESIRE’10: Creativity and Innovation in Design. Aurhus, Denmark.
Maher, M.L., and Josiah Poon. 1996. Modeling design exploration as co-evolution. Computer-Aided
Civil and Infrastructure Engineering 11 (3): 195–209. https://doi.org/10.1111/j.1467-8667.1996.
tb00323.x.
Martin, Roger. 2009. The Design of Business: Why Design Thinking is the Next Competitive
Advantage. Boston, MA: Harvard Business Press.
Mc Neill, T., J.S. Gero, and J. Warren. 1998. Understanding conceptual electronic design using
protocol analysis. Research in Engineering Design—Theory, Applications, and Concurrent
Engineering 10 (3): 129–140.
McKim, Robert H. 1973. Experiences in Visual Thinking. Monterey, CA: Brooks/Cole.
Miller, George A. 2003. The cognitive revolution: A historical perspective. Trends in Cognitive
Sciences 7 (3): 141–144.
Mouter, Niek, and D.M. Vonk Noordegraaf. 2012. Intercoder reliability for qualitative research:
You win some, but do you lose some as well? In Proceedings of the 12th TRAIL Congress, 30–31
oktober 2012, Rotterdam, Nederland. TRAIL Research School.
Mulet, Elena, Vicente Chulvi, Marta Royo, and Julia Galán. 2016. Influence of the dominant thinking
style in the degree of novelty of designs in virtual and traditional working environments. Journal
of Engineering Design 27 (7): 413–437. https://doi.org/10.1080/09544828.2016.1155697.
Neumeier, Marty. 2009. The Designful Company: How to build a culture of nonstop innovation.
Berkeley, CA: New Riders.
Newman, Stephen. 1999. Constructing and critiquing reflective practice 1. Educational Action
Research 7 (1): 145–163. https://doi.org/10.1080/09650799900200081.
Osborn, Alex F. 1963. Applied Imagination: Principles and Procedures of Creative Thinking. New
York: Scribner.
Ostwald, Michael, and J. 2010. Ethics and the auto-generative design process. Building Research
and Information. 38 (4): 390–400.
Ostwald, Michael J. 2012. Systems and enablers: Modeling the impact of contemporary computa-
tional methods and technologies on the design process. In Computational Design Methods and
Technologies: Applications in CAD, CAM and CAE Education, ed. Gu. Ning and Xiangyu Wang,
1–17. Hershey, PA: IGI Global.
Phare, Darin, Gu. Ning, and Michael J. Ostwald. 2016. Representation in collective design: Are
there differences between expert designers and the crowd? In Cooperative Design, Visualization,
and Engineering, ed. Yuhua Luo, 59–68. Cham: Springer International Publishing.
Phare, Darin, Gu. Ning, and Michael J. Ostwald. 2018. Representation in design communication:
meaning-making in a collective context. Frontiers in Built Environment 4 (36): 1–9. https://doi.
org/10.3389/fbuil.2018.00036.
Pinker, S. 1994. The Language Instinct: The New Science of Language and Mind. New York:
William Morrow and Company.
Rhodes, Mel. 1961. An analysis of creativity. The Phi Delta Kappan 42 (7): 305–310.
Rosenman, M.A., and J.S. Gero. 1993. Creativity in design using a design prototype approach.
In Modeling Creativity and Knowledge-Based Creative Design, ed. J.S. Gero and M.L. Maher,
119–148. Hillsdale, New Jersey: Lawrence Erlbaum.
Rowe, Peter G. 1987. Design Thinking. Cambridge, MA: The MIT Press.
Rowland, I.D., and T.N. Howe. 1999. Vitruvius: Ten books on Architecture. Cambridge, UK:
Cambridge University Press.
Runco, Mark A., and Selcuk Acar. 2012. Divergent thinking as an indicator of creative potential.
Creativity Research Journal 24 (1): 66–75. https://doi.org/10.1080/10400419.2012.652929.
References 29

Santandreu, D., and A.F. Safiullin. 2015. Can culturally, disciplinarily and educationally diverse
(D3) teams function and be creative? A case study in a Korean university. Educational Studies
41 (4): 369–392. https://doi.org/10.1080/03055698.2015.1018871.
Sarkar, Prabir, and Amaresh Chakrabarti. 2011. Assessing design creativity. Design Studies 32 (4):
348–383. https://doi.org/10.1016/j.destud.2011.01.002.
Sawyer, R.K. 2006. Explaining Creativity. The Science of Human Innovation. Oxford: Oxford
University Press.
Schön, Donald A. 1983. The Reflective Practitioner: How Professionals Think in Action. New York:
Basic Books.
Schön, Donald A. 1984. The architectural studio as an exemplar of education for reflection-in-action.
Journal of Architectural Education 38 (1): 2–9.
Schön, Donald A., and Glenn Wiggins. 1992. Kinds of seeing and their functions in designing.
Design Studies 13 (2): 135–156. https://doi.org/10.1016/0142-694X(92)90268-F.
SHoP/Sharples Holden Pasquarelli. 2002. Versioning: Evolutionary Techniques in Architecture.
London: Wiley.
Simon, Herbert. 1969. Sciences of the Artificial. Cambridge, MA: MIT Press.
Simonton, D.K. 2003. Scientific creativity as constrained stochastic behavior: The integration of
product, person, and process perspectives. Psychological Bulletin 129 (4): 475–494.
Singh, Vishal, and Ning Gu. 2012. Towards an integrated generative design framework. Design
Studies 33 (2): 185–207. https://doi.org/10.1016/j.destud.2011.06.001.
Spiller, Neil. 2008. Digital Architecture Now: A Global Survey of Emerging Talent. London: Thames
and Hudson.
Stempfle, Joachim, and Petra Badke-Schaub. 2002. Thinking in design teams—An analysis of team
communication. Design Studies 23 (5): 473–496. https://doi.org/10.1016/S0142-694X(02)000
04-2.
Sternberg, Robert J., and Todd I. Lubart. 1999. The concept of creativity: Prospects and paradigms.
In Handbook of Creativity, ed. Robert J. Sternberg, 3–15. Cambridge University Press.
Suwa, Masaki, and Barbara Tversky. 1997. What do architects and students perceive in their design
sketches? A protocol analysis. Design Studies 18 (4): 385–403. https://doi.org/10.1016/S0142-
694X(97)00008-2.
Suwa, M., T. Purcell, and J. Gero. 1998. Macroscopic analysis of design processes based on a
scheme for coding designers’ cognitive actions. Design Studies 19 (4): 455–483.
Tang, H.H., Y.Y. Lee, and J.S. Gero. 2011. Comparing collaborative co-located and distributed
design processes in digital and traditional sketching environments: A protocol study using the
function–behaviour–structure coding scheme. Design Studies 32 (1): 1–29. https://doi.org/10.
1016/j.destud.2010.06.004.
Tenbrink, Thora, and Marco Ragni. 2012. Linguistic principles for spatial relational reasoning. In
Spatial Cognition VIII, ed. Cyrill Stachniss, Kerstin Schill, and David Uttal, 279–298. Lecture
Notes in Computer Science. Berlin, Heidelber: Springer.
Thang, Binh, Wouter Sluis-Thiescheffer, Tilde Bekker, Berry Eggen, Arnold Vermeeren, and Huib
de Ridder. 2008. Comparing the creativity of children’s design solutions based on expert assess-
ment. In Proceedings of the 7th International Conference on Interaction Design and Children.
Chicago, Illinois: ACM.
Valkenburg, Rianne, and Kees Dorst. 1998. The reflective practice of design teams. Design Studies
19 (3): 249–271. https://doi.org/10.1016/S0142-694X(98)00011-8.
van der Zee, E., and J.M. Slack. 2003. Representing Direction in Language and Space. Oxford
University Press.
Wallas, G. 1926. The Art of Thought. New York: Harcourt Brace.
Ware, C. 2008. Visual Thinking for Design. Burlington, MA: Morgan Kaufmann.
Williams, Anthony, Michael J. Ostwald, and Hedda Haugen Askland. 2010. Creativity, Design
and Education: Theories, Positions and Challenges. Sydney: Australian Learning and Teaching
Council.
30 1 Introduction: Exploring Design Thinking

Wittgenstein, Ludwig. 1922. Tractatus Logico-Philosophicus. New York: Harcourt, Brace &
Company.
Woodbury, Robert. 2010. Elements of Parametric Design. London, New York: Routledge.
Yu, Rongrong, Ning Gu, and Ju Hyun Lee. 2013. Comparing designers’ behavior in responding to
unexpected discoveries in parametric design environments and geometry modeling environments.
International Journal of Architectural Computing 11 (4).
 Part I
Creativity
Chapter 2
Design Strategies and Creativity

Abstract This chapter uses the results of two studies to develop an understanding
of different types of design strategies and their connection to creativity in design.
Two sets of experimental data are used to capture these strategies and then correlate
them to readings of novice or expert practices, and the production of conventional or
creative designs. The first study identifies three effective design strategies during the
conceptual design stage: drawing-reflection, graphical-goal forwarding and textual-
goal forwarding. The second study identifies two generative strategies in para-
metric design for developing creative solutions and products: problem-forwarding
and solution-reflecting. The chapter explains these strategies and links them to past
research about design cognition and creativity.

2.1 Introduction

This chapter is primarily concerned with two themes. The first is the design process,
viewed in cognitive and strategic terms. This encompasses the tactics, processes
or sequences of activities designers employ to solve problems. The second theme
is creativity, which refers to the level of innovation or originality implicit in the
product of a design process. These two themes are often connected in past research,
because one sign of a successful design process is that its products are demonstrably
creative. Furthermore, a core value of design thinking is its capacity to support
creativity. Given the importance of this connection, it is not surprising that multiple
empirical studies have attempted to link broad design strategies or specific problem-
solving techniques to creative outcomes (Goldschmidt and Smolkov 2006; Koko-
tovich 2008; Lee et al. 2015). Despite such research into the connection between
design thinking and creativity (Amabile and Pillemer 2012; Batey 2012; So and Joo
2017), evidence about the precise strategies obor techniques to achieve this connec-
tion remains limited. In response, this chapter reports the results of two studies using
protocol analysis and Consensual Assessment Techniques (CAT) to identify design
strategies that may enhance creativity.
The first study in this chapter is concerned with the cognitive processes of
designers working in “traditional” pen-and-paper environments, using sketching as

© Springer Nature Switzerland AG 2020 33

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_2
34 2 Design Strategies and Creativity

their main tool. By examining these processes, different cognitive and problem-
solving strategies are identified and discussed. The study identifies that three of these
cognitive strategies––drawing-reflection, graphical-goal forwarding and textual-goal
forwarding—correlate to expert design and problem-solving processes. The results
are analysed through reference to past theories and studies of problem-solving and
goal-setting behaviours in design (Weth 1999).
The second study in this chapter is about the cognitive processes of designers
working in parametric design environments. Using protocol analysis, this study
assesses the types and levels of creativity present in design outcomes. This allows
for a direct correlation to be made between cognitive processes and creative prod-
ucts. The successful strategies identified in this study are problem-forwarding and
solution-reflecting. Both studies in this chapter examine cognitive strategies in
design, considering the complex and diverse combinations of tactics and techniques
used by designers. Before the two studies are explained and the data presented, this
introduction briefly outlines some key concepts and ideas.
In design thinking, problem-solving is the process where solutions are found to
the needs of clients, which are typically encapsulated in the functional performance
requirements of products. Problem-solving involves “a sequential search, making
small successive accretions to the store of information about the problem” (Simon
1978, p. 274). It can be thought of as the process of moving from one problem-
solving state to another, by the way of evaluation functions (Goel and Pirolli 1992).
Problem-solving involves combinations of analytic and synthetic thinking, divergent
and convergent thinking or intuitive and lateral thinking. Newell and Simon’s (1972)
information-processing theory defines problem-solving as the interaction between
an information-processing system, the problem-solver and the task environment. The
task environment is conceptualised in recent models as the “problem space” because
it is where the problem is first defined or understood. The problem space can also be
explained as the volume of cognitive activity that is bounded first by divergent and
later convergent, thinking. In practice, undertaking a specific design task for a client
or user involves understanding the design brief (the problem space) and formulating
an effective response to this (the solution space).
Multiple problem-solving approaches and techniques have been identified in the
fields of psychology, cognitive science, computational intelligence, and design. Some
of the most well-known of these use analogies, heuristics, synectics, lateral thinking
and directional thinking. Each of these approaches is briefly summarised hereafter.
• Problem-solving by analogy involves the adoption or adaption of solutions devel-
oped for analogous situations, treating these situations as a body of knowledge to
draw on. This approach is common in both traditional and computational design
processes (Choi and Kim 2017; Gero 2000; Rosenman and Gero 1993).
• Heuristic problem-solving uses self-discovery or pragmatic “rules of thumb” to
identify obvious or pragmatic solutions (Wang and Chiew 2010). Such solutions
are rarely optimal or creative, although they can be refined and improved over
time.
2.1 Introduction 35

• Synectic, collaborative and brainstorming problem-solving techniques all require

groups to be actively involved in the process (Gordon 1961). Members of the
groups may use analogical or heuristic problem-solving techniques; however,
by working in a group the weaknesses inherent in these approaches are easy to
overcome. “Idea seeding” (Blosiu 1999) is a variation of collaborative problem-
solving, where a group of people are involved, but initially work separately
proposing, developing and refining solutions. These solutions are swapped
between other group members at regular intervals, before they finally work
together to rank options and select the right solution.
• Lateral problem-solving, like lateral thinking, uses reasoning, situational thinking
or concepts that are not immediately obvious, as a catalyst for solution-finding.
Lateral thinking approaches can lead to creative solutions, and especially when
working in teams, although the solutions they generate are not always optimal
(Karlusch et al. 2018).
• The “working-forward” problem-solving techniques start with a detailed and
directed consideration of problems, before proposing and testing solutions. In
practice, they often combine goal and sub-goal setting and problem segmentation
or decomposition. For example, the “hill-climbing” technique involves hierarchi-
cally setting sub-goals (arrayed from larger to smaller scale) and solving them,
step-by-step, to reach an overarching solution. The “divide-and-conquer” tech-
nique decomposes a whole problem into a set of sub-problems which are then
prioritised and solved (Wang and Chiew 2010). In this approach, priority may
be given to sub-problems with obvious or parsimonious solutions (the “Occam’s
razor” model), leaving more complex or wicked sub-problems to later.
• The “working-backward” problem-solving technique starts with solutions that
are then tested to see if they solve the problem. This is sometimes called a “non-
directed” approach. One example of a non-directed approach is “trial and error”,
which is sometimes called “backward reasoning”. Such strategies are often found
in novice design processes (Ahmed et al. 2003; Lee et al. 2015). However, the
combination of “hill-climbing” and “working-backward” approaches, known as
a “means-ends analysis” (Simon 1981), is common and effective in some fields,
including Artificial Intelligence (AI).
The processes of defining and setting goals are often subsumed in problem-solving
models, even though they are significant in themselves (Weth 1999). Furthermore,
goal segmentation or the setting of sub-goals can be an effective approach to complex
problems. The process of defining and sequencing sub-goals can be undertaken hier-
archically (“most to least” important or “biggest to smallest” scale) or sequentially
(“first to last”, as defined in a project brief).
A “design strategy” is made up of the (i) type, (ii) combination and (iii) sequence
of techniques and activities that occurs during the design process. Potentially, each
and every designer has their own individual strategy that is unique to them (Lee et al.
2014). Designers from different backgrounds, cultures and environments will also
approach problems differently (Lee et al. 2019). Similarly, designers with diverse
levels of experience will often adopt different strategies. Past research has begun
36 2 Design Strategies and Creativity

to qualitatively and quantitatively map some of these design strategies to types of

outcomes and levels of experience. Kruger and Cross (2006), for example, demon-
strate that design strategy is linked to both design quality and creativity, and they
specifically investigate problem- and solution-driven strategies. They argue that the
problem-driven strategy often produces a better balance between creativity and
quality, while the solution-driven strategy tends to result in a more creative, but
lower quality outcomes. Ho’s (2001) study of expert and novice designers’ strate-
gies identifies multiple differences in their applications of problem decomposition,
working-forward and working-backward strategies. Ahmed et al. (2003) also reveal
that novice designers often rely on “trial and error” and are rarely aware that they
possess, or are using a design strategy.
The two studies that make up the core of this chapter examine design strategies
used by participants in design experiments. They specifically identify the (i) types,
(ii) combinations and (iii) sequence of activities in the design process. In this way,
they characterise various strategies in terms of their problem-solving, goal setting,
solution-finding and other characteristics. From there, it is possible to differentiate
expert and novice strategies, and map specific strategies to creative outcomes.

2.2 Study I

The first study highlights the results of a protocol analysis of designers’ sketching
activities in a “traditional”, manual (non-digital) design environment. The study
uses the protocol data to explore two important components of a creative design
process: “problem-structuring” and “problem-solving” activities. This introduction
to the first study provides a brief background as to why sketching is significant and
why problem-solving and structuring are important processes.
As previously noted, design is not a straightforward, linear and predictable
process. It is often subjective, recursive and heuristic. It requires designers to engage
with multiple complex cognitive processes and the concepts that arise through the
processes of visually and verbally communicating an idea or proposition. As such,
design cognition can be thought of as a product of the interplay between two distinct
subsystems, visual imagery and verbal systems (Paivio 1971). The former subsystem
comprises the representational language of sketches, drawings and models, whereas
the latter is the spoken or written word. Probably, the most basic example of design
representation is the sketch, and it is not surprising that it has been the regular
subject of past research in design thinking (Tovey 1989). For example, Suwa and
Tversky (1997) argue that sketches support the design process by crystallising ideas
and concepts. Goldschmidt and Smolkov (2006) uncover the role sketching plays
in design problem-solving, and Mathias (1993) highlights the importance of draw-
ings and models as representations of the stages in design thinking. Importantly, the
sketches produced in the design process provide a basis for empirical investigation
of design using protocol analysis (1993). This is why the first study in this chapter
is focussed on sketch-based design processes.
2.2 Study I 37

Mathias (1993) identifies four categories or classifications of drawings and models

in the design problem-solving process. The first, “drawing for analysis and initial
ideas”, deals with the preliminary assembly of concepts and the rapid exploration of
ideas through free-hand drawing. The second, “drawing for synthesis”, is concerned
with detailed considerations to aid decision-making. The next, “style and nature of
the drawing”, is concerned with the shift in representational content from concept to
format that can occur in a drawing. The last, “the use of drawing to support thinking”,
covers representations intended to model and test ideas. Mathias (1993) also presents
a framework for considering design strategy, which considers modes of thinking
(convergent, divergent, holistic, serial, analytic, synthetic) and levels of capacity
or ability (novice or expert). This framework, which supports the interpretation of
various modes of thinking in design stages, has parallels with several other design
models, including two that emphasise the strategies adopted by designers.
Jones’ (1963) ASE framework distinguishes three stages in the design process—
analysis, synthesis and evaluation—to capture both logical and creative thought
patterns. Akin (1978) identifies eight staged information-processing mechanisms
used in developing design solutions. They are (i) information acquisition, (ii) problem
interpretation, (iii) problem representation, (iv) solution generation, (v) solution inte-
gration, (vi) solution evaluation, (vii) perception and (viii) sketching. Akin also
acknowledges that not all solutions arise from an analysis of all relevant aspects
of a problem. As such, Akin (1979) questions the sequential assumptions in ASE
and also argues that “analysis” can be a part of all phases of the design process.
Gero also proposes that in the early stages of the design process “formulation” is a
better description of the cognitive process, and “analysis” is more correctly “used
to refer to a precursor of evaluation” (Gero 1999, p. 49). Gero’s division between
formulation and analysis is also instrumental in several related versions of the design
process modelling. For example, Hay et al. (2017) suggest that the conceptual design
stage can be understood as being focussed on search and exploration. They argue that
the “design as search” strategy addresses solution-seeking and problem-structuring
behaviours, while the “design as exploration” strategy involves (i) co-evolutionary
design, (ii) visual reasoning processes, (iii) cognitive actions and (iv) unexpected
discovery or situated invention.
The design and problem-solving models of Jones, Akin, Gero and Hay all connect
design strategies to processes and outputs. Significantly, several of these design
strategies have also been tied to the production of creative solutions (Kokotovich
2008). The first study in this chapter looks at the cognitive and behavioural strategies
of a small group of designers. It uses protocol analysis to differentiate approaches to
problem-solving and structuring activities in a traditional design environment, and
comments on the alignment between strategies and theorised levels of expertise and
creativity.
38 2 Design Strategies and Creativity

2.2.1 Research Procedure

The first study in this chapter has six participants (Ko1–Ko6), all Masters-level
students in the Department of Interior Architecture Design at Hanyang University in
South Korea. All participants had successfully completed at least one major design-
studio project and had sufficient English skills to clearly understand a design brief
written in English, as well as to enable general written communication. Despite
all being enrolled in the same level of study, the participants could not be classi-
fied as either “experts” or “novices”, with all six having some professional design
experience.
Before each experiment was conducted, a research coordinator explained to all
participants the design brief (Box 2.1) and project site for the design (Fig. 2.1),
which provide the stimulus for the study. All participants were also given time before
the design experiment started to practice “thinking aloud”. The experiments were
conducted serially, one after the other, with a gap between each.

Fig. 2.1 Building site

2.2 Study I 39

Box 2.1 Design Brief for Study I

Multi-Generational Housing Design
This project asks participants to examine a vacant block in a typical suburban
street (drawings and support documents attached) and explore how to accom-
modate a multi-generational house. Participants are asked to illustrate their
conceptual ideas for the design of a multi-generational house for inhabitants
who range from young children to seniors. The concept design should suggest
how a multi-generational housing community would better support sustainable
living for inhabitants.
Design Brief
Design a house to accommodate and meet the needs of a multi-generational
family in a vacant suburban block.
Considerations
• Environments should be created to be accessible for all users—including
those who are challenged with mobility, dexterity or sensory/cognitive
processing.
• Environments should be considered to be safe for all users—including those
who are very young or senior occupants.
• Environments should balance both communal living requirements and the
need for privacy.
• Planning considerations to maximise limited space
• Indoor environments should have physical and visible connection to outdoor
environments.
Occupants
• Couple aged 70–90 with limited mobility
• Couple aged 35–45
• 2 children aged 0–12
• Occupants can be from diverse linguistic and cultural backgrounds.

Deliverable: Sketch(s) of four design stages/components of the multi-

generational housing design:
1. Initial conceptual design
2. Site layout(s) of the multi-generational house
3. Plan(s)
4. Elevation(s) and/or section(s)

Time limit: 45 min.

40 2 Design Strategies and Creativity

Camera 1 Camera 2

Fig. 2.2 Perspectives of two cameras

(i) Experimental setting. Participants were asked to take part in a 45-min experi-
ment in their own studio space. They undertook a specified design task, “multi-
generational housing design”, with pencil-and-paper tools on a normal desk.
While participants were allowed to use their own tools, they were all provided
with yellow tracing paper, A3 white paper and pencils. Two video cameras
were installed: one (Camera 1) focusing on the desk to capture sketch activi-
ties or processes, and the other (Camera 2) on the participant from the side, to
give a clear view of the designer’s activities (see Fig. 2.2). One voice recorder
was also installed to capture the designer’s “thinking aloud” data. Participants
were able to stop their design session at any time, but they were encouraged to
complete at least one of the deliverables in the 45-min period. The deliverables
were initial conceptual design(s), site layout(s), plan(s), elevation(s) and/or
section(s). After each design experiment, all deliverables were collected.
(ii) Verbalisation. Participants were asked to verbalise or describe their thoughts
and activities using their own language (Korean). The only time the research
coordinator spoke during a session was to remind participants, if needed, to
“please keep talking” in order to achieve effectively “think-aloud”.
(iii) Transcription and segmentation. Two native Korean speakers transcribed and
segmented the recorded data, according to cognitive activities (codes) specified
in the coding scheme (Table 3.1). The first coder, who had over 5 years of
experience in art and design practice, transcribed Camera 1’s video recording
data with the support of voice recordings and segmented the data for analysis.
The second coder having over 7 years of experience in protocol analysis in
design undertook segmentation of the transcribed protocol data by the first
coder, mainly using Camera 2’s video recording to capture further changes in
cognitive activities. Consequently, the second coder produced more segments
for each protocol.
(iv) Encoding with a coding scheme. The coding scheme (Table 2.1) used for this
study is a modified version of Suwa et al.’s (1998) scheme, consisting of five
categories (representation, perception, function, evaluation and goal setting).
The representation category consists of three physical design activities, R-
drawing, R-writing and R-label, which often occur during a sketch-based design
2.2 Study I 41

Table 2.1 Coding scheme for exploring the traditional design process
Category Subclass Description
Representation R-drawing Make drawings (geometries) and/or geometric ideas
R-writing Write texts
R-label Make labels (or titles)
Perception P-drawing Attend to drawings
P-brief Attend to a design brief or its summary or a (given) site
P-writing Attend only to participant’s own writing
Function F-interaction Explore the issues of interactions between artefacts, people and
nature
F-reaction Consider psychological reactions of people
Evaluation E-drawing Evaluate existing drawings
E-writing Evaluate existing texts
E-label Evaluate existing labels
Goal setting G-initial Introduce goals (ideas) based on a given design brief
G-sub Introduce new goals extended from a previous one
G-repeat Repeated goals from a previous segment

process. The perception category consists of three subclasses, P-drawing, P-

brief and P-writing. In particular, P-brief captures reviewing a design brief as
well as any site information made available, which is sometimes followed by
introducing new goals (ideas). In the function category, F-interaction is related
to exploring the interactions between artefacts, people and nature, while F-
reaction records activities associated with the psychological and emotional
reactions of people. The evaluation and goal-setting categories are related to
Suwa et al.’s conceptual level of cognitive actions. Both categories facilitate
identification of cognitive patterns and design strategies. If a segment can be
encoded using more than one code, it is encoded using the lower category code,
because processing at an upper level is based on that at a lower level (Suwa
et al. 1998).
(v) Arbitration and intercoder reliability. The final protocol data used for the anal-
ysis were developed through an arbitration process where both coders agreed
on the segmentation and coding of each protocol. Table 2.2 describes the results
of segmentation and intercoder reliability of each protocol. The average time
duration of a design protocol was 42 min 29.4 s, and the average number of
segments was 173.67. The average time per segment was 15.03 s. The average
score of percent agreement in the table is slightly lower than the overall target
of 80%. Krippendorff’s α values—measuring intercoder reliability for nominal
data with missing value—are in the range of “tentative conclusions” for items
with α values between 0.80 and 0.67 (Krippendorff 2011, 2018). While it is
customary to seek results were α ≥ 0.800, the scores in this first study are
42 2 Design Strategies and Creativity

Table 2.2 Results of segmentation and intercoder reliability of each protocol

Protocol code Time Num. of Time per Intercoder reliability
duration segments segment Percent Krippendorff’s
(second) agreement (%) alpha
Ko1 45 min 1 s 137 19.72 76.83 0.677
Ko2 30 min 2 s 136 13.25 77.17 0.590
Ko3 44 min 178 15.13 77.40 0.676
52.8 s
Ko4 46 min 227 12.20 78.35 0.670
9.3 s
Ko5 46 min 1 s 159 17.36 73.17 0.686
Ko6 42 min 205 12.54 87.65 0.826
50 s
Average 42 min 173.67 15.03 78.43 0.688
29.4 s

common for a cognitive protocol analysis identifying individual design strate-

gies. As Stevens et al. (2014) note, in many studies Krippendorff’s upper and
lower standards must be reduced due to the presence of items that are either
highly discrete (0.70 and 0.58) or highly distributed (0.40 and 0.33).
(vi) Normalisation of Time Duration. Although participants were notionally given
45 min to complete the design task, one designer (Ko2) took less time, and two
(Ko4 and Ko5) took longer time (Table 2.2) To accommodate these differences,
the coding results refer to the percentage of the frequency weighted by time
duration of each code (Table 2.3).

2.2.2 Coding Results

The protocol data for the first study reveal that, on average in the design experiment,
representation activities account for 53.6% (R-drawing: 45.4%, R-writing: 3.4%, R-
label: 4.8%); perception activities account for 23.0% (P-drawing: 16.3%, P-brief :
4.7%, P-writing: 2.0%); function activities account for 12.3% (F-interaction: 11.4%,
F-reaction: 0.8%); evaluation activities account for 6.8% (E-drawing: 6.5%, E-
writing: 0.4%); and goal-setting activities account for 4.3% (G-initial: 1.7%, G-sub:
2.2%, G-repeat: 0.3%) (Table 2.3).
All participants produced relatively large amounts of representation (or physical)
activities (particularly, R-drawing, 45.4%) and only relatively small amounts of goal-
setting (4.3%) activities. On average, they also produced smaller volumes of the lower
category of cognitive activities than the upper category in the scheme hierarchy. This
pattern might seem significant, but individual cognitive allocations are quite different
in Table 2.3. For example, Ko1 produced the largest amount of both R-drawing and
F-interaction and the smallest amount of P-drawing (the fifth-ranked subclass in
2.2 Study I 43

Table 2.3 The distributions of the coding results (the percentage of time duration)
Category Subclass Designer Mean SD
Ko1 Ko2 Ko3 Ko4 Ko5 Ko6
Representation R-drawing 54.5 53.3 51.7 49.2 23.2 40.6 45.4 12.0
R-writing 2.4 0.0 2.5 4.1 9.6 1.9 3.4 3.3
R-label 6.7 3.4 6.7 5.0 2.9 3.8 4.8 1.7
Perception P-drawing 6.3 19.1 14.6 19.7 12.1 26.1 16.3 6.9
P-brief 2.1 0.7 6.2 4.0 10.5 4.7 4.7 3.4
P-writing 0.0 0.0 0.0 0.0 12.0 0.0 2.0 4.9
Function F-interaction 16.4 15.3 7.9 5.1 12.6 11.2 11.4 4.3
F-reaction 1.2 0.9 0.0 0.0 1.6 1.3 0.8 0.7
Evaluation E-drawing 7.0 5.3 2.6 11.7 3.4 8.8 6.5 3.4
E-writing 0.0 0.0 0.0 0.0 2.1 0.0 0.4 0.9
E-label 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
Goal setting G-initial 1.4 0.5 2.2 0.5 4.2 1.6 1.7 1.4
G-sub 1.8 1.4 3.6 0.7 5.8 0.0 2.2 2.1
G-repeat 0.0 0.0 2.0 0.0 0.0 0.0 0.3 0.8
Sum 100.0 100.0 100.0 100.0 100.0 100.0 100.0 –

her protocol), while the others generally focussed on drawings (the second-ranked
subclass in their protocols, except for Ko5). The results indicate the presence of
multiple cognitive patterns and preferences, or different thinking styles and design
strategies.

2.2.3 Sketch-Based Design Strategies

The results in study one indicate that half the participants (Ko1, Ko3 and Ko5)
adopted specific design strategies, while the remainder used ad hoc or more idiosyn-
cratic methods. For example, Ko1, Ko3 and Ko5 produced more goal-setting activi-
ties than the other three (see Table 2.3). Goal-setting activities are identified in both
Simon’s (1973) and Jones’ (1963) models as being part of a conventionally struc-
tured design strategy. Notably, experts often establish sub-problems or reformulate
problems in the design process, while novices tend not to employ this approach (Lind-
ström 2006). As such, it could be argued that Ko1, Ko3 and Ko5 exhibit signs of a
more advanced or experienced design strategy. In contrast, Ko2, Ko4 and Ko6 spent
more time “attending” to their drawings (P-drawing). A relatively small proportion
of the perception code for “attend to drawing” can be associated with “incuba-
tion” moments, when contemplating a drawing leads to creative insights. However,
too much P-drawing activity may indicate a “trial-and-error” strategy or a “lack of
confidence” in a decision, both behaviours typical of novice designers (Ahmed et al.
44 2 Design Strategies and Creativity

2003). Thus, the remainder of this study focusses largely on the sketch-based and
deliberative design strategies of Ko1, Ko3 and Ko5, as examples of problem-solving
strategies and cognitive activities during the design process.
Figure 2.3 depicts the three selected designers’ representations over time for
the design sessions. In the data Ko1 adopts a design strategy that commences with
drawing a plan and then develops solutions and refinements through iterative graph-
ical reflection (“drawing-reflection”). Where novice designers (Ko2, Ko4 and Ko6)
also drew and contemplated or evaluated drawings (P-drawing, E-drawing), they
did so without goal setting or clear direction. In contrast, Ko1 continuously evolved

Ko1

Ko3

Ko5

Fig. 2.3 Designers’ evolving representations over time

2.2 Study I 45

her designs and produced sequential deliverables at each step. Ko3’s data initially
develop a bubble diagram encapsulating initial and sub-goals and subsequently
generate detailed solutions (“graphical-goal”). He also develops a further graphical-
goal (G-sub) for a site layout at the end of his design session. Ko5’s cognitive strategy
commences with a written summary of goals which he continues to use to inform
the generation of designs (“textual goals”). He also uses bubble diagrams to develop
design solutions, rather than for goal-setting purposes.
It is also informative to review the data describing the behaviours and cognitive
strategies of these designers (Fig. 2.4). Ko1, for example, continuously explores
the interactions between artefacts, people and nature (F-interaction) in plan layouts
focussing on related deliverables until her design session ends. She exhibits lower
proportions of perception and evaluation activities, but they are well distributed
across the session. Her initial goal also evolves over time. Thus, her protocol

Ko1

Start Time End

Ko3

Start Time End

Ko5

Start Time End

Fig. 2.4 Individual patterns of cognitive activities of different categories over time
46 2 Design Strategies and Creativity

produces the highest percentage of representation” (63.7%) and function (17.7%)

and the lowest percentages of perception” (8.4%) (Table 2.3). She adopts a so-called
“drawing-reflection” strategy and continuously generates sequential design solu-
tions. Thus, her design strategy is close to an expert’s “drawing-reflection forwarding
strategy” which differs from a novice’s less directed “drawing-reflection behaviour”.
Both Ko3 and Ko5 develop clear goals at the beginning of the design session
and continue to use initial goals for the generation of designs. Both designers
also subsequently generate detailed designs based on “graphical goals” or “tex-
tual goals”, respectively. Their cognitive patterns show that they start by exhibiting
more goal-setting and perception activities—mainly G-initial and P-brief , respec-
tively (Fig. 2.4). They then produce more function and evaluation activities in the
middle and towards the end of their sessions. Thus, Ko3’s strategy is characterised
as a “graphical-goal forwarding strategy”, while Ko5’s is a “textual-goal forwarding
strategy”.
By comparing not just the frequency of behaviours, but also their timing, it is
possible to construct a picture of different design strategies and cognition processes
(Fig. 2.4). For example, across all three strategies (as exhibited in Ko1, Ko3 and
Ko5, respectively) there is a relatively consistent level of representation activities,
and perception activities also take place throughout all three protocols. There is,
however, a different density or frequency of perception activities, with Ko3 being
more reliant on this behaviour, and Ko1, far less. Despite being able to clearly identify
patterns in cognitive activity in this data, and categorise these as expert or novice,
the results cannot be directly tied to creativity. Rather the link to creativity is by
way of past research, outlined previously in this chapter, which connects particular
problem-solving strategies to potentially innovative outcomes.

2.3 Study II

The “environment” for the first study was a traditional design context, involving
“pencil and paper” on a desk in a studio. For the second study, the environment
is parametric design, or algorithmic scripting in a computer using software. In a
traditional design environment drawn objects only have shape and are only seen and
manipulated as shapes. Thus, for example, a sketch of a rectangle might represent a
door, but the sketch itself has only its visible properties, few of which are actually
associated with the door. In contrast, in a parametric design environment objects are
defined by innate rules, properties, configurations and dependencies which are visible
and able to be manipulated (Monedero 2000). For example, a parametric object may
be coded as a door, being at least human height and width, having an associated frame,
hinges and handle position, all of which will alter if the basic dimensions of the door
are changed. The fact that the door has a rectangular shape is just a by-product of
the rules that define its parameters. As such, objects are defined in this environment
in terms of mathematically formulated properties, rather than shapes on a page. The
major software tools used for this purpose in parametric design are Grasshopper (a
2.3 Study II 47

plug-in for Rhino), CATIA and Generative Components. Each of the programs uses
“scripting” to enable design (Holzer et al. 2007).
Scripting, sometimes called “algorithmic activity”, is the process of writing or
graphically programming parameters, rules and their topological relationships (Salim
and Burry 2010; Burry 2011). In the parametric design, process scripting replaces
sketching as a mode of design production, although its impact is more profound than
this. A script is a formulation of the parameters that need to be employed to create
an effective design. Authoring a script potentially emphasises the problem-definition
part of the design process. A script can also automatically generate many hundreds
of variations of a compliant design solution, providing unprecedented opportunities
for design exploration. It is not surprising, given these properties, that parametric
design has been repeatedly linked to creativity. It supports a potentially innovative
approach to designing, and especially insofar as it is capable of generating creative
design solutions at the conceptual design stage (Iordanova 2007; Lee et al. 2015).
However, parametric design activities and their impacts on the design process and
product are not well understood. Furthermore, relatively little is known about the
generative and algorithmic strategies used by parametric designers and their impact
on creativity.
The second study in this chapter responds to this gap in our knowledge by
analysing the activities and strategies used by designers in a parametric environ-
ment. It then correlates the outcomes of these strategies to expert assessments of the
levels of creativity apparent in the products of these strategies. In this way, the study
examines the relationship between design strategies and creative outcomes. Before
describing the experiment and its results, the following section briefly considers some
theories connecting design thinking, parametric design and creativity.

2.3.1 Parameter and Rule

In design thinking, a divergent approach is one that widens the scope or breadth of
consideration of an issue, encompassing more factors, possibilities and responses. In
contrast, a convergent approach is one that narrows the scope or breadth of possibil-
ities being considered, focussing on potential solutions. In parametric design, there
are aspects of both of these cognitive dimensions embedded in the scripting processes
used by designers. In a script, for example, the “parameter” component promotes
and acknowledges divergence, whereas the “rule” supports and even necessitates
convergence. These divergent and convergent processes, in design thinking and in
parametric design, are tied to different aspects of creativity and problem-solving.
A “parameter” is a factor defining a range of variation (Monedero 2000; Cardenas
(2008). While the concept of the parameter originates in mathematics, in design it
has been used to describe the scope or extent of design possibilities being investi-
gated. This is especially the case when considering the generation of design varia-
tions, where a divergent approach is often associated with creativity (Kolarevic 2003;
Iordanova 2007). Parametric design not only makes this divergent generative activity
48 2 Design Strategies and Creativity

relatively effortless (Roberto 2006), it allows designers to rapidly explore different

ideas without reliance on their own drawing skills (Lawson 2002). The capacity to
generate variations is not only seen as a precursor to pursuing creative solutions, but
also to extending the boundaries of knowledge (Gero 1996; Liu and Lim 2006). The
generated variations, however, can be abstract and difficult to grasp, and identifying
a creative outcome from hundreds of alternatives remains a challenge in the field
(Hanna and Turner 2006).
A “rule” defines the relations between elements that must be maintained when
generating variations (Lee and Kim 1996; Sutherland 1963). As such, rules can be
conceptualised as tools supporting convergent strategies as they serve to narrow the
set of possible solutions. A rule also implies the presence of knowledge, reflecting an
understanding of the operations or dependencies implicit in a relationship. Knowl-
edge is one of the important factors supporting creativity in design (Li et al. 2007).
For example, encoded architectural knowledge, linked to rules pertaining to struc-
ture, climate and composition, provides a new way of exploring options and using
architectural expertise (Iordanova 2007). The associative modelling, for example,
considers the parametric constraints of user needs, functional requirements and
structural demands (Park et al. 2004). While such rules in the traditional design
environment might hinder creativity, in a parametric environment they may support
it through control of variations.
The parameters and rule functions in parametric design are akin to Roberto’s
(2006) variable and fixed attributes, and have parallels to Jones’ (1992) observation of
the coexistence of intuition and rationality in a design process. These three conceptual
pairings—parameter/rule, variable/fixed attributes and intuition/rationality—point to
ways the strategies used by parametric designers may be understood or interpreted.
Significantly, whereas in a traditional design environment there is a level of inter-
pretation required to differentiate such pairs, in a parametric environment the script
actually requires the identification of the two. This means parameters and rules are
relatively straightforward to detect, encode and analyse. Figure 2.5 illustrates exam-

Parameters

Rules

Parameters Rules

a. Graphical algorithm editor (Grasshopper) b. Text-based algorithm editor (Python)

Fig. 2.5 Examples of parameters and rules in two different parametric design environments
2.3 Study II 49

ples of parameters and rules in two different parametric design environments (a

graphical algorithm editor and a text-based algorithm editor).

2.3.2 Research Method

The second study in this chapter combines protocol analysis with the Consensual
Assessment Technique (CAT) to explore individual design strategies and creativity.
Protocol analysis is used in a similar way to the first study in the chapter, to identify the
behaviours, activities and strategies of participants taking part in a design experiment.
CAT is then used to assess the creativity present in the products of the experiments.
This study is significant because the vast majority of claims made about the creativity-
enhancing properties of parametric design have little empirical evidence to support
them.

2.3.2.1 Protocol Analysis

Four postgraduate architectural design students in Australia—two with over 5 years’

parametric design experience (Au1 and Au3) and two with only 1 year of experience
(Au2 and Au4)—took part in this study. Each of the participants was given one
hour (although the actual timing was flexible) to undertake a specified design task
(see Box 2.2). The task was the conceptual design of a high-rise building that was
required to meet performance requirements for building function, floor area, height,
structural expression and visual expression. Deliverables of the experiment were
design representations of the high-rise building to satisfy the brief, including (i) three-
dimensional model(s) and (ii) three rendered or captured images clearly representing
the design concept and building. Before each experiment, a research coordinator
explained the design brief to participants and all were given time to practice “thinking
aloud”.

Box 2.2 Design Brief for Study II

Design Brief: Conceptual Design of a High-rise Building
You are asked to provide a conceptual design of a high-rise building. This is a
form generation design task for a landmark high-rise building.
The following can be considered:
(1) The high-rise building will have two main functional areas: Office and
hotel.
(2) Maximum floor area is 2,500 m2 (50 m × 50 m) per floor.
(3) The high-rise building will have over 40 storeys, and it will be designated
as a regional landmark.
50 2 Design Strategies and Creativity

(4) You can reflect transformations of structural forces using external data.
The basic structural design may be represented in the conceptual design
(e.g. columns at the exterior or interior of the building).
You should consider Novelty (e.g. originality, complexity and evolution),
Value (e.g. function, usefulness and understandable form) and Aesthetics (e.g.
aesthetic form, elegance and well transformation) for the conceptual design.
A new creative architectural vision depends on you. At this early design phase,
no site or construction constraints have been stipulated, with a client seeking
a highly creative quality outcome as a priority.
Deliverable: Design representation(s) of the high-rise building that should
satisfy the brief and produce (1) three-dimensional model(s). In order for you
to clearly represent the conceptual design, (2) three rendered or captured images
showing the strength of your design should be saved on your desktop.
Timeline: One hour.

(i) Experimental setting: Each design session was video recorded using two
cameras, one providing a view of the student’s overall activities and the other
recording the computer screen. Participants were allowed to choose their own
parametric modelling tools. Two (Au1 and Au2) of the four participants chose
Grasshopper, a graphical algorithm editor, while the other two (Au3 and Au4)
used a text-based editor (Python and Maya script editor, respectively). The
design models produced in each session were collected at its conclusion.
(ii) Verbalisation: A concurrent verbalisation (“think-aloud”) method was used.
After the experiment, each designer also participated in a post-experiment
interview leading to a retrospective protocol, to explain their thoughts and
activities in the experiments.
(iii) Transcription and segmentation: Each video session was directly transcribed
using NVivo software and segmented into smaller episodes.
(iv) Encoding with a coding scheme: One coder encoded each protocol using the
coding scheme (Table 2.4). For this study, three levels from Suwa et al.’s (1998)
scheme (physical, perceptual and conceptual) were adapted and revised to be
more suitable for a parametric design environment. Designing in parametric
environments involves both geometric and algorithmic activities, which are
used as two categories to inform the physical and perceptual levels in the
coding scheme.
The physical level in the coding scheme represents generative activities and
activities in parametric design. Geometry activities are the modelling activ-
ities for generating digital forms, while algorithmic activities represent the
generative processes that describe the parameters and rules used, as well as
the “reference”. The perception level represents the activities related to visual
imagery in the design process. This level also has two categories of geometry
2.3 Study II 51

Table 2.4 Coding scheme for exploring the parametric design process
Level Category Subclasses Description
Physical Geometry G-Geometry Create geometries without an
algorithm
G-Change Change existing geometries
Algorithm A-Parameter Create initial parameters
A-Change Parameter Change existing parameters
A-Rule Create initial rules
A-Change Rule Change existing rules
A-Reference Retrieve or get references
Perceptual Geometry P-Geometry Attend to existing geometries
Algorithm P-Algorithm Attend to existing algorithms
Conceptual Problem-finding F-Initial Goal Introduce new ideas (or goals) based
on a given design brief
F-Geometry Sub Goal Introduce new geometric ideas
F-Algorithm Sub Goal Introduce new algorithmic ideas
Solution-generating G-Generation Make generation (or variation)
Solution-evaluating E-Geometry Evaluate primitives or existing
geometries
E-Parameter Evaluate existing parameters
E-Rule Evaluate existing rules
E-Reference Evaluate existing references

and algorithm. The conceptual level consists of three categories of activities:

problem-finding, solution-generating and solution-evaluating (Gero and Neill
1998). This final level is closely related to problem-solving processes (Dorst
and Dijkhuis 1995; Ho 2001) as well as selection and evaluation (Dacey and
Lennon 1998), which may support design creativity. The solution-generating
category is also essential for capturing the generative aspects of parametric
design.
(v) Arbitration and reliability: To ensure consistency and validity, the coder
segmented and encoded each protocol twice with a 3-month interval between
the processes. A final protocol coding was achieved using a process of arbi-
tration. There was little conflict between the first and second coding results
because most subclasses in the coding scheme developed for this study repre-
sent only algorithmic or geometric activities that are easy to identify and
interpret. There was also a high degree of clarity in activities captured by the
video which could be readily interpreted with the assistance of the designer’s
verbalisations.
52 2 Design Strategies and Creativity

2.3.2.2 Consensual Assessment Technique

While the protocal analysis results can be used to identify cognitive patterns and
strategies in each participant’s design process, a different technique must be used to
measure the creativity arising from these processes. The technique used in this study
is CAT. For this study, a panel consisting of seven expert judges provided assessment
of the four outputs. Each expert had more than 10 years’ experience as a designer
and assessor of design in a tertiary institution (university). All assessments were
undertaken individually, and all design outputs were “blinded”. Participants in the
experiments were not made aware of the identities of the expert judges.
Each design was presented to the experts as a collage of images on A4 size paper,
all design products being similarly scaled for consistency of evaluation (Fig. 2.6). The
judges assessed the designs using three evaluation frameworks: (i) independent non-
criteria-based assessment of creativity; (ii) comparative non-criteria-based assess-
ment of creativity; and (iii) criteria-based assessment of creativity using novelty,
usefulness, complexity and aesthetics. Each assessment used a seven-point Likert
scale (from 1 to 7, with 7 being the highest).

Fig. 2.6 Four design outcomes generated by the participants

2.3 Study II 53

2.3.3 Results

2.3.3.1 Coding Result for Each Protocol

The average value of the number of segments of the four protocols is 297.8 (Au1: 220,
Au2: 319, Au3: 364, Au4: 287). Over 90% of each protocol was encoded using the
coding scheme, regardless of the two types of applications: graphical algorithm editor
or text-based algorithm editor. This confirms that the coding scheme is effective for
capturing the protocol data. Only Au1 completed the design in less than the given time
(48 min), while the others each took longer (approximately 90 min). The overrun
in time was allowed because of the need for additional troubleshooting processes
required by participants Au2, Au3 and Au4. Parametric design tools also require the
user to wait for the design to be generated, a factor which differs from one computer
to another. For example, there were 17 segments of “waiting” in Au3’s protocols,
contributing to the need for extra time.

Table 2.5 The percentage of coding results

Level Category Subclass Au1 Au2 Au3 Au4 Mean SD
Physical Geometry G-Geometry 2.0 6.6 0.0 1.3 2.5 2.9
G-Change 0.5 9.3 0.0 3.0 3.2 4.3
Algorithm A-Parameter 4.9 2.6 0.1 2.9 2.6 2.0
A-Change 10.0 1.7 4.7 1.0 4.4 4.1
Parameter
A-Rule 24.7 20.5 20.0 19.0 21.1 2.5
A-Change 6.0 4.3 12.2 26.9 12.4 10.3
Rule
A-Reference 0.0 0.0 0.2 2.8 0.8 1.4
Perceptual Geometry P-Geometry 2.8 1.3 1.8 0.4 1.5 1.0
Algorithm P-Algorithm 3.2 2.9 4.0 8.6 4.7 2.7
Conceptual Problem-finding F-Initial Goal 3.0 0.9 0.2 1.1 1.3 1.2
F-Geometry 7.0 8.7 2.5 2.3 5.1 3.2
Sub Goal
F-Algorithm 4.1 4.0 4.4 1.0 3.4 1.6
Sub goal
Solution-generating G-Generation 4.0 2.0 13.0 3.0 5.5 5.1
Solution-evaluating E-Geometry 19.4 18.8 17.3 8.8 16.1 5.0
E-Parameter 0.8 0.0 0.0 0.0 0.2 0.4
E-Rule 7.6 16.3 19.5 17.9 15.3 5.3
E-Reference 0.0 0.0 0.1 0.0 0.0 0.1
Sum 100 100 100 100 100 –
54 2 Design Strategies and Creativity

Table 2.5 shows the percentage of the frequency weighted by time (calculated by
the time of the duration of each coded protocol). This allows for the time devoted to
each component of the design strategy to be determined, as well as time devoted to
each level of design thinking for each participant. On average, physical behaviours
account for 46.8% (geometry: 5.7%, algorithm: 41.1%), perceptual account for 6.3%
and conceptual account for 46.9% (problem-finding: 9.8%, solution-generating:
5.5%, solution-evaluating: 31.6%). Solution-evaluating processes dominate in the
conceptual level, while algorithmic activities are dominant in the physical level for
each protocol. A-Rule (writing an algorithmic rule) and A-Change Rule (changing
an algorithmic rule) were the dominant activities in the study. These algorithmic
representations could be a preferred medium for progressing design in a parametric
environment, being akin to producing a sketch, and then modifying it in a traditional
design environment.
The second most dominant activity is E-Geometry (visually evaluating the
outcome of a rule) and E-Rule (evaluating the rule itself). A large amount of E-
Geometry appeared in the protocol of Au1, whereas E-Rule was dominant in the
other protocols for the verification of their design algorithms. Because parametric
design can produce solutions that are unexpected, the outcomes must be regularly
evaluated in the 3D view as well as in the scripting view. There were also some other
notable differences between the participants, including that the first three designers
regularly created new rules (A-Rule) rather than changing existing ones (A-Change
Rule), while the last designer (Au4) tended to change existing rules. The number of
coded segments of the last protocol (Au4) was also the highest, and his problem-
finding activities were also relatively lower. This participant tended to solve problems
by introducing small variations. This is consistent with the trial-and-error procedure
of a novice problem-solver who may lack analytical or scoping skills and strategies.
Au4 was also the least experienced of the designers who took part. Au3, who was
a more experienced designer, also used some trial-and-error or backward reasoning,
although his generative synthesis was significantly higher than the other participants.
Au1 and Au2 contributed relatively higher percentages of their time to the
problem-finding activity. Both used a graphical algorithm editor, which led to more
F-Initial Goal and F-Geometry Sub Goal activities for problem-finding. Au1 in
particular, who was identified as a more experienced designer, applied strategic rules
from the initial states of the problem, which is a feature of the working-forward
search strategy. One finding of the study related to Au3’s performance, which demon-
strated a high level of both generative and evaluation activities (algorithm level). His
approach may have provided an advantage in generating creative design variations,
because these activities are related to both the problem and solution spaces in the
“co-evolution process” (Dorst and Cross 2001; Maher and Poon 1996).

2.3.3.2 Level of Creativity of Individual Design Solution

The results for CAT are reported in Tables 2.6 and 2.7. Table 2.6 shows the inde-
pendent non-criteria-based assessment of creativity by the judges (J1–J7). In these
2.3 Study II 55

Table 2.6 Independent non-criteria-based assessment of creativity (J: Judge)

Designer J1 J2 J3 J4 J5 J6 J7 Mean SD
Au1 5 5 2 5 3 5 2 3.86 1.46
Au2 5 5 1 4 1 6 5 3.86 2.04
Au3 6 7 5 7 5 6 4 5.71 1.11
Au4 4 2 2 4 1 3 5 3.00 1.41

Table 2.7 Criteria-based assessment of creativity (J: Judge)

Designer J1 J2 J3 J4 J5 J6 J7 Mean SD
Novelty
Au1 6 7 4 5 4 4 1 4.43 1.90
Au2 6 5 3 4 1 5 5 4.14 1.68
Au3 7 7 7 7 6 6 6 6.57 0.53
Au4 4 2 2 3 2 1 2 2.29 0.95
Usefulness
Au1 5 7 3 5 3 4 5 4.57 1.40
Au2 4 5 3 4 2 6 7 4.43 1.72
Au3 6 7 2 7 4 2 2 4.29 2.36
Au4 4 3 3 4 4 4 6 4.00 1.00
Complexity
Au1 3 6 3 5 5 6 5 4.71 1.25
Au2 4 5 4 3 4 4 3 3.86 0.69
Au3 5 7 6 7 7 7 6 6.43 0.79
Au4 3 2 2 4 2 1 3 2.43 0.98
Aesthetics
Au1 5 7 1 5 3 3 1 3.57 2.23
Au2 2 2 1 2 1 4 3 2.14 1.07
Au3 3 2 5 6 5 7 4 4.57 1.72
Au4 6 1 4 4 4 1 5 3.57 1.90

results, Au3’s design receives the highest mean score for creativity (5.71), while Au4
receives the lowest (3.00). Six judges identified Au1’s design as the most creative
relative to the criteria and most judges assessed Au2’s and Au4’s designs as the least
creative.
Table 2.7 shows the results of the criteria-based assessment of creativity using
novelty, usefulness, complexity and aesthetics. For the novelty criteria, there is a
wide range of results, and the rank order from highest to lowest is Au3 (6.57),
Au1 (4.43), Au2 (4.14) and Au4 (2.29). For the usefulness criterion, the range of
results is relatively narrow, perhaps reflecting the difficulty of assessing the functional
properties of a concept or schematic design. The rank order from most to least is
56 2 Design Strategies and Creativity

Au1 (4.57), Au2 (4.43), Au3 (4.29) and Au4 (4.00). For the complexity criteria, the
rank order matches that of novelty, being Au3 (6.43), Au1 (4.71), Au2 (3.86) and
Au4 (2.43). Finally, the rank order for aesthetics is Au3 (4.57), Au1 and Au4 (equal,
3.57) and Au2 (2.14).
The CAT data identify the most creative designs as those of Au3 and Au1, who
were also the most experienced of the participants. The question then arises, which
cognitive activities and strategies did they use?

2.3.4 Parametric Design Strategies

As noted previously, the problem-finding and solution-generating codes that might

be regarded as signs of an experienced and creative designer are not strongly present
in two of the protocols: Au2 and Au4. The solution-generating codes only appear
towards the end of Au2’s protocol, while the problem-finding codes rarely occur in
the protocol of the last designer Au4. Instead, Au4’s protocol highlights physical
level activities like A-Change Rule and algorithmic solution-evaluating codes like
E-Rule (Table 2.5). These more novice behaviours do not conform to any of the
predicted indicators of creativity, and the results were not assessed as being in the
top two in terms of this criterion.
Focussing on the design strategies of the experienced designers, who were also
ranked as the most creative, reveals several patterns. Figure 2.7 illustrates the five

Au1
Problem-finding

Physical level
SoluƟon-generaƟng
SoluƟon-evaluaƟng
(geometry)
SoluƟon-evaluaƟng
(algorithm) Start Time End

Au3
Problem-finding

Physical level
SoluƟon-generaƟng

SoluƟon-evaluaƟng
(geometry)
SoluƟon-evaluaƟng
(algorithm) Start Time End

Fig. 2.7 Patterns of different categories of parametric design thinking over time by Au1 and Au3
2.3 Study II 57

categories of cognitive activities in the parametric design experiments of Au1 and

Au3. These are presented in sequence, in order to compare the changes over time.
The primary similarities are that in both protocols the physical level and solution-
evaluating (geometry) activities were dominant (in volume) and consistent (over
time). Codes for problem-finding and solution-generating were, however, less consis-
tent. For example, Au1 uses solution-generating behaviours only rarely, and mostly
near the end of the time. Au3, however, uses them consistently from start to finish.
The problem-finding codes also appear regularly in the protocols of Au1 and
Au3. The data show that Au3 tends to use a design strategy highlighting both the
problem and the solution spaces, a strategy that may have enhanced his creativity
and high scores in the CAT assessment. Au1’s protocol shows the regular use of
problem-finding with a small number of activities encoded as solution-generating (for
example, G-Generation). These features imply that Au1 adopted a problem-driven
strategy rather than a solution-driven one to produce the design.
Figure 2.8 shows the three cognitive activities of problem-finding over time.
It highlights problem decomposition strategies in detail. F-Initial Goal refers to
the explicit introduction of the main problem, while the other two codes intro-
ducing sub-goals deal with sub-problems. A closer examination of Fig. 2.8 reveals
that Au1 produced F-Initial Goal activities at the beginning, middle and end of
the protocol. Au1 also sequentially decomposed the problem into geometric sub-
problems and then algorithmic sub-problems. This is consistent with “the explicit
problem-decomposing strategy” identified in past research (Ho 2001). This strategy
has similarities to Kruger and Cross’ (Kruger and Cross 2006) problem-driven
strategy that produces results which may be more highly rated in terms of solu-
tion quality and creativity. Significantly, the level of usefulness of Au1’s design is
higher than that of Au3’s design. This analysis of the data suggests that both the
problem-decomposing strategy and the problem-driven strategy may facilitate the
generation of creative solutions.

Au1
F-IniƟal Goal
F-Geometry SubGoal
F-Algorithm SubGoal
Start Time End

Au3
F-IniƟal Goal
F-Geometry SubGoal
F-Algorithm SubGoal
Start Time End

Fig. 2.8 Three cognitive problem-finding activities over time by Au1 and Au3
58 2 Design Strategies and Creativity

The results of the study also indicate that Au1 and Au3 explicitly decomposed the
initial problem into both geometric and algorithmic sub-problems (Fig. 2.8). Further-
more, Au3 regularly used the solution-generating activity (Fig. 2.7). He adopted a
design strategy considering both the problem and the solution, creating solutions
using sub-problems (Fig. 2.7, see also sequential sub-problems in Fig. 2.8). As such,
the data for Au3 show a solution-driven strategy coupled with a reflective process,
possibly reminiscent of Schön’s theory (1984). In contrast, Au2 and Au4 tended
to use an implicit problem-decomposing strategy which is more common in novice
designers. For example, Au2’s protocol demonstrates F-Initial Goal at the begin-
ning, which does not sequentially relate to geometric and algorithmic sub-problems.
Instead, it shows a “stop-start” pattern where each problem is solved as it emerges,
rather than following a more creative working-forward search strategy. Au4 produced
the code F-Initial Goal at the beginning and end of the protocol but rarely dealt
with sub-problems. Overall, Au4 worked unsystematically using a backward-search
strategy.
From the coding results (Table 2.5) and the graphical analysis, two creative para-
metric design strategies can be identified: problem-forwarding and solution-reflecting
generative strategies (Fig. 2.9). The problem-forwarding generative strategy, as
a problem-driven strategy, explicitly decomposes the initial problem into both
geometric and algorithmic sub-problems. Au1’s problem-decomposing strategy,
combined with a working-forward search strategy, resulted in a sequential analytic-
synthesis procedure to achieving a final design solution. The strategy shows the
“inventiveness” (Lindström 2006) typical in an expert design process. It is also
followed sequentially by a number of activities encoded as making generation (or
variation). The solution-reflecting generative strategy, as a solution-driven strategy,
highlights the generation of variations. Au3’s use of a solution-reflecting strategy
received the highest score from the judges in terms of creativity. This is consistent
with the results of Kruger and Cross’ (2006) research that suggests a solution-driven
strategy produces higher creativity scores. In order to achieve a comprehensive
solution, the solution-reflecting strategy often evolves into making variations and
reflecting variations activities recursively (Fig. 2.9). In practice, however, designers
probably use both of these parametric design strategies—problem-forwarding and
solution-reflecting—to produce creative solutions, but each strategy may be more
effective in achieving different qualities in the outcomes.

2.4 Design Strategies

This chapter has identified individual design strategies that can support creativity
in two different design environments (traditional and parametric design). Study I
exploring sketch-based design activities reveals three forwarding strategies.
• The drawing-reflection forwarding strategy involves representing an initial goal
directly into a draft solution (drawing) and then continuing refinements through
2.4 Design Strategies 59

Goal Goal Goal

Sub-Problems Sub-Problems Sub-Problems

Synthesis Synthesis Synthesis
Generation Generation Generation

forwarding forwarding

SoluƟon + SoluƟon SoluƟon

Problem-forwarding generaƟve strategy

Goal

Sub-Problems Sub-Problems Sub-Problems

GeneraƟon GeneraƟon GeneraƟon

reflecting reflecting
SoluƟon SoluƟon SoluƟon

SoluƟon-reflecƟng generaƟve strategy

Fig. 2.9 Two generative strategies for creative parametric design

iterative graphic reflection. This generates sequential design solutions over the
initial design solution or it evolves the solution.
• The graphical-goal forwarding strategy starts by exhibiting graphical goal setting
based on design requirements and then decomposing the initial goal into a set of
sub-problems to deliver a series of design solutions.
• The textual-goal forwarding strategy is similar to the graphical-goal forwarding
strategy, but it expresses initial ideas in text. The textual-goal is often followed
by graphical goals or draft solutions (drawings).
60 2 Design Strategies and Creativity

In contrast, study II investigating parametric design activities reveals two

generative strategies:
• The problem-forwarding generative strategy combines a problem-decomposing
strategy with a working-forward search strategy, dividing the initial goals (ideas)
into both geometric and algorithmic goals and then synthesising them using step-
by-step generations to reach an overarching solution.
• The solution-reflecting generative strategy starts by generating a quick solution
from an initial goal and then develops new sub-problems reflecting the previous
solution until a solution meets the initial goal or given design requirements.

2.5 Conclusion

This chapter has examined the relationship between cognitive and problem-solving
strategies and creativity. The first study in this chapter identifies three effective
design strategies in a sketch-based environment: drawing-reflection, graphical-goal
forwarding and textual-goal forwarding. In the second study, two parametric design
strategies, the problem-forwarding and solution-reflecting generative strategies, are
directly mapped to theories about, and assessments of, creativity. The results of these
two studies provide insights into both traditional and parametric design processes
that help to differentiate novice from expert practices and competent from creative
outputs.
The two coding schemes used for these studies successfully capture design strate-
gies that supported problem-solving processes in sketch-based and parametric envi-
ronments, but also support the co-evolution of problem and solution spaces. However,
as is common in protocol analysis, the sample sizes for both studies are limited, even if
large bodies of data are generated and used in the analysis. In addition, the raw scores
of the intercoder reliability test for the first study were only in the range of “tentative
conclusions”. However, such measures were developed largely for content analysis
not cognitive analysis, and different benchmarks would be expected. Nonetheless, in
this chapter at least two coding processes and an arbitration process between them are
used to ensure the reliability of results. Acknowledging these limitations, the results
in this chapter reveal important design strategies for the generation or production of
creative designs. These strategies can assist both professional designers and design
students alike to better understand and reflect on their practices.
References 61

References

Ahmed, Saeema, Ken M. Wallace, and Luciënne M. Blessing. 2003. Understanding the differences
between how novice and experienced designers approach design tasks. Research in Engineering
Design 14 (1): 1–11.
Akin, Ömer. 1978. How do architects design? In Artificial Intelligence and Pattern Recognition
in Computer Aided Design, ed. J.-C. Latombe, 65–98. Amsterdam: North-Holland Publishing
Company.
Akin, O. 1979. An exploration of the design process. Design Methods and Theory 13 (3/4): 115–119.
Amabile, Teresa M., and Julianna Pillemer. 2012. Perspectives on the Social psychology of
creativity. The Journal of Creative Behavior 46 (1): 3–15. https://doi.org/10.1002/jocb.001.
Batey, Mark. 2012. The measurement of creativity: From definitional consensus to the introduction
of a new heuristic framework. Creativity Research Journal 24 (1): 55–65. https://doi.org/10.1080/
10400419.2012.649181.
Blosiu, J. O. 1999. Use of synectics as an idea seeding technique to enhance design creativity.
In1999 IEEE International Conference on Systems, Man, and Cybernetics, 1999. IEEE SMC ‘99
Conference Proceedings. Tokyo.
Burry, M. 2011. Scripting Cultures: Architectural Design and Programming. Wiley.
Cardenas, Carlos Andres. 2008. Modeling Strategies: Parametric Design for Fabrication in
Architectural Practice. Harvard University.
Choi, Han Hee, and Mi Jeong Kim. 2017. The effects of analogical and metaphorical reasoning
on design thinking. Thinking Skills and Creativity 23: 29–41. https://doi.org/10.1016/j.tsc.2016.
11.004.
Dacey, John S., and Kathleen H. Lennon. 1998. Understanding Creativity: The Interplay of
Biological, Psychological, and Social Factors. San Francisco: Jossey-Bass.
Dorst, K., and J. Dijkhuis. 1995. Comparing paradigms for describing design activity. Design Studies
16 (2): 261–274.
Dorst, Kees, and Nigel Cross. 2001. Creativity in the design process: Co-evolution of problem-
solution. Design Studies 22 (5): 425–437.
Gero, John S. 1996. Creativity, emergence and evolution in design. Knowledge-Based Systems 9:
435–448.
Gero, John. 1999. Towards a model of designing which includes its situatedness. In Universal
Design Theory, ed. H. Grabowski, S. Rude, and G. Green, 47–56. Aachen: Shaker Verlag.
Gero, John S. 2000. Computational models of innovative and creative design processes. Technolog-
ical Forecasting and Social Change 64 (2): 183–196. https://doi.org/10.1016/S0040-1625(99)001
05-5.
Gero, John S., and Thomas Mc Neill. 1998. An approach to the analysis of design protocols. Design
Studies 19 (1): 21–61.
Goel, Vinod, and Peter Pirolli. 1992. The structure of design problem spaces. Cognitive Science 16
(3): 395–429. https://doi.org/10.1207/s15516709cog1603_3.
Goldschmidt, Gabriela, and Maria Smolkov. 2006. Variances in the impact of visual stimuli on
design problem solving performance. Design Studies 27 (5): 549–569. https://doi.org/10.1016/j.
destud.2006.01.002.
Gordon, W.J.J. 1961. Synectics: The Development of Creative Capacity. New York: Harper.
Hanna, Sean, and Alasdair Turner. 2006. Teaching parametric design in code and construction.
Paper presented at the SiGraDi2006/Educacion y Desarrollo Academico,
Hay, Laura, Alex H. B. Duffy, Chris McTeague, Laura M. Pidgeon, Tijana Vuletic, and Madeleine
Grealy. 2017. A systematic review of protocol studies on conceptual design cognition: Design as
search and exploration. Design Science 3: e10. https://doi.org/10.1017/dsj.2017.11.
Ho, Chun-Heng. 2001. Some phenomena of problem decomposition strategy for design thinking:
Differences between novices and experts. Design Studies 22 (1): 27–45.
62 2 Design Strategies and Creativity

Holzer, Dominik, Richard Hough, and Mark Burry. 2007. Parametric design and structural opti-
misation for early design exploration. International Journal of Architectural Computing 5 (4):
625–643. https://doi.org/10.1260/147807707783600780.
Iordanova, Ivanka. 2007. Teaching digital design exploration: Form follows. International Journal
of Architectural Computing 5 (4): 685–702.
Jones, John Christopher. 1992. Design Methods. New York: John Wiley & Sons Inc.
Jones, John Christopher. 1963. A method of systematic design. In Conference on Design Methods,
ed. John Christopher Jones, and D.G. Thornley, 53–73. Oxford: Pergamon Press.
Karlusch, Albrecht, Wolfgang Sachsenhofer, and Kathrin Reinsberger. 2018. Educating for the
development of sustainable business models: Designing and delivering a course to foster creativity.
Journal of Cleaner Production 179: 169–179. https://doi.org/10.1016/j.jclepro.2017.12.199.
Kokotovich, Vasilije. 2008. Problem analysis and thinking tools: An empirical study of non-
hierarchical mind mapping. Design Studies 29 (1): 49–69. https://doi.org/10.1016/j.destud.2007.
09.001.
Kolarevic, Branko. 2003. Architecture in the Digital Age: Design and Manufacturing. New York &
London: Spon Press-Taylor & Francis Group.
Krippendorff, Klaus. 2011. Agreement and information in the reliability of coding. Communication
Methods and Measures 5 (2): 93–112.
Krippendorff, Klaus. 2018. Content Analysis: An Introduction to Its Methodology. Thousand Oaks,
CA: Sage publications.
Kruger, Corinne, and Nigel Cross. 2006. Solution driven versus problem driven design: Strategies
and outcomes. Design Studies 27 (5): 527–548.
Lawson, Bryan. 2002. CAD and creativity: Does the computer really help? Leonardo 35 (3): 327–
331. https://doi.org/10.1162/002409402760105361.
Lee, Jae Yeol, and Kwangsoo Kim. 1996. Geometric reasoning for knowledge-based parametric
design using graph representation. Computer-Aided Design 28 (10): 831–841.
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014. Parametric design strategies for the generation
of creative designs. International Journal of Architectural Computing 12 (3): 263–282. https://
doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2015. Creativity and parametric design? Comparing
designer’s cognitive approaches with assessed levels of creativity. International Journal of Design
Creativity and Innovation 3 (2): 78–94. https://doi.org/10.1080/21650349.2014.931826.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2019. Cognitive and linguistic differences in
architectural design. Architectural Science Review 62 (3): 248–260. https://doi.org/10.1080/000
38628.2019.1606777.
Li, Yan, Jian Wang, Xianglong Li, and Wu Zhao. 2007. Design creativity in product innovation.
The International Journal of Advanced Manufacturing Technology 33 (3): 213–222. https://doi.
org/10.1007/s00170-006-0457-y.
Lindström, Lars. 2006. Creativity: What is it? Can you assess it? Can it be taught? International
Journal of Art & Design Education 25 (1): 53–66.
Liu, Yu-Tung., and Chor-Kheng Lim. 2006. New tectonics: A preliminary framework involving
classic and digital thinking. Design Studies 27 (3): 267–307.
Maher, M.L., and Josiah Poon. 1996. Modeling design exploration as co-evolution. Computer-Aided
Civil and Infrastructure Engineering 11 (3): 195–209. https://doi.org/10.1111/j.1467-8667.1996.
tb00323.x.
Mathias, J.R. 1993. A study of the problem solving strategies used by expert and novice designers.
Birmingham, UK: University of Aston.
Monedero, Javier. 2000. Parametric design: A review and some experiences. Automation in
Construction 9 (4): 369–377.
Newell, A., and H.A. Simon. 1972. Human Problem Solving. Englewood Cliffs, NJ: PrenticeHall.
Paivio, Allan. 1971. Imagery and Verbal Processes. Oxford: Holt, Rinehart & Winston.
References 63

Park, Sang Min, Mahjoub Elnimeiri, David C. Sharpe, and Robert J. Krawczyk. 2004. Tall building
form generation by parametric design process. Paper presented at the CTBUH 2004 Seoul
Conference,
Roberto, Barrios Hernandez Carlos. 2006. Thinking parametric design: Introducing parametric
Gaudi. Design Studies 27 (3): 309–324.
Rosenman, M.A., and J.S. Gero. 1993. Creativity in design using a design prototype approach.
In Modeling Creativity and Knowledge-Based Creative Design, ed. J.S. Gero and M.L. Maher,
119–148. Hillsdale, New Jersey: Lawrence Erlbaum.
Salim, Flora, and Jane Burry. 2010. Software openness: Evaluating parameters of parametric
modeling tools to support creativity and multidisciplinary design integration. In Computational
Science and Its Applications—ICCSA 2010, ed. David Taniar, Osvaldo Gervasi, Beniamino
Murgante, Eric Pardede, and Bernady Apduhan, 483–497. Lecture Notes in Computer Science.
Berlin/Heidelberg: Springer.
Schön, Donald A. 1984. The architectural studio as an exemplar of education for reflection-in-action.
Journal of Architectural Education 38 (1): 2–9.
Simon, H.A. 1973. The structure of ill structured problems. Artificial Intelligence 4 (3): 181–201.
https://doi.org/10.1016/0004-3702(73)90011-8.
Simon, H.A. 1978. Information-processing theory of human problem solving. In Handbook of
Learning and Cognitive Processes, ed. W.K. Estes, 271–295. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Simon, H.A. 1981. The Sciences of the Artificial. Cambridge, MA: MIT Press.
So, Chaehan, and Jaewoo Joo. 2017. Does a persona improve creativity? The Design Journal 20
(4): 459–475. https://doi.org/10.1080/14606925.2017.1319672.
Stevens, Mark, Ward Lyles, and Philip Berke. 2014. Measuring and reporting intercoder reliability
in plan quality evaluation research. Journal of Planning Education and Research 34 (1): 77–93.
https://doi.org/10.1177/0739456X13513614.
Sutherland, Ivan E. 1963. Sketch pad: a man-machine graphical communication system. In The
AFIPS Spring Joint Computer Conference. ACM.
Suwa, Masaki, and Barbara Tversky. 1997. What do architects and students perceive in their design
sketches? A protocol analysis. Design Studies 18 (4): 385–403. https://doi.org/10.1016/S0142-
694X(97)00008-2.
Suwa, M., T. Purcell, and J.S. Gero. 1998. Macroscopic analysis of design processes based on a
scheme for coding designers’ cognitive actions. Design Studies 19 (4): 455–483.
Tovey, Michael. 1989. Drawing and CAD in industrial design. Design Studies 10 (1): 24–39. https://
doi.org/10.1016/0142-694X(89)90022-7.
von der Weth, Rudiger. 1999. Design instinct? The development of individual strategies. Design
Studies 20 (5): 453–463.
Wang, Yingxu, and Vincent Chiew. 2010. On the cognitive process of human problem solving.
Cognitive Systems Research 11 (1): 81–92. https://doi.org/10.1016/j.cogsys.2008.08.003.
Chapter 3
Creative Micro-processes in Parametric
Design

Abstract This chapter presents a detailed analysis of the actual, rather than theo-
rised, relationship between cognitive activities and creativity in design. Focussing on
parametric design, the chapter uses protocol data developed from four cognitive activ-
ities—changing parameters, perceiving geometries, introducing algorithmic ideas
and evaluating geometries—to investigate creative problem-solving processes. This
analysis of parametric design activities and their sequential patterns not only informs
a better understanding of process-based creativity, it also identifies two important
creative micro-processes in parametric design. Through this analysis, this chapter
contributes to the evolution of new knowledge about the relationship between design
and creativity.

3.1 Introduction

The rapid rise of interest in parametric design that occurred over the past two decades
arguably reached its popular apogee in “parametricism”, a biomorphic design style or
movement (Schumacher 2009). At that time, influential architects, like Zaha Hadid,
NOX and UNstudio, used dynamic factors for design generation and fabrication,
employing methods that were indebted to parametric design to propose creative
or novel works. Such works naturally encouraged speculation about the capacity
of parametric design environments to support creativity, although the evidence for
this has only gradually been tested (Lee et al. 2014b, 2015). For example, some
researchers (Blosiu 1999; Iordanova 2007; Iordanova et al. 2009) argue that para-
metric design supports creativity through its capacity to enhance design exploration
during the conceptual design stage. Such propositions, however, tend to be supported
by theories, personal observations or reflections, not empirical evidence. In contrast,
this chapter combines protocol analysis from design experiments with the Consen-
sual Assessment Technique (CAT) to develop a deeper understanding of parametric
design’s actual potential to support creativity.
As discussed previously in Chap. 2, the act of designing, regardless of whether
it relies on traditional or emerging digital tools, is understood as both a type of
problem-solving activity and a series of cognitive processes (Bilda and Demirkan

© Springer Nature Switzerland AG 2020 65

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_3
66 3 Creative Micro-processes in Parametric Design

2003; Casakin and Kreitler 2011). In addition to these two—problem-solving and

cognition—designing has always been associated with creativity, but the relationship
between them remains highly contested. Further complicating our understanding of
this tripartite relationship is the environment or toolset used to support designing.
For example, previous studies have examined designers’ sketches to identify creative
activities during the conceptual design stage (Akin and Moustapha 2004; Benami
and Jin 2002; Hasirci and Demirkan 2007; Kim et al. 2010). The flexible and intu-
itive nature of the sketch has been repeatedly linked to creativity, and it would
appear to offer clear advantages over CAD tools in this respect (Bilda and Demirkan
2003; Verstijnen et al. 1998). In contrast, both sketching and the use of conventional
CAD tools have limitations as problem-solving devices (Ibrahim and Pour Rahimian
2010). For example, for constructing, a complex design sketching has serious defi-
ciencies. Conversely, conventional CAD tools lack flexibility and a capacity for
intuitive exploration. To overcome these limitations, Bilda and Demirkan (2003)
suggest employing more adaptive and fluid digital design media. Interactive imagery
may also be used to better visualise design features and organisational relations
when generating alternative solutions. Furthermore, the careful management of part-
whole relationships, design hierarchies and topology-geometry relationships, along
with alternative approaches to handling ill-structured problems and problem-related
parameters, are all strategies that may overcome the limitations of specific design
toolsets (Akin and Moustapha (2004). These properties, which are thought to mitigate
the misalignment between the need for creativity and the capacity of CAD tools, are
noticeably present in recent parametric design environments (Madkour et al. 2009;
Schumacher 2009; Aish and Woodbury 2005).
Unlike previous CAD and geometric modelling tools, parametric design appears
to support both intuitive and flexible design explorations (Lee et al. 2014a). Advanced
parametric design tools also provide an increased capacity to first encode and under-
stand the generative process and second, to evaluate the generated product. Through
its use of scripting, parametric design environments appear to provide designers
with instinctual and adaptable tools, thereby allowing them to potentially engage
in higher levels of complexity in design (Aish and Woodbury 2005; Madkour et al.
2009). Somewhat ironically, it has also been argued that because parametric design
environments rely on constrained relationships between objects to generate solutions,
they are more flexible and responsive (Sakamoto and Ferré 2008). This position main-
tains that such constraints encapsulate complex structures in design logic, freeing
the designer to invest their energies in creative thinking (Coorey and Jupp 2011).
Furthermore, scripting or coding activities in parametric design can be understood
as a channel for creativity and a means of representing design ideas (Salim and Burry
2010).
These theorised parametric properties are well documented in the literature
(Kolarevic 2003; Javier 2000; Sakamoto and Ferré 2008; Schnabel 2007), although
there is a much more rudimentary understanding of what is actually happening in a
design process in respect to creativity. Past research does identify that the generative
modelling processes in parametric design can contribute to creativity (Iordanova et al.
3.1 Introduction 67

2009; Lee et al. 2014a; Lee et al. 2013). Such generative, parametric design activ-
ities are relatively novel and differ from traditional design activities (for example,
sketches, drawings and models). The type of “divergent and convergent thinking
processes” that can occur in parallel with creativity is also observed in parametric
design (Cross 2008; Pugh 1991; Guilford 1967; Lawson 1980; Lee et al. 2015). Obvi-
ously, enabled by the computer, the generative capacity of parametric design supports
the production of seemingly unlimited divergent options, while parametric knowl-
edge and constraints clearly support convergent thinking. From this point of view,
parametric design enhances not only the generation of ideas, but also their evolu-
tion, thereby possibly supporting creativity. Nonetheless, there is a general lack of
empirical evidence linking specific parametric design activities to creativity. Funda-
mentally, we do not know if parametric environments enhance or hinder creativity.
At the very least, some parametric design activities have flexibly generated divergent
design possibilities and modified design alternatives, which should support creativity.
To begin to clarify this situation, this chapter examines four cognitive activities that
have been theorised as promoting creativity in parametric design.
The four parametric design activities discussed in this chapter are (i) changing
existing parameters, (ii) attending to existing geometry, (iii) introducing new algo-
rithmic ideas from a previous idea and (iv) evaluating primitives or existing geome-
tries. Using a coding scheme (Table 2.4), they are identified as A-Change Parameter,
P-Geometry, F-Algorithm Sub Goal and E-Geometry, respectively. The next section
reviews cognitive activities in the protocol data (developed in Study II in Chap. 2)
and then presents an in-depth analysis of the four activities. Through a considera-
tion of iterative cognitive patterns, this chapter concludes with a presentation of two
important creative micro-processes in parametric design.

3.2 Four Parametric Design Activities

The basis for the analysis conducted in this chapter is protocol data developed from a
design experiment (see Chap. 2 for details). This chapter uses the data to examine the
relationship between four parametric design activities and creativity. As discussed
previously, the “making generation or variation” activity (G-Generation) is crit-
ical for developing the type of generative design strategies in parametric design that
produce creative outcomes. In addition to analysing the generative activity, the cogni-
tive activities of the design process (Chap. 2. Table 2.5) are also correlated to expert’s
rating of the creativity implicit in the design product using CAT (Chap. 2. Table 2.7).
To effectively deal with different types of data and values, the normalised coverage
values of cognitive activities (subclasses)—A = (A – mean)/standard deviation—are
calculated (Bilda and Gero 2007; Lee et al. 2014a, 2015). The average values of
the criteria-based assessment scores are also normalised in order to compare each
designer’s cognitive activities (see Table 3.1).
Possible types of patterns emerging from these comparisons are listed in the left-
hand-side column of Table 3.1. For example, “Au1, 3, 4 < Au2” means that three
68 3 Creative Micro-processes in Parametric Design

Table 3.1 Main types of patterns identified in the comparison between the four designers
(normalised values of the coding coverages of each subclass and the average values of rating
scores)
Type Subclass Au1 Au2 Au3 Au4
Process Au1, 3, 4 < Au2 G-Geometry −0.17 1.44 −0.86 −0.41
G-Change −0.63 1.43 −0.75 −0.05
F-Geometry 0.58 1.11 −0.81 −0.88
Sub Goal
Au1, 2, 3 < Au4 A-Change Rule −0.62 −0.78 −0.01 1.42
A-Reference −0.55 −0.55 −0.40 1.50
P-Algorithm −0.56 −0.67 −0.25 1.48
Au1 > Au4 > Au2 > Au3 A-Parameter 1.16 −0.01 −1.28 0.14
F-Initial Goal 1.42 −0.33 −0.92 −0.17
Au1 > Au2 > Au4 > Au3 E-Rule −1.45 0.18 0.79 0.48
Au1 > Au2 > Au3 > Au4 E-Geometry 0.67 0.55 0.25 −1.48
Au1, 3 > A-Change 1.38 −0.65 0.09 −0.82
Au2, 4 Parameter
P-Geometry 1.22 −0.27 0.22 −1.17
F-Algorithm 0.46 0.39 0.64 −1.49
Sub Goal
G-Generation −0.30 −0.69 1.48 −0.49
Product Au1, 3 > Average of 0.19 −0.47 1.30 −1.02
Au2, 4 criteria-based
assessment
scores

designers (Au1, 3 and 4) have lower normalised coverage values than Au2. “Au 1
> Au2 > Au3 > Au4” means that designer Au1’s coding coverage value is higher
than Au2’s, and Au2’s is higher than Au3’s, and Au3’s is higher than Au4’s. “Au1,
3 > Au2, 4” means that Au1’s and Au3’s values are similarly higher than Au2’s and
Au4’s. This mapping convention allows for the comparison of characteristics of the
design process with rating results of the design product. Because of this, the last type
(“Au1, 3 > Au2, 4”) in process is consistent with the rating results of creativity in
product. That is, four activities, A-Change Parameter, P-Geometry, F-Algorithm Sub
Goal and G-Generation can be identified as creative design activities in parametric
design.
In conjunction with the four creative activities, E-Geometry (evaluating geome-
tries) in a three-dimensional (3D) view is also linked to parametric design activities
that support creativity (Lee et al. 2014a, 2015). In Table 3.1, E-Geometry conforms
to a “Au1 > Au2 > Au3 > Au4” type. Designers Au1 and Au2, who used a graph-
ical algorithm editor, tended to produce more geometry-based activities, which may
explain why Au2 produced relatively higher amounts of E-Geometry. In parametric
3.2 Four Parametric Design Activities 69

A-Change Parameter P-Geometry RaƟng Score

F-Algorithm Sub goal E-Geometry G-GeneraƟon
1.5

1
Normalised value

0.5

-0.5

-1

-1.5
Au1 Au2 Au3 Au4

Fig. 3.1 Normalised values of the coding coverages of cognitive activities (process) over the rating
scores on creativity (product)

design environments, the design products should be repeatedly examined in the 3D-
view mode, a visualisation process that fundamentally supports conceptual design.
Thus, this chapter regards E-Geometry as a further important creative activity in para-
metric design. Visualised patterns in Fig. 3.1 not only inform the mapping process
(Table 3.1), but also identify broadly similar allocations of five cognitive activi-
ties between the designers in the experiment. In summary, excluding G-Generation
(already explored in Chap. 2) this chapter investigates the four cognitive activities,
A-Change Parameter, P-Geometry, F-Algorithm Sub Goal and E-Geometry identi-
fied as being significant, so as to better understand each activity in relation to creative
problem-solving.

3.2.1 Changing Parameters

In parametric design processes, design variations are generated using constraints such
as topological relationships, design rules and parameter controls. In contrast, parame-
ters facilitate a range of activities, enabling divergence in design. For example, param-
eters support restructuring and regulating processes, which are commonly found in
sketching activities. They are typically related to functions defining a range of vari-
ations (Cardenas 2008). Parametric design enables the production of large numbers
of design that all comply with the given constraints. Thus, creativity in parametric
design is likely to be related to generative activities such as G-Generation and A-
Change Parameter, an argument that is clearly supported by the correlation with the
creativity scores shown in Fig. 3.1. This finding is further evident in the summary of
generative activities (see Fig. 3.2). As discussed in Chap. 2, the expert designers, Au1
and Au3, generally had significantly higher frequency coverage of G-Generation and
70 3 Creative Micro-processes in Parametric Design

20 G-GeneraƟon A-Change Parameter

18

Coding coverage (%)

16
14
12
10
8
6
4
2
0
Au1 Au2 Au3 Au4

Fig. 3.2 The coding coverage of two generative activities

A-Change Parameter than the novice designers. These results suggest that experts
have a clear advantage when engaged in the generative aspects of parametric design.
All four designers generated their design schemata and solutions and then modi-
fied them by creating and changing parameters. To generate different design solu-
tions, they produced and edited different classes of objects by changing parameters.
Au1 and Au2, using the same graphical algorithm editor (Grasshopper), explicitly
defined and then changed parameters, whereas for the other two designers the coding
process wasn’t able to distinguish between their use of parameters or rules. This is
because Au3 and Au4 used a script editor and wrote using the system’s program-
ming language to progress the design. Nonetheless, Table 3.2 shows that the designers
tended to change parameters to generate different variations of the original solution,
while Au3 changed sparameters to fix problematic outcomes (“troubleshooting”) or
unintended consequences.
All of the protocols indicate that parametric modelling is a process in which
the generation of new design variations from existing ones is relatively effortless
(Roberto 2006). Furthermore, while the designers had limited programming exper-
tise, by utilising design constraints and parameters, they were able to explore a large
variety of ideas without being restricted by their own drawing skills (Lawson 2002).
Making variations is therefore the key to pursuing creativity as well as extending the
boundaries of the designers’ existing knowledge and the state space of possible solu-
tions (Gero 1996; Liu and Lim 2006). From this perspective, “changing parameters”
is closely related to embodying “divergent thinking”, which has been identified as
one of the most important factors in creative models (Cross 2008; Guilford 1967;
Pugh 1991).
The twin activities of creating and changing parameters also combine convergent
and divergent thinking. For example, activities encoded by A-Change Parameter
in Au1 and Au3 in Table 3.2 not only generate design alternatives, but also signal
the exploration and identification of more appropriate solutions. In addition, Table
3.2 shows how variations were generated using the “changing parameters” activity
and then evaluated and selected by each designer in the 3D view (E-Geometry) to
3.2 Four Parametric Design Activities 71

Table 3.2 Segments and subclasses related to “changing parameters”

Segment No Transcription Subclass
Au1 14 I am just going to create one by one. What I am F-Geometry Sub Goal
going to do is I am going to work on structure
solution first. And, what I want to do is to put a grid
around the outside
15 (V: dragging and copying some parameters and A-Parameter
rules) so I am going to create a series of grids
16 On the outside, (V: changing parameters on the A-Change Parameter
copied sliders)
17 I am just creating a border of the grid, I copy that A-Rule
again. … (V: inserting and connecting rules)
18 Fifty, too much on them, 40 (V: changing A-Change Parameter
parameters on the inserted rules)
19 Okay (V: Examining forms in the 3D view mode) E-Geometry
Au3 255 (V: Examining forms in the 3D view mode) why is E-Geometry
this huge, isn’t it?
256 What agents do I want to use, 50? (V: Examining E-Rule
rules in the script mode). I am going to rescale…
257 Tower by factors of 10…-3, … 140, 10, 40… A-Change Parameter
258 problems solved… (V: Examining rules in the script E-Rule
mode)
259 (V: Executing the agents…input numbers one by A-Change Parameter
one)
260 Frame skins to one, two…a kind of guarantee… E-Geometry
(V: Examining forms in the 3D view mode)
Au4 25 (V: modifying rules) A-Change Rule
26 Just trying to fix errors, I gained (V: Examining E-Geometry
forms in the 3D view mode) objects are in a wrong
plane
27 (V: changing parameters) A-Change Parameter
28 (V: Examining forms in the 3D view mode) Okay E-Geometry
(V:…): Visual cue in the video recording

arrive at a meaningful and feasible outcome. Considering Hanna and Turner’s (2006)
criticism that parametric design variations may be too abstract, only making sense in
a virtual design environment, “changing parameters” is clearly an important design
abstraction and realisation activity.
72 3 Creative Micro-processes in Parametric Design

3.2.2 Perceiving Geometries

The perception level of cognitive activities is related to visual imagery in the design
process as well as the “incubation” stage in Wallas’s (1926) creative stage model.
Thus, it is theorised that perception facilitates creativity, and it also involves seeing
relationships between elements or components (Flowers and Garbin 1989). Although
there are two different perception activities, P-Geometry and P-Algorithm, the coding
coverage of P-Geometry (perceiving or attending to existing geometries) is the one
consistent with the results of the product-based evaluation of creativity in Table 3.1.
That is, perceiving geometries in parametric design may play a more significant role
in creativity than reviewing or contemplating existing algorithms.
In traditional “pen-and-paper” design environments (see Study I in Chap. 2), the
encoding of perception refers to the act of concentrating on the visuospatial features
of depicted elements (Suwa et al. 1998). Sensing visual cues is an essential part of
the creative problem-solving processes associated with the formulation of mental
imagery. In contrast, cognitive activities around P-Geometry and P-Algorithm in
parametric design frequently signal a significant problem-solving event or situation
in the design process. Importantly, the moment when designers switch to the 3D view
to attend to existing geometries (thereby evoking the P-Geometry activity) often coin-
cided with, or heralded, a mental insight. For example, designer Au1’s protocol high-
lights a correspondence between P-Geometry and a number of sudden realisations
(so-called “A-ha” moments) related to particular problem-solving activities.
Table 3.3 shows examples of Au1’s and Au2’s “A-ha” responses. Further design
activities occurring in parallel with the P-Algorithm activity were also related to the
occurrence of unexpected discoveries. In two instances, for example, Au4 was trou-
bleshooting an algorithmic problem in the script editor when the design solution was
re-framed and discovered. These sudden insights or discoveries were made during
consideration of visual compositions and using knowledge of the design domain
(Akin and Akin 1996). A further insight into how problems are solved can be seen
in connection to the P-Geometry subclass, when designers were waiting for design
variations to be generated by the computer. For example, perceiving geometries for
designer Au3, as shown in Table 3.3, occurred as the designer was waiting for the
generation of design outcomes. This visualisation process from a generative activity
may enable the designer to critically reflect on their own design ideas and to construct
mental imagery of outcomes.
Akin and Akin (1996) examine the discovery of a creative solution that corre-
sponds to the sudden insight in the sketching process for a design problem, structured
with several restricting frames of reference. Akin and Akin (1996) describe the “A-
ha” response as a reference to the moment when a creative flash occurs. In parametric
design, a number of sudden mental insights and unexpected discoveries were identi-
fied in the protocols as occurring in parallel with P-Geometry and P-Algorithm activ-
ities. These “A-ha” responses also relate to switching behaviours between different
representations of the design model. This finding highlights the role that multiple
3.2 Four Parametric Design Activities 73

Table 3.3 Segments and subclasses related to “perceiving geometries”

Segment no. Transcription Subclass
Au1 95 What I will do is to settle with all of that A-Change Parameter
96 I will just (V: just attending to the monitor screen)… E-Geometry
97 … A-ha … This is what happened P-Geometry
98 I am still moving original points, I will move new E-Rule
points. (V: Scrolling the rule view)
99 Excuse me, hold things at that E-Geometry
(V: Examining forms in the 3D view mode)
Au2 205 Okay, That’s rotating a few objects (V: Examining E-Geometry
forms in the 3D view mode)
206 Change that from one to lists (V: changing A-Change Parameter
parameters)
207 (V: Examining rules in the 3D view mode) E-Geometry
208 (V: Examining forms in the script mode) E-Rule
209 What I’ve done, I wasn’t intended to, I am E-Geometry
wondering, it all works. (V: examining forms in the
3D view mode)
210 (V: Attending to forms) … it’s a happy accident P-Geometry
(A-ha)
Au3 314 (V: Executing the agent to generate a cloud of R-Generation
points…input numbers one by one)
315 (V: Examining forms in the 3D view mode) E-Geometry
316 (V: just attending to the monitor screen) P-Geometry
317 (V: just waiting for completing the generation) (No code)
318 (V: Examining forms in the 3D view mode) Oh, E-Geometry
what the hell…
(V:…): Visual cue in the video recording

representations and interfaces play in design processes and their potential signifi-
cance to creative activities and outcomes. From this perspective, parametric design
environments may open up more and greater possibilities for sudden mental insights
and unexpected discoveries than traditional “pen-and-paper” or CAD environments.
This may be due to the ability of designers to use the different interfaces (geometric
and algorithmic modes) as “triggers” for breaking out of their frames of reference.
74 3 Creative Micro-processes in Parametric Design

3.2.3 Introducing Algorithmic Ideas

As discussed in Chap. 2, three cognitive “problem-finding” activities, F-Initial Goal,

F-Geometry Sub Goal and F-Algorithm Sub Goal, play an important role in devel-
oping generative design strategies that can enhance creativity in parametric design.
Figure 3.3 illustrates the coding coverages of the three “problem-finding” activities.
Au4 produced the most frequent coverage in the physical level but the lowest
coverage of “problem-finding” activities. His design solution also received the lowest
scores from the judges. Conversely, Au2, the other novice designer, had the highest
frequency coverage of “problem-finding” (18.5%) suggesting that these activities
alone may not necessarily have a positive impact on creativity. Conversely, generative
activities including G-Generation and A-Change Parameter may be more important
for supporting creativity in parametric design. In addition, the two designers (Au1
and Au2) who used graphical algorithm editors produced more F-Initial Goal and F-
Geometry Sub Goal activities for problem-finding compared to the text-based editors.
They also tended to produce more geometric activities in the physical representation
level than the text-based editors. This suggests that the graphical editor is more likely
to lead designers to conduct geometry-based activities, regardless of the application
of a creative, generative design strategy. Thus, among the three activities, introducing
algorithmic ideas or sub-goals, encoded as F-Algorithm Sub Goal, is explored as an
important creative activity in parametric design.
“Introducing algorithmic ideas” is related to the mathematical representation of
design goals and the collection of schemata. Table 3.4 presents protocols related to the
F-Algorithm Sub Goal, showing a series of representation and evaluation activities
(e.g. A-Change Rule, E-Rule, E-Geometry) leading to the introduction of algorithmic
ideas. Furthermore, “introducing algorithmic ideas” often occurs in conjunction with
the F-Geometry Sub Goal. That is, new ideas at the geometric level are extensions

F-Algorithm Sub goal

20 F-Geometry Sub Goal
18 F-IniƟal Goal
Coding coverage (%)

16
14
12
10
8
6
4
2
0
Au1 Au2 Au3 Au4

Fig. 3.3 The coding coverages of “problem-finding” activities

3.2 Four Parametric Design Activities 75

Table 3.4 Segments and subclasses related to “introducing algorithmic sub-goals”

Segment No Transcription Subclass
Au2 69 (V: Examining forms in the 3D view mode) SE-Geometry
70 we’ll say give a core for this F-Geometry Sub Goal
71 an arbitrary value of one quarter (V: drawing a box G-Geometry
in the 3D view mode)
72 and so (V: dragging and changing a box in the 3D G-Change
view mode)
73 I don’t know if its accurate or not, this is a basic E-Geometry
track available to put the hotel rooms and the office
space
Maybe that core can be reconfigured later
74 We need a random point F-Algorithm Sub Goal
75 (V: inserting a component in the script mode) A-Rule
76 checking the component E-Rule
77 (V: linking components in the script mode) A-Rule
Au3 264 (V: Examining forms in the 3D view mode) E-Geometry
265 (V: Deleting all previous model traces) (no code)
266 Okay. What time am I going to look? (no code)
267 I am just lofting things F-Geometry Sub Goal
268 I am … the actual rotation of the agent wants to set F-Algorithm Sub Goa
aside to go…
269 Bake Agents a…. (V: Examining rules) cool… E-Rule
270 Points…don’t need it (V: inserting rules) I am A-Rule
…range…Bake Agent…in range…
(V:…): Visual cue in the video recording

of previous ideas at the algorithmic level and vice versa. The introduction of algo-
rithmic sub-goals can be regarded as an editing process of previous geometric or algo-
rithmic design ideas. Thus, “introducing algorithmic ideas” can be seen to support
the evolution of design ideas.

3.2.4 Evaluating Geometries

Designers in digital environments typically rely on the 3D view to support perception

and exploration of design ideas (Aish and Woodbury 2005). “Evaluating geometries”
in designers’ transcripts (Tables 3.2, 3.3 and 3.4) indicates that there are potential
cognitive sequences in parametric design and that this activity is closely related to
several others in parametric design. For example, designers tend to change parameters
and rules (A-Change Parameter and A-Change Rule) both before and after evaluating
geometries (E-Geometry).
76 3 Creative Micro-processes in Parametric Design

E-Geometry, that is, visually evaluating the outcome of a rule, is the second most
dominant activity in the coding results (Table 2.5), and E-Rule (evaluating the rule
itself) is the third. A large number of E-Geometry activities, with a relatively small
number of E-Rule, appear in Au1’s protocol, whereas E-Rule occurs more frequently
in the other protocols for the verification of design algorithms. The logic behind
this sequence is that, because parametric design often produces solutions that are
unexpected, the outcomes must be regularly evaluated in the 3D view as well as in
the scripting view mode (Lee et al. 2015). These evaluation activities can be seen to
further refine and potentially restructure the design solution, which could in some
cases lead to a re-conception of the design problem.
Multiple interactions between the script editor interface and the 3D view also
highlight the iterative patterns of evaluation activities consisting of E-Rule code
followed by E-Geometry (see Tables 3.2 and 3.3). These two evaluation activities are
frequently linked to each other which reinforces the importance of utilising multiple
representations and interfaces. In essence, in a parametric environment, designers
are able to model at both geometric and algorithmic levels and also make evaluations
across them. These multi-level interactions enable multi-level evaluations, which
may be a catalyst for a designer to break out of an existing mindset and explore new
ideas. Thus, like “perceiving geometries”, E-Geometry can create opportunities for
“A-ha” moments associated with unexpected discoveries.
Au3 and Au4, who employed scripting as the main method to generate their
designs, produced a relatively high number of instances of E-Rule in the verification
of their algorithms. Au3’s and Au4’s design processes also produced more unex-
pected outcomes than the others using the graphical algorithm editor. Bilda and
Demirkan (2003) claim that it is time-consuming to switch between the different
representations in digital design tools. In parametric design, however, the switching
between geometric and algorithmic modes is necessary to assess and explore both
the visuospatial features and algorithmic ones of generated design outcomes. These
iterative cognitive activities may be considered akin to Salim and Burry’s (2010)
complementary creative method, which supports creativity in parametric design.

3.3 Creative Micro-processes in Parametric Design

3.3.1 Patterns of Parametric Design Activities

Thus far in this chapter, it has been observed that the four parametric design activities
develop iterative, sequential patterns consisting of two or more activities. Through
correlation analysis, it would appear that these sequences of mental activities are also
closely related to creativity and therefore, they could be treated as a creative thinking
pattern. Chusllp and Jin (2004) propose a cognitive activity model consisting of three
iterations to conceptualise creative thinkin6g patterns. The three iterations in their
loop model are problem redefinition, idea stimulation and concept reuse. Generative
3.3 Creative Micro-processes in Parametric Design 77

and parametric design solutions conform to this pattern, as they evolve with exten-
sive iterations by modifying parameters and rules. The “solution-reflecting generative
strategy” (identified in Chap. 2) also features a series of iterative activities, wherein
form generation is followed by the development or refinement of sub-problems
(including evaluation activities). The iterative series of activities in parametric design
are also reminiscent of Stempfle and Badke-Schaub’s (2002) “process 1”, which is
a sequence of solution ideas followed by immediate evaluation. The sequence of
cognitive activities surrounding these four activities facilitates identifying creative
design processes in the context of parametric design and the co-evolution process of
problem and solution spaces (Maher and Poon 1996; Dorst and Cross 2001). Thus,
exploring the relationship between the four cognitive activities, their occurrence and
sequence could reveal how they actually support creativity in a design process.
To investigate the significance of sequences of activities, two segments before
and after each segment encoded as F-Algorithm Sub Goal in a design protocol were
examined to identify recurring sequential relationships. Because “evaluating geome-
tries” occurred relatively frequently (on average, 16.1%) and “perceiving geometry”
only rarely occurred (1.5%) in Table 2.5, the five-activity sequences include one of
two algorithmic activities (A-Change Parameter and F-Algorithm Sub Goal) and
thereby provide a considerable set of data for exploration. In addition, “introducing
algorithmic ideas” as a “problem-finding” activity plays an identical role in refining
and restructuring both design problem and solution in parametric design. This is why
this section selects F-Algorithm Sub Goal as the centre code of a series of cogni-
tive activities. Table 3.5 presents the different types of iterations of design activities
identified by the sequential patterns of the three activities, A-Change Parameter (A),
P-Geometry (P) and E-Geometry (E), surrounding F-Algorithm Sub Goal (F).
An E–F repetition is a dominant pattern across all four designers’ protocols (14
times as a pair, and a further 11 times as part of sets of three activities). That is, the
evaluation of existing geometries (E-Geometry) was frequently followed by the intro-
duction of algorithmic ideas (F-Algorithm Sub Goal). This implies that the design
problem and solution are being refined or restructured in a continuous manner in
parametric design, which conforms to Chusllp and Jin’s (2004) “problem redefinition
loop”.

Table 3.5 Sequential patterns of the four activities

Protocol Au1 Au2 Au3 Au4
Pattern E–F-A (3) E–F (6) E–F (4) E–F (2)
(Frequency) F-A-E (3) E–F-E (2) E–F-F (2) E–F-E (1)
E–F (2) F-F (2) F-F (2) F-E (1)
E–F-E (1) F-A-E (1) F-A-E (2)
E–F-P (1) P-F (1) A-E–F (1)
P-F-E (1) F-E (1)
(A: A-Change Parameter, P: P-Geometry, F: F-Algorithm Sub Goal, E: E-Geometry)
78 3 Creative Micro-processes in Parametric Design

An additional significant pattern is E–F-A in designer Au1’s protocol and its

inverse, the F-A-E pattern in the first three designers’ protocols (Au1, Au2 and Au3)
(Table 3.5). These patterns indicate that the F-Algorithm Sub Goal often results in A-
Change Parameter. That is, the introduction of a new algorithmic idea is the catalyst
for changing an existing parameter. Obviously, making design variations is a rela-
tively effortless process in parametric design due to the ease of changing parameters.
Furthermore, the occurrence of both the E–F-A and F-A-E patterns can be seen as part
of a cyclic pattern (i.e. E–F-A-E–F-A-…) of evaluating geometries, refining algo-
rithmic (or sometimes geometric) goals, updating parameters (or sometimes rules)
and then returning to evaluating geometries again.
An A-E pattern also occurs in the protocol data. For example, 15 out of 34 A-
Change Parameter codes in Au1’s protocol preceded E-Geometry. In Au2’s and
Au3’s protocols, many A-Change Parameter codes were followed by the evaluation
of generated geometries, 75.0% (6/8) and 59.1% (13/22), respectively. Furthermore,
the incidence of both E-A or A-E patterns is notable, occurring 18 times in Au3’s
protocol. Au3 adopted an agent-based approach to resolving the design problem
which typically generated programmatic relationships and an architectonic response
to the design problem. Since Au3’s design variations were generated from emergent
relationships between program and form, the pattern of E-A or A-E appears to be a
crucial sequence of activities in his creative problem-solving process.
Figure 3.4 presents the patterns of cognitive activities in the work of the two
designers (Au1 and Au3) who generated the most creative design solutions. The
activity of perceiving geometries (G-Geometry) was disregarded in this analysis
due to its rare occurrence. G-Geometry may also be conflated with E-Geometry
activities that supersede its role at the geometric level of design cognition. While E

Au1
E

A
Start Time End

Au3
E

A
Start Time End

Fig. 3.4 Patterns of the three cognitive activities over time (A: A-Change Parameter, F: F-Algorithm
Sub Goal, E: E-Geometry)
3.3 Creative Micro-processes in Parametric Design 79

is a dominant activity, occurrences of both F and A differ significantly between the

two designers. F in Au3’s protocol occurs at more regular intervals throughout his
design session, and many A activities appear late in the design session. Au3’s design
activities focused on the evolution of ideas at the algorithmic level which steadily
increased over time. In contrast, Au1’s protocol rarely develops F and A codes later
in his design session. These features are also related to individual generative design
strategies (see Study II of Chap. 2): Au1’s “problem-forwarding generative strategy”
and Au3’s “solution-reflecting generative strategy”. Interestingly, in Au3’s protocol F
activities (or E–F repetition) happen in the early stage of his session, with A activities
in the later stage. Thus, he not only adopted the solution-reflecting generative strategy,
but also employed a “problem redefinition loop” at the beginning of his design session
and an “idea simulation loop” at the end. From this analysis, it appears that these
micro-iterative patterns of cognitive activities may support creativity in a parametric
design process.

3.3.2 Creative Micro-processes

Design process models are typically based on theorised sets of cognitive patterns that
occur during the act of designing. For example, Jones’ (1963) Analysis–Synthesis–
Evaluation (ASE) framework, Oxman’s (2008) “representation–generation–evalua-
tion–performance” model and Stempfle and Badke-Schaub’s (2002) four cognitive
operations (generation, exploration, comparison, selection) all offer a macro-level
description of designing. Specifically, Oxman (2006) identifies five paradigmatic
classes of digital design models: CAD, formation, generation, performance and inte-
grated models (Table 3.6). In order to explain the design processes embodied in
each digital design model, she proposes a schematic framework consisting of four
components: representation, generation, evaluation and performance. Because these

Table 3.6 Oxman’s five types of digital design models (Oxman 2006)
Type Digital design model
CAD model • CAD descriptive model
• Generation-evaluation CAD model
• CAD descriptive model and its evolution to dual-directional
digital processes
Formation model • Topological formation model
• Associative design formation model
• Motion-based formation model
Generative model • Grammatical transformative design model
• Evolutionary design model
Performance model • Performance-based formation model
• Performance-based generation model
Integrated compound model • Integrated compound model
80 3 Creative Micro-processes in Parametric Design

four are basically developed for traditional design activities, she argues that digital
design thinking (DDT) equivalents are (i) representational digital media, (ii) gener-
ation and interaction with digital form, (iii) analytical and judgmental processes,
and (iv) programmatic and contextual considerations, respectively. In this way, DDT
supports the adoption of new design approaches such as morphogenesis and para-
metric formation. From these alternatives, Oxman (2006) singles out the “integrated
compound model” as the most ideal for supporting DDT (see Table 3.6).
Compared to these macro-level models, Botella et al. (2018) suggest that the
micro-level is more important for identifying the mechanisms underlying the creative
process. This chapter identifies two cyclic mechanisms (patterns) of parametric
design activities, which enable the development of creative micro-processes. The
first micro-process is a cyclic pattern of E–F-A-E activities, which is similar to the
conventional ASE model. However, the introduction of new algorithmic ideas is a
critical feature of this process, and it is important to note that the cyclic pattern
crosses both the algorithmic and geometric levels. The second micro-process is an
E-A-E cycle which appears as a solution-reflecting activity. It is reminiscent of the
restructuring of components (Verstijnen et al. 1998) and the regulation of elements
into a comprehensive solution (Akin and Moustapha 2004). However, the iterative
E-A-E pattern in parametric design often occurred when algorithmic rules were being
documented and “chunked” into scripts, such that the cyclic process is understood as
a recurring generative activity. Based on these cyclic patterns of creative parametric
design activities, this chapter identifies two creative micro-processes (see Fig. 3.5).
The first creative micro-process is effectively a combined process of E–F-A-
E and E-A-E cycles. In this creative micro-process, the major loop consists of E-
Geometry, F-Algorithm Sub Goal and A-Change Parameter, and it also includes a
minor loop comprising E-Geometry and A-Change Parameter. The second creative
micro-process is as expansion or variation of the first. In it, the E-Geometry activity
links (or swaps) to P-Geometry, while the A-Change Parameter is similar to G-
Generation in terms of its generative role in parametric design. Many A-Rule and

F F

E A P E A G

CreaƟve micro-process 1 CreaƟve micro-process 2

Fig. 3.5 A creative micro-process model for parametric design (A: A-Change Parameter, F: F-
Algorithm Sub Goal, E: E-Geometry, P: P-Geometry, G: G-Generation)
3.3 Creative Micro-processes in Parametric Design 81

F-Geometry Sub Goal activities are present in the protocol before and after the E-
Geometry activity. The evaluation activity can also be extended to the inclusion of
E-Rule. Thus, this model can be used to explore general “parametric micro-process”
as well as creative micro-processes.
These two creative micro-processes highlight the significance of revisiting gener-
ative components in parametric design. In traditional design environments, visual
representations, such as a sketch or drawing, can enhance idea-generation, struc-
turing thinking and augmenting problem-solving abilities (Bresciani 2019). Visual
elements (both textual and graphic) immediately organise information in prepara-
tion for communication or to provide clarity. In this context, designers continu-
ously draw on their imagination and cognitive skills to develop new visualisations
during the conceptual design stage that are often based on previous representations.
Collectively, in the traditional design process (see sketch-based design strategies
in Chap. 2), problem analysis and decomposition are important problem-solving
activities and a starting point of creative design processes and strategies. However,
designers in parametric environments frequently revisit a set of scripts or algorithmic
codes, revising them until they fulfil expectations (the generative solution depicted in
the 3D-view mode). Furthermore, designers often examine familiar scripts and even
reuse previous algorithms to solve new design problems. However, even experts
are not always prepared for the results (visualisations) produced when algorithms
are modified. There is, therefore, a cognitive gap between a designer’s informa-
tion processing through scripting and their capacity to visualise the results. Thus,
the creative micro-processes in design start with “changing parameters” or “making
generation”, activities are behind many “A-ha” moments. In addition, the major loop
consisting of E, F and A is also similar to the “problem redefinition loop” which
captures another important creative problem-solving process, “co-evolution”, where
“problem leads to solution” and “solution refocusses the problem” (Maher and Poon
1996).

3.4 Conclusion

This chapter has presented an in-depth analysis of four parametric design activi-
ties that have been identified as potentially supporting creativity. In particular, two
algorithmic activities (A-Change Parameter and F-Algorithm Sub Goal) appear to
provide a significant channel for creativity. Both can also be correlated to divergent
and convergent thinking in a parametric design process. Two geometric activities
(P-Geometry and E-Geometry) are associated with the exploration of design ideas in
3D modelling views, which support digital and visual synthetic geometric modelling.
This is significant because it emphasises that even though parametric design high-
lights algorithmic approaches to creativity, geometric activities provide close support
for this.
82 3 Creative Micro-processes in Parametric Design

The chapter further identifies two sequential patterns of cognitive activities that
constitute distinct creative micro-processes. The creative micro-processes are E–F-
A-E and E-A-E, as well as the crossover of algorithmic and geometric levels of cogni-
tive activities. The micro-processes are easily extended to explain other parametric
design patterns as well. However, only the cyclic operations of both the geometric and
algorithmic activities can support “unexpected discoveries” (Akin and Akin 1996)
and thereby this type of creativity. Thus, while parametric design environments may
still have limitations in supporting intuitive design activities, their generative activ-
ities effortlessly support flexible design visualisation and through this heightened
creativity.
The results of these empirical studies in laboratory settings certainly won’t
perfectly reflect every designers’ quotidian activities or individual idiosyncrasies.
Nonetheless, this in-depth analysis reveals how designers actually work in parametric
design. Furthermore, the identification of the creative micro-process model for para-
metric design offers a plausible and valuable explanation for past theorised models
of creativity, contributing to a better understanding of process-based creativity in
design.

References

Aish, Robert, and Robert Woodbury. 2005. Multi-level interaction in parametric design. In Smart
Graphics, ed. Andreas Butz, Brian Fisher, Antonio Krüger, and Patrick Olivier, 924-924. Lecture
Notes in Computer Science. Berlin/Heidelberg: Springer.
Akin, Ömer, and Cem Akin. 1996. Frames of reference in architectural design: Analysing the
hyperacclamation (A-h-a-!). Design Studies 17 (4): 341–361.
Akin, Ömer, and Hoda Moustapha. 2004. Strategic use of representation in architectural massing.
Design Studies 25 (1): 31–50.
Benami, O., and Y. Jin. 2002. Creative stimulation in conceptual design. In Proceedings of the
ASME Design Engineering Technical Conference.
Bilda, Z., and H. Demirkan. 2003. An insight on designers’ sketching activities in traditional versus
digital media. Design Studies 24 (1): 27–50.
Bilda, Z., and J.S. Gero. 2007. The impact of working memory limitations on the design process
during conceptualization. Design Studies 28 (4): 343–367.
Blosiu, J. O. 1999. Use of synectics as an idea seeding technique to enhance design creativity.
In1999 IEEE International Conference on Systems, Man, and Cybernetics, 1999. IEEE SMC ‘99
Conference Proceedings. Tokyo.
Botella, Marion, Franck Zenasni, and Todd Lubart. 2018. What are the stages of the creative process?
What visual art students are saying. Frontiers in Psychology 9. https://doi.org/10.3389/fpsyg.
2018.02266.
Bresciani, Sabrina. 2019. Visual design thinking: a collaborative dimensions framework to profile
visualisations. Design Studies 63: 92–124. https://doi.org/10.1016/j.destud.2019.04.001.
Cardenas, Carlos Andres. 2008. Modeling strategies: Parametric design for fabrication in architec-
tural practice. Harvard University.
Casakin, Hernan, and Shulamith Kreitler. 2011. The cognitive profile of creativity in design.
Thinking Skills and Creativity 6 (3): 159–168.
Chusllp, P., and V. Jin. 2004. Cognitive modeling of iteration in conceptual design. In Proceedings
of the ASME Design Engineering Technical Conference, 473–485.
References 83

Coorey, B., and J. Jupp. 2011. Parametric modelling and design processes: Exploring synthesis
and evaluation using a function-behaviour-structure perspective. In Circuit Bending, Breaking
and Mending: CAADRIA2011, ed. C.M. Herr, N. Gu, S. Roudavski, and M.A. Schnabel, 39–48.
Australia: The University of Newcastle.
Cross, N. 2008. Engineering Design Methods: Strategies for Product Design. Wiley.
Dorst, Kees, and Nigel Cross. 2001. Creativity in the design process: Co-evolution of problem-
solution. Design Studies 22 (5): 425–437.
Flowers, John H., and Calvin P. Garbin. 1989. Creativity and Perception. In Handbook of Creativity.
Perspectives on Individual Differences, ed. John A. Glover, Royce R. Ronning, and Cecil R.
Reynolds, 135–145. New York, NY, US: Plenum Press.
Gero, J.S. 1996. Creativity, emergence and evolution in design. Knowledge-Based Systems 9 (7):
435–448.
Guilford, Joy P. 1967. The Nature of Human Intelligence. New York: McGraw-Hill.
Hanna, Sean, and Alasdair Turner. 2006. Teaching parametric design in code and construction.
Paper presented at the SiGraDi2006/Educacion y Desarrollo Academico,
Hasirci, D., and H. Demirkan. 2007. Understanding the effects of cognition in creative decision
making: A creativity model for enhancing the design studio process. Creativity Research Journal
19 (2–3): 259–271.
Ibrahim, Rahinah, and Farzad Pour Rahimian. 2010. Comparison of CAD and manual sketching
tools for teaching architectural design. Automation in Construction 19 (8): 978–987. https://doi.
org/10.1016/j.autcon.2010.09.003.
Iordanova, Ivanka. 2007. Teaching digital design exploration: Form follows. International Journal
of Architectural Computing 5 (4): 685–702.
Iordanova, I., T. Tidafi, M. Guité, G. De Paoli, and J. Lachapelle. 2009. Parametric methods of
exploration and creativity during architectural design: A Case study in the design studio. In Joining
Languages, Cultures and Visions: CAADFutures 2009, ed. T. Tidafi, and T. Dorta, 423–439.
Javier, Monedero. 2000. Parametric design: A review and some experiences. Automation in
Construction 9 (4): 369–377.
Jones, John Christopher. 1963. A method of systematic design. In Conference on Design Methods,
ed. John Christopher Jones, and D.G. Thornley, 53–73. Oxford: Pergamon Press.
Kim, J.E., C. Bouchard, J.F. Omhover, and A. Aoussat. 2010. Towards a model of how designers
mentally categorise design information. CIRP Journal of Manufacturing Science and Technology
3 (3): 218–226.
Kolarevic, B. 2003. Architecture in the Digital Age: Design and Manufacturing. New York and
London: Taylor and Francis.
Lawson, Bryan. 1980. How Designers Think. Architectural Press; Westfield, N.J.: Eastview Editions.
Lawson, Bryan. 2002. CAD and creativity: Does the computer really help? Leonardo 35 (3): 327–
331. https://doi.org/10.1162/002409402760105361.
Lee, Ju Hyun, Ning Gu, Michael J. Ostwald, and Julie Jupp. 2013. Understanding cognitive activities
in parametric design. In Global Design, ed. Jianlong Zhang and Chengyu Sun, 38–49. Berlin,
Heidelberg: Springer.
Lee, Ju Hyun, Ning Gu, Julie Jupp, and Sue Sherratt. 2014a. Evaluating creativity in parametric
design processes and products: A pilot study. In Design Computing and Cognition ‘12, ed. John
S. Gero, 165–183. Dordrecht: Springer Netherlands.
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014b. Parametric design strategies for the gener-
ation of creative designs. International Journal of Architectural Computing 12 (3): 263–282.
https://doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2015. Creativity and parametric design? Comparing
designer’s cognitive approaches with assessed levels of creativity. International Journal of Design
Creativity and Innovation 3 (2): 78–94. https://doi.org/10.1080/21650349.2014.931826.
Liu, Yu-Tung., and Chor-Kheng. Lim. 2006. New tectonics: A preliminary framework involving
classic and digital thinking. Design Studies 27 (3): 267–307.
84 3 Creative Micro-processes in Parametric Design

Madkour, Yehia, Oliver Neumann, and Halil Erhan. 2009. Programmatic formation: Practical appli-
cations of parametric design. International Journal of Architectural Computing 7 (4): 587–604.
https://doi.org/10.1260/1478-0771.7.4.587.
Maher, M.L., and Josiah Poon. 1996. Modeling design exploration as co-evolution. Computer-Aided
Civil and Infrastructure Engineering 11 (3): 195–209. https://doi.org/10.1111/j.1467-8667.1996.
tb00323.x.
Oxman, Rivka. 2006. Theory and design in the first digital age. Design Studies 27 (3): 229–265.
https://doi.org/10.1016/j.destud.2005.11.002.
Oxman, Rivka. 2008. Digital architecture as a challenge for design pedagogy: Theory, knowl-
edge, models and medium. Design Studies 29 (2): 99–120. https://doi.org/10.1016/j.destud.2007.
12.003.
Pugh, S. 1991. Total Design: Integrated Methods for Successful Product Engineering. Co: Addison-
Wesley Pub.
Roberto, Barrios Hernandez Carlos. 2006. Thinking parametric design: Introducing parametric
Gaudi. Design Studies 27 (3): 309–324.
Sakamoto, T., and A. Ferré. 2008. From Control to Design: Parametric/Algorithmic Architecture.
Barcelona: Actar-D.
Salim, Flora, and Jane Burry. 2010. Software openness: Evaluating parameters of parametric
modeling tools to support creativity and multidisciplinary design integration. In Computational
Science and Its Applications—ICCSA 2010, ed. David Taniar, Osvaldo Gervasi, Beniamino
Murgante, Eric Pardede, and Bernady Apduhan, 483–497. Lecture Notes in Computer Science.
Berlin/Heidelberg: Springer.
Schnabel, M.A. 2007. Parametric designing in architecture: A parametric design studio. In Inte-
grating Technologies for Computer-Aided Design, CAADfutures 2007, ed. aaa, 237–250. Sydney,
Australia: Springer Academic Press.
Schumacher, Patrik. 2009. Parametricism: A new global style for architecture and urban design.
Architectural Design 79 (4): 14–23. https://doi.org/10.1002/ad.912.
Stempfle, Joachim, and Petra Badke-Schaub. 2002. Thinking in design teams—An analysis of team
communication. Design Studies 23 (5): 473–496. https://doi.org/10.1016/S0142-694X(02)000
04-2.
Suwa, M., T. Purcell, and J. Gero. 1998. Macroscopic analysis of design processes based on a
scheme for coding designers’ cognitive actions. Design Studies 19 (4): 455–483.
Verstijnen, I.M., C. van Leeuwen, G. Goldschmidt, R. Hamel, and J.M. Hennessey. 1998. Sketching
and creative discovery. Design Studies 19 (4): 519–546.
Wallas, G. 1926. The Art of Thought. New York: Harcourt Brace.
Chapter 4
Measuring Cognitive Complexity

Abstract Past research has theorised that high levels of individual cognitive
complexity may result in heightened design thinking and creativity. The precise rela-
tionship, however, between cognitive complexity and creativity in design remains
largely unexplored. This chapter develops two measures of cognitive complexity
in design: content complexity and structural complexity. Using a combination of
protocol analysis and linkography, it demonstrates how these two can be measured
and studied. The demonstration uses two sets of protocol data developed from exper-
iments in parametric design. The results indicate that (i) content complexity can
be used to explain individual differences of cognitive complexity and (ii) structural
complexity using decile growth plots of linkographs can reveal cognitive patterns
over time. This method for measuring cognitive complexity contributes to advancing
fundamental knowledge about design cognition and thinking.

4.1 Introduction

Cognitive complexity is a measure of an individual’s capacity to conceptualise,

organise, perceive or communicate. The greater the number of Mental Models (MMs)
or conceptual structures available to an individual, along with a more nuanced or
sensitive their capacity to apply them, the higher their cognitive complexity. A person
with a low level of cognitive complexity might respond to a given situation with only
limited, rigid, naïve or inappropriate actions or behaviours. They might also be unable
to communicate alternative perspectives or understand personal and cultural differ-
ences. Conversely, a person with a high level of cognitive complexity will have the
capacity to employ intricate abstraction and communication skills, while using and
testing alternative theories or constructs.
Design thinking has clear synergies with cognitive complexity as both rely on
conceptual thinking, perception, abstraction and communication. Design has histor-
ically been described as both an art and a science, requiring a substantial under-
standing of human needs and a capacity to respond to them in creative ways. Even
the oldest architectural treatises call for the designer to have a rich and deep aware-
ness of multiple fields of knowledge. In the first century BC, for example, Vitruvius

© Springer Nature Switzerland AG 2020 85

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_4
86 4 Measuring Cognitive Complexity

argued that architectural education must not only include geometry and drawing,
but also history, philosophy, law and music. His core message is that an effective
designer must be equally well versed in theory as in practice. He dismisses mere
copyists and technicians, and those who work without an adequate grasp of optics,
physics, strategy, symbolism and mathematics. In his Ten Books of Architecture,
Vitruvius effectively argues that a successful designer must possess a high level of
cognitive complexity.
In recent years, the need for advanced design skills and increased cognitive
capacity has become more pronounced. With the rise of data-driven approaches,
mathematical models, algorithmic thinking and interactive simulation, design has
become both multi-dimensional and multi-disciplinary (Weinstock 2010; Aish and
Woodbury 2005). For example, the application of parameters and rules in parametric
design may support advanced problem-solving, but it also potentially increases the
cognitive complexity required of a designer. In addition, generative design requires
the use of abstract constructs to program and assess outputs (Chien and Flemming
2002). In parametric and generative environments, designers must work with uncon-
ventional algorithm editors and employ intricate design thinking and functions, which
collectively increase the complexity of the process.
Past research has confirmed that design is not only a complex process (Almeida
et al. 2005; Earl et al. 2005; Khan and Angeles 2007), it relies on emergence, self-
organisation and hologrammaticity (Almeida et al. 2005). In other words, design
is a system wherein constituent elements interact nonlinearly and interdependently,
requiring increased cognitive capacity to manage effectively. Given this background,
it is not surprising that measuring and understanding cognitive complexity is poten-
tially significant in design. Throughout history, it has been theorised that a height-
ened capacity to use conceptual thinking, abstraction and communication is linked
to creativity. As such, developing insights into cognitive complexity may provide a
pathway to more innovative, original and visionary designs.
This chapter explores cognitive complexity in design using two indicators or
measures: content complexity and structural complexity. The theoretical and math-
ematical origins of both measures are derived from Information Theory (Shannon
1948; Krus 2013; Khan and Angeles 2007). Like Kolmogorov’s complexity theory,
Shannon’s information theory seeks to measure the “information” in an “object”.
While there are differences between the two theories, conceptually they both compare
quality of information with volume of information. This chapter uses entropy¸ a quan-
titative measure for “disorder” or “order compared with disorder” (Baranger 2001),
for quantifying individual cognitive complexity. To explain how cognitive complexity
may be captured, measured and compared, this chapter revisits the protocol data
presented in Chap. 2.
This chapter is divided into three main parts. This first section introduces cogni-
tive complexity and its content and structure in design. Following this, the entropy
measure is described along with the use of a coding scheme and linkography to
develop two measures of cognitive complexity. In the third section, protocol data
is used to investigate cognitive processes and measure the content and structural
complexity of parametric design.
4.2 Background 87

4.2 Background

The concept of cognitive complexity was first proposed in the 1950s in the fields
of psychology and communication, and multiple definitions and assessment tech-
niques have since been established (Bieri 1955; Kelly 1955). Despite continuing
developments in this field, one of the earliest theories to explain it, Kelly’s “per-
sonal construct psychology” (PCP), remains in wide use today. PCP describes a
person’s capacity to construct, construe or interpret the world. It argues that each
person anticipates events by interpreting their properties, categorising them and then
determining a course of behaviour (Kelly 1955; Pervin 1975). The extent to which
a person has the capacity to construe widely and strategically is a measure of their
cognitive simplicity or complexity (Bieri 1955; Pervin 1975). PCP is effectively an
MM of cognitive construction. Such “models play a central and unifying role in
representing objects, states of affairs, sequences of events, the way the world is,
and the social and psychological actions of daily life” (Johnson-Laird 1983, p. 397).
MMs of this type are representations of reality that people use to understand specific
phenomena (Norman 1983). The more conceptually intricate, abstract and nuanced
the representational model, the higher the level of cognitive complexity.
O’Keefe and Sypher (1981) define cognitive complexity as a variable that
describes the limits of a person’s socio-cognitive system. The degree of differen-
tiation, articulation and abstraction within the socio-cognitive system is measured
using “individual difference variables” (Burleson and Caplan 1998). For example,
Benet-Martínez et al. (2006) examine cultural representations finding that a higher
level of cognitive complexity is related to increased information clustering (more
differentiation and integration) and abstraction (less tangible and structured descrip-
tions). Collectively, these psychological models of cognitive complexity are reliant on
the process of categorising socio-cognitive systems and communicative behaviours
using complexity measures (O’Keefe and Sypher 1981). This last point is significant
as psychological, social and semiotic researchers have repeatedly noted that ideas are
intrinsically tied to the language in which they are both constructed and expressed
(Bonvillain 2010; Lewis 2012; Lopez 2003).
The reciprocal relationship between ideation and conceptualisation on one side,
and language and communication on the other, has been repeatedly noted in past
research. A key example of this is the construction and communication of spatial and
formal relationships and reasoning (Herskovits 1986; Tenbrink and Ragni 2012; van
der Zee and Slack 2003). Linguistic studies, like “naming tasks” (Munnich et al. 2001)
and “descriptive tests” (Tenbrink and Ragni 2012), also confirm this relationship. The
close connection between linguistic complexity and cognitive complexity is signifi-
cant because the former can be measured with relative ease. For this reason, linguistic
analysis is often used to reveal cognitive complexity and illuminate a range of other
characteristics (Bowerman 1996; Levinson 1996). For example, Tenbrink’s (2015)
cognitive discourse analysis technique uses language protocols to assess complex
cognitive processes such as problem-solving, cognitive strategies and heuristics.
88 4 Measuring Cognitive Complexity

Linguistic or syntactic complexity can be defined and measured in several ways.

Syntactic complexity is an indicator of the range of forms uncovered in language
production and the degree of sophistication of these forms (Ortega (2003). It is
used to describe linguistic maturity or competency, and many measures of it have
been developed, including length of sentence, length of “T-unit” (a unit of “thought”
consisting of one main clause with all subordinate clausal and non-clausal elements
embedded in it), length of clause, clauses per T-unit (C/T ) and dependant clauses per
clause (Hunt 1970). “T-unit complexity ratio” allows for linguistic comparisons of
relative complexity in grammatical constructions (Lu 2010). This linguistic measure
is useful in the present context because designers’ spoken words are so variable in
length in the design process that a comparable value is required to measure individual
differences. Furthermore, Kelly’s (1955) PCP theory and Bieri’s (1955) cognitive
complexity theory both propose possible mathematical grounds for comparisons
between design and linguistics. While such studies are largely associated with social
cognition and communication, they allow researchers to conceptualise individual
difference in terms of complexity. Moreover, they draw on advances in science,
mathematics and social science which have emphasised the importance of measuring
and understanding the complexity in systems (Baranger 2001).
In contrast to the psychological, psychosocial, communication and linguistic
approaches to cognitive complexity, in the design domain the terms cognition and
complexity are rarely used together (Lee and Ostwald 2019). Design process and
cognition, for example, are recurring themes in design research (Chai and Xiao
2012), whereas complexity tends to be used in the design domain to interrogate
notions of novelty or as a means of imbuing a design with unique or distinct proper-
ties (Almeida et al. 2005). Complexity has also been used as a criterion for assessing
“levels of creativity” (Amabile 1983). As such, complexity is used to describe or
assess design as a product, not as a process, and there is no established use of the
concept of cognitive complexity to describe the actions and behaviours of designers.
One possible reason why cognitive complexity is not commonly used or discussed
in design is that there are not any methods to measure it. Protocol analysis, for
example, has been limited in the past to describing and categorising design activities
through various cognitive methods (Chan 2015). However, design problem-solving
behaviour is a core and measurable cognitive indicator in the design process. More-
over, several coding schemes used in protocol analysis do have the potential to be
adapted to measure cognitive complexity. Such coding schemes identify cognitive
micro-activities and associated behaviours that occur during design experiments and
categorise them in such a way as to match measures of cognitive complexity in
psychosocial research.
As discussed in previous chapters, researchers in the design domain have devel-
oped multiple coding schemes corresponding to various research agendas (Gero and
Neill 1998; Kim et al. 2010; Kim and Maher 2008; Stempfle and Badke-Schaub
2002; Suwa et al. 1998; Akin 1986; Eastman 1969). For example, Gero’s Func-
tion–Behaviour–Structure (FBS) framework has frequently been applied to measure
cognitive properties in design, and Gero and McNeill (1998) identify five categories
for understanding models of engineering design. These coding schemes capture
4.2 Background 89

various cognitive events and associated behaviours in design, and thereby allow
for the exploration of individual designer’s cognitive systems. This approach has
much in common with Burleson’s and Caplan’s (1998) individual difference vari-
able, which is used in psychosocial experiments to measure cognitive complexity.
Suwa et al.’s (1998) coding scheme enables the examination of physical, perceptual,
functional and conceptual processes of human cognition. Such schemes support the
collection of diverse information that can be used for measuring, among other things,
cognitive complexity.

4.3 Measuring Cognitive Complexity

One of the earliest methods used in psychosocial research for measuring cogni-
tive complexity was based on Kelly’s repertory grid (1955). This method requires
the capturing of data using transcribed interviews, questionaries or written descrip-
tions. Participants’ responses are then rated on a tailored complexity scale to quan-
titatively identify individual cognitive traits. This method can be used to capture
and compare cognitive activities that occur during complex multi-variable problem-
solving processes. Such psychological processes are strongly reminiscent of the
cognitive activities that occur while designing.
In psychological research, two dimensions of cognitive complexity have been
measured and analysed. Benet-Martínez et al. (2006) describe these two dimensions
as pertaining to the content (properties and features) of a cognitive process and
its underlying structure (relationships and dynamics). The design equivalents that
are measured in this chapter are degrees of differentiation of individual cognitive
content (categorisations of cognitive activities) and cognitive structure (relationships
between cognitive activities). Collectively, these two contribute to a determination
of levels of cognitive complexity (Fig. 4.1). This two-part definition has parallels to
those used in psychology, but with the focus here being solely on design as a series
of cognitive activities and associated behaviours.

Cognitive complexity Content complexity

Coding scheme
 CategorisaƟons of cogniƟve acƟviƟes
 Weighted coverage value

Structural complexity
Linkography
 RelaƟonships between cogniƟve acƟviƟes
 Complexity over Ɵme

Fig. 4.1 Content and structure of cognitive complexity

90 4 Measuring Cognitive Complexity

Content complexity relates to the individual categories of cognitive activities

evident in a design process. These categories can be understood as the set of all
possible states of the design cognition system, and therefore the presence or absence
of various categories can be used to measure the uncertainty of cognitive activities in
a system. When considered in the context of classical information theory (Shannon
1948), this means that cognitive categorisations are the “variables” or “events” that
generate the dynamical system. These events are equivalent to “letters” in a typical
message, or “bits” in a stream of data. In this chapter, the data for analysing content
complexity in design is developed using a coding scheme to capture the cognitive
content of the process (that is, its cognitive properties). Because different design
sessions have different lengths, the frequency of each code is weighted by its time
duration, or weighted coverage value.
Structural complexity is concerned with the relations between cognitive activities
in a design process. Whereas content complexity is about the presence, absence or
frequency of cognitive categories, structural complexity is a measure of the connec-
tions between these categories. To capture structural complexity in protocol data, this
chapter uses a variation of linkography (Goldschmidt 1990, 1995, 2014). Combining
linkography with distance graphs supports the exploration of the effect of inspira-
tion in design (Cai et al. 2010). Measuring the degree of development through a
distance graph can also reveal the individual level of abstraction present. Because
the variables in a dynamical system can vary with time (Baranger 2001), the struc-
tural complexity measure considers changing connectivity and dynamics over time.
In summary, if there is cognitive disorder in a design process, individual categorisa-
tions of cognitive activities and their structures can be used as critical indicators of
subjective operations governing cognitive complexity.

4.4 Method

4.4.1 Entropy Measure

In Shannon’s (1948) mathematical theory of communication, entropy is used as a

measure of complexity, describing the average amount of information in a message
or the degree of randomness in the system. Conceptually, the greater the entropy, the
greater the uncertainty and the higher the complexity. This chapter uses “information
entropy” to quantify complexity in the design process. In theoretical physics and
mathematics, complexity is explained in terms of entropy, which refers to a state that
lacks order or predictability. While “disordered” or “unpredictable” states may not
appear desirable, in a mathematical sense these properties simply denote a system
with a higher level of information or a higher rate of connectivity. A higher level
of entropy indicates increased complexity in a system, or in the case of the design
process, higher cognitive complexity. Thus, in the discussion and explanation that
4.4 Method 91

follows, references to the mathematical property “uncertainty” can be generally read

as indicators of potential complexity in a system.
There have been multiple uses of entropy in design in the past. For example,
entropy has been used as a measure of the “design space” in the early stages of the
design process, quantifying the information content of a design (Frey and Jahangir
1999; Krus 2013; LaFleur 1996; Suh 1999). LaFleur (1996) employs entropy as a
measure of multi-functional problems, wherein the complexity of the design problem
is a product of the number of functions it encompasses. Suh (1999) also defines
complexity as a measure of uncertainty in achieving a set of specific functions
or functional requirements. Frey and Jahangir (1999) suggest the use of differen-
tial entropy as a measure of information content in axiomatic design, transforming
design parameters into functional characteristics. Khan and Angeles (2007) propose
using entropy to define diversity in design solution alternatives. Krus (2013) also
suggests using design information entropy as a state—reflecting both complexity
and refinement—to quantitatively describe various aspects in the design process.
The logic being that through concept generation, concept selection and parameter
optimisation, complexity and uncertainty are reduced, while information is gained.
The mathematical formula for calculating entropy (H) follows Shannon’s (1948)
equation combining information, choice and uncertainty:


n
H =− pi log pi (4.1)
i=1

where H is named after Boltzmann’s H-theorem and pi is the probability of a complex

system being in state i. The choice of a logarithmic base determines the unit used
to measure information. When the base is 2, the unit is called binary digits (bits).
While sometimes adopting base e can be useful (resulting in natural units), decimal
or Briggs logarithms are also used to measure H (Khan and Angeles 2007). As base
2 is widely used in applications in information theory (Bansiya et al. 1999; Frey and
Jahangir 1999; Kan and Gero 2008; Khan and Angeles 2007; Krus 2013; LaFleur
1996), this chapter focuses on bits of information content. Shannon describes this
Eq. (4.1) as measuring the entropy of the set of probabilities pi , …, pn . Thus, in the
case of two possibilities with probabilities p and q = 1–p, the entropy is

H = −( p log p + q log q) (4.2)

This can also be plotted (Fig. 4.2).

When calculated in this way, H is a measure of choice or information (Shannon
1948). In a design protocol, each segment can be encoded using the initials of the
five codes of Gero’s FBS framework (Gero 1990; Gero and Kannengiesser 2014)—
function (F), behaviour (B), structure (S), requirement (R) and design description
(D)—and two behaviours (Be and Bs). However, designers’ cognitive choices are,
of course, independent in the model. Thus, R precedes F and F is often followed
by B and then S. S often returns to F and leads to D. Thus, an example of a typical
92 4 Measuring Cognitive Complexity

0.9

0.8

0.7

Entropy 0.6
(H)
0.5

0.4

0.3

0.2

0.1

0
0 0.5 1
Probability(p)

Fig. 4.2 Entropy in the case of two probabilities (p and q = 1 – p)

sequence of FBS codes in a design process might be

R F B S F B S F B S F B S F B S F B S D.

In this example, both R (requirement) and D (design description) occur once in

the process over 20 segments (design activities). Thus, there is a possibility of 1/20
of the designer undertaking R or D activities. The possibilities of F, B and S are 6/20,
6/20 and 6/20, respectively. Collectively, the total information content of the given
design process is

H = −H R −HF −H B −HS −H D
1 1 6 6 6 6 6 6 1 1
H = − log2 − log2 − log2 − log2 − log2
20 20 20 20 20 20 20 20 20 20
H = − (−0.2161)− (−0.5211)− (−0.5211)− (−0.5211)− (−0.2161)
H = 1.9955 bits

Because the FBS framework is a process-oriented coding scheme, it is not entirely

appropriate for measuring cognitive complexity. There are, for example, both content
and process components of a designer’s action that also need to be considered (Dorst
and Dijkhuis 1995; Lee and Ostwald 2019; Lee et al. 2014a). Thus, a content-oriented
coding scheme is required to identify the information content of a design process.
4.4 Method 93

4.4.2 A Coding System for Content Complexity

Measuring content complexity requires a content-oriented coding scheme that

captures both the content of cognitive activities and the uncertainty of informa-
tion processing in human cognition. As in previous chapters, Suwa et al.’s (1998)
coding scheme is used as the basis, because it provides a general taxonomy for
the content of designers’ cognitive processes as well as information processing. In
this scheme, physical actions correspond to sensory level, perceptual actions to the
perceptual level, and conceptual actions to semantic. Its cognitive categories are also
derived from information categories (Suwa and Tversky 1997). This coding system
is used to reveal the sequence of cognitive activities in a design protocol. This vari-
ation uses three categories—physical, perceptual and conceptual—but with revised
activity classifications that are more suitable for geometric and algorithmic activities
in parametric design (Table 4.1). As discussed previously, the geometric activities
are regarded as the modelling activities for creating digital geometries, while the
algorithmic activities represent the programming processes using “parameters” and
“rules” as well as the “reference” code.
The physical level in the coding scheme represents descriptive activities.
The perception level represents the activities related to visual imagery in the
design process (P-Geometry and P-Algorithm), while the conceptual level high-
lights problem-solving processes, such as problem-finding, solution-generating and
solution-evaluating evaluating. This chapter aims to measure cognitive complexity
through the coding results of subclasses in Table 4.1, because the micro-activities
can be regarded as equivalent to the events in a complex system.
The different frequencies of cognitive activities observed in an experiment can
be used to indicate the individual uncertainty of information processing, which is
the set of all possible states of the cognitive system. Thus, the frequencies of micro-
cognitive categories (subclasses) are the key cognitive complexity variables because
they can be thought of as degrees of uncertainty or complexity. Design protocols are
initially divided into smaller segments (episodes or events), which are encoded as at
least one of the subclasses in Table 4.1. An example of this type of coded protocol
data is shown in Table 4.2.
A typical information entropy measure uses the frequencies of sub-codes to calcu-
late the probabilities of events. However, as described in Table 4.2, segments (cogni-
tive events) can have different time spans, which must be taken into account to

Table 4.1 Subclasses (codes) of the coding scheme for parametric design
Level Subclass
Physical G-Geometry, G-Change, A-Parameter, A-Change Parameter, A-Rule, A-Change
Rule, A-Reference
Perceptual P-Geometry, P-Algorithm
Conceptual F-Initial Goal, F-Geometry Sub Goal, F-Algorithm Sub Goal, G-Generation,
E-Geometry, E-Parameter, E-Rule, E-Reference
94 4 Measuring Cognitive Complexity

Table 4.2 Example of an encoded protocol (V: Visual cue in the video recording)
Segment No Time span (s) Transcription of Design Level Subclass
Experiment
14 13.1 What I am going to do is Conceptual F-Geometry Sub Goal
work on a structural (or F-GSG)
solution first. What I want
to do is to put a grid
around the outside …
15 10.0 (V: dragging and copying Physical A-Rule
some rules) so I am
creating a series of grids
…
16 15.2 … on the outside, (V: Physical A-Change Parameter
changing parameters on (or A-CP)
the copied sliders)
17 32.8 I am just creating a border Physical A-Rule
of the grid, I copy that
again. … (V: inserting and
connecting rules)
18 13.1 Fifty, too many of them, Physical A-Change Parameter
forty (V: changing (or A-CP)
parameters on the inserted
rules)
19 1.4 Okay (V: Examining Conceptual E-Geometry
forms in the 3D view
mode)

calculate the given probabilities of cognitive activities. The weighted frequency by

time duration of each subclass and the probability of the subclass of cognition (P) is

T (a)
P(a) =  (4.3)
x∈S T (x)

In this formula, T(a) is the time duration of the code in the protocol and S is the set
of subclasses. That is, T(a) is the summary of time spans of cognitive events being
the subclass. For example, consider a sample protocol consisting of six segments
(Table 4.2). The probability of the F-Geometry Sub Goal (shortly, F-GSG) is

T (F − G SG)
P(F − G SG) =
T (F − G SG) + T (A − Rule) + T (A − C P) + T (E − Geometr y)
13.1
=
13.1 + (10.0 + 32.8) + (15.2 + 13.1) + 1.4
= 0.1530
4.4 Method 95

Since the lengths of design protocols are often different, probabilities based on
weighted frequencies must be determined to compare individual cognitive cate-
gories (subclasses) in the set of protocols. Thus, the percentage of the weighted
coverage value of each subclass in a design protocol can also be regarded as the
corresponding probability for measuring the content complexity of the protocol. The
content complexity (H C ) that is the summary of H values of all subclasses is then
calculated (Eq. 4.1).
In addition to these complexity values, this chapter calculates the normalised
coverage values of cognitive levels (macro-thinking categories), where normalised
A = (A – mean)/standard deviation (Bilda and Gero 2007; Lee et al. 2014a, 2015).
Specifically, the normalised data over time can be used to refer to individual patterns
of cognitive complexity in parametric design, because they show individual changes
and differences of the content of cognitive activities as well as the levels of infor-
mation processing in parametric design. In summary, the coding scheme for content
complexity is a fundamental component for measuring cognitive complexity.

4.4.3 Linkography Measures for Structural Complexity

Linkography is a graphical and topographic analysis technique that is used to inves-

tigate the way one idea, action or event interlinks with another (Goldschmidt 1990,
2014, 1995; Kan and Gero 2008; Lee and Ostwald 2019; Goldschmidt and Tatsa
2005). Linkography allows for recording, visualising and analysing the structure of
connections between things. For the purpose of measuring cognitive complexity, the
structural relationship of components in a linkograph is used.
Although linkography has been used to map idea-generation process in design
research, the task of objectively constructing a linkograph with “moves” and “links”
remains a significant challenge. The primary means by which a link can be derived
from protocol data is through the application of a coding scheme that identifies
explicit connections. To develop a linkograph from protocol data, Kan and Gero
(2008) identify three critical procedures: segmentation, linking and analysis. Only a
few studies (Lee et al. 2016; Lee and Ostwald 2019), however, have identified moves
and links in parametric design. The segment or “move”, is the step, act or operation
that transforms a design situation, while the “link” is the connection between the
segments (Goldschmidt 1995). Goldschmidt also suggests that rational links should
be based on “common sense” reasoning in the analysis of a protocol.
In this chapter, algorithmic scripts connecting design moves are used for under-
standing the parametric design process. Scripting activities are regarded as a channel
for creativity and a means of representing design ideas in parametric design (Salim
and Burry 2010). Furthermore, different types of moves and links develop in the
design process. For example, van der Lugt (2000) categorises links into three types:
supplementary, modification and tangential, and Perttula and Sipilä (2007) identify
an alternative set of three types: parts sharing, same principle and modification.
“Parts sharing” deals with primary parts of examples used in a follow-up concept,
96 4 Measuring Cognitive Complexity

and “same principle” is regarded as a follow-up concept that is based on the same
overall principle as an example. “Modification” is a follow-up concept that comprises
a minor modification of an example at the embodiment level.
Figure 4.3 shows examples of algorithmic components (units) and their connec-
tions (links in the graphic-based scripts and adjacencies in the text-based scripts).
Scripts in parametric design, like sketches in traditional design environments, act

a. Graphic-based scripts (Grasshopper)

b. Text-based scripts (Python)

Fig. 4.3 Algorithmic components/units and their connections
4.4 Method 97

as blueprints for the identification of algorithmic components or units and their

sequence. Designers create and change parameters and rules on both the graph-
ical and text-based scripting editors. For example, Fig. 4.3a shows the sequence of
algorithmic components (or units) with numbering and clear graphical links between
components. In the case of a text-based scripting editor (Fig. 4.3b), a group of related
scripts that were created or modified together are defined as a unit (or a block of
codes). Each numbered unit can then be linked to adjacent unit(s) within a program-
ming component (see units 11 and 15 in Fig. 4.3b). This scripted structure allows us
to identify algorithmic components and their sequence in parametric design, when
designers create and change parameters or rules. From these algorithmic scripts, a
linkograph can be constructed using four significant types of moves: (i) introducing
geometric/algorithmic ideas, (ii) creating algorithmic components (as a unit), (iii)
modification activities and (iv) evaluation activities.
Once the moves and links are defined, then the next stage analyses the four signif-
icant moves. These moves correlate to Goldschmidt’s (1995) “link-intensive” or
“critical moves” (CMs). CMs also serve as an effective indicator of productivity.
Figure 4.4 illustrates part of a linkograph including the example of coded protocol
data (Table 4.2). As shown in this graph, introducing geometric ideas (F-Geometry
Sub Goal) and evaluation activities (E-Geometry) develop the CMs. After developing
the linkograph, statistical and other measures, including entropy, can be calculated
(Kan and Gero 2008).
Individuals with high cognitive complexity have the capacity to analyse a situation
to discern constituent elements and explore connections and possible relationships
between elements (Lee and Ostwald 2019). The linkograph is especially useful in this
context because it captures the structural complexity of parts and connections (the
dynamic complexity of behaviour), allowing for information entropy in the design
process to be measured. Kan and Gero (2008) use entropy as a means of inter-
preting linkographic results, focusing on acquiring abstracted information from the
graphs. In contrast, information entropy relating to relationships between cognitive

Fig. 4.4 A part of a

linkograph of cognitive
categories and connections
98 4 Measuring Cognitive Complexity

activities in a linkograph is a structural aspect of cognitive complexity. The struc-

tural complexity enables capturing cognitive information processing in parametric
design. Furthermore, this measurement considers the complexity of connectivity and
dynamics over time because the set of levels of cognition can vary during the design
process. Entropy can also be used to measure idea development opportunity (Kan and
Gero 2008). Since the probability of a link being present complies with an On/Off
rule, the entropy of each row in a graph is

H = − p(O N ) log2 p(O N ) − p(O F F) log2 p(O F F) (4.4)

The “row” in this context refers to Goldschmidt’s (1995) two types of links:
“backlinks” and “forelinks” (see Fig. 4.5). Both forelink entropy and backlink
entropy can be determined using an Eq. (4.4). Forelink entropy measures reflect
idea-generation opportunities, while backlink measures relate to opportunities asso-
ciated with enhancements or responses. Horizonlink (horizontal-link) entropy can
also be measured as the distance between the linked moves, and it measures the
opportunities relating to cohesiveness and incubation (Kan and Gero 2008). Thus,
individual values derived from these three types of entropy can be used to identify a
variety of cognitive features in a set of design protocols.
The total cumulative entropy in the system indicates the volume of idea develop-
ment opportunities (Kan and Gero (2008), which is similar to the link index calcu-
lated by dividing the number of links by the number of moves (Goldschmidt and
Tatsa 2005). In this chapter, cumulative value is regarded as the degree of individual
structural complexity. Because the links of a linkograph indicate the relationships
between cognitive activities they conform to the second characteristic of cognitive
complexity in design (structural complexity). In order to compare different lengths
of design sessions, the structural complexity (H S ) of each protocol is defined as the
cumulative total per move.
A final consideration when looking at these three forms of links is that, concep-
tually at least, entropy as part of a design process increases as a system evolves
and time passes (Baranger 2001). While it might be possible to imagine a back-
ward evolution process in design, the more common model sees entropy increase
over time (Eddington’s so-called “arrow of time” phenomenon). Using statistical
methods, additional measures of complexity can be derived using this assumption.
The measuring of changes in coded activities or links over time is often shown in

a. Forelinks b. Backlinks c. Horizonlinks

Fig. 4.5 Forelinks, backlinks and horizonlinks
4.4 Method 99

protocol analysis (Gero and Neill 1998; Lee et al. 2015, 2014b; Suwa et al. 1998).
Earl et al. (2005) indicate that complex systems are also dynamic, changing and
evolving over time, like design processes. In order to deal with these complexities of
connectivity and dynamics over time, this chapter introduces time-based entropies
through decile growth plots of a linkograph that allows for a measure of the variation
of entropies in a design protocol to be produced.

4.5 Application

This chapter uses the protocol data developed from experiments in Chap. 2 to examine
the content and structural complexity in individual designer’s cognitive systems and
associated design strategies.

4.5.1 Content Complexity

Table 4.3 presents the coding results for the experiments of two designers (from
Study II in Chap. 2) divided into subclasses and expressed as the percentage of the
frequency weighted by time span (calculated by time duration of each code) as well
as its entropy value. If the percentage is 0, its entropy cannot be calculated because
log 0 is undefined. Individual cognitive complexity (content) uses the summary of all
entropy values in a protocol. Thus, the content complexity (H C ) of Au1’s protocol
is 3.3339 and Au3’s is 2.9897. The data record that Au1 adopts a clear problem-
decomposing strategy with a working-forward search strategy, while Au3 uses trial-
and-error sequences or backward reasoning. These different design processes may
explain the different content complexity results. In addition, the content complexities
of the novice designers, Au2 and Au4, are 3.2879 and 3.0611, respectively.
These results illuminate the theorised connection between cognitive complexity
and creativity in an unexpected way. For example, while Au3’s design solution
received the highest score for creativity from the judges, it was the lowest in terms of
content complexity. One possible explanation for this anomaly is that both Au1 and
Au2 contributed relatively higher percentages of their time to the “problem-finding”
activity (Au1 is the second highest and Au2 is the highest percentage). Au3 and Au4,
however, struggled with trial and error in their design sessions and produced relatively
higher percentages of A-Change Rule and P-Algorithm in Table 2.5 (Chap. 2.). Thus,
content complexity may be a better indicator of the probability of problem-finding
or problem-decomposing in a design session, than of creativity.
This finding is also reflected in the six “pen-and-paper” design protocols presented
previously (from Study I in Chap. 2). The two designers (Ko3 and Ko5) who used
a clear problem-forwarding strategy produced relatively higher complexity values
(2.3733, 3.2241, respectively) than the others. Although Ko1 developed a productive
design strategy, the complexity value of her protocol (2.1994) is the second lowest.
100 4 Measuring Cognitive Complexity

Table 4.3 The percentage of the coverage of each subclass and its entropy value by Au1 and Au3
Category Subclass Percentage Entropy
Au1 Au3 Au1 Au3
Physical G-Geometry 2.0 0.0 0.1129 –
G-Change 0.5 0.0 0.0382 –
A-Parameter 4.9 0.1 0.2132 0.0100
A-Change Parameter 10.0 4.7 0.3322 0.2073
A-Rule 24.7 20.0 0.4983 0.4644
A-Change Rule 6.0 12.2 0.2435 0.3703
A-Reference 0.0 0.2 – 0.0179
Perceptual P-Geometry 2.8 1.8 0.1444 0.1043
P-Algorithm 3.2 4.0 0.1589 0.1858
Conceptual F-Initial Goal 3.0 0.2 0.1518 0.0179
F-Geometry Sub Goal 7.0 2.5 0.2686 0.1330
F-Algorithm Sub goal 4.1 4.4 0.1889 0.1983
G-Generation 4.0 13.0 0.1858 0.3826
E-Geometry 19.4 17.3 0.4590 0.4379
E-Parameter 0.8 0.0 0.0557 –
E-Rule 7.6 19.5 0.2826 0.4599
E-Reference 0.0 0.1 – –
Summary (H C ) 100.0 100.0 3.3339 2.9897

This result may be because her design process is based on “drawing-reflection”, a

type of novice behaviour. Specifically, Ko2’s design process, which produced only
one rough plan layout and was the shortest design session, resulted in the lowest
complexity (1.9806). A further interpretation of the content complexity result might
be that it is dependent on a coding scheme itself. In other words, the relatively broad
subclasses—such as R-drawing in Table 2.3 and A-Rule in Table 2.4—may produce a
lower result. These results do, however, show that the content complexity measure not
only provides a meaningful basis for analysis, it also reveals individual differences
and cognitive properties.
The other way of examining the protocol data in terms of information content
is to consider changes in cognitive levels over time. The summed coverage data in
each decile interval can be visualised using the normalised coverage value to explore
sequential patterns. Figure 4.6 shows the normalised coverage of three levels of cogni-
tion over time. It enables the representation of physical, perceptual and conceptual
activities in sequence and comparison of their evolution. A close examination reveals
different patterns in levels of cognition within each protocol. For example, both
designers produced multiple physical activities in the early stages of their protocols,
while Au1’s protocol has a further peak at the sixth decile of the timeframe, whereas
the normalised coverage of physical activities in Au3’s protocol decreases after the
4.5 Application 101

2.5

2.0
Au1
Normalised coverage

1.5

1.0

0.5 Physical
Perceptual
0.0
1 2 3 4 5 6 7 8 9 10 Conceptual
-0.5

-1.0

-1.5 Decile Time

-2.0

2.5

2.0
Au3
Normalised coverage

1.5

1.0
Physical
0.5
Perceptual
0.0
1 2 3 4 5 6 7 8 9 10
Conceptual
-0.5

-1.0

-1.5 Decile Time

-2.0

Fig. 4.6 Normalised changes of three cognitive levels (physical, perceptual and conceptual) over
time by Au1 and Au3

fourth decile. The value of conceptual activities is relatively high in the middle of
the timeframe of Au3’s protocol, while the same activities in Au1’s protocol often
occurred later in the design process. Thus, the set of these normalised coverage
values over time in a protocol can be understood as an individual pattern of complex
cognitive content.
The coding results identify the design activities undertaken by each participant.
They also indicate different usages of cognitive categories (levels and subclasses)
that potentially allow for the two to be differentiated in terms of content complexity.
However, the frequency coverage of categories of design cognition has its limits,
as categorising cognitive levels in this way is both subjective and sensitive to the
research perspective or researcher. However, the segmentation of design protocols
and their structures may better support the quantitative measurement of cognitive
complexity in terms of moves and links in a linkograph (structural complexity).
102 4 Measuring Cognitive Complexity

4.5.2 Structural Complexity

The linkography analysis commences by addressing the four distinct move types
identified previously in this chapter. Because these move types tend to generate a
high number of forward or backward links (over five links), they can be said to
constitute Goldschmidt’s (1995) “link-intensive” or “critical moves” (CMs). In this
test, the two designers (Au1 and Au3) generally started by examining the design
brief before considering ideas that respond to the brief at both the geometry and the
algorithm thinking levels, coded as introducing geometric/algorithmic ideas (the first
move type). Thus, 84.4% (Au1) and 68.3% (Au3) of three codes in the “Finding”
category (F-Initial Goal, F-Geometry Sub Goal and F-Algorithm Sub Goal) in Table
4.4 link to more than five moves, which result in CMs (forelinks) that earn their
designation.
The third type of moves, modification activities, occurs when designers revisit and
change algorithmic units. When a designer modifies an algorithmic unit, it is linked
to the original move that created the unit. This process also tends to generate more
than five links, being a CM (forelink). Many algorithmic units in parametric design
are related to such modification activities (e.g. A-Change Parameter and A-Change
Rule). The final type of significant move in parametric design is the “evaluation
activity” that can form “backlinks”, recording the path of generating the move. For
example, a move wherein designers evaluate geometries in the 3D view is often linked
to all previous “physical” moves before “evaluating geometries (E-Geometry)”.
For this chapter, Excel and Linkoder (Pourmohamadi and Gero 2011) are used
to produce the linkographs. Figure 4.7 illustrates the linkographs of the two design
protocols. Au1 presents a design strategy that commences with problem-finding
activities (F-Geometry Sub Goal and F-Algorithm Sub Goal) in both the geometry
and algorithm activities before creating rules and then ending with evaluation (E-
Geometry) in the geometry category. Conversely, Au3 starts using physical synthesis
(e.g. A-Rule) and generation (G-Generation) activities in the algorithm category, even
at the start of the design session. Au3 then evaluates and revisits the algorithmic
units. Au3’s approach may be related to either a “troubleshooting” approach or a
“solution-driven” strategy of design.

Table 4.4 The frequency of

Subclass Move CM (%)
move and CM of three
subclasses in the “Finding” Au1 F-Initial Goal 4 2 (50.0%)
category by Au1 and Au3 F-Geometry Sub Goal 17 15 (88.2%)
F-Algorithm Sub Goal 11 10 (90.9%)
Sum 32 27 (84.4%)
Au3 F-Initial Goal 2 1 (50.0%)
F-Geometry Sub Goal 16 12 (75.0%)
F-Algorithm Sub Goal 23 15 (65.2%)
Sum 41 28 (68.3%)
4.5 Application 103

F-Geometry Sub Goal

F-Algorithm Sub goal

A-Change Parameter
A-Change Rule
A-Parameter

E-Geometry
A-Rule

E-Rule
Pa1

A-Change Rule

A-Change Rule

A-Change Rule
G-Generation

G-Generation

G-Generation
E-Geometry

E-Geometry

E-Geometry
Pa3

E-Rule
Fig. 4.7 Linkographs of the two design protocols (and details of a selected section of each as
circled)

The total number of design moves undertaken by Au1 and Au3 were 220 and
363, respectively. However, the former’s link index (2.75) is marginally higher than
the latter’s (2.68). The percentages of all three types of critical moves—according to
Goldschmidt (1995), CM5 , CM6 and CM7 which each have more than five, six and
seven links, respectively—in Au1’s protocol are also higher than in Au3’s. According
to the rationale of Goldschmidt (1995) and Kan and Gero (2008), these figures
suggest that Au1’s design protocol is more productive than Au3’s. However, to better
understand and compare the structural characteristics of the two cognitive processes,
this test measures forelink and backlink entropies which are indicators of structural
complexity.
Table 4.5 shows the results of entropy calculations derived from the two design
protocols. In both protocols, backlink entropy is higher than forelink entropy. This

Table 4.5 Entropy of the two design protocols

Forelink total H Backlink total H Horizonlink total H Cumulative total
Au1 35.23 52.56 14.17 101.96
Au3 36.18 62.07 16.18 114.43
104 4 Measuring Cognitive Complexity

Table 4.6 Entropy per move and structural complexity (H S ) of the two design protocols
Forelink H Backlink H Horizonlink H HS
Au1 0.161 0.239 0.064 0.464
Au3 0.100 0.171 0.045 0.315

suggests that design activities in parametric environments are focused on responding

to or restructuring existing components. Backlink entropy measures the opportunities
arising from enhancements or responses (Kan and Gero 2008). Table 4.6 records the
structural entropy per move of the two design protocols. The total cumulative entropy
per move agrees with the results of each individual link index as well as the results of
the content complexity in Table 4.3. Consequently, for comparing design protocols
with different durations, the structural-entropy-per-move measure can be effective
at identifying individual structural complexity. Thus, the cumulative total per move
would be regarded as the structural complexity (H S ) of each design process (Au1:
0.464 and Au3: 0.315).
Decile growth plots of linkographs developed in this chapter further illuminate
the variation of entropies that arise from growing complexity over time. Table 4.7
contains the data for entropies per move over decile time in the two design proto-
cols. The decile entropy value is also normalised to effectively show its variation.
Figure 4.8 represents the normalised value of the three types of entropies (i.e. fore-
links, backlinks and horizonlinks) on the decile timeline and the sequential changes
that also describe their complexity patterns over time. Although the values of the
first two entropy types would increase, corresponding to the ten growths of each
graph, “horizonlink entropy” illustrates similar patterns (a single peak over the
session) between the two protocols. Because Kan and Gero (2008) propose that
horizonlink entropy could reveal opportunities relating to cohesiveness and incuba-
tion in design, it may be conceptually related to the perceptual level of cognition in
Fig. 4.6. However, the relation is weak considering the relatively small amount of
the frequency coverage of perceptual activities in Table 4.3.
The results of the entropy calculations (Table 4.7 and Fig. 4.8) are derived from
the structures of linkographs in Fig. 4.7. For example, the peaks of the normalised
coverages of both horizonlink entropies in Fig. 4.8 are closely mapped to the largest
“chunks” of densely interlinked segmentations in both protocols. In addition, some-
what smaller “chunks” of link concentrations at the later stage of Au1’s protocol in
Fig. 4.8 may cause forelink entropy coverage to be higher than backlink entropy.
This may imply that the backlinks relate to dense, larger “chunks” in linkographs.
In addition, both protocols show that the coverage of forelink entropy in Fig. 4.8
is often higher than the backlink entropy, when the coverage of conceptual activ-
ities in Fig. 4.6 is higher than in physical activities. Forelink entropy measures
idea-generation opportunities (Kan and Gero 2008). One of the dominant activ-
ities in the conceptual level of cognition is E-Geometry (visually evaluating the
outcome of a rule), and it usually contributes to the construction of backlinks. In
contrast, problem-finding activities at the conceptual level of cognition relate to the
 4.5 Application

Table 4.7 Entropies per move over decile time in the two design protocols
Au1 Au3
Decile time Forelink Backlink Horizonlink Cumulative Forelink Backlink Horizonlink Cumulative
1 0.0209 0.0284 0.0186 0.0679 0.0317 0.0512 0.0263 0.1092
2 0.0593 0.0826 0.0427 0.1846 0.0426 0.0760 0.0397 0.1583
3 0.0873 0.1224 0.0579 0.2676 0.0599 0.1053 0.0535 0.2187
4 0.1094 0.1609 0.0667 0.3370 0.0679 0.1215 0.0505 0.2399
5 0.1190 0.1918 0.0898 0.4006 0.0807 0.1291 0.0484 0.2582
6 0.1236 0.2042 0.0798 0.4076 0.0779 0.1404 0.0488 0.2671
7 0.1371 0.2175 0.0752 0.4298 0.0882 0.1539 0.0490 0.2911
8 0.1551 0.2271 0.0699 0.4521 0.095 0.1605 0.0474 0.3029
9 0.1658 0.2338 0.0675 0.4671 0.0983 0.1669 0.046 0.3112
10 0.1601 0.2389 0.0644 0.4634 0.0997 0.171 0.0446 0.3153
105
106 4 Measuring Cognitive Complexity

1.5

1.0
Au1
Normalised coverage

0.5

0.0
1 2 3 4 5 6 7 8 9 10
-0.5 Forelink
-1.0 Backlink
Horizonlink
-1.5 Decile Time entropy
-2.0

-2.5

1.5

1.0
Au3
Normalised coverage

0.5

0.0
1 2 3 4 5 6 7 8 9 10
-0.5
Forelink
-1.0
Backlink
-1.5
Horizonlink
-2.0 Decile Time entropy
-2.5

-3.0

Fig. 4.8 Changes of normalised values of three types of entropies over time by Au1 and Au3

coverage of forelink entropy, supporting Kan and Gero’s hypothesis. Ultimately,

these results, while limited in scope, identify that the entropy measures are able to
capture complex combinations of cognitive activities and reveal various sequential
patterns of cognition and their structural complexities in the design process.

4.6 Conclusion

This chapter presents an entropy measure to quantify cognitive complexity in the

design process. Cognitive complexity has two characteristics, cognitive properties
and relationships, that can be measured using the content and structure of a design
process. This chapter introduces these concepts and a way of understanding the
design process in terms of complexity. It also offers a combined approach of protocol
analysis and linkography for its quantitative measurement.
Two complexity measures, content complexity and structural complexity (both
based on information entropy), are examined in this chapter. Both content and struc-
tural complexity values developed from our data are consistent with the individual
4.6 Conclusion 107

design strategies and design thinking processes exhibited by participants in the design
experiments. Furthermore, the normalised coverage values of cognitive activities and
entropies over time allow for capturing patterns of cognitive content and structure.
The complexity values developed in this way can also be easily compared to other
findings (for example, the results of product evaluation). In this way, the proposed
measures presented here are applicable to broader design research and practice.
Finally, although this chapter presents and examines a method for quantitatively
measuring cognitive complexity, these measures are limited to indicating the degree
of individual cognitive difference. Design thinking is dynamic and individualistic,
and it is difficult to interpret the measures of complexity developed here without a
clear scale and standard to compare them with. Categorisation of cognitive activi-
ties is also a challenge when measuring information entropy. Thus, further research
is required to verify variables and categorisations. For this reason, the complexity
measures developed in this chapter are only suitable for comparative purposes. While
the method is consistent and repeatable, further studies are required to develop
meaningful scaled content and structural complexity.

References

Aish, Robert, and Robert Woodbury. 2005. Multi-level interaction in parametric design. In Smart
Graphics, ed. Andreas Butz, Brian Fisher, Antonio Krüger, and Patrick Olivier, 924-924. Lecture
Notes in Computer Science. Berlin/Heidelberg: Springer.
Akin, Ö. 1986. Psychology of Architectural Design. London: Pion.
Almeida, Clarissa Ribeiro Pereira de, Anja Pratschke, and Renata La Rocca. 2005. In-between and
through: Architecture and complexity. International Journal of Architectural Computing 3 (3):
335–354. https://doi.org/10.1260/147807705775377311.
Amabile, Teresa M. 1983. The Social Psychology of Creativity. Springer series in social psychology.
New York: Springer.
Bansiya, Jagdish, Carl Davis, and Letha Etzkorn. 1999. An entropy-based complexity measure for
object-oriented designs. Theory and Practice of Object Systems 5 (2): 111–118.
Baranger, M. 2001. Chaos, Complexity, and Entropy: A Physics Talk for Non-physicists. Cambridge,
MA: New England Complex Systems Institute.
Benet-Martínez, Verónica, Fiona Lee, and Janxin Leu. 2006. Biculturalism and cognitive
complexity: Expertise in sations. Journal of Cross-Cultural Psychology 37 (4): 386–407. https://
doi.org/10.1177/0022022106288476.
Bieri, James. 1955. Cognitive complexity-simplicity and predictive behavior. The Journal of
Abnormal and Social Psychology 51 (2): 263.
Bilda, Z., and J.S. Gero. 2007. The impact of working memory limitations on the design process
during conceptualization. Design Studies 28 (4): 343–367.
Bonvillain, Nancy. 2010. Language, Culture and Communication: The Meaning of Messages. New
York: Pearson.
Bowerman, M. 1996. Which way to the present? Cross-linguistic differences in thinking about time.
In Rethinking linguistic relativity, ed. J. Gumperz and S. Levinson, 145–176. Cambridge, MA:
Cambridge University Press.
Burleson, B.R., and S.E. Caplan. 1998. Cognitive complexity. In Communication and Personality:
Trait Perspectives, ed. J.C. McCroskey, J.A. Daly, M.M. Martin, and M.J. Beatty, 233–286.
Cresskill, NJ: Hampton Press.
108 4 Measuring Cognitive Complexity

Cai, Hui, Ellen Yi-Luen. Do, and Craig M. Zimring. 2010. Extended linkography and distance graph
in design evaluation: An empirical study of the dual effects of inspiration sources in creative
design. Design Studies 31 (2): 146–168. https://doi.org/10.1016/j.destud.2009.12.003.
Chai, Kah-Hin., and Xin Xiao. 2012. Understanding design research: A bibliometric analysis of
design studies (1996–2010). Design Studies 33 (1): 24–43.
Chan, Chiu-Shui. 2015. Introduction of design cognition. In Style and Creativity in Design, ed.
Chiu-Shui. Chan, 9–78. Cham: Springer International Publishing.
Chien, Sheng-Fen., and Ulrich Flemming. 2002. Design space navigation in generative design
systems. Automation in Construction 11 (1): 1–22. https://doi.org/10.1016/S0926-5805(00)000
84-4.
Dorst, Kees, and Judith Dijkhuis. 1995. Comparing paradigms for describing design activity. Design
Studies 16 (2): 261–274. https://doi.org/10.1016/0142-694X(94)00012-3.
Earl, C., J. Johnson, and C. Eckert. 2005. Complexity. In Design Process Improvement. A Review
of Current Practice, ed. J. Clarkson, and C. Eckert, 174–197. London: Springer.
Eastman, C. M. 1969. Cognitive processes and ill-defined problems: a case study from design.
In Proceedings of the 1st international Joint Conference on Artificial Intelligence, 669–690.
Washington, DC: Morgan Kaufmann Publishers Inc.
Frey, Dan, and Ebad Jahangir. 1999. Differential entropy as a measure of information content in
axiomatic design. Las Vegas, USA: Paper presented at the ASME Design engineering technical
conference.
Gero, John S., and Thomas Mc Neill. 1998. An approach to the analysis of design protocols. Design
Studies 19 (1): 21–61.
Gero, John S. 1990. Design prototypes: a knowledge representation schema for design. AI Magazine
11 (4): 26–36.
Gero, John S., and Udo Kannengiesser. 2014. The function-behaviour-structure ontology of design.
In An Anthology of Theories and Models of Design: Philosophy, Approaches and Empirical
Explorations, ed. Amaresh Chakrabarti, and Lucienne T.M. Blessing, 263–283. London: Springer
London.
Goldschmidt, Gabriela. 1990. Linkography: Assessing Design productivity. In Cyberbetics and
System ‘90, World Scientific, ed. R. Trappl, 291–298. Singapore: World Scientific.
Goldschmidt, Gabriela. 2014. Linkography: Unfolding the Design Process. Cambridge, MA: MIT
Press.
Goldschmidt, Gabriela. 1995. The designer as a team of one. Design Studies 16 (2): 189–209.
https://doi.org/10.1016/0142-694X(94)00009-3.
Goldschmidt, Gabriela, and Dan Tatsa. 2005. How good are good ideas? Correlates of design
creativity. Design Studies 26 (6): 593–611.
Herskovits, A. 1986. Language and Spatial Cognition: A Interdisciplinary Study of the Prepositions
in English. Cambridge, England: Cambridge University Press.
Hunt, Kellogg W. 1970. Syntactic Maturity in schoolchildren and adults. Monographs of the Society
for Research in Child Development 35 (1): iii–67. https://doi.org/10.2307/1165818.
Johnson-Laird, P.N. 1983. Mental Models: Towards a Cognitive Science of Language, Inference,
and Consciousnes. Cambridge, MA: Harvard University Press.
Kan, Jeff W. T., and John S. Gero. 2008. Acquiring information from linkography in protocol studies
of designing. Design Studies 29 (4): 315–337. https://doi.org/10.1016/j.destud.2008.03.001.
Kelly, G.A. 1955. Personal Construct Psychology. New York: Norton.
Khan, W.A., and J. Angeles. 2007. The role of entropy in design theory and methodology. Paper
presented at the CDEN, C2E2. Alberta: Winnipeg.
Kim, J.E., C. Bouchard, J.F. Omhover, and A. Aoussat. 2010. Towards a model of how designers
mentally categorise design information. CIRP Journal of Manufacturing Science and Technology
3 (3): 218–226.
Kim, Mi Jeong, and Mary Lou Maher. 2008. The impact of tangible user interfaces on spatial
cognition during collaborative design. Design Studies 29 (3): 222–253. https://doi.org/10.1016/
j.destud.2007.12.006.
References 109

Krus, Petter. 2013. Information entropy in the design process. In ICoRD’13: Glocal Product Devel-
opment, ed. Amaresh Chakrabarti and Raghu V. Prakash, 101–112. New Delhi, Heidelberg, New
York: Springer India.
LaFleur, Ronald S. 1996. Entropy measures in engineering design. In Mechanical Design: Theory
and Methodology, ed. Manjula B. Waldron, and Kenneth J. Waldron, 275–298. New York, NY:
Springer New York.
Lee, Ju Hyun, Ning Gu, Julie Jupp, and Sue Sherratt. 2014a. Evaluating creativity in parametric
design processes and products: A pilot study. In Design Computing and Cognition ‘12, ed. John
S. Gero, 165–183. Dordrecht: Springer Netherlands.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2015. Creativity and parametric design? Comparing
designer’s cognitive approaches with assessed levels of creativity. International Journal of Design
Creativity and Innovation 3 (2): 78–94. https://doi.org/10.1080/21650349.2014.931826.
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014. Parametric design strategies for the generation
of creative designs. International Journal of Architectural Computing 12 (3): 263–282. https://
doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, and Michael J. Ostwald. 2019. Measuring cognitive complexity in parametric design.
International Journal of Design Creativity and Innovation 7 (3): 158–178. https://doi.org/10.
1080/21650349.2018.1555492.
Lee, Ju Hyun, Michael J. Ostwald, and Gu. Ning. 2016. The language of design: Spatial cognition
and spatial language in parametric design. International Journal of Architectural Computing 14
(3): 277–288. https://doi.org/10.1177/1478077116663350.
Levinson, S.C. 1996. Frames of reference and Molyneux’s question: Crosslinguistic evidence. In
Language and Space, ed. P. Bloom and M. Peterson, 109–169. Cambridge, MA: MIT Press.
Lewis, Rhodri. 2012. Language, Mind and Nature. Cambridge: Cambridge University Press.
Lopez, J. 2003. Society and Its Metaphors: Language, Social Theory and Social Structure. New
York: Continuum.
Lu, Xiaofei. 2010. Automatic analysis of syntactic complexity in second language writing.
International Journal of Corpus Linguistics 15 (4): 474–496.
Munnich, Edward, Barbara Landau, and Barbara Anne Dosher. 2001. Spatial language and spatial
representation: A cross-linguistic comparison. Cognition 81 (3): 171–208. https://doi.org/10.
1016/S0010-0277(01)00127-5.
Norman, D.A. 1983. Some observations on mental models. In Mental Models, ed. D. Gentner and
A.L. Stevens, 7–14. Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
O’Keefe, Daniel J., and Howard E. Sypher. 1981. Cognitive complexity measures and the relation-
ship of cogntive complexity to communication. Human Communication Research 8 (1): 72–92.
https://doi.org/10.1111/j.1468-2958.1981.tb00657.x.
Ortega, Lourdes. 2003. Syntactic complexity measures and their relationship to L2 proficiency: A
research synthesis of college-level L2 writsing. Applied Linguistics 24 (4): 492–518. https://doi.
org/10.1093/applin/24.4.492.
Perttula, Matti, and Pekka Sipilä. 2007. The idea exposure paradigm in design idea generation.
Journal of Engineering Design 18 (1): 93–102. https://doi.org/10.1080/09544820600679679.
Pervin, L.A. 1975. Personality: Theory, Assessment, and Research. Canada: Wiley.
Pourmohamadi, M., and J. S. Gero. 2011. LINKOgrapher: An analysis tool to study design protocols
based on FBS coding scheme. Paper presented at the DS 68–2: Proceedings of the 18th Inter-
national Conference on Engineering Design (ICED 11), Impacting Society through Engineering
Design, Vol. 2: Design Theory and Research Methodology, Lyngby/Copenhagen, Denmark.
Salim, Flora, and Jane Burry. 2010. Software openness: Evaluating parameters of parametric
modeling tools to support creativity and multidisciplinary design integration. In Computational
Science and Its Applications—ICCSA 2010, ed. David Taniar, Osvaldo Gervasi, Beniamino
Murgante, Eric Pardede, and Bernady Apduhan, 483–497. Lecture Notes in Computer Science.
Berlin, Heidelberg: Springer.
Shannon, C.E. 1948. A mathematical theory of communication. Bell System Technical Journal 27
(3): 379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x.
110 4 Measuring Cognitive Complexity

Stempfle, Joachim, and Petra Badke-Schaub. 2002. Thinking in design teams—An analysis of team
communication. Design Studies 23 (5): 473–496. https://doi.org/10.1016/S0142-694X(02)000
04-2.
Suh, Nam P. 1999. A theory of complexity, periodicity and the design axioms. Research in
Engineering Design 11 (2): 116–132. https://doi.org/10.1007/pl00003883.
Suwa, M., T. Purcell, and J. Gero. 1998. Macroscopic analysis of design processes based on a
scheme for coding designers’ cognitive actions. Design Studies 19 (4): 455–483.
Suwa, Masaki, and Barbara Tversky. 1997. What do architects and students perceive in their design
sketches? A protocol analysis. Design Studies 18 (4): 385–403. https://doi.org/10.1016/S0142-
694X(97)00008-2.
Tenbrink, Thora. 2015. Cognitive Discourse analysis: Accessing cognitive representations and
processes through language data. Language and Cognition 7 (1): 98–137. https://doi.org/10.
1017/langcog.2014.19.
Tenbrink, Thora, and Marco Ragni. 2012. Linguistic principles for spatial relational reasoning. In
Spatial Cognition VIII, ed. Cyrill Stachniss, Kerstin Schill, and David Uttal, 279-298. Lecture
Notes in Computer Science. Berlin/Heidelberg: Springer.
Van der Lugt, Remko. 2000. Developing a graphic tool for creative problem solving in design
groups. Design Studies 21 (5): 505–522. https://doi.org/10.1016/S0142-694X(00)00021-1.
van der Zee, E., and J.M. Slack. 2003. Representing Direction in Language and Space. Oxford
University Press.
Weinstock, Michael. 2010. The Architecture of Emergence: The Evolution of Form in Nature and
Civilisation. Chichester, UK: Wiley.
 Part II
Collaboration
Chapter 5
Collaborative Design: Team Cognition
and Communication

Abstract This chapter examines two cognitive issues in collaborative design: team
cognition and communication. It commences with a detailed review of past research,
before developing a new framework for design team cognition. This framework is
built around an understanding of individual, team and distributed mental models and
cognitive behaviours, including a consideration of different modes of communication
and working. This framework is elaborated, and its implications for team cognition
and communication are examined, through a discussion of the results of two exper-
iments. The first uses the results of a protocol study of cognitive traits to reveal
the importance of distributed mental models. The second uses protocol data from
a study of cross-national teams to investigate design communication. This chapter
contributes to the development of new methodologies for examining design teams and
to advanced knowledge about design cognition and representation. These advances
are potentially significant for supporting creativity and cross-cultural collaboration
in teams.

5.1 Introduction

A team can be defined as “a distinguishable set of two or more people who interact,
dynamically, interdependently, and adaptively toward a common and valued goal,
[and] who have each been assigned specific roles or functions to perform” (Salas
et al. 1992, p. 4). Design practice is typically a team process wherein a collabo-
rative environment is the setting for complex problem-solving. A design team is a
designated group who are operating collaboratively and synergistically to achieve
a singular vision or outcome. The team typically comprises two or more profes-
sionals with different areas of expertise, levels of experience, innate and developed
design strategies and working practices. In the design industries, the team typically
work together on a project-by-project basis to produce solutions for clients. These
solutions are achieved through the managed application of individual and collec-
tive strengths and abilities. A related definition describes collaborative design as the
process wherein team members “from different disciplines share their knowledge
about both the design process and the design content” (Kleinsmann 2006, p. 30). By

© Springer Nature Switzerland AG 2020 113

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_5
114 5 Collaborative Design: Team Cognition and Communication

definition, the creation of a “shared understanding” in a design team is an essential

component for collaborative design and as such, developing a deeper understanding
of knowledge creation, integration and communication between team members, is
essential for improving collaborative design (Kleinsmann 2006).
Design thinking in collaborative environments encompasses multiple, simulta-
neous cognitive activities and processes. As discussed in previous chapters, cognitive
design processes are not always easy to identify or study. They are largely intuitive,
contingent and subliminal, because design itself is a conceptual, creative and strategic
process that evolves out of an internalised cognitive system. That is, individual cogni-
tion is not always clear, not even to the designer. A further complicating factor occurs
in teamwork, where cognitive processes and their results must be communicated to,
or shared with, team members. To be effective, the team needs to engage collec-
tively in this process of cognitive transference and processing, which may require a
change in individual or collective Mental Models (MMs). As such, improving team
performance and effectiveness in design practice requires a dynamic, collective and
situated understanding of design cognition (Marshall 2007). It also necessitates a
consideration of the construction and communication of socially held representa-
tions (Dong 2005). This is why past research in this field emphasises the importance
of understanding team cognitive and communication processes (Badke-Schaub et al.
2007; Bierhals et al. 2007; Dong et al. 2013; Du et al. 2012). Indeed, these two
issues, team cognition and design communication, are so closely connected in past
research that it is difficult to completely separate them. Thus, this chapter explores
both concepts in the operations of the design team.
It must also be acknowledged that there is a tension implicit in the relation-
ship between cognition and communication in design teams. In most team-based
circumstances, it might be assumed that transparent communication and achieving
perfect shared understanding are core to effective team processes, in part because
they support the management of tasks and resources (Du et al. 2012). Such assump-
tions, not unreasonably, are the catalyst for Valkenburg and Dorst (1998) to argue
that a lack of shared understanding in a team causes unnecessary iterative loops and
delays. Consequently, “team performance improves if team members have a shared
understanding of the task that is to be performed and of the involved team work”
(Jonker et al. 2011, p. 132). While this might appear to be a reasonable proposition,
a counter argument is that possessing an overly shared understanding can actually
hinder creativity.
Creative problem-solving techniques typically encourage and reward diverse
thought processes. This is because collective cognitive structures, such as shared
cognitive maps, belief structures and MMs in a team, may cause “individuals to
ignore discrepant information and may inhibit creative problem solving” (Klimoski
and Mohammed 1994, p. 405). In order to explain the tension between communica-
tion and cognition in teams, Langan-Fox et al. (2001) contrast the notions of a “Team
Mental Model” (TMM) and a “Shared Mental Model” (SMM). The former refers
to a diverse, rich, collaborative cognitive process, whereas the second describes a
more homogenous cognitive model, like “group think”, which they call a “collectiv-
ity”. Whereas TMMs contribute to social creativity, SMMs can stifle expression and
5.1 Introduction 115

novelty (Fischer et al. 2005). Such arguments lead to the view that diverse, collab-
orative cognitive processes can be more creative than shared or collective ones. As
a counterpoint to this position, Banks and Millward (2009) call for the development
of a “Distributed Mental Model” (DMM), which negates the effects of collectivity
through effective communication. The DMM is also an effective reminder that the
situation is not a binary (either/or) one. A theme in many of these models is that
in order to support team performance and creativity, design teams should possess a
balanced mixture of shared and distributed team cognitions. But beyond this general
consensus, there remains debate about how we can develop a deeper understanding
of what this means.
The development of new knowledge to support improved design teamwork skills
and abilities has been the subject of several empirical studies (Chiu 2002; Cross
and Cross 1995; Mulet et al. 2016; Stempfle and Badke-Schaub 2002; Eppler and
Kernbach 2016). Despite this, there remain substantial gaps in our knowledge of
design team cognitive processes. Furthermore, there are two fundamental challenges
that research into design teamwork faces. First, the environment where the team
undertakes a given design task has an impact on both the way a team works and on
its capacity to be creative (Bilda and Demirkan 2003; Ibrahim and Rahimian 2010;
Mulet et al. 2016). Second, communication barriers—not only visual and verbal but
also cognitive and linguistic—complicate any empirical approach to the topic (Dong
2005; Lee et al. 2016; Valkenburg and Dorst 1998). For this reason, research into team
cognition should not only consider the impacts of emerging design environments,
but also individual differences in communication skills. These factors have become
important in recent years as digital design environments have become the common
interface for large multi-national design teams.
The design environment comprises the set of creative and communicative tools,
platforms, media, systems and locations that support the operations of individuals
and teams. While historically, the design team was almost always co-located, physi-
cally sharing the same space, tools and systems, over the last few decades advances in
Information and Communication Technology (ICT) have broadened the definitions
of collaboration and teamwork, to include a range of Computer Supported Coop-
erative Work (CSCW) tools. Such CSCW tools tend to be optimised for efficient
teamwork, and advances in this field have tended to be focussed on creating appro-
priate environments for collaboration and supporting asynchronous communication
(Chiu 2002; Cross and Cross 1995; Grudin 1994; Ibrahim and Rahimian 2010). As a
result of the success of CSCW applications, they have become core to the cognitive
and communicative operations of design teams around the world.
Against this backdrop, the present chapter undertakes a review of literature on
cognition and representation, before proposing a mixed Design Team Cognition
(DTC) model to conceptualise both mental and communicative processes in design.
Thereafter protocol data (from Study II in Chap. 2) is revisited to explore individual
cognitive patterns in different design spaces and modes and then expanded to consider
their impact on DTC. The spaces are the problem and solution spaces of design
strategy and cognition, and the modes or environments are geometric and algorithmic
modelling in parametric design. The chapter then employs data from two additional
116 5 Collaborative Design: Team Cognition and Communication

design protocols, one from an Australian design session and another from a Swedish
session, to investigate individual communicative differences in design information
and spatial language.
The studies in this chapter are not, however, its core contribution, they serve as
a counterpoint and as a catalyst for discussion, rather than for drawing conclusions.
These are studies of individuals, working in environments that are optimised for
team processes and to enable the development of CSCW applications for sharing
and managing design information, coding and visualisation (Holzer et al. 2007;
Kolarevic 2003). The results of these studies are used to highlight cognition and
communication in design teams, and their implications are analysed using the DTC
model.

5.2 Cognition and Representation for Teamwork

5.2.1 Team Cognition

The phrase “team cognition”, not unexpectedly, refers to the set of team members’
cognitive operations, perceptions, reasoning, conscious thought and MM, among
other things. The relationship between an individual’s cognitive operations and that
of a team they are a member of, is a rich one, which has been linked to creativity
(Fischer et al. 2005). The emergence and sharing of creative activities and cognitions
in a social environment is also known as “co-creation” (Giaccardi 2004, 2005).
The topic of team cognition has been extensively researched in the fields
of psychology, sociology and education. From this past research DeChurch and
Mesmer-Magnus (2010) identify two major cognitive constructs of team cogni-
tion: the Mental Model (MM) and Transactive Memory (TM). The former refers
to the knowledge commonly held by team members, while the latter describes the
knowledge distributed among team members.
A MM is a “mechanism whereby humans generate descriptions of system purpose
and form, explanations of system functioning and observed system states, and predic-
tions of future system states” (Rouse and Morris 1986, p. 360). MMs contribute to
“representing objects, states of affairs, sequences of events, the way the world is,
and the social and psychological actions of daily life”. They support people to fore-
cast results, draw inferences, “to understand phenomena, to decide what action to
take and to control its execution” (Johnson-Land 1983, p. 397). An effective MM
can be used by team members to describe, explain and predict events and situations
(Mathieu et al. 2000). Importantly, MMs are not universal: “there can be (and prob-
ably would be) multiple MMs co-existing among team members at a given point in
time”. (Klimoski and Mohammed 1994, p. 432). A MM must be shared among team
members before it can be tested, adopted or applied, and complex tasks will often
require the application of multiple MMs (Cannon-Bowers et al. 1993).
5.2 Cognition and Representation for Teamwork 117

The second cognitive construct, TM, refers to the collective capacity of each
member of a team to store and share information. It encapsulates an awareness of
different areas of expertise that can be accessed as required, and as such it has
been described as being analogous to memory in a computer (Wegner et al. 1991).
It is also closely related to notions of “cognitive diversity” (Sauer et al. 2006) and
DMMs (Banks and Millward 2009) in a complex task environment. Austin (2003), for
example, reveals that TM impacts on group performance in terms of both volume of
knowledge available (both stored and specialisation) and capacity for consensus and
accuracy (transactive processes). TM also operates in support of diversity, because
access to specialised knowledge and a capacity to share it with team members are
often prerequisites for complex problem-solving.
As these descriptions of MM and TM suggest, they can, and probably should,
occur simultaneously in effective teams. MM provides a basis for convergent knowl-
edge and the management of tasks of processes. TM supports access to its divergent
counterpart, specialisation and diversity, which potentially leverages team produc-
tivity and creativity. This realisation, that MM and TM are both operative in a team
process, also highlights that they are co-dependent. Teams consisting of interdepen-
dent, diverse individuals face significant challenges in terms of shared responses to
management and performance (Santandreu Calonge and Safiullin 2015). Managing
or developing a team’s MMs is even “more important whenever diverse knowledge
has to be coordinated” (Badke-Schaub et al. 2007, p. 14). Thus, in a balanced combi-
nation, MM and TM enhance team cognition in collaborative environments and
especially those involving multidisciplinary and multifunctional teams and complex
design tasks.
In cognitive science, SMMs and TMMs have been employed to examine the
influence of MM and TM on teamwork processes and performance (DeChurch and
Mesmer-Magnus 2010; Holyoak 1984; Klimoski and Mohammed 1994; Langan-Fox
et al. 2001; Mathieu et al. 2000; Johnson-Land 1983). Problematically though, in past
research, the difference between SMMs and TMMs is not always clear and they are
often used interchangeably. As such, for the purposes of the present chapter, precise
definitions of these are less important than understanding the themes they articulate.
The TMM, for example, can be used to refer to an agreed MM or TM, akin to a type
of “collectivity”, whereas the SMM can capture “shared cognition among dyads of
individuals” (Langan-Fox et al. 2001, p. 99). This means that the SMM can be used to
describe the way multiple individuals can possess similar cognitive representations
of a situation or phenomenon, without a formal agreement between them. Unlike
a pure “collectivity” or “group mind”, SMMs function to “allow team members to
draw on their own well-structured knowledge as a basis for selecting actions that
are consistent and coordinated with those of their teammates” (Mathieu et al. 2000,
p. 274). SMMs allow team members to predict what their teammates are going to do
or require, allowing them to effectively conduct a task without the need for formal
communication (Cannon-Bowers et al. 1993). They may not represent all individual
MMs, but SMMs can at least partially describe the individual information processes
conceptually involved in “group information processing” (Hinsz et al. 1997; Tindale
1989). Whereas TMMs may be employed to describe agreed cognitive states, like a
collective MM, SMMs describe differences, transferences and negotiations, akin to
118 5 Collaborative Design: Team Cognition and Communication

TM. Such distinctions are not always agreed in other fields, and in design research,
as the next few sections reveal, they are often not considered at all.
In the design domain, the notion of “team cognition” has been used to explain
the relationship between TMMs and collective processes and performances (Badke-
Schaub et al. 2007; Bierhals et al. 2007; Boos 2007; Casakin and Badke-Schaub 2015;
Dong et al. 2013; Marshall 2007). Past research argues that teams are “more suit-
able for complex tasks because they allow members to share the workload, monitor
the work behaviours of other members, and develop and contribute expertise on
subtasks” (Mathieu et al. 2000, p. 273). Early definitions of team cognition in design
tended to emphasise properties and behaviours of teams akin to the “group mind”
or “collectivity” models of psychology and sociology. In a narrow sense, a collec-
tive approach of this type, supported by transactional memory and socio-cognition, is
effective because it has a common purpose and shared frames of reference (Klimoski
and Mohammed 1994). However, if the team’s purpose is to undertake complex and
creative, multi-dimensional problem-solving, say for a design, then such variations
of TMM are not enough.
Effective collaborative design naturally requires both diverse participants and the
close coordination of design information and tasks, because its purpose is to share
expertise, ideas, resources and responsibilities (Chiu 2002). Team organisation also
has an impact on communication and performance, and Bierhals et al. (2007) reveal
that design team performance is enabled by SMMs. Marshall’s (2007) practice-based
approach further highlights social, dynamic and emergent characteristics of shared
knowledge in design teams. These factors and many others have made TMMs and
MMs in design more difficult to understand and model.
In a pioneering study on TMMs in design, Badke-Schaub et al. (2007) identify
five types of MMs: team, process, task, context and competence.
1. Team-MMs deal with coordination and communication and are most closely
associated with operational efficiency. They are important to ensure that resources
and expertise are appropriately deployed and managed to achieve a particular
outcome in a given timeframe or within defined performance parameters.
2. Process-MMs could be regarded as a subset of team-MMs, but rather than empha-
sising holistic structures and systems, they help manage distributed tasks in the
design process. For example, they might serve to share individual designers’
steps and strategies with the team. Conceptually, process-MMs highlight indi-
vidual cognitive operations in design in a manner reminiscent of a DMM (Banks
and Millward 2009).
3. Task-MMs share specific problems or challenges across the team with the goal
of producing a common solution. Although using diverse knowledge to solve a
given design problem can lead to divergent thinking and creativity, task-MMs
are shared within the scope of the information process in a team, making them
more akin to a convergent thought process.
These first three MM types have been the subject of further research, whereas of
the last two identified by Badke-Schaub et al. (2007), context and competence have
tended to be undervalued or ignored.
5.2 Cognition and Representation for Teamwork 119

4. Context-MMs assist in managing and sharing knowledge about an organisational

setting or condition. They are used to communicate, store or retrieve a deep
understanding of, for example, clients, users, availability of materials, approval
processes, legal time-lines or market forces.
5. Competence-MMs relate to a capacity to understand the relative levels of experi-
ence or types of expertise in a team. A shared sense of the capacity and skills avail-
able to support a design process can be essential for motivating and coordinating
a team.

These five MM types provide a foundation for the DTC model proposed later in the
chapter.
Past research has used multiple team cognition attributes to understand problem-
solving behaviours (Stempfle and Badke-Schaub 2002) and collaborative design
attributes (Ostergaard and Summers 2003) (Table 5.1). The emergence of cogni-
tion has also been identified as an important indicator of effective teamwork. “Team
cognition is a bottom–up emergent construct, originating in the cognition of indi-
viduals” (DeChurch and Mesmer-Magnus 2010, p. 35). Cognition is said to emerge
in a team process when either “composition” or “compilation” occurs. The first of
these refers to the accuracy or congruity of a SMM. The second, “compilation” arises
when a SMM encapsulates relevant specialisation and task coordination in the team.
Across the multiple models of collective cognition, there are also three recur-
ring MM types: perceptual, structural and interpretive. Perceptual-MMs encompass
team members’ values, beliefs, attitudes and expectations. Structural-MMs highlight
existing and evolving knowledge patterns within a team. Interpretive-MMs support
the collective development of SMMs or DMMs.
If we compare the three main team cognition models—from Stempfle and Badke-
Schaub (2002), Ostergaard and Summers (2003) and DeChurch and Mesmer-Magnus
(2010)—there are several common themes (Table 5.1). For example, both the
problem-solving process in design (Stempfle and Badke-Schaub 2002) and collab-
orative processes (Kleinsmann 2006) are recurring themes. Task-related and team-
related cognitive processes also appear in Badke-Schaub et al.’s (2007) and DeChurch
and Mesmer-Magnus’ (2010) models. Further commonalities are present in the lists
of attributes for the content and process of team cognition. For example, Stempfle
and Badke-Schaub’s (2002) “content” includes goal clarification, solution genera-
tion, analysis, evaluation, decision and control, while Ostergaard and Summers’s
(2003) “nature of the problem” addresses design sub-tasks, abstraction, scope and
complexity. The overlaps between the multiple models also reflect the concept of the
“compatibility in cognitions”, which refers in part to those factors which are not only
common between models, but are complementary (DeChurch and Mesmer-Magnus
2010).
There are also some differences between the three models. For example, the
first three categories of collaborative design attributes identified in Ostergaard and
Summers’s (2003) research are associated with effective interactions between team
members. Team composition deals with team member relations and leadership, while
the distribution category covers personnel and information in collaborative design.
120 5 Collaborative Design: Team Cognition and Communication

Table 5.1 Team cognition attributes identified in the literature

Study Stempfle and Ostergaard and DeChurch and
Badke-Schaub Summers (2003) Mesmer-Magnus (2010)
(2002)
Focus Design team Collaborative Cognitive underpinnings
problem-solving design attributes
behaviour
Categories and Content Team composition Nature of emergence of
attributes Goal clarification, Group, Individual, cognition
Solution Team member Composition (congruence or
generation, relations, accuracy),
Analysis, Leadership Compilation (specialisation or
Evaluation, Communication global of
Decision, Control Mode, Quantity, complementarity/transactive
Process Syntax memory)
Planning, Analysis Proficiency of team, Form of cognition
Evaluation, Dependability of Perceptual, Structured,
Decision resources, Interpretive
Control Intent Content of cognition
Residual Distribution Task-related cognition,
Residual Personnel, Team-related cognition
Information
Design approach
Design tools,
Evaluation of
progress,
Degree of structure,
Process approach,
Stage
Information
Form, Management,
Level of criticality,
Dependability
Nature of the
problem
Type of design,
Coupling of
sub-tasks,
Level of
abstraction,
Scope, Complexity

Importantly, the communication category recognises the intent, mode and quantity of
the communication, the proficiency of the team, and the reliability or predictability of
resources. Different types of communication “may facilitate or hinder” the design or
team process. “It follows that some communication forms (verbal, written, graphic,
or gestures) may be better suited for use in particular tasks or phases of the design
process” (Ostergaard and Summers 2003, p. 759). Both task-related and team-related
5.2 Cognition and Representation for Teamwork 121

cognitions, or content and process of MMs, are fundamental to understanding how

team cognition for collaborative design is developed and shared.

5.2.2 Cognitive Representation

A MM is a representation of the way a system, process or event operates or occurs.

They are necessary abstractions or simplifications and tend to be based more on expe-
rience and intuition than facts. They do, however, allow individuals to understand
complex phenomena, draw inferences, and “experience events by proxy” (Johnson-
Land 1983). A MM can be regarded as a “psychological representation of the envi-
ronment and its expected behaviour” (Holyoak 1984, p. 193). It can be understood
both as a type of knowledge and as “a general class of cognitive constructs that
have been invoked to explain how knowledge and information are represented in the
mind” (Klimoski and Mohammed 1994, p. 405). MMs support team members to
describe, explain, predict and communicate future system states (Rouse and Morris
1986). TMMs capture representations that support collaborative (both collective and
distributed) cognitive processes (Klimoski and Mohammed 1994).
TMMs are fundamentally involved in, or instrumental to, communication. Paivio
(1971), for example, extends the concept of the TMM’s “imagery and verbal
processes” of representation to encapsulate “dual coding theory”, which describes
the situation that occurs when an encounter with an image and a word develops both
non-verbal and verbal representations and associations. He classifies the “non-verbal
(symbolic) subsystem” as the site of imagery and representation, “because its crit-
ical functions include the analysis of scenes and the generation of mental images”
(Paivio 1983, pp. 53–54). In contrast, spoken linguistic systems, or “verbal systems”
transmit representations and can be used to achieve confirmation of understanding.
Collectively the non-verbal subsystem and the verbal system (Paivio 1991), support
an in-depth understanding of individuals’ MMs and their cognitive representations.
These cognitive representations, which are expressed by visual and verbal language,
are the focus of the next two sections.

5.2.2.1 Visual Representation

After individual MMs are established, they must be exchanged and synchronised with
TMMs for effective collaboration (Badke-Schaub et al. 2007; Boos 2007). Sonnen-
wald (1996) indicates that the processes of exchange and synchronisation facilitate
“boundary sharing”, which plays a vital role in knowledge exchange, exploration
and task coordination. While language clearly supports design communication in
teamwork, visual representations, such as diagrams and models, make a signifi-
cant contribution to the development of SMMs or DMMs. Visual representations
are also intrinsic to “designerly ways of knowing” (Cross 1982). As such, it is
122 5 Collaborative Design: Team Cognition and Communication

not unexpected that the role of sketching in visual design thinking has been exten-
sively explored (Goldschmidt 1991, 1994; van der Lugt 2005). Furthermore, there
has been a sustained interest in design thinking in comparisons between sketching
(traditional design environment) and various computer-aided design environments
(Bilda and Demirkan 2003; Ibrahim and Rahimian 2010; Won 2001). The findings
of these studies often emphasise the differences between visual thinking in the two
environments (Won 2001).
In addition to a consideration of MMs in collaborative design, social processes,
technical tools and environments also have an impact on the design team (Cross
and Cross (1995). For example, Oxman (2004) highlights that an “explicit shared
representational schema” can be employed “to represent the conceptual content in
design precedents” in such a way that “helps to organise knowledge and to provide
structure” (p. 71). In addition, Gonçalves et al. (2014) identify fourteen approaches—
brainstorming, function analysis, scenarios, mind map, checklists, analogies, how-
to’s, storyboards, metaphors, collages, context mapping, morphological charts, role
playing and synectics—to help designer’s idea-generation tasks. The first three tech-
niques are common ones used by professional designers. Brainstorming enables idea
generation, while scenarios facilitate an overall understanding of design options.
“Function analysis represents a systematic analysis of the relationship between the
functions and the different parts of the future product” (Gonçalves et al. 2014, p. 43).
These visual thinking techniques potentially support both divergent and convergent
collaborative tasks (Eppler and Kernbach 2016). That is, design thinking in teams
relies on visual representations for idea generation as well as collaborative anal-
ysis and decision-making. These processes also reflect the team cognition attributes
identified in the previous discussion of TMMs, SMMs and DMMs (Table 5.1). In
summary, visual representation has a significant influence on the content and process
of collaborative design, as well as the development and application of TMMs.
The role of visual representation in cognition has been studied in multiple projects,
four of which are compared in this section (Table 5.2). The first examines the cogni-
tive dimensions of notations that are typically communicated using visuals. Green
and Petre (1996) suggest that seven dimensions are required to capture the cogni-
tive aspects of non-textual reasoning in Visual Programming (language) Environ-
ments (VPEs). The first of these, closeness of mapping, deals with the mapping of
program entities to problem entities, to ensure that usable languages for end-users
reflect task-specific entities. The second, viscosity or resistance to local change,
highlights the need for VPEs to be readily revised. The third, hidden dependen-
cies in programs may result in unexpected outcomes or flaws. VPEs (for example,
the box-and-wire representation of program and data flow in graphic-based scripts
in Fig. 4.3) can communicate the consequences of changes in textual programming
environments, and also increase cognitive load, leading to the fourth dimension, hard
mental operations, through their use of complex visual notations (“brain-twisters”).
Although VPEs provide more flexibility in programming than standard linear text-
based languages, they still require guess-ahead strategies, the next dimension, to
avoid creating “visual spaghetti”, a common flaw in “box-and-wire” VPEs. The
sixth dimension, secondary notation, describes visual systems that communicate
5.2 Cognition and Representation for Teamwork 123

Table 5.2 Visual representation attributes identified in the literature

Study Green and Petre Blackwell et al. Eppler (2004) Bresciani (2019)
(1996) (2001)
Focus Cognitive Cognitive Knowledge Visualisation
dimensions of dimensions of communication characteristics
notations notations
Attribute Closeness of Creative ambiguity Focus Structural
mapping Specificity Coordination restrictiveness
Viscosity Detail in context Documentation Content
Hidden Indexing Consistency modifiability
dependencies Synopsis Accountability Directed focus
Hard mental Free rides Traceability Perceived
operations Useful finishedness
Imposed awkwardness Outcome clarity
guess-ahead Unevenness Visual appeal
Secondary Lability Collaboration
notation Permissiveness support
Visibility

supplementary information. The final, visibility dimension, supports mapping of

effective search strategies to improve cognitive representations in VPEs. Blackwell
et al. (2001) extend Green and Petre’s (1996) cognitive dimensions to describe the
usability of notational systems and information artefacts, suggesting ten dimensions
which have some overlap, but also include additional nuanced readings of factors
like awkwardness and unevenness, which can support or prohibit transmission of
knowledge (Table 5.2).
The third visual representation model, from Eppler (2004), examines “knowledge
communication” in visualisation and the fourth, from Bresciani (2019), explores
its characteristics. Knowledge visualisation systems facilitate team communica-
tion and reasoning processes. Such systems are used to shape knowledge—to
abstract, diverge, converge, structure, elaborate, and evaluate—as part of the process
of building a TMM (Briggs et al. 2001). In this context, Eppler (2004) exam-
ines software-based, collaborative visual communication tools that serve cognitive
purposes through systematic interactive visualisation. Three visual communication
tools—the OnTrack visual protocol tool, the Parameter Ruler application and the
Synergy Map—are presented which support knowledge focussing, coordination,
documentation, accountability and tracking. Bresciani (2019) presents an alterna-
tive set of seven knowledge visualisation attributes of collaborative visualisation
systems—structural restrictiveness, content modifiability, directed focus, perceived
finishedness, outcome clarity, visual appeal and support for collaboration—that
control and enable specific cognitive and collaborative actions. Collectively these
attributes support the sequential tracking, progressive evaluation, revision and simul-
taneous modification of knowledge visualisation. Bresciani’s (2019) last dimension,
collaboration support describes “the ability of a visual representation’s format and
environment to facilitate the group interaction in the co-creation process” (Bresciani
2019, p. 106).
124 5 Collaborative Design: Team Cognition and Communication

These four models—Green and Petre (1996); Blackwell et al. (2001); Eppler
(2004); Bresciani (2019)—can be used to explain how visual representation works in
design communication as well as the externalisation of an individual MM. Thus, they
support knowledge sharing between teammates and facilitate team design cognitive
processes in collaborative design.

5.2.2.2 Verbal Representation

Past research on design communication has largely focused on visual representation,

with only a limited body of research concerned with verbal representation. However,
visualisation and verbalisation are both fundamental forms of design communication
and external representation of design thinking. Other, non-verbal representations, like
hand gestures, can also impact on design collaboration, but the visual and the verbal
systems are the primary means of expressing shared or reasoned information.
The structure of language-based communication is founded on socially shared
representations of understanding (Dong 2005). The degree of psychological correla-
tion between a “designer’s own cognitive representation and the socially held repre-
sentation of the designed artefact is reflected in the semantic coherence between
words in the way they co-occur in dialog and other language-based communication”
(Dong 2005, p. 447). This correlation of collective mental models and modes of
work, “facilitates the coordinated action that is required for successful team-based
design” (Dong 2005, p. 448). Language, in this sense, provides the “mental space” for
the creation or evocation of knowledge domains. Furthermore, language “builds up
mental spaces, relations between them, and relations between elements within them”
(Fauconnier 1985, p. 2). Communication is, therefore, dependent on the extent to
which people may “build up similar space configurations from the same linguistic and
pragmatic data” (Fauconnier 1985, p. 2). That is, people represent objects and their
relations within mental spaces and language expresses and articulate these represen-
tations (Bloom 1993). Thus, in cognitive science, these representations constitute a
type of “working memory” or “consciousness”.
As noted previously, Casakin and Badke-Schaub (2015) use three categories of
MMs to investigate different types of verbal activities in design team meetings:
Task-MMs, Process-MMs and Team-MMs. Interestingly, their research reveals that
a repeated verbal communication loop of task-MM and Team-MM, can reflect
a behavioural structural pattern in a design team. That is, effective design team
problem-solving processes require structured communication between members.
Such a structure accommodates both verbal and visual communication simulta-
neously. “Moreover, the language system is peculiar in that it deals directly with
linguistic input and output … while at the same time serving a symbolic function
with respect to nonverbal objects, events, and behaviours” (Paivio 1983, p. 53).
Therefore, understanding the relationship between spatial language and cognition is
a crucial process for design teams, and especially if there are differences in languages
spoken in the team (Lee et al. 2017).
5.2 Cognition and Representation for Teamwork 125

In summary, visual and verbal representations play a vital role in both design
communication and cognition. As visual and verbal representations also support
and sustain individual MMs, they must be clearly expressed between members for
effective design team performance. One aspect of this verbal and visual communi-
cation that is especially important in the construction and transmission of TMMs is
spatial-relation reasoning. Communicating spatial relations requires sufficient and
determinate information to define a relationship qualitatively (Tenbrink and Ragni
2012). Thus, the following sections of this chapter introduce a new approach to
exploring visual and verbal representations in the design process.

5.2.3 Design Team Cognition (DTC) Model

In past research, TMMs have been used to understand performance differences

between teams and also to describe knowledge shared by team members (Mohammed
and Dumville 2001). The research suggests that some specific TMMs, like “group
minds” and “knowledge structures”, can enhance “interpretive processes by enabling
individuals to screen out information in order to prevent information overload
and intolerable levels of uncertainty” (Klimoski and Mohammed 1994, p. 405).
In parallel, individual belief structures also enable the development of socially
shared cognitive processes (Resnick 1991). As previously noted, individual processes
are core to group processes (Hinsz et al. 1997). Notwithstanding distinctions
between individual, collective and distributed MMs, communication and informa-
tion processing across a group have a substantial impact on the development and
application of TMMs. The theory of DMMs, in particular, accommodates different
roles or expertise that cannot be easily shared. A DMM provides a means of under-
standing the capacity of a team to accommodate different perspectives (Banks and
Millward 2009). In this context, a DMM presents a “basis for understanding high-
level cognition in distributed systems, specifically cognition which relies on mental
models including reasoning, naturalistic decision making [and] the cognitive basis
of teamwork” (Banks and Millward 2009, p. 259). Thus, the combination of DMMs
and SMMs would appear to provide a foundation for exploring complex cognitions
in diverse design teams. Past research has considered this combination and how it
sustains team activities.
Dong et al. (2013) present a conceptual TMM for design teamwork that consists of
two levels: the cognitive realm and the realm of action. The first of these is concerned
with TMM quality, which is a factor of two, secondary concepts: (i) the emergence
of “sharedness” or “accordance” arising from the TMM; and (ii) the accuracy of the
TMM in respect to the real-world phenomenon, occurrence or event which it seeks
to model. The second realm, the realm of action, is concerned with the way shared
cognition is implemented or actioned. It addresses the goal-directed behaviour of
the TMM. The cognitive realm can be investigated using “latent semantic analysis”
(Dong 2005) to assess the quality of a TMM in terms of accuracy and the emergence of
sharedness. The realm of action can be examined using “reflective practice analysis”
126 5 Collaborative Design: Team Cognition and Communication

(Valkenburg and Dorst 1998) to reveal the relationships between frames of reference
and a team’s design process. Consequently, the combined method of latent semantic
analysis and reflective practice analysis can be effective for investigating design team
cognition and allow for the assessment of design TMMs. It also enables the analysis
of both the content and the process of team cognition and communication.
As the last few sections in this chapter have demonstrated, the foundations for a
model of design team cognition were laid in previous research on SMMs and TMMs.
While there are many subtle distinctions in past research into team cognition, for the
purposes of the present chapter it is clear that a TMM in design is a combined model
involving a SMM, for transactive memory, and a DMM, to encapsulate the emergent
phenomenon and collective, “bottom-up” development that arises from individual
MMs. Such a model has not previously been articulated in this way or presented.
Thus, this chapter proposes a new DTC model, for representing mixed cognitive,
collaborative and creative processes (Fig. 5.1).
At the centre of the DTC model is transactive memory, the common, shared
or distributed knowledge in the design team. “Transactive memory systems are
a form of cognitive architecture that encompasses both the knowledge uniquely
held by particular group members with a collective awareness of who knows what”
(DeChurch and Mesmer-Magnus 2010, p. 33). The transactive memory is a “shared
system for encoding, storing, and retrieving information” (Wegner et al. 1991,
p. 923). Like an individual memory, a team transactive memory evolves through three
processes: encoding, storing, and retrieving. Thus, both the shared and distributed
knowledge of design team members can be considered transactive. Furthermore,
like a human memory system, the transactive memory has both Long-Term Memory
(LTM) and Short-Term Memory (STM). For example, team-related cognition is
regarded as a component of LTM because it is sustained or exists over an extended

ProducƟvity, Design informaƟon,

Design representaƟon

Performance Emergence and Sharedness CommunicaƟon

Visual & verbal
representaƟons
TransacƟve
Task SoluƟon
Memory
CogniƟon
CreaƟvity Distributed Knowledge Content & process

Design strategies, ComplexiƟes,

Design spaces, Design modes

Fig. 5.1 A Design Team Cognition (DTC) model for the mixed cognitive, collaborative and creative
design processes
5.2 Cognition and Representation for Teamwork 127

period. Task-related cognition is part of STM because it is limited to the given

situation.
The DTC model (Fig. 5.1) has two knowledge stores: (i) emergence and shared-
ness and (ii) distributed knowledge. According to the given design task, the trans-
active memory continuously retrieves appropriate information from the two stores
until the team produces the final solution. Thus, the two knowledge stores also grow
and change during the team design process.
The first knowledge store, emergence and sharedness, is closely related to the
emergent construct of the collection of individual cognitions and also to the shared
knowledge of SMMs. TMMs are considered “emergent characteristics” of a “group
which reflect organised knowledge and the tendency of individuals to categorise
what they know” (Klimoski and Mohammed 1994, p. 417). The nature of emer-
gence is well described in DeChurch and Mesmer-Magnus’ study (2010) in terms
of either composition or compilation (Table 5.1). SMMs also provide guidance for
team members about how to progress, coordinate and adapt actions required by the
task or team (Cannon-Bowers et al. 1993). “As the design progresses, team members
interact with each other to exchange opinions and ideas, while their mental models
are modified, adapted, and eventually shared within the team”. (Casakin and Badke-
Schaub 2015, p. 158). SMMs of this type also shape the future responses of design
teams to new problems and situations.
The second store in the DCT, distributed knowledge, encompasses “cognitive
artefacts, the environment, culture, social factors and other elements which might be
adopted into the unit of analysis” (Banks and Millward 2009, p. 259). Whereas the
first store is closely associated with team performance, the second deals with infor-
mation that is dispersed throughout the team, and relationships between individual
cognitions, both of which are connected to creativity.
Performance-related aspects of the emergence and sharedness store are captured
by measuring productivity, design information and design representation. In contrast,
the creativity of the distributed knowledge store may be explored using individual
design strategies, complexities, design spaces and design modes. These categories in
both knowledge stores reflect the dominant themes in design thinking and cognition
research set out in the present chapter and in previous ones.
The DCT, through its communication and cognition components, also begins
to accommodate the types of individual thinking and reasoning styles that shape
each person’s capacity to communicate with, and understand, teammates (Mulet
et al. 2016). According to Herrmann (1991, 1989), people who have different
thinking styles communicate and collaborate in different ways. Herrmann’s famous
model of “thinking dominances” identifies four different modes: analytical, imagina-
tive, sequential and interpersonal. Analytical people have logical, cogent reasoning
styles, while imaginative people have abstract, metaphorical or conceptual styles.
A common way of differentiating analytical and imaginative modes identifies the
former with a mathematical or engineering approach, and the latter with an artistic
or aesthetic predilection. Despite this simplification (an example of an artificial,
although useful MM), neither category is exclusive, and mathematicians can be
creative and artists can be logical. Herrmann’s third mode, sequential thinking,
128 5 Collaborative Design: Team Cognition and Communication

governs structured cognitive processes, while his fourth, interpersonal thinking,

captures processes that require dynamic interaction and expression. These thinking
styles have an impact on creativity or novelty arising from design, with rational
thinking being potentially more effective in contemporary virtual environments, and
the interpersonal style better suited to face-to-face environments (Mulet et al. 2016).
This is because some collaborative design environments, and especially computa-
tional and parametric ones, may reward people who possess a higher level of system-
atic preparedness, whereas, in traditional design environments, more spontaneous
interpersonal styles are not only possible, but may even be beneficial.
Despite Herrmann’s model, and the findings of some past studies, cognition in
design is not so easily categorised, and it is not likely to perfectly conform to either
the TMM or DMM paradigms. For example, Chap. 2 explored individual design
strategies and processes in both traditional and parametric design environments. An
individual design strategy defines sub-goals which limit or enable certain cognitive
operations, which are also linked to personal preferences and habits and contribute to
defining designers’ thought processes (Lee et al. 2014; von der Weth 1999). Diverse
team membership (for example, expert and novice designers in a team) can increase
complexity due to heterogeneous design strategies and thought processes (Eppler
and Kernbach 2016; Ho 2001). Integrating different individual cognitive strategies
into a team design process is a significant challenge for teamwork (Lee et al. 2017).
Thus, the five individual design strategies identified in Chap. 2 can be understood as
part of the distributed store of knowledge, offering complimentary, but not collective,
approaches to problem-solving in the design team.
A further complicating factor is that, as previously noted, productivity and
“economy of thought” can be associated with increased levels of creativity (Perkins
1981). Efficiency in a design team has been linked with both creativity and expertise
and, as Chap. 4 demonstrates, can reveal “design productivity” (Goldschmidt 1995).
This productivity is associated with the effectiveness, efficiency and creativity of
different cognitive approaches to design, whether individual or collaborative (van der
Lugt 2000). Indeed, generating ideas and defining problems at a system level results
in the highest positive correlation with increased productivity in a team (Costa and
Sobek 2004). Individual content and structural complexities (see Chap. 4) can also
be used to understand individual cognitive differences as well as potential measures
of team performance.
In addition to the individual cognitive patterns and strategies discussed previously
in Part I of this book, the next section of this chapter addresses two components
of distributed knowledge, design spaces (problem and solution spaces) and design
modes (geometric and algorithmic designing). The following section investigates
design representation, in terms of design information and spatial language. These
representations are indicators of emergence and sharedness in design teamwork.
Although this chapter is the first in Part II (Collaboration), its design cognition section
is closely related to Part I (Creativity), and design representation is also connected
to Part III (Culture). That is, this chapter is an intermediate stage in developing an
understanding of the relationship between collaboration, creativity and culture in
design thinking.
5.3 Design Cognition 129

5.3 Design Cognition

In this section, protocols of individual designers, derived from experiments in a

parametric design environment (Study II in Chap. 2), are explored from a different
perspective to understand two components of distributed knowledge in the mixed
DTC model (Fig. 5.1).
The first is associated with the argument that design is an iterative process of
transition between two design spaces—problem and solution spaces—that evolve
over time (Maher and Poon 1996). Creativity in design is normally accepted as
arising from the process of cyclically developing and refining both the formulation
of a problem and ideas for a solution (Dorst and Cross 2001). The co-evolutionary
design process is the first component investigated in this section to explain individual
cognition. By understanding individual cognition, inferences and lessons can be
drawn about team cognition.
The second component is associated with the different design environments that
shape DMMs. As discussed in Chaps. 2 and 3, one of the more advanced collaborative
environments, parametric design, enables two types of design activities: algorithmic
and geometric. One specific operation in parametric design, switching behaviour
between operations and three-dimensional (3D) views of the design model, can be
understood as the transition between geometric and algorithmic spaces, which is
a type of co-evolutionary activity, albeit between the two different representation
modes (Bilda and Demirkan 2003). The switching is significant for several reasons.
It supports self-reflection, which is key to successful designing for both individuals
and teams (Schön 1988; Valkenburg and Dorst 1998). It signals a change from analyt-
ical (algorithmic) to imaginative (geometric) MMs, from mathematical to aesthetic
dimensions of individual and team design processes.
In combination, the investigation into co-evolutionary design spaces and modes in
this section draws on the theory developed and described in Part I of this book while
highlighting the connection to distributed knowledge for effective design teamwork
in the DTC model.

5.3.1 Problem and Solution Spaces

Problem-solving is a core and measurable cognitive indicator within the design

process (Coley et al. 2007). Multiple cognitive operations in design team communica-
tion can serve to widen (divergent thinking) or narrow (convergent thinking) problem-
solving potential. Significantly, the transitions between these cognitive operations
support individual reflection that contributes to social inquiry and creativity (Fischer
et al. 2005). A similar idea is implicit in the theory of a co-evolutionary process for
explorative design which, arguably, results in creative products or outcomes. This
130 5 Collaborative Design: Team Cognition and Communication

Table 5.3 The percentage of the coverage of each category in the conceptual level
Category Au1 Au2 Au3 Au4 Mean SD
Problem-finding 14.1 13.6 7.1 4.4 9.8 4.8
Solution-generating 4.0 2.0 13.0 3.0 5.5 5.1
Solution-evaluating 27.8 35.1 36.9 26.7 31.6 5.1

Table 5.4 Mean ratings for creativity, novelty, usefulness, complexity and aesthetics
Category Au1 Au2 Au3 Au4 Mean SD
Creativity 3.86 3.86 5.71 3.00 4.11 1.14
Novelty 4.43 4.14 6.57 2.29 4.36 1.75
Usefulness 4.57 4.43 4.29 4.00 4.32 0.24
Complexity 4.71 3.86 6.43 2.43 4.36 1.67
Aesthetics 3.57 2.14 4.57 3.57 3.46 1.00
Mean 4.23 3.69 5.51 3.06 4.12 1.04

cooperative co-evolutionary activity in between problem and solution spaces facili-

tates mutual interactions and explicit relationships between designers. The concep-
tual level of the coding scheme (Table 2.4) enables the investigation of both problem
and solution spaces in the parametric design process. Table 5.3 shows coding results
for the experiments of four designers divided into problem- or solution-related cate-
gories and expressed as the percentage of the frequency weighted by time span
(calculated by time duration of each category). As a point of comparison, Table 5.4
summarises expert ratings for each designer’s works, in terms of creativity, novelty,
usefulness, complexity and aesthetics.
The problem-finding category identifies cognitive activities in the problem
space, while the solution space can be explored by the solution-generating and
solution-evaluating categories. The frequency coverage of problem-finding shows
that designers using the graphical algorithm editor adopted a problem-driven strategy
(Au1: 14.8% and Au2- : 13.6%) more than those using the text-based algorithm editor
(Au3: 7.1% and Au4: 4.4%). It has previously been argued that a problem-driven
design strategy (see Chap. 2) can result in the best balance between quality and inno-
vation (Kruger and Cross 2006). In contrast, the expert panel assessment identifies
Au1’s and Au2’s designs as among the most useful designs whereas Au2’s design
received relatively lower scores for other criteria.
Au3 demonstrates a solution-driven process for design, as his protocol had the
highest frequency coverage of solution-generating and solution-evaluating activ-
ities (combination of 49.9%). The frequency coverage of G-Generation for Au2
was the highest and his protocol had the second highest frequency coverage of A-
Change Parameter. Complementing the generative activities in the solution space,
G-Generation and A-Change Parameter may have enabled Au2 to generate more
creative solutions. Consequently, his model had the highest creativity score as
5.3 Design Cognition 131

assessed by the expert panel, indicating that a solution-driven process can lead to
outcomes with higher creativity scores (Kruger and Cross 2006).
In the design experiments, participants typically switch backwards and forwards
between problem and solution spaces until they deliver a final design outcome. In
order to explore these switching activities, the summed coverage data in each 20-
segment interval is visualised using the normalised coverage value (normalised A
= (A − mean)/SD) to facilitate the exploration of cognitive activities in problem
and solution spaces over time. Figure 5.2 shows the normalised coverage of Au1
and Au3’s conceptual activities in the two spaces over time. Au1 tended to produce
design activities in the problem space at the start of the session, and then solution-
related activities at the end. In contrast, A3 switched repeatedly between problem
and solution spaces, displaying the iterative co-evolution of the two.
The solution-generating activities in the solution space are the most important
for facilitating creative cognition. Furthermore, teamwork should have an advantage
over individual work in terms of a number and variety of concepts generated (Cross
and Cross 1995; Visser 1993). The solution-evaluating activities are also aligned
to Schön’s “reflection-in-action” (Schön 1984). Consequently, it is suggested that
both conceptual activities (generating and evaluating) in the solution space are key

Problem SoluƟon
Au1
Normalised coverage

2
1.5
1
0.5
0
-0.5
-1

Start Time End

Au3
3
2.5
Normalised coverage

2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
Start Time End

Fig. 5.2 The normalised coverage of Au1’s and Au3’s conceptual activities in problem and solution
spaces over time
132 5 Collaborative Design: Team Cognition and Communication

to support creativity in both an individual and a team creative process in parametric

design.
Individual cognitive activities in the problem and solution spaces collectively
develop a “task-MM”. The problem-finding activities in the problem space arise
from the problem definition of the task-MM, while solution-generating and solution-
evaluating activities can be associated with new ideas, solution evaluation, explana-
tion and decision-making. In particular, the last two activities (explanations and deci-
sions) are not strongly present in the coding schemes for individual design processes,
whereas both should be considered when exploring design team processes.
The cognitive patterns shown in Fig. 5.2 also broadly support the results of Study II
in Chap. 2, confirming two parametric design strategies for enhancing creativity: the
problem-forwarding generative strategy (Au1) and the solution-reflecting generative
strategy (Au3). That is, the different co-evolutionary patterns of problem and solution
spaces are closely related to individual design strategies and designers may use
both problem- and solution-driven strategies to produce creative outcomes. However,
focusing on one or the other strategy could be more effective in achieving different
qualities in a design, and combining different cognitive and learning styles in a
team. In order to achieve a comprehensive solution, the solution-reflecting generative
strategy often leads to the production of variations and to reflecting on variations
recursively. These reflective and self-evaluative processes are difficult to instil in
young designers. Especially given, as Stempfle and Badke-Schaub (2002) observe,
criticism of a team’s approach to design mostly produces unfavourable reactions,
which can undermine useful self-reflection. Collaborative design, therefore, needs to
emphasise the importance of continuous self-reflection, or formative guidance and
assessment, rather than its more summative variant.

5.3.2 Geometric and Algorithmic Modes

Parametric modelling in design employs two simultaneous “modes”—geometric

and algorithmic modelling—to propose and test decisions (Holzer et al. 2007). The
first of these is akin to the standard CAD interface, where outcomes are created by
changing geometry, but the second is more abstract and mathematical, as it comprises
the scripted rules used to generate options. The design process in a parametric
environment requires switching between these two modes.
Figure 5.3 illustrates the coding coverage (%) of geometric and algorithmic codes
in three cognitive levels (physical, perceptual and conceptual). The two designers
(Au1 and Au2) using graphical algorithm editors, produce more geometric activities
in the conceptual level than the text-based editors, while the others (Au3 and Au4)
using text-based algorithm editors produce more algorithmic activities across three
levels. This confirms that the text-based algorithm editor is likely to lead designers to
conduct more algorithmic activities. In particular, P-Algorithm activities dominate
the experiments of designers using algorithm editors and both Au3 and Au4 invested
time in trouble-shooting their scripts (A-Change Rule).
5.3 Design Cognition 133

Geometry Algorithm

60 10 60

50 50
8
Coding coverage (%)

40 40
6
30 30
4
20 20

2
10 10

0 0 0
Au1 Au2 Au3 Au4 Au1 Au2 Au3 Au4 Au1 Au2 Au3 Au4
Physical Perceptual Conceptual

Fig. 5.3 The coding coverage (%) of geometric and algorithmic codes in three cognitive levels
(physical, perceptual and conceptual)

Parametric design processes vary when adopting the problem- versus solution-
driven strategies as well as geometric versus algorithmic approaches. Participant
Au1 produced a relatively large amount of F-Geometry Sub Goal and E-Geometry
activities in the conceptual level of design cognition, which appear to be the core
activities in the geometric approach. Au1’s algorithmic approach in the physical level
was also dominated by A-Change Parameter which is one of the significant activities
required to produce design variations. As discussed in Chap. 3, the use of A-Change
Parameter tends to be related to design generation.
In contrast, participant Au3’s protocol produced a relatively large amount of F-
Algorithm Sub Goal and E-Rule activities in the conceptual level, which appear to
be the core activities in the algorithmic approach. Au3’s geometric approach was
dominated by E-Geometry (evaluating geometries) which has also been linked to
supporting creativity (see Chap. 3). It has been argued that this type of switching
between geometric and algorithmic modes in parametric design has the potential to
support creativity (Salim and Burry 2010), and even constitutes a type of co-evolution
of modes. Compared to conventional design environments, which highlight the co-
evolution of problem and solution spaces, the co-evolution of the two different design
modes is a critical factor for creative cognition in parametric design. This realisation
also draws attention to the way a team of designers comprising people with different
backgrounds and expertise, can support the type of co-creative situated experience,
that is usually evolved by a spiral process of socialisation, externalisation, combi-
nation and internalisation (Nonaka and Konno 1998). Thus, it would be essential to
bridge the gaps of distributed knowledge between the conceptually different team
members to enhance creative cognition in parametric design as well as teamwork
performance.
Aish and Woodbury (2005) identify that a key design feature of parametric systems
is the requirement for multiple views of the design model and simultaneous interac-
tion across these views. Moments of insight and discovery arising from this capacity
134 5 Collaborative Design: Team Cognition and Communication

may be triggered by the complexity of both the representation and the interface. At
the representation level, a designer must understand the graph, node compilation,
intentionality and a set of mathematical ideas related to descriptive geometry and
linear algebra (Aish and Woodbury 2005). The interface level requires the designer to
understand hierarchical operations, navigate menus and linked file structures. Collec-
tively, the complexity of multiple representations and interfaces not only provide a
foundation for insight and discovery, they may offer opportunities for generating
creative solutions.
Perceiving and evaluating geometries in the 3D modelling view may, therefore, not
only enable designers to identify new visuo-spatial features but also algorithmic ones,
and vice versa, thus discovering relationships between multiple representations of
the model and generating creative outcomes from unexpected consequences. While
current parametric tools may highlight creative thinking at the algorithmic level,
perceiving and evaluating geometries—an activity traditionally reported as signifi-
cant to design cognition—continue to be critical for creativity in parametric design.
Furthermore, in the creative micro-processes discussed in Chap. 3, “changing param-
eters” naturally lead to “evaluating geometries”. That is, generating alternative design
solutions is frequently linked to the evaluation or perception of visuo-spatial features.
Switching between different design modes (interfaces) and utilising multiple repre-
sentations, appear to be critical activities for creative problem-solving processes in
parametric design.
Csikszentmihalyi (1997, 1994) proposes a creativity model, consisting of person,
field and domain. He argues that creativity is the result of the dynamic operation
of a system comprising three components: “a culture that contains symbolic rules,
a person who brings novelty into the domain, and a field of experts who recognise
and validate the innovation” (Csikszentmihalyi 1997, p. 6). An individual designer’s
cognition is a unitary system that resembles the team (Goldschmidt 1995). Switching
between design spaces and/or design modes is a co-evolutionary process and at the
same time related to widening and narrowing (or divergent and convergent) cogni-
tive processes for collaborative design. Thus, it contributes to advances in creative
cognition for a design team.

5.4 Design Communication

Design teamwork is a complex process made up of multiple designers’ cognitive

representations and communications. Design expertise is typically spread across
project teams and locations and uses collaborative design management processes
(London and Singh 2013), where design communication is supported by addi-
tional media such as sketches and drawings, and/or digital teamwork tools. Burleson
and Caplan (1998) define a communication process as entailing four components:
(i) perceiving others and defining the social context, (ii) message production, (ii)
message reception, interpretation and response, and (iv) coordinating interaction
with others. Cognitive and linguistic abstraction in design communication is also, as
5.4 Design Communication 135

these past models imply, related to individual linguistic and cultural experiences and
preferences (Lee et al. 2019). That is, for designers, language is not just a spoken
or written system, it also shapes how they use and understand design representation
and communication. The language of design is explored in detail in Part III of this
book, Culture, while this section highlights the way designers develop ideas and
solve problems and communicate solutions.
Regardless of the content of philosophical debates about the relationship between
visuals and verbals—including the “picture theory of language” (Wittgenstein 1922)
and “picture theory” (Mitchell 1994)—design cognition is a product of the interplay
between two distinct subsystems, visual imagery and verbal systems (Paivio 1971).
There are two aspects of representations, cognitive externalisation and communica-
tion, which compositely emerge in teamwork. Specifically, emergence and shared-
ness in the DTC model (Fig. 5.1) can be regarded as “shared knowledge”, which is
common information developed through design communication. In this context, this
section introduces two coding schemes to capture “design information” and “spatial
language” in the design process. Both components of emergence and sharedness are
then explored through cross-linguistic (one Australian and one Swedish) protocol
data. The following “design information” section seeks to develop a better under-
standing of designers’ imagery in MMs, while “spatial language” deals with their
verbal representations.

5.4.1 Design Information

Suwa and Tversky (1997) argue that external representations—diagrams, sketches,

charts, graphs and hand-written memos—“not only serve as memory aids, but also
facilitate and constrain inference, problem solving and understanding” (p. 385).
Beyond the communicative role of visual representation, Goldschmidt (1991) claims
that sketching supports design reasoning through a special kind of visual imagery. She
further emphasises the optimisation of intuitive visuality, because visual reasoning
plays a significant role in problem-solving (Goldschmidt 1994). Interactive imagery
produced through sketching can be understood as a systematic exchange between
cognitive and visual representations. van der Lugt (2005) for example, proposes that
sketching can serve as a visible graphic memory and thereby affect idea generation
in design groups. Thus, an investigation of image-based-information in design can
reveal how designers externalise and share their ideas and design solutions.
Suwa and Tversky (1997) use four information categories—emergent proper-
ties, spatial relations, functional relations and background knowledge—to investi-
gate how external representation delivers meanings and concepts. With the excep-
tion of their last category, background knowledge, information categories and their
subclasses in designers’ external representation are described in Table 5.5. Emergent
properties are associated with depicted elements, such as shapes and sizes, visual
properties and descriptions or names given to them. Spatial relations capture two
spatial arrangements in design, local and global relations, which also relate to spatial
136 5 Collaborative Design: Team Cognition and Communication

Table 5.5 Information

Category Subclasses
categories and subclasses
(Suwa and Tversky 1997, Emergent properties Spaces, Things, Shapes/angles, Sizes
p. 388) Spatial relations Local relation, Global relation
Functional relations Practical roles, Abstract
features/reactions, Views, Lights,
Circulation of people/cars

relational reasoning in linguistic representation (Tenbrink and Ragni 2012). Func-

tional relations, as non-visual representations, identify “interactions among spaces,
things, people visiting or using them, and/or environments” (Suwa and Tversky 1997,
p. 389).
In order to determine if individual designers from different backgrounds deal
with design information differently, two cross-national designers (one Australian
expert and one Swedish novice) were asked to conduct a conceptual design task
that was the same as Study II of Chap. 2, both “thought aloud” in English. The
Swedish designer was not an English-native speaker, but fluent in English. The design
protocols involved analysing 30-min periods (from the designers’ verbal introduction
to the design brief) of the protocol data to get the same time-duration protocols from
each test subject, which varied in length. Table 5.6 reports the coding results of two
design protocols using the “information categories” coding scheme. Considering
the different volumes of verbalisation between the two designers, the frequencies
are also normalised using the number of “T-units” in each protocol. “T-units” or
“thought units” (Hunt 1970) and often used for linguistic comparisons (Cheung

Table 5.6 The coding results of two design protocols about “information categories” (F:
Frequency)
Category Subclass Australian designer Swedish designer
F F/T-unit F F/T-unit
Emergent Spaces 26 0.123 5 0.037
properties Things – – – –
Shapes/angles 22 0.104 10 0.074
Sizes 24 0.114 27 0.199
Spatial Local relation 27 0.128 7 0.051
relations Global relation – – – –
Functional Practical roles 2 0.009 – –
relations Abstract 19 0.090 2 0.015
features/reactions
Views – – 1 0.007
Lights – – – –
Circulation of – – – –
people/cars
5.4 Design Communication 137

and Kemper 1992). The number of clauses per T-unit (or per utterance) is also
related to level of language proficiency (Nippold et al. 2008, 2014). The number
of T-units was 211 and 136, respectively, and the Australian designer’s linguistic
complexity (C/T ) is slightly higher than the Swedish designer’s: 1.81 and 1.73,
respectively. This result indicates that both designers have no difficulty in concurrent
verbalisation in English. However, Table 5.5 reveals there are differences between
the two in the production of information categories when they are designing. The
Australian designer considered spaces and shapes/angles more than the Swedish
designer. The expert also produced more terms describing spatial relations (local)
and functional relations (abstract features/reactions) than the novice. That is, the
experienced designer attended to visual elements and spatial and functional relations
more frequently than the student designer. The results of this analysis are similar
to Suwa and Tversky’s (1997) comparison between architects and students. In our
experiment, however, the student dealt with sizes more than the expert, frequently
thinking about the visual attribute of their 3D models. This disparity might indicate
that both had different design concepts and/or strategies, because the given design
task was a conceptual design for a high-rise building with a lack of design constraints
(see Box 2.2).
In summary, there are clear differences in external representation between the
two designers. These differences may come from their linguistic and cultural back-
grounds, but their different levels of design expertise and experience may provide a
simpler explanation. As Suwa and Tversky (1997) note, experienced designers are
more capable of considering design information in the design process than novice
designers. In a design team, “team dynamics” (Eppler and Kernbach 2016) is both
a natural phenomenon and a challenge for effective collaborative design. Multiple
techniques are available to overcome this challenge in design teamwork, including
concept maps (Yin et al. 2005), visual stimuli (Goldschmidt and Smolkov 2006), non-
hierarchical mind maps (Kokotovich 2008), think-maps (Oxman 2004) and dyna-
grams (Eppler and Kernbach 2016). These visual devices provide structured repre-
sentation of concepts as well as constraints, which facilitate shared understanding of
task-based MMs for collaborative design.

5.4.2 Spatial Language

Like a visual representation, a verbal representation plays a significant role in sharing

and constructing MMs and in the ability to use language. Thus, the construction of
elaborate representations is the process of cognitive development (Bloom 1993).
Spatial MMs are an integrated representation of premise information and analogical
representation of space (Knauff and Johnson-Laird 2002). Spatial language includes
qualitative, function-based distinctions (Talmy 2000). Importantly, Tenbrink and
Ragni (2012) identify four linguistic principles – (i) spatial representation, (ii) spatial
directness, (iii) trajectory, (iv) syntactic format and information structure—for spatial
relational reasoning in a verbal representation. “Spatial representation” refers to the
138 5 Collaborative Design: Team Cognition and Communication

general mapping of one entity (locatum) against another (relatum), distinguishing

between 1-object, 2-object, and multi-object descriptions. “Spatial directness” deals
with compass-based or cardinal terms (north of, south of), comparative terms and
projective terms (left, right, front, behind, above, below). “Trajectory” describes the
sequential order of successive spatial descriptions using motion verbs as, for example,
the speaker guides the listener through a sequence of spaces (Linde and Labov 1975).
The last principle, “syntactic format and information structure”, is related to the allo-
cation of the roles of locatum and relatum in the verbal representation. Such spatial
descriptors and prepositions are very influential in spatial cognition (Coventry and
Garrod 2004; Herskovits 1986).
Using these linguistic categories, this chapter presents spatial language categories
that identify spatial relationships in English (Table 5.7). The first six categories are
related to general spatial relationships. The locative prepositions are used to express
the location of one entity in relation to one or two others (so-called “relatum”).
The directional prepositions are further categorised into three types: (i) expresses a
direction (e.g. move the house forward); (ii) direction in which an object is located
(e.g. the house is to the north); (iii) and a change in position (e.g. drive across the
bridge to the house). The spatial adverbs involve two terms, here and there, while
the spatial deixis captures two types of spatial terms, near (this, these) and far (that,
those). The motion verbs identify those incorporating spatial/motion relationships
(path encoding motion verbs), such as enter, insert, exit, leave and ascend or descend.
The contact category examines if an entity is in contact with a reference object,
such as support or adhesion (on) and non-support (above). In contrast, the last six
categories deal with the syntactic structures of spatial relationships, which are similar
to Tenbrink and Ragni’s (2012) “syntactic format and information structure”.
Table 5.8 presents the coding results of the two designers’ spatial language proto-
cols (Australian and Swedish designers). Except for the spatial adverb category, the
Australian designer used more spatial language terms to describe spatial relation-
ships than the Swedish designer. In particular, the expert developed more spatial
representations using locative prepositions than the novice, which is related to the

Table 5.7 Spatial language

Category Example Category Example
categories (Pr: Preposition,
Adv: Adverb, Adj: Adjective, Locative Pr. the door is in Pr N in Sydney
N: Noun, D: Demonstrative) the wall
Directional Pr. the house is to Pr D N in the box
the north
Spatial Adv. here, there Pr Adj N above large
openings
Spatial deixis this, these, Pr D Adj N above the big
that, those window
Motion verbs enter, extract, Pr Adv in here
insert, exit
Contact on, above Pr D in that
5.4 Design Communication 139

Table 5.8 The coding results of two design protocols in terms of the spatial language categories
(Pr: Preposition, Adv: Adverb, Adj: Adjective, N: Noun, D: Demonstrative, F: Frequency)
Category Australian designer Swedish designer
F F/T-unit F F/T-unit
Locative Pr. 31 0.147 5 0.037
Directional Pr. 9 0.043 5 0.037
Spatial Adv. 17 0.081 22 0.162
Deixis – – – –
Motion verbs 39 0.185 16 0.118
Contact (on) – – 1 0.007
Pr N 6 0.028 3 0.022
Pr D N 20 0.095 4 0.029
Pr Adj N 2 0.009 0 0.000
Pr D Adj N 10 0.047 4 0.029
Pr Adv 6 0.028 3 0.022
Pr D 1 0.005 1 0.007

Australian’s frequent production of “spatial relations” in the “information categories”

(Table 5.4). This result, importantly, indicates the relationship between verbal and
visual representations. The frequent use of motion verbs in the Australian protocol
is also interesting, because it may have an impact on both “kinaesthetic imagery”
(Rabahi et al. 2013) and “spatial semantics” (Sen and Janowicz 2005).
Although this protocol study deals with design information and spatial language,
the results differentiate patterns in usages of visual and spatial terms which can be
used as an indicator of “productivity” and of individual cognitive differences. The
advantage the team has over the individual, which is often stressed in descriptions
of the DMM, is the capacity to combining members’ abilities and MMs (Fischer
et al. 2005; Prather and Middleton 2002; Safoutin and Thurston 1993). In parallel,
there is a process loss in the group because of its reliance on imperfect visual and
verbal communication (Hackman and Morris 1975; Steiner 1972). As such, devel-
oping a “communication-based technique” (Safoutin and Thurston 1993) for inter-
disciplinary design teams may be a core factor for improving both productivity and
performance in design teams.

5.5 Conclusion

This chapter has developed a Design Team Cognition (DTC) model that builds on
important messages in the literature about team TMs, MMs and cognitive representa-
tions. The DTC model is centred on transactive memory dealing with emergence and
sharedness and distributed knowledge or SMMs and DMMs, respectively, promoting
140 5 Collaborative Design: Team Cognition and Communication

both performance and creativity in collaborative design. Significantly, four compo-

nents of the model—design spaces, design modes, design information and design
representation—are explored using two sets of protocol data relating to commu-
nication and cognition in the design process. Furthermore, this chapter introduced
two coding schemes to identify information categories and spatial language in the
design process. The coding schemes are further developed in Part III of this book to
investigate the cultural aspect of design thinking.
Design cognition is not fixed, it evolves as it moves from problem space to solu-
tion space, as well as between different modes of representation. Both past research,
and the simple protocol study discussed in this chapter, suggest that the switching
between geometric and algorithmic modes in parametric design could support co-
creation between team members to enhance creativity. If this is true, it would consti-
tute a new type of co-evolution of cognition in design. In contrast, design commu-
nication is a process of combined visual and verbal representation. Combining past
research and theories with data from a pilot protocol study shows the importance of
different modes of linguistic expression and spatial relational reasoning. Considering
that communication is the most important factor for design teams to increase both
creativity and performance, the approaches developed in this chapter can be extended
to examine further collaborative issues, such as social cognition, coordination and
design management. This approach can be used to provide a new understanding of
multi-cultural differences in spatial design thinking and communication. Without
an appreciation of the differences in visual and verbal design thinking, supporting
effective design teamwork may be impossible. This topic is further explored by Part
III of this book addressing the relationships between language and design.
Even though this chapter describes some significant cognitive contributors to a
creative teamwork process, the data presented herein is only used as a catalyst for
reviewing theories, not for providing “proof”. The cognitive issues examined in
this chapter should be the subject of future research, using an appropriate sampling
method and design, and a variety of team members. In particular, while task-related
MMs can be investigated using individual design protocols, team-related MMs must
naturally be explored through team design experiments. Nevertheless, a precursor to
conducting such a series of studies of design thinking in teams is developing a model
and a method that can be effectively used for this purpose. These are the goals of the
present chapter.

References

Aish, Robert, and Robert Woodbury. 2005. Multi-level interaction in parametric design. In Smart
Graphics, ed. Andreas Butz, Brian Fisher, Antonio Krüger, and Patrick Olivier, 924-924. Lecture
Notes in Computer Science. Berlin/Heidelberg: Springer.
Austin, J.R. 2003. Transactive memory in organizational groups: The effects of content, consensus,
specialization, and accuracy on group performance. Journal of Applied Psychology 88 (5): 866–
878. https://doi.org/10.1037/0021-9010.88.5.866.
References 141

Badke-Schaub, Petra, Andre Neumann, Kristina Lauche, and Susan Mohammed. 2007. Mental
models in design teams: a valid approach to performance in design collaboration? CoDesign 3
(1): 5–20. https://doi.org/10.1080/15710880601170768.
Banks, Adrian P., and Lynne J. Millward. 2009. Distributed mental models: Mental models in
distributed cognitive systems. The Journal of Mind and Behavior 30 (4): 249–266.
Bierhals, R., I. Schuster, P. Kohler, and P. Badke-Schaub. 2007. Shared mental models—Linking
team cognition and performance. CoDesign 3 (1): 75–94. https://doi.org/10.1080/157108806011
70891.
Bilda, Z., and H. Demirkan. 2003. An insight on designers’ sketching activities in traditional versus
digital media. Design Studies 24 (1): 27–50.
Blackwell, A. F., C. Britton, A. Cox, T. R. G. Green, C. Gurr, G. Kadoda, M. S. Kutar et al.
2001. Cognitive dimensions of notations: Design tools for cognitive technology. In Cognitive
Technology: Instruments of Mind. CT 2001. Lecture Notes in Computer Science, ed. Beynon M.,
Nehaniv C.L., and Dautenhahn K, 325–341. Cognitive Technology: Instruments of Mind. Berlin,
Heidelberg: Springer Berlin Heidelberg.
Bloom, Lois. 1993. The Transition from Infancy to Language: Acquiring the Power of Exprsession.
New York: Cambridge University Press.
Boos, Margarete. 2007. Optimal sharedness of mental models for effective group performance.
CoDesign 3 (1): 21–28. https://doi.org/10.1080/15710880601170776.
Bresciani, Sabrina. 2019. Visual design thinking: A collaborative dimensions framework to profile
visualisations. Design Studies 63: 92–124. https://doi.org/10.1016/j.destud.2019.04.001.
Briggs, R.O., G. de Vreede, J.F. Nunamaker Jr, and D. Tobey. 2001. ThinkLets: Achieving
predictable, repeatable patterns of group interaction with Group Support Systems (GSS). In
Proceedings of the 34th Hawaii International Conference on System Sciences. Los Alamitos,
CA: IEEE Computer Society Press.
Burleson, B.R., and S.E. Caplan. 1998. Cognitive complexity. In Communication and Personality:
Trait Perspectives, ed. J.C. McCroskey, J.A. Daly, M.M. Martin, and M.J. Beatty, 233–286.
Cresskill, NJ: Hampton Press.
Cannon-Bowers, J. A., E. Salas, and S. A. Converse. 1993. Shared mental models in expert team deci-
sion making. In Current Issues in Individual and Group Decision Making, ed. Jr. N.J. Castellan,
221–246. Hillsdale, NJ: Erlbaum.
Casakin, Hernan, and Petra Badke-Schaub. 2015. Mental models and creativity in engineering and
architectural design teams. In Design Computing and Cognition ‘14, ed. John S. Gero and Sean
Hanna, 155–171. Cham: Springer International Publishing.
Cheung, Hintat, and Susan Kemper. 1992. Competing complexity metrics and adults’ production
of complex sentences. Applied Psycholinguistics 13 (1): 53–76. https://doi.org/10.1017/S01427
16400005427.
Chiu, Mao-Lin. 2002. An organizational view of design communication in design collaboration.
Design Studies 23 (2): 187–210. https://doi.org/10.1016/S0142-694X(01)00019-9.
Coley, F., O. Houseman, and R. Roy. 2007. An introduction to capturing and understanding the
cognitive behaviour of design engineers. Journal of Engineering Design 18 (4): 311–325.
Costa, Ramon, and Durward K. Sobek. 2004. How process affects performance: An analysis of
student design productivity. In ASME 2004 International Design Engineering Technical Confer-
ences and Computers and Information in Engineering Conference. Salt Lake City, Utah, USA:
American Society of Mechanical Engineers.
Coventry, K.R., and S.C. Garrod. 2004. Saying, Seeing and Acting: The Psychological Semantics
of Spatial Prepositions. Hove and New YorK: Psychology Press.
Cross, Nigel. 1982. Designerly ways of knowing. Design Studies 3 (4): 221–227.
Cross, Nigel, and Anita Clayburn Cross. 1995. Observations of teamwork and social processes in
design. Design Studies 16 (2): 143–170. https://doi.org/10.1016/0142-694X(94)00007-Z.
Csikszentmihalyi, M. 1994. The domain of creativity. In Changing the World: A Framework for the
Study of Creativity, ed. D.H. Feldman, M. Csikszentmihalyi, and H. Gardner, 135–158. Westport,
CT: Praeger.
142 5 Collaborative Design: Team Cognition and Communication

Csikszentmihalyi, M. 1997. Creativity: Flow and the Psychology of Discovery and Invention. New
York: HarperCollins.
DeChurch, Leslie A., and Jessica R. Mesmer-Magnus. 2010. The cognitive underpinnings of effec-
tive teamwork: A meta-analysis. Journal of Applied Psychology 95 (1): 32–53. https://doi.org/
10.1037/a0017328.
Dong, Andy. 2005. The latent semantic approach to studying design team communication. Design
Studies 26 (5): 445–461.
Dong, Andy, Maaike S. Kleinsmann, and Fleur Deken. 2013. Investigating design cognition in the
construction and enactment of team mental models. Design Studies 34 (1): 1–33. https://doi.org/
10.1016/j.destud.2012.05.003.
Dorst, Kees, and Nigel Cross. 2001. Creativity in the design process: Co-evolution of problem-
solution. Design Studies 22 (5): 425–437.
Du, Junpeng, Shikai Jing, and Jihong Liu. 2012. Creating shared design thinking process for collab-
orative design. Journal of Network and Computer Applications 35 (1): 111–120. https://doi.org/
10.1016/j.jnca.2011.02.014.
Eppler, M.J. 2004. Facilitating knowledge communication through joint interactive visualization.
Journal of Universal Computer Science 10 (6): 683–690.
Eppler, Martin J., and Sebastian Kernbach. 2016. Dynagrams: Enhancing design thinking through
dynamic diagrams. Design Studies 47: 91–117. https://doi.org/10.1016/j.destud.2016.09.001.
Fauconnier, G. 1985. Mental Spaces. Cambridge, MA: MIT Press.
Fischer, Gerhard, Elisa Giaccardi, Hal Eden, Masanori Sugimoto, and Yunwen Ye. 2005. Beyond
binary choices: Integrating individual and social creativity. International Journal of Human-
Computer Studies 63 (4–5): 482–512. https://doi.org/10.1016/j.ijhcs.2005.04.014.
Giaccardi, Elisa. 2004. Principles of Metadesign: Processes and Levels of Co-creation in the New
Design Space. Plymouth, UK: University of Plymouth.
Giaccardi, Elisa. 2005. Metadesign as an emergent design culture. Leonardo 38 (4): 342–349.
Goldschmidt, Gabriela. 1991. The dialectics of sketching. Creativity Research Journal 4 (2): 123–
143. https://doi.org/10.1080/10400419109534381.
Goldschmidt, Gabriela. 1994. On visual design thinking: The vis kids of architecture. Design Studies
15 (2): 158–174. https://doi.org/10.1016/0142-694X(94)90022-1.
Goldschmidt, Gabriela. 1995. The designer as a team of one. Design Studies 16 (2): 189–209.
https://doi.org/10.1016/0142-694X(94)00009-3.
Goldschmidt, Gabriela, and Maria Smolkov. 2006. Variances in the impact of visual stimuli on
design problem solving performance. Design Studies 27 (5): 549–569. https://doi.org/10.1016/j.
destud.2006.01.002.
Gonçalves, Milene, Carlos Cardoso, and Petra Badke-Schaub. 2014. What inspires designers? Pref-
erences on inspirational approaches during idea generation. Design Studies 35 (1): 29–53. https://
doi.org/10.1016/j.destud.2013.09.001.
Green, T.R.G., and M. Petre. 1996. Usability Analysis of Visual Programming Environments: A
‘Cognitive Dimensions’ Framework. Journal of Visual Languages & Computing 7 (2): 131–174.
https://doi.org/10.1006/jvlc.1996.0009.
Grudin, J. 1994. Computer-supported cooperative work: History and focus. Computer 27 (5): 19–26.
https://doi.org/10.1109/2.291294.
Hackman, J Richard, and Charles G. Morris. 1975. Group tasks, group interaction process, and group
performance effectiveness: A review and proposed integration. In Advances in Experimental
Social Psychology, ed. L. Berkowitz, 1–55. New York: Academic Press.
Herrmann, Ned. 1989. The Creative Brain. Buffalo, NY: Brain books.
Herrmann, Ned. 1991. The creative brain*. The Journal of Creative Behavior 25 (4): 275–295.
https://doi.org/10.1002/j.2162-6057.1991.tb01140.x.
Herskovits, A. 1986. Language and Spatial Cognition. Cambridge: Cambridge University Press.
Hinsz, Verlin B., R. Scott Tindale, and David A. Vollrath. 1997. The emerging conceptualization
of groups as information processors. Psychological Bulletin 121 (1): 43–64. https://doi.org/10.
1037/0033-2909.121.1.43.
References 143

Ho, Chun-Heng. 2001. Some phenomena of problem decomposition strategy for design thinking:
Differences between novices and experts. Design Studies 22 (1): 27–45.
Holyoak, K.J. 1984. Mental models in problem solving. In Tutorials in Learning and Memory, ed.
J.R. Anderson and S.M. Kosslyn, 193–218. New York: W. H. Freeman and Company.
Holzer, Dominik, Richard Hough, and Mark Burry. 2007. Parametric design and structural opti-
misation for early design exploration. International Journal of Architectural Computing 5 (4):
625–643. https://doi.org/10.1260/147807707783600780.
Hunt, Kellogg W. 1970. Syntactic maturity in schoolchildren and adults. Monographs of the Society
for Research in Child Development 35 (1): iii–67. https://doi.org/10.2307/1165818.
Ibrahim, R., and F. Pour Rahimian. 2010. Comparison of CAD and manual sketching tools for
teaching architectural design. Automation in Construction 19 (8): 978–987.
Johnson-Land, P.N. 1983. Menlal Models: Towards a Cognitive Science of Language, Inference,
and Consciousness. Cambridge, MA: Harvard University Press.
Jonker, Catholijn M., M. Birna van Riemsdijk, and Bas Vermeulen. 2011. Shared mental models: A
conceptual analysis. In Coordination, Organizations, Institutions, and Norms in Agent Systems
VI, ed. Marina De Vos, Nicoletta Fornara, Jeremy V. Pitt, and George Vouros, 132–151. Berlin,
Heidelberg: Springer Berlin Heidelberg.
Kleinsmann, M.S. 2006. Understanding Collaborative Design. Delft, Netherlands: Delft Unversity
of Technology.
Klimoski, Richard, and Susan Mohammed. 1994. Team mental model: Construct or metaphor?
Journal of management 20 (2): 403–437.
Knauff, Markus, and P.N. Johnson-Laird. 2002. Visual imagery can impede reasoning. Memory &
Cognition 30 (3): 363–371. https://doi.org/10.3758/BF03194937.
Kokotovich, Vasilije. 2008. Problem analysis and thinking tools: An empirical study of non-
hierarchical mind mapping. Design Studies 29 (1): 49–69. https://doi.org/10.1016/j.destud.2007.
09.001.
Kolarevic, B. 2003. Architecture in the Digital Age: Design and Manufacturing. London, New York:
Spon Press.
Kruger, Corinne, and Nigel Cross. 2006. Solution driven versus problem driven design: Strategies
and outcomes. Design Studies 27 (5): 527–548.
Langan-Fox, Janice, Anthony Wirth, Sharon Code, Kim Langfield-Smith, and Andrew Wirth. 2001.
Analyzing shared and team mental models. International Journal of Industrial Ergonomics 28
(2): 99–112. https://doi.org/10.1016/S0169-8141(01)00016-6.
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014. Parametric design strategies for the generation
of creative designs. International Journal of Architectural Computing 12 (3): 263–282. https://
doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2016. The language of design: Spatial cognition
and spatial language in parametric design. International Journal of Architectural Computing 14
(3): 277–288. https://doi.org/10.1177/1478077116663350.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2017. Cognitive challenges for teamwork in design.
In Collaboration and Student Engagement in Design Education, 55–75. Hershey, PA, USA: IGI
Global.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2019. Cognitive and linguistic differences in
architectural design. Architectural Science Review 62 (3): 248–260. https://doi.org/10.1080/000
38628.2019.1606777.
Linde, Charlotte, and William Labov. 1975. Spatial networks as a site for the study of language and
thought. Language 51 (4): 924–939. https://doi.org/10.2307/412701.
London, K., and V. Singh. 2013. Integrated construction supply chain design and delivery solutions.
Architectural Engineering and Design Management 9 (3): 135–157. https://doi.org/10.1080/174
52007.2012.684451.
Maher, M.L., and Josiah Poon. 1996. Modeling Design Exploration as Co-Evolution. Computer-
Aided Civil and Infrastructure Engineering 11 (3): 195–209. https://doi.org/10.1111/j.1467-8667.
1996.tb00323.x.
144 5 Collaborative Design: Team Cognition and Communication

Marshall, Nick. 2007. Team mental models in action: A practice-based perspective. CoDesign 3
(1): 29–36. https://doi.org/10.1080/15710880601170784.
Mathieu, John E., Tonia S. Heffner, Gerald F. Goodwin, Eduardo Salas, and Janis A. Cannon-
Bowers. 2000. The influence of shared mental models on team process and performance. Journal
of Applied Psychology 85 (2): 273–283. https://doi.org/10.1037/0021-9010.85.2.273.
Mitchell, W.J.T. 1994. Picture Theory: Essays on Verbal and Visual Representation. Chicago,
London: The University of Chicago Press.
Mohammed, Susan, and Brad C. Dumville. 2001. Team mental models in a team knowledge
framework: Expanding theory and measurement across disciplinary boundaries. Journal of
Organizational Behavior 22 (2): 89–106.
Mulet, Elena, Vicente Chulvi, Marta Royo, and Julia Galán. 2016. Influence of the dominant thinking
style in the degree of novelty of designs in virtual and traditional working environments. Journal
of Engineering Design 27 (7): 413–437. https://doi.org/10.1080/09544828.2016.1155697.
Nippold, Marilyn A., Paige M. Cramond, and Christine Hayward-Mayhew. 2014. Spoken language
production in adults: Examining age-related differences in syntactic complexity. Clinical
Linguistics & Phonetics 28 (3): 195–207. https://doi.org/10.3109/02699206.2013.841292.
Nippold, Marilyn A., Tracy C. Mansfield, Jesse L. Billow, and J. Bruce Tomblin. 2008. Expos-
itory discourse in adolescents with language impairments: Examining syntactic development.
American Journal of Speech-Language Pathology 17 (4): 356–366.
Nonaka, Ikujiro, and Noboru Konno. 1998. The concept of “Ba”: Building a foundation for
knowledge creation. California Management Review 40 (3): 40–54. https://doi.org/10.2307/411
65942.
Ostergaard, Karen J., and Joshua D. Summers. 2003. A Taxonomy for collaborative design. In ASME
2003 International Design Engineering Technical Conferences and Computers and Information
in Engineering Conference. Chicago, IL: American Society of Mechanical Engineers.
Oxman, Rivka. 2004. Think-maps: Teaching design thinking in design education. Design Studies
25 (1): 63–91. https://doi.org/10.1016/S0142-694X(03)00033-4.
Paivio, Allan. 1971. Imagery and Verbal Processes. Oxford: Holt, Rinehart & Winston.
Paivio, Allan. 1983. Mental Representations: A Dual Coding Approach. Oxford: Oxford University
Press.
Paivio, Allan. 1991. Dual coding theory: Retrospect and current status. Canadian Journal of
Psychology/Revue canadienne de psychologie 45 (3): 255–287. https://doi.org/10.1037/h00
84295.
Perkins, D.N. 1981. The Mind’s Best Work. Cambridge, MA: Harvard University Press.
Prather, Larry J., and Karen L. Middleton. 2002. Are N+1 heads better than one?: The case of
mutual fund managers. Journal of Economic Behavior & Organization 47 (1): 103–120. https://
doi.org/10.1016/S0167-2681(01)00172-X.
Rabahi, Tahar, Patrick Fargier, Ahmad Rifai Sarraj, Cyril Clouzeau, and Raphael Massarelli. 2013.
Effect of Action verbs on the performance of a complex movement. PLoS ONE 8 (7): e68687.
https://doi.org/10.1371/journal.pone.0068687.
Resnick, L. B. 1991. Shared cognition: Thinking as social practice. In Perspectives on Socially
Shared Cognition, ed. L.B. Resnick, J.M. Levine, and S.D. Teasley. Washington, DC: American
Psychological Association.
Rouse, William B., and Nancy M. Morris. 1986. On looking into the black box: Prospects and
limits in the search for mental models. Psychological Bulletin 100 (3): 349–363. https://doi.org/
10.1037/0033-2909.100.3.349.
Safoutin, M.J., and D.J. Thurston. 1993. A communications-based technique for interdisciplinary
design team management. Engineering Management, IEEE Transactions on 40 (4): 360–372.
https://doi.org/10.1109/17.257728.
Salas, E., T.L. Dickinson, S.A. Converse, and S.I. Tannenbaum. 1992. Toward an understanding
of team performance and training. In Teams: Their Training and Performance, ed. R.W. Swezey
and E. Salas, 3–29. Norwood, NJ: Ablex.
References 145

Salim, Flora, and Jane Burry. 2010. Software openness: Evaluating parameters of parametric
modeling tools to support creativity and multidisciplinary design integration. In Computational
Science and Its Applications—ICCSA 2010, ed. David Taniar, Osvaldo Gervasi, Beniamino
Murgante, Eric Pardede, and Bernady Apduhan, 483–497. Lecture Notes in Computer Science.
Berlin/Heidelberg: Springer.
Santandreu Calonge, David, and Askhat F.. Safiullin. 2015. Can culturally, disciplinarily and educa-
tionally diverse (D3) teams function and be creative? A case study in a Korean university.
Educational Studies 41 (4): 369–392. https://doi.org/10.1080/03055698.2015.1018871.
Sauer, Jürgen, Tobias Felsing, Holger Franke, and Bruno Rüttinger. 2006. Cognitive diversity and
team performance in a complex multiple task environment. Ergonomics 49 (10): 934–954. https://
doi.org/10.1080/00140130600577502.
Schön, D.A. 1984. The architectural studio as an exemplar of education for reflection-in-action.
Journal of Architectural Education 38 (1): 2–9.
Schön, D.A. 1988. Designing: Rules, types and words. Design Studies 9 (3): 181–190.
Sen, S., and K Janowicz. 2005. Semantics of motion verbs. In Workshop on Spatial Language and
Dialogue (WOSLAD-05). Delmenhorst, Germany.
Sonnenwald, Diane H. 1996. Communication roles that support collaboration during the design
process. Design Studies 17 (3): 277–301. https://doi.org/10.1016/0142-694X(96)00002-6.
Steiner, I.D. 1972. Group Processes and Productivity. New York: Academic Press.
Stempfle, Joachim, and Petra Badke-Schaub. 2002. Thinking in design teams—An analysis of team
communication. Design Studies 23 (5): 473–496. https://doi.org/10.1016/S0142-694X(02)000
04-2.
Suwa, Masaki, and Barbara Tversky. 1997. What do architects and students perceive in their design
sketches? A protocol analysis. Design Studies 18 (4): 385–403. https://doi.org/10.1016/S0142-
694X(97)00008-2.
Talmy, L. 2000. Toward a Cognitive Semantic. Cambridge: MIT Press.
Tenbrink, Thora, and Marco Ragni. 2012. Linguistic principles for spatial relational reasoning. In
Cyrill Stachniss, Kerstin Schill, and David Uttal, ed. V.I.I.I. Spatial Cognition, 279–298. Lecture
Notes in Computer Science: Springer, Berlin Heidelberg.
Tindale, R. Scott. 1989. Group vs individual information processing: The effects of outcome feed-
back on decision making. Organizational Behavior and Human Decision Processes 44 (3):
454–473. https://doi.org/10.1016/0749-5978(89)90019-8.
Valkenburg, Rianne, and Kees Dorst. 1998. The reflective practice of design teams. Design Studies
19 (3): 249–271. https://doi.org/10.1016/S0142-694X(98)00011-8.
van der Lugt, Remko. 2000. Developing a graphic tool for creative problem solving in design groups.
Design Studies 21 (5): 505–522. https://doi.org/10.1016/S0142-694X(00)00021-1.
van der Lugt, Remko. 2005. How sketching can affect the idea generation process in design group
meetings. Design Studies 26 (2): 101–122. https://doi.org/10.1016/j.destud.2004.08.003.
Visser, W. 1993. Collective design: A cognitive analysis of cooperation in practice. In ICED 93, 9th
International Conference on Engineering Design (Volume 1). Zurich, Switzerland: HEURISTA.
von der Weth, Rüdiger. 1999. Design instinct?—The development of individual strategies. Design
Studies 20 (5): 453–463. https://doi.org/10.1016/S0142-694X(99)00021-6.
Wegner, Daniel M., Ralph Erber, and Paula Raymond. 1991. Transactive memory in close
relationships. Journal of Personality and Social Psychology 61: 923–929.
Wittgenstein, Ludwig. 1922. Tractatus Logico-Philosophicus. New York: Harcourt, Brace &
Company.
Won, P.H. 2001. The comparison between visual thinking using computer and conventional media
in the concept generation stages of design. Automation in Construction 10 (3): 319–325. https://
doi.org/10.1016/S0926-5805(00)00048-0.
Yin, Yue, Jim Vanides, Maria Araceli Ruiz-Primo, Carlos C. Ayala, and Richard J. Shavelson.
2005. Comparison of two concept-mapping techniques: Implications for scoring, interpretation,
and use. Journal of Research in Science Teaching 42 (2): 166–184. https://doi.org/10.1002/tea.
20049.
Chapter 6
Design Thinking and Building
Information Modelling

Abstract Building Information Modelling (BIM) is an approach to computer-

enabled multi-disciplinary collaboration, communication and coordination. A BIM
model is a consolidated digital data repository of a design, which facilitates seamless
information exchange between stakeholders during a project lifecycle. BIM is the
industry standard in the design and construction sector. It has strong synergies with
the emerging concept of a “Digital Twin”, and BIM is pivotal to implementing innova-
tions arising from this digital ecosystem. This chapter presents a new BIM knowledge
framework which is founded on a collaborative design thinking approach. This new
framework adopts emerging ontologies and technologies that enable collaborative
design thinking and decision-making through better support for design representation
and communication among users.

6.1 Introduction

In the last two decades, Building Information Modelling (BIM) has played a signifi-
cant role in the transformation of the design, collaboration and management processes
across the Architecture, Engineering and Construction (AEC) sector (Daniotti et al.
2020). BIM is defined as “a modelling technology and associated set of processes to
produce, communicate and analyse building models” (Eastman et al. 2011, p. 16). In
essence, BIM creates a combined digital representation of a building (Kiviniemi et al.
2008; Rinella 2008; Simeone et al. 2019) and an information management system
for its components (Alizadehsalehi et al. 2020; Jang and Collinge 2020). BIM is the
most pervasive example of a design platform that enables both “interaction” (indi-
vidual interface with technology) and “collectivity” (collective use of technology)
(Verstegen et al. 2019). The former function, interaction, encapsulates the ways indi-
viduals use technology, while the latter addresses the ways they organise its use.
Within the digital ecology, these platforms are the services, systems and architec-
tures that provide the infrastructure for businesses to operate. They also address the
need for a collaborative, real-time, co-present system in a multi-user digital design
ecology (Martinez-Maldonado et al. 2017). Given that design thinking for business
strategy and transformation addresses “the visualisation of concepts and the actual

© Springer Nature Switzerland AG 2020 147

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_6
148 6 Design Thinking and Building Information Modelling

delivery of new products and services” (Cooper et al. 2009, p. 48), this chapter intro-
duces BIM as an approach to facilitating a particular type of collaborative digital
design thinking process.
A BIM model comprises a single, consolidated digital data repository of a design,
which allows integrated digital representation of all information about that design
throughout the entire project lifecycle (Gu and London 2010). In a practical sense,
BIM is an object-oriented Computer-Aided Design (CAD) package with a heightened
capacity to handle complexity and automation in contemporary building projects. It
is, importantly, responsible for fostering the uptake and exchange of 3D data during
the collaborative design process in the AEC sector (Singh et al. 2011). That is, BIM
is a “digital design medium” (Bilda and Demirkan 2003; Ibrahim and Rahimian
2010; Won 2001), which enables the type of visual representation (or visualisa-
tion) discussed previously in this book. Furthermore, BIM platforms use parametric
modelling to support the types of divergent and convergent thinking described in
Part I of this book, along with enhanced design analysis and decision-making (Ning
et al. 2018; Tang et al. 2020; Utkucu and Sözer 2020).
Through its production of a singular, consistent and coordinated digital model,
BIM assists users to break down the communication barriers between different
team members and stakeholders, supporting new ways of generating, managing and
sharing building data. The advanced visualisation and communication capabilities of
BIM have also facilitated collaboration in building design. Moreover, successful BIM
implementation is core to the digital ecosystem in design, where related products,
processes and people can co-evolve (Gu et al. 2015). Thus, exploring BIM applica-
tions and technologies not only facilitates wider adoption in the AEC industry but
provides a better understanding of collective and interactive design and management
processes in dynamic design practices (Verstegen et al. 2019).
This chapter positions BIM as a domain-specific example of design collabora-
tion in the digital ecosystem. It starts by presenting a background to BIM research
and practice, focussing in part on collaboration. Thereafter, the chapter develops an
advanced BIM knowledge framework. This framework adopts emerging ontologies
and technologies that allow “collaborative design thinking” to better support hetero-
geneous design representations and communications among different stakeholders.
Importantly, the framework complies with the philosophy of the “BIM ecosystem”,
which considers BIM-related products, processes and people in parallel (Gu et al.
2015). The framework also takes into account both new (“as-planned”) designs and
existing (“as-built”) buildings.

6.2 Digital Design Collaboration and BIM

As described in Chap. 5, research in Computer-Supported Cooperative Work

(CSCW) provides support for design teams, by creating appropriate digital envi-
ronments for collaboration and communication (Chiu 2002; Grudin 1994; Ibrahim
and Rahimian 2010). Chiu (2002) further suggests that CSCW can support three
6.2 Digital Design Collaboration and BIM 149

levels of communication: the individual, group and project. BIM is a type of CSCW
that is specifically optimised for AEC projects across these three levels. The imple-
mentation and reinforcement of BIM through commercial modelling software (like
Autodesk Revit) has made it one of the most widely adopted and professionally
aligned CSCW applications in the sector.

6.2.1 Design Collaboration Through Digital Platforms

The catalysts for BIM can be traced to multiple simultaneous pressures in the AEC
industry. Design collaboration in the AEC industry has long involved the develop-
ment and exchange of 2D drawings and documents, featuring both the visual and
descriptive specifications that make up a design proposition. It is also common for
designers to use scaled “mock-ups”, analogue models and renderings from 3D models
to supplement these 2D representations. 2D documentation, despite its many limi-
tations, still plays an important role in conventional collaboration practices across
the AEC sector. Furthermore, uncertainty surrounding the readiness of digital tech-
nologies means that parts of the sector continue to rely on 2D documentation for
completeness, accuracy and legalities (Gu and London 2010). One solution to the
problem of collaboration and its reliance on 2D drawings and documents is the
Document Management System (DMS), which stores and catalogues 2D data for
controlled access. In addition to DMS, a range of strategies and enablers—such
as agreed protocols, version-control nomenclature, standardised quality-assurance
procedures and tools—have been introduced to improve the quality and account-
ability of representations for design collaboration. Despite such developments, over
the last few decades there has been a growing need for more responsive and immediate
design tools. In addition, with the increased degree of complexity and collaboration
required in a contemporary building project, an integrated digital platform is required
that is capable of representing design and building data in different forms and multiple
disciplinary formats. In parallel with this growing pressure to improve collaboration
and document exchange, the AEC industry has also begun to take responsibility for
the creation and exchange of information across a project lifecycle. BIM is the AEC
industry’s solution to these challenges.
As a generic interactive, collaborative system, BIM provides a new design thinking
and management platform for creating and sharing 3D digital models and their
embedded knowledge. This does, however, have implications for modelling, work-
sharing and coordinating. A client–server model (cloud collaboration and data
management) may be required for adapting to this new context, where multiple parties
seamlessly contribute to a centralised model at any time and from anywhere. In addi-
tion, this type of emerging digital collaborative platform is typically the catalyst for
new roles and evolving relationships within a project team. For key stakeholders,
in addition to technical competencies, cultural issues and disciplinary differences
require careful attention in order to establish shared understandings within a team.
For example, design team members with CAD backgrounds, like architects, often
150 6 Design Thinking and Building Information Modelling

prioritise features that support design visualisation and model navigation and which
are comparable with other familiar CAD applications. In contrast, team members
with DMS backgrounds, such as contractors and project managers, may only expect
these visualisation and navigation features as add-ons (Singh et al. 2011). Thus, a
mature CSCW platform like BIM integrates CAD, DMS and multiple other compo-
nents and functionalities into a single product. It offers an effective computational
approach to design collaboration which meets the requirements of diverse stake-
holders and also provides support features to assist users in assessing, planning and
implementing the BIM approach in their projects.

6.2.2 BIM Collaboration Essentials

BIM, as a model-centric platform, supports “information modelling” in a building

project. BIM also refers to a digital design and management process that visualises
the project lifecycle of a building in an integrated digital information environment.
Thus, it represents the built environment (both a new proposition and an existing one)
virtually as a consistent data model enriched with added knowledge and intelligence
(Biagini et al. 2016; Eastman et al. 2011; Tang et al. 2010). ISO standard 29481–1
defines BIM as an approach that creates and maintains “a shared digital representation
of a built object (including buildings, bridges, roads, process plants, etc.) to facilitate
design, construction and operation processes to form a reliable basis for decisions”
(ISO 2016, np). The phrase “shared digital representation” refers to both the phys-
ical and functional characteristics of a built object (ISO 2016). BIM manages the
diverse sets of data and provides a platform for describing and displaying information
required for the planning, design, construction and operation of built works.
In the AEC industry, BIM is typically described as an object-oriented CAD
system enabling parametric representation of building components (Cerovsek 2011;
Volk et al. 2014). Objects can have geometric or non-geometric attributes including
functional, semantic or topological information (Eastman et al. 2011). Functional
attributes may include construction duration, staging or cost. Semantic attributes are
concerned with the connectivity, aggregation, containment or intersection of building
data. Topological attributes store information regarding objects’ relative locations
in a system (Biagini et al. 2016). As such, responding to these three attributes,
BIM platforms have features that support data management and integration, building
component libraries and general functionalities related to visualisation and analysis
(Eastman et al. 2011). Furthermore, because BIM can be used across an entire project
lifecycle, a much larger range of stakeholders need to interact with the data model
(Gu et al. 2015).
Despite such definitions, attitudes to BIM and expectations of its performance
vary across AEC disciplines (Jiao et al. 2013a; Shirowzhan et al. 2020). For example,
in the design disciplines, BIM is sometimes described as an extension of CAD. In
contrast, the non-design disciplines tend to regard BIM as an intelligent DMS that
can extract data from CAD packages. Because of these conflicting attitudes, BIM
6.2 Digital Design Collaboration and BIM 151

developers and ICT service providers often integrate and balance these two require-
ments. Furthermore, because there is a need for multiple parties to simultaneously
contribute to a single coordinated BIM model, a “concurrent engineering approach”
is often used for BIM model development (Eppinger 1991; Evbuomwan and Anumba
1996). Recently, the cloud approach to unified lifecycle data has become an industry
standard for design collaboration (Chen et al. 2020; Jiao et al. 2013a, b; Liu et al.
2017). Customised business strategy and transformation have also been considered
to suit varying implementation needs. For example, a BIM model can be maintained
in-house or outsourced to a service provider, each having different capacities that
address diverse organisational and financial needs and limitations.
Effective information management and integration in BIM requires the close coop-
eration of the entire project team. Quality communication and knowledge exchange
between different stakeholders (both within and beyond the stand-alone BIM model)
is also essential for successful BIM implementation. It is also critical for collab-
orative or collective participation from different team members (and sometimes
different teams) in a building project lifecycle. In practice, however, many BIM
models are developed in local repositories and are typically accessible by individual
team members through proprietary stand-alone software. This means that the scope
for collaboration is confined to a single isolated BIM model and to ad hoc decen-
tralised and traditional forms of communication such as email, analogue outputs or
other traditional channels of information exchange (El-Diraby et al. 2017). Further-
more, it can be costly to implement a full BIM model, and often the budget is not
available to support its use as a multi-disciplinary lifecycle model. Thus, despite the
aspirations of the AEC sector, and the growing capacity of BIM platforms, in practice
a proportion of BIM models have only partial functionality.
Research and development in BIM is ongoing, and advances in technology and
data interoperability are essential for creating an enhanced understanding of BIM’s
potential and for exploring emerging roles and team dynamics within this platform.
A critical examination of existing project workflows and resourcing capabilities can
also help to define and support these roles and dynamics as well as determine whether
they would be internally or externally resourced. Research also identifies that there
is a need to develop BIM adoption strategies tailored for specific stakeholder needs,
contingent upon the capabilities of their collaborators (Singh et al. 2011). Such
strategies must take into account diverse readiness levels and the challenges and
costs of training for BIM applications and upgrades (Gu and London 2010).
Past research emphasises that the complexities of achieving successful collab-
oration in BIM are not entirely techno-centric. An integrated socio-technological
approach to managing the complex interdependencies across key BIM players is
perhaps of even greater importance for collaboration. The Design Team Cogni-
tion (DTC) model in Chap. 5 highlights this issue and helps us to understand the
dynamic cognitive, collaborative and creative design processes that occur in the
BIM ecosystem.
152 6 Design Thinking and Building Information Modelling

6.3 An Advanced BIM Knowledge Framework

The need for an integrated socio-technological approach to digital design collab-

oration in the AEC industry has been identified in past research (Gu et al. 2015).
In the BIM ecosystem, traditional 2D and 3D CAD exists alongside and is trans-
forming into multi-dimensional modelling platforms (Bortolini et al. 2019; Charef
et al. 2018; Lu et al. 2016; Rinella 2008; Sheikhkhoshkar et al. 2019). The multi-
dimensional platform functionalities, as the previous section notes, include planning
and scheduling (4D), cost estimation (5D), sustainability (6D) and FM (7D) (Charef
et al. 2018; Rinella 2008). The “level of development” (LOD) (sometimes called
“level of detail”) of the BIM ecosystem can also be mapped to reflect its evolutionary
state (Table 6.1) (Eastman et al. 2011; Leite et al. 2011; Volk et al. 2014). This is
significant in the present context, because this chapter seeks to assist design teams
in the implementation of BIM in an evolved ecosystem. In the following sections,
this chapter develops an advanced BIM framework that considers BIM-related prod-
ucts, processes and people in the co-evolutionary collaboration phases of a design
process. These correspond to the range from pre-design (LOD-100) to maintenance
(LOD-500) in Table 6.1.
The new framework developed in this chapter is focussed on two dimensions:
diverse project application and advanced knowledge capture. The first requires a

Table 6.1 LOD definitions and collaboration phases (Source BIMForum 2019)
LOD Definition Related data Collaboration phase
100 The element is represented Non-geometric Pre-design
in the model with a symbol representations
or other generic
representation
200 The element is represented Approximate quantities, Schematic design
within the model as a size, shape, location and
generic system, object or orientation
assembly
300 The element is represented Quantity, size, shape, Design development
within the model as a location and orientation
350 specific system, object or Quantity, size, shape, Construction documentation
assembly location, orientation and
interfaces with other
building systems
400 Quantity, size, shape, Construction
location and orientation
with detailing, fabrication,
assembly and installation
information
500 The Model Element is a Quantity, size, shape, As-built or Maintenance
field verified representation location and orientation
6.3 An Advanced BIM Knowledge Framework 153

consideration of a diversity of BIM applications, from new designs to existing build-

ings. While BIM research and practice have largely addressed new building design, its
application to existing buildings is equally important (Volk et al. 2014). Furthermore,
“Historic BIM” has been proposed to address the integration of historical, cultural
and social parameters into object-oriented models (Biagini et al. 2016; Lee and Gu
2019; Quattrini et al. 2017). “As-built BIM” accommodates additional collabora-
tion phases such as building maintenance, refurbishment and deconstruction, which
are often overlooked in “as-planned” or “as-designed” BIM development. The inte-
gration of as-built BIM alongside as-planned reflects the importance of the whole
lifecycle in an advanced BIM framework. The second dimension of the knowledge
framework is that it adopts emerging ontologies and technologies for enhancing
systems to enable collaborative design thinking and decision-making.

6.3.1 A BIM Knowledge Framework

The BIM knowledge framework (Fig. 6.1) supports digital design collaboration in the
evolving BIM ecosystem. Initially, such collaboration commences by the way of as-
planned BIM, facilitated through the design of a single coordinated building model.
Collaboration is core to BIM, as it serves multiple purposes throughout the design and
construction phases. From community interaction and client engagement to design
modelling, analysis and review, BIM supports inter and multi-disciplinary collab-
oration. After the schematic design stage, BIM supports subcontractor tendering,
construction management and site management. FM operation and maintenance then
uses as-built BIM.
In contrast, traditional as-built or historic BIM is often limited to the maintenance
stage of existing buildings and only occasionally includes a consideration of retrofit
potential or logistics (Logothetis et al. 2015; Volk et al. 2014). That is, both as-
planned and as-built BIM do not always consider the full lifecycle, while the advanced
BIM knowledge framework deals with the project lifecycle spanning either over an
extended period or even on a continuing basis. Thus, the information flow of the
framework is both as-planned and as-built.
BIM operation and collaboration can also be enhanced by adopting new and
improved communication media and technologies (Shafiq et al. 2013). A wide range
of emerging ontologies and technologies have potential for advancing BIM platforms.
Some of these include geomatics, Virtual Reality (VR), Augmented Reality (AR) and
Mixed Reality (MR) technologies, mobile and smartphone applications, groupware,
server technologies and cloud computing (Alizadehsalehi et al. 2020; Lee and Gu
2019). In comparison with as-planned BIM, the support for communication is even
more critical in as-built BIM. This is because support must not only be provided for
building and construction professionals but also for other users including the general
public. The new ontologies and technologies mentioned above are especially useful
when considering social and cultural information in BIM. For example, VR and AR
applications have already been utilised for heritage conservation (Acierno et al. 2017;
 154

InformaƟon modeling Maintenance Refurbishment DeconstrucƟon Design ConstrucƟon

Data collecƟon InformaƟon Planning Procurement/

maintenance fabricaƟon

3D modeling/ Quality control/ Refurbishment EsƟmaƟon/ Design 4D/5D

pre-processing monitoring opƟons recycling workflow scheduling

Data integraƟon DocumentaƟon/simulaƟon/analysis

BIM Plaƞorm

BIM Social Network Public Data InformaƟon

CommunicaƟon Analysis Engagement Management Exchange

Fig. 6.1 Advanced BIM knowledge framework

6 Design Thinking and Building Information Modelling
6.3 An Advanced BIM Knowledge Framework 155

Napolitano et al. 2017). Murphy et al. (2017) propose a semantically enriched frame-
work for historic BIM using a web-based game engine for coordinating visualisation,
communication and participation. These emerging technologies are key to accessing,
engaging and supporting diverse stakeholder groups as well as their communication
and collaboration in the advanced BIM knowledge framework (Fig. 6.1).
The framework is divided into six main phases: information modelling, mainte-
nance, refurbishment, deconstruction, design and construction. While many BIM
frameworks start with a design phase, this framework spans from information
modelling to construction. However, this doesn’t mean the framework necessarily
commences with information modelling. BIM creation can start from any phase, and
the information modelling phase would be the conventional starting point for as-built
BIM. The six phases synthesise the cyclical information flows abstracted from BIM
creation for both new and existing building projects. This defines a formal structure
for creating, maintaining and sharing information of both new and existing buildings.
These phases are supported by the BIM platform (Fig. 6.1) which contains five inde-
pendent but interchangeable modules: (i) BIM-enabled communication, (ii) social
network analysis, (iii) public engagement, (iv) data management and (v) information
exchange. These modules, which are informed by advances in ontology and tech-
nology, are closely connected with the key processes in BIM creation. The following
two sections describe the phases and modules in more detail.

6.3.2 Six Phases of the BIM Knowledge Framework

In the information modelling phase, the goal is to create a BIM model from an
existing building, typically using survey data. Key tasks in this phase can be optimised
through a range of computational techniques such as terrestrial laser scanning and
photogrammetry. The raw survey data produced can be further processed to create
3D digital models from which BIM models can be developed (Dore and Murphy
2017). An alternative approach to as-built models is to use automated modelling
techniques. This requires that data are structured in a machine-readable format, such
as “Resource Data Framework” or “Ontology Web Language”. Modelling the domain
ontology is essential here, and it should carefully consider how end users may read
and query the data set (Quattrini et al. 2017). Once the 3D digital model is complete,
non-geometric information including current building usage and performance, or
heritage and historical records can then be structured and integrated to form the
complete BIM model. In summary, this first phase of the framework involves three
collaborative tasks: data collection, 3D modelling and/or pre-processing, and data
integration. The information modelling phase is typically the starting point for as-built
BIM projects, while new BIM projects typically start with the design phase.
The second phase, maintenance, is important for both new and existing build-
ings. It encapsulates operations, FM, retrofitting and monitoring processes along
with multi-disciplinary activities and information requirements (Becerik-Gerber et al.
2012). As such, the first priority of this phase is to address the management of data
156 6 Design Thinking and Building Information Modelling

that is continuously accumulating and changing during the operation of a building.

Singh et al. (2011) define three operational requirements for a BIM server to manage
this data: (i) BIM model management-related requirements, (ii) design review-related
requirements and (iii) data security-related requirements. The first of these, the BIM
model management-related requirements, is related to the storage, operation and
preservation of a model which contains 3D geometries, 2D documents and non-
geometric building information. The second requirement, design review-related,
includes the functions needed for visualisation and evaluation, as well as for team
communication and interaction during the review process. Finally, data security-
related requirements are important for maintaining the veracity of the overall BIM
model. To connect the maintenance and construction phases in the framework, an
overarching third task group—documentation, simulation and analysis—is used to
aid decision-making. Thus, the maintenance phase of the BIM framework highlights
information maintenance, quality control or monitoring along with documentation,
simulation and analysis.
The refurbishment, deconstruction and design phases start with another overar-
ching task—planning—that connects these three together. Planning considers the
diverse needs and perspectives of the data stakeholders. Through the planning
process, the whole of the building is considered and evaluated, through consulta-
tion and participation of interdisciplinary experts with communities and end users.
In addition, building-specific factors relating to preservation, refurbishment, adap-
tive reuse and deconstruction or demolition are considered. After the planning task
is complete, the team responsible for a refurbishment or reuse task can select various
solutions. Conventional, as-planned BIM, already supports advanced design and
drafting standards for refurbishment or reuse using hierarchical data in the BIM
model (Giuda et al. 2015). To develop an optimal refurbishment or reuse solution,
BIM tools also need to address properties related to time and costs (Khaddaj and Srour
2016). The BIM framework supports decisions of this type for both as-planned and
as-built BIM. This phase ends with the overarching documentation, simulation and
analysis task.
The deconstruction phase in BIM supports the visualisation and estimation of
demolition options as well as the related costs and environmental impacts. The
latter dimension includes identification and implementation of recycling and waste
management solutions, material removal deployment, as well as simulation and
management of building end-of-life alternatives (Akinade et al. 2017). From a
sustainability perspective, efficient building and construction waste management
strategies should be incorporated into this phase of BIM (Jaillon and Poon 2014;
Kibert 2012). Akinade et al. (2017) suggest the use of a “closed material-loop”
process that can eliminate the linear pattern of material movement in demolition,
thereby minimising building and construction waste. In the future, this phase could
also collaborate more effectively with the following design phase to effectively
“designing out waste”—designing and optimising for deconstruction.
The fourth phase, design, consists of three key collaborative works: (i) planning,
(ii) design workflow and (iii) documentation, simulation and analysis. The focus
of this phase is assisting design development, review and evaluation. In contrast,
6.3 An Advanced BIM Knowledge Framework 157

the final construction phase comprises different collaborative tasks concerned with
procurement and fabrication. BIM in this phase adopts 4D/5D scheduling, as well
as documentation, simulation and analysis for monitoring, evaluating and ensuring
on-time delivery.

6.3.3 Five Modules of the BIM Platform

The BIM knowledge framework contains five independent but interchangeable

modules (Fig. 6.1). These five are based on El-Diraby et al.’s (2017) modules for
BIM-based collaborative design and socio-technical analytics. For the new BIM
framework, however, they are revised to accommodate both as-planned and as-built
models.
The first module, BIM-enabled communication, provides support for online
communication and collaboration. Users commonly employ 3D visualisation tools
to explore, identify and reference specific elements in a digital building model (El-
Diraby et al. 2017). Sharing a multi-disciplinary BIM model typically requires the
use of open-standard Industry Foundation Classes (IFC). IFC is a transferrable and,
notionally, universal data format. Most communication features of BIM are derived
from DMS. In practice, standard DMS provides multiple modes of communication
from SMS and email to voice recording and transcription. DMS in BIM supports both
synchronous and asynchronous communication during the project lifecycle. This
communication may include direct knowledge transmission and documentation, as
well as automated notifications and reminders.
The social network analysis module supports social interactions between users
during collaboration, and it also gathers valuable information for subsequent anal-
ysis (Newman 2001). This module accommodates large-scale multi-user scenarios
and is able to address complex relations during mass participation. Many existing
Social Network Service (SNS) applications support high levels of interactivity and
information sharing and collaboration, as well as providing personalisation through
individual queries and the delivery of user-specific content. The subsequent analysis
is used to enhance user experience and support decision-making. In BIM projects,
this module is responsible for the visualisation of project networks and interfaces
(El-Diraby et al. 2017). It is especially significant for as-built BIM projects, because
of the increase of user types and participant numbers.
Public engagement is a special module in the BIM platform for improving engage-
ment with the non-AEC community. These people include experts from other disci-
plines as well as the general public. The latter may be especially interested in cultural
and social information in as-built BIM data. To facilitate this level of engagement,
a range of multi-modal media and technologies can be considered and incorporated
to suit different user needs and behavioural patterns. For example, Keil et al.’s AR-
based architectural history model (2011) and web-based game-engine BIM plat-
form (Murphy et al. 2017) and Napolitano et al.’s virtual tours of cultural sites
(2017) all demonstrate the effectiveness of media-rich technologies for supporting
heritage conservation. As discussed previously in this chapter, Geographic Informa-
tion Systems (GIS), Global Positioning Systems (GPS), Location-Based Services
158 6 Design Thinking and Building Information Modelling

(LBS), VR, AR, and mobile or cloud computing are associated with this module. In
contrast, the data management module is largely used by building and construction
professionals. It supports the monitoring, exploration, optimisation and evaluation
of a new design, or of different intervention strategies and their implementation
processes.
The information exchange module serves information sharing and exchange func-
tions across different building data sets and different disciplines. Ontologies play an
important role in this module because they enable the creation of formal structures,
principles and rules for information exchange. They also provide the capacity to
identify and overcome constraints in BIM tools and systems (Li et al. 2017). In other
words, sound ontologies in BIM provide the foundation for effective communication
between stakeholders and enhanced interoperability between systems and data sets
(Succar 2009).
Collectively, these five modules in a BIM platform are not only essential for digital
design collaboration but applicable to collaborative design thinking in a broader
digital ecosystem. Thus, they contribute to the development of technology-enabled
collaboration as well as increased understanding of design thinking arising from
digital technologies. The following section discusses the use of the BIM framework
and the conceptual transformation of BIM into a Digital Twin.

6.4 BIM Futures

1. The BIM framework introduced in this chapter has several potential applications.
For example, it can be used to understand and structure a project team in order
to promote improved design thinking and collaboration. It also has the capacity
to support an integrated socio-technological approach to BIM, which enables
the adoption and implementation of advances relating to products, processes
and people. Such advances include cloud computing, machine learning, artificial
intelligence and cognitive computing. In addition, advances in “extended reality”
(Alizadehsalehi et al. 2020), “nD BIM” (Ghaffarian Hoseini et al. 2017), “Digital
Twins” (Grieves 2014; Qi and Tao 2018; Tao et al. 2018) and mobile computing
(Cook and Das 2007) can be conceptualised using this framework. The most
important of these developments is the Digital Twin.
2. The concept of a Digital Twin was first proposed by Michael Grieves in the
early 2000s for product design. In essence, a Digital Twin is a digital or virtual
replica of a physical product or process. Since being proposed as a product
lifecycle management concept (Grieves 2014), the Digital Twin has been adopted
by technology developers such as IBM and Dassault Systèmes as well as major
organisations like NASA. Recently, the application of Digital Twins has extended
beyond product design to other domains for simulating a wide range of physical,
social and economic systems (Batty 2018). From buildings, precincts and cities
to commercial organisations and communities, the concept of a Digital Twin has
grown to describe a category of data model of which BIM could be regarded as
a growing subset.
6.4 BIM Futures 159

Digital Twins can be distinguished from common simulation models through

their capacity for real-time input, which relies on the close and rapid interaction
between digital and physical counterparts. As such, through real-time data integra-
tion, analytics and machine learning, Digital Twins are moving “closer and closer
to the real thing” (Batty 2018, p. 819). The benefit of this, unlike BIM which either
lags behind real-time or is only used for simulation, is that the Digital Twin can
better inform our decision-making processes. This poses a challenge for BIM, to
be able to accommodate real-time data and adaptive or responsive modelling, which
extend beyond design and construction. The BIM framework presented in this chapter
provides a starting point for considering a Digital Twin–BIM hybrid, for both new
designs and existing buildings.
Looking into the future, the enduring value of BIM may be associated with
its capacity to support design collaboration. The following are just three thoughts
about new research directions that could enhance BIM to better support design and
collaboration.
• Critical knowledge about design cognition and representation—the cognitive,
visual and verbal—for design teams (as discussed in Chap. 5) should be carefully
integrated into BIM systems. In combination with Herrmann’s model (1989,1991)
of “thinking dominances”, it may provide a better design environment for different
designers and design team dynamics.
• In terms of communication, current BIM systems are largely concerned with
Chiu’s (2002) project level of communication. Current BIM collaboration is also
mostly limited to professional teams. With the design context becoming more
diverse and the broadening of the project team and end-user group, communication
on the individual and group levels should be more adequately considered and inte-
grated into the overall ecosystem. This is important for promoting “collaborative
design thinking”.
• This book identifies several strategies and processes for supporting “creative
design thinking” in parametric design (Chaps. 2 and 3) and essential factors for
supporting cognitive complexity (Chap. 4). There are significant opportunities
for enhancing the BIM ecosystem to more explicitly support and encourage such
strategies and activities, and to effectively execute them in both individual and
collaborative modes.

6.5 Conclusion

This chapter has presented a new integrated BIM knowledge framework for facili-
tating digital design collaboration in the BIM ecosystem. Acknowledging that digital
technologies and our social systems are continuously evolving, emerging BIM-
related technologies are unified in this framework. In particular, the five modules—
BIM-enabled communication, social network analysis, public engagement, data
management and information exchange—can be extended to accommodate multi-
dimensional BIM creation, ranging from 4D/5D to nD BIM. Moreover, the Digital
160 6 Design Thinking and Building Information Modelling

Twin is a clear future direction of BIM as well as of the wider digital ecosystem, and
modules could be developed to begin to understand its implications.
The new framework proposed in this chapter contributes to addressing the entire
BIM lifecycle, specifically focusing on sharing, preserving and reusing building data
beyond the design, construction and maintenance phases. However, in proposing this
framework, it highlights that further research on interoperability and data exchange
across multiple collaboration phases within a BIM platform is needed.
Finally, the future of BIM, and its hybrid Digital Twin-BIM, is as a multi-
disciplinary collaboration platform that allows interactive creative processes with
an increased number and range of participants. Through the involvement of a more
diverse group of participants, and the subsequent knowledge sharing this entails, the
integrated collaboration platform can support new commercial, cultural and social
values beyond the intelligent model or the building project. These properties of
interactive and collective digital collaboration platforms are explored in the next
chapter.

References

Acierno, Marta, Stefano Cursi, Davide Simeone, and Donatella Fiorani. 2017. Architectural heritage
knowledge modelling: An ontology-based framework for conservation process. Journal of
Cultural Heritage 24: 124–133. https://doi.org/10.1016/j.culher.2016.09.010.
Akinade, Olugbenga O., Lukumon O. Oyedele, Kamil Omoteso, Saheed O. Ajayi, Muhammad
Bilal, Hakeem A. Owolabi, Hafiz A. Alaka, Lara Ayris, and John Henry Looney. 2017. BIM-
based deconstruction tool: Towards essential functionalities. International Journal of Sustainable
Built Environment 6 (1): 260–271. https://doi.org/10.1016/j.ijsbe.2017.01.002.
Alizadehsalehi, Sepehr, Ahmad Hadavi, and Joseph Chuenhuei Huang. 2020. From BIM to extended
reality in AEC industry. Automation in Construction 116: 103254. https://doi.org/10.1016/j.aut
con.2020.103254.
Batty, Michael. 2018. Digital twins. Environment and Planning B: Urban Analytics and City Science
45 (5): 817–820. https://doi.org/10.1177/2399808318796416.
Becerik-Gerber, Burcin, Farrokh Jazizadeh, Nan Li, and Gulben Calis. 2012. Application areas and
data requirements for bim-enabled facilities management. Journal of Construction Engineering
and Management 138 (3): 431–442. https://doi.org/doi:10.1061/(ASCE)CO.1943-7862.000043
3.
Biagini, Carlo, Pietro Capone, Vincenzo Donato, and Nora Facchini. 2016. Towards the BIM imple-
mentation for historical building restoration sites. Automation in Construction 71: 74–86. https://
doi.org/10.1016/j.autcon.2016.03.003.
Bilda, Z., and H. Demirkan. 2003. An insight on designers’ sketching activities in traditional versus
digital media. Design Studies 24 (1): 27–50.
BIMForum. 2019. Level of Development (LOD) Specification Part I & Commentary: For Building
Information Models and Data.
Bortolini, Rafaela, Carlos Torres Formoso, and Daniela D. Viana. 2019. Site logistics planning and
control for engineer-to-order prefabricated building systems using BIM 4D modeling. Automation
in Construction 98: 248–264. https://doi.org/10.1016/j.autcon.2018.11.031.
Cerovsek, Tomo. 2011. A review and outlook for a ‘Building Information Model’ (BIM): A multi-
standpoint framework for technological development. Advanced Engineering Informatics 25 (2):
224–244. https://doi.org/10.1016/j.aei.2010.06.003.
References 161

Charef, Rabia, Hafiz Alaka, and Stephen Emmitt. 2018. Beyond the third dimension of BIM:
A systematic review of literature and assessment of professional views. Journal of Building
Engineering 19: 242–257. https://doi.org/10.1016/j.jobe.2018.04.028.
Chen, Yi-Jao., Yong-Shan. Lai, and Yen-Han. Lin. 2020. BIM-based augmented reality inspection
and maintenance of fire safety equipment. Automation in Construction 110: 103041. https://doi.
org/10.1016/j.autcon.2019.103041.
Chiu, Mao-Lin. 2002. An organizational view of design communication in design collaboration.
Design Studies 23 (2): 187–210. https://doi.org/10.1016/S0142-694X(01)00019-9.
Cook, Diane J., and Sajal K. Das. 2007. How smart are our environments? An updated look at the
state of the art. Pervasive and Mobile Computing 3 (2): 53–73. https://doi.org/10.1016/j.pmcj.
2006.12.001.
Cooper, Rachel, Sabine Junginger, and Thomas Lockwood. 2009. Design thinking and design
management: A research and practice perspective. Design Management Review 20 (2): 46–55.
https://doi.org/10.1111/j.1948-7169.2009.00007.x.
Daniotti, Bruno, Marco Gianinetto, and Stefano Della Torre. 2020. Digital Transformation of the
Design, Construction and Management Processes of the Built Environment. Cham: Springer
International Publishing.
Dore, C., and M. Murphy. 2017. Current state of the art historic building information modelling. The
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
XLII-2/W5:185–192. https://doi.org/10.5194/isprs-archives-XLII-2-W5-185-2017.
Eastman, Chuck, Paul Teicholz, Rafael Sacks, and Kathleen Liston. 2011. BIM Handbook: A Guide
to Building Information Modeling for Owners, Managers, Designers, Engineers and Contractors.
Hoboken, NJ: Wiley.
El-Diraby, Tamer, Thomas Krijnen, and Manos Papagelis. 2017. BIM-based collaborative design
and socio-technical analytics of green buildings. Automation in Construction 82: 59–74. https://
doi.org/10.1016/j.autcon.2017.06.004.
Eppinger, Steven D. 1991. Model-based approaches to managing concurrent engineering. Journal
of Engineering Design 2 (4): 283–290. https://doi.org/10.1080/09544829108901686.
Evbuomwan, Nosa, and Chimay J. Anumba. 1996. Towards a concurrent engineering model for
design and build projects. Structural Engineer 74 (5): 73–78.
GhaffarianHoseini, Ali, Tongrui Zhang, Okechukwu Nwadigo, Amirhosein GhaffarianHoseini,
Nicola Naismith, John Tookey, and Kaamran Raahemifar. 2017. Application of nD BIM Integrated
Knowledge-based Building Management System (BIM-IKBMS) for inspecting post-construction
energy efficiency. Renewable and Sustainable Energy Reviews 72: 935–949. https://doi.org/10.
1016/j.rser.2016.12.061.
Giuda, Giuseppe Martino, Valentina Villa Di, and Paolo Piantanida. 2015. BIM and energy efficient
retrofitting in school buildings. Energy Procedia 78: 1045–1050. https://doi.org/10.1016/j.egy
pro.2015.11.066.
Grieves, Michael. 2014. Digital twin: manufacturing excellence through virtual factory replication.
White Paper: 1–7.
Grudin, J. 1994. Computer-supported cooperative work: History and focus. Computer 27 (5): 19–26.
https://doi.org/10.1109/2.291294.
Gu, Ning, and Kerry London. 2010. Understanding and facilitating BIM adoption in the AEC
industry. Automation in Construction 19 (8): 988–999. https://doi.org/10.1016/j.autcon.2010.
09.002.
Gu, Ning, Vishal Singh, and Kerry London. 2015. BIM ecosystem: The coevolution of products,
processes, and people. In Building Information Modeling, 197–210. Wiley.
Herrmann, Ned. 1989. The Creative Brain. Buffalo, NY: Brain books.
Herrmann, Ned. 1991. The creative brain*. The Journal of Creative Behavior 25 (4): 275–295.
https://doi.org/10.1002/j.2162-6057.1991.tb01140.x.
Ibrahim, R., and F. Pour Rahimian. 2010. Comparison of CAD and manual sketching tools for
teaching architectural design. Automation in Construction 19 (8): 978–987.
162 6 Design Thinking and Building Information Modelling

ISO, 29481–1. 2016. Building information models—Information delivery manual—Part 1: Method-

ology and format.
Jaillon, Lara, and C.S. Poon. 2014. Life cycle design and prefabrication in buildings: A review and
case studies in Hong Kong. Automation in Construction 39: 195–202. https://doi.org/10.1016/j.
autcon.2013.09.006.
Jang, Ryan, and William Collinge. 2020. Improving BIM asset and facilities management processes:
A Mechanical and Electrical (M&E) contractor perspective. Journal of Building Engineering:
101540. https://doi.org/10.1016/j.jobe.2020.101540.
Jiao, Yi., Yinghui Wang, Shaohua Zhang, Yin Li, Baoming Yang, and Lei Yuan. 2013a. A cloud
approach to unified lifecycle data management in architecture, engineering, construction and
facilities management: Integrating BIMs and SNS. Advanced Engineering Informatics 27 (2):
173–188. https://doi.org/10.1016/j.aei.2012.11.006.
Jiao, Yi., Shaohua Zhang, Yongkui Li, Yinghui Wang, and BaoMing Yang. 2013b. Towards cloud
Augmented Reality for construction application by BIM and SNS integration. Automation in
Construction 33: 37–47. https://doi.org/10.1016/j.autcon.2012.09.018.
Keil, J., M. Zollner, M. Becker, F. Wientapper, T. Engelke, and H. Wuest. 2011. The House of
Olbrich—An augmented reality tour through architectural history. In 2011 IEEE International
Symposium on Mixed and Augmented Reality—Arts, Media, and Humanities.
Khaddaj, Maggie, and Issam Srour. 2016. Using BIM to retrofit existing buildings. Procedia
Engineering 145: 1526–1533. https://doi.org/10.1016/j.proeng.2016.04.192.
Kibert, C.J. 2012. Sustainable Construction: Green Building Design and Delivery. Wiley.
Kiviniemi, Arto, Jan Karlshøj, Väino Tarandi, Håvard Bell, and Ole Jørgen Karud. 2008. Erabuild:
Review of the Development and Implementation of IFC compatible BIM.
Lee, Ju Hyun, and Ning Gu. 2019. Historical BIM+: Sharing, preserving and reusing archi-
tectural design data. In Partners for Preservation: Advancing Digital Preservation Through
Cross-Community Collaboration, ed. Jeanne Kramer-Smyth, 124–144. London: Facet Publishing.
Leite, Fernanda, Asli Akcamete, Burcu Akinci, Guzide Atasoy, and Semiha Kiziltas. 2011. Anal-
ysis of modeling effort and impact of different levels of detail in building information models.
Automation in Construction 20 (5): 601–609. https://doi.org/10.1016/j.autcon.2010.11.027.
Li, Xiao, Wu. Peng, Geoffrey Qiping Shen, Xiangyu Wang, and Yue Teng. 2017. Mapping the knowl-
edge domains of Building Information Modeling (BIM): A bibliometric approach. Automation
in Construction 84: 195–206. https://doi.org/10.1016/j.autcon.2017.09.011.
Liu, Rui, Du. Jing, Raja R.A. Issa, and Brittany Giel. 2017. BIM cloud score: Building informa-
tion model and modeling performance benchmarking. Journal of Construction Engineering and
Management 143 (4): 04016109. https://doi.org/doi:10.1061/(ASCE)CO.1943-7862.0001251.
Logothetis, S., A. Delinasiou, and E. Stylianidis. 2015. Building information modelling for cultural
heritage: A review. In The ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial
Information Sciences II-5/W3:177–183. https://doi.org/10.5194/isprsannals-II-5-W3-177-2015.
Lu, Qiqi, Jongsung Won, and Jack C.P. Cheng. 2016. A financial decision making framework for
construction projects based on 5D Building Information Modeling (BIM). International Journal
of Project Management 34 (1): 3–21. https://doi.org/10.1016/j.ijproman.2015.09.004.
Martinez-Maldonado, Roberto, Peter Goodyear, Lucila Carvalho, Kate Thompson, Davinia
Hernandez-Leo, Yannis Dimitriadis, Luis P. Prieto, and Dewa Wardak. 2017. Supporting collab-
orative design activity in a multi-user digital design ecology. Computers in Human Behavior 71:
327–342. https://doi.org/10.1016/j.chb.2017.01.055.
Murphy, M., A. Corns, J. Cahill, K. Eliashvili, A. Chenau, C. Pybus, R. Shaw, G. Devlin, A. Deevy,
and L. Truong-Hong. 2017. Developing historic building information modelling guidelines and
procedures for architectural heritage in Ireland. In The International Archives of the Photogram-
metry, Remote Sensing and Spatial Information Sciences XLII-2/W5:539–546. https://doi.org/
10.5194/isprs-archives-XLII-2-W5-539-2017.
Napolitano, Rebecca K., George Scherer, and Branko Glisic. 2017. Virtual tours and informational
modeling for conservation of cultural heritage sites. Journal of Cultural Heritage. https://doi.org/
10.1016/j.culher.2017.08.007.
References 163

Newman, M.E.J. 2001. The structure of scientific collaboration networks. Proceedings of the
National Academy of Sciences 98 (2): 404–409. https://doi.org/10.1073/pnas.98.2.404.
Ning, Gui, He Kan, Qiu Zhifeng, Gui Weihua, and Deconinck Geert. 2018. e-BIM: A BIM-centric
design and analysis software for building integrated photovoltaics. Automation in Construction
87: 127–137. https://doi.org/10.1016/j.autcon.2017.10.020.
Qi, Q., and F. Tao. 2018. Digital twin and Big Data towards smart manufacturing and industry 4.0:
360 degree comparison. IEEE Access 6: 3585–3593. https://doi.org/10.1109/ACCESS.2018.279
3265.
Quattrini, Ramona, Roberto Pierdicca, and Christian Morbidoni. 2017. Knowledge-based data
enrichment for HBIM: Exploring high-quality models using the semantic-web. Journal of
Cultural Heritage 28: 129–139. https://doi.org/10.1016/j.culher.2017.05.004.
Rinella, Tony. 2008. Computer Technology in Practice. In The Architect’s Handbook of Professional
Practice, ed. Joseph A. Demkin, 417–428. Hoboken, NJ: Wiley.
Shafiq, M.T., J. Matthews, and S.R. Lockley. 2013. A study of BIM collaboration requirements and
available features in existing model collaboration systems. Journal of Information Technology in
Construction 18: 148–161.
Sheikhkhoshkar, Moslem, Farzad Pour Rahimian, Mohammad Hossein Kaveh, M. Reza Hosseini,
and David J. Edwards. 2019. Automated planning of concrete joint layouts with 4D-BIM.
Automation in Construction 107: 102943. https://doi.org/10.1016/j.autcon.2019.102943.
Shirowzhan, Sara, Samad M.E. Sepasgozar, David J. Edwards, Heng Li, and Chen Wang. 2020.
BIM compatibility and its differentiation with interoperability challenges as an innovation factor.
Automation in Construction 112: 103086. https://doi.org/10.1016/j.autcon.2020.103086.
Simeone, Davide, Stefano Cursi, and Marta Acierno. 2019. BIM semantic-enrichment for built
heritage representation. Automation in Construction 97: 122–137. https://doi.org/10.1016/j.aut
con.2018.11.004.
Singh, Vishal, Ning Gu, and Xiangyu Wang. 2011. A theoretical framework of a BIM-based multi-
disciplinary collaboration platform. Automation in Construction 20 (2): 134–144. https://doi.org/
10.1016/j.autcon.2010.09.011.
Succar, Bilal. 2009. Building information modelling framework: A research and delivery foundation
for industry stakeholders. Automation in Construction 18 (3): 357–375. https://doi.org/10.1016/
j.autcon.2008.10.003.
Tang, Pingbo, Daniel Huber, Burcu Akinci, Robert Lipman, and Alan Lytle. 2010. Automatic
reconstruction of as-built building information models from laser-scanned point clouds: A review
of related techniques. Automation in Construction 19 (7): 829–843. https://doi.org/10.1016/j.aut
con.2010.06.007.
Tang, Fanlong, Tao Ma, Yongsheng Guan, and Zhixiang Zhang. 2020. Parametric modeling and
structure verification of asphalt pavement based on BIM-ABAQUS. Automation in Construction
111: 103066. https://doi.org/10.1016/j.autcon.2019.103066.
Tao, Fei, Jiangfeng Cheng, Qinglin Qi, Meng Zhang, He Zhang, and Fangyuan Sui. 2018. Digital
twin-driven product design, manufacturing and service with big data. The International Journal
of Advanced Manufacturing Technology 94 (9): 3563–3576. https://doi.org/10.1007/s00170-017-
0233-1.
Utkucu, Duygu, and Hatice Sözer. 2020. Interoperability and data exchange within BIM platform
to evaluate building energy performance and indoor comfort. Automation in Construction 116:
103225. https://doi.org/10.1016/j.autcon.2020.103225.
Verstegen, Luuk, Wybo Houkes, and Isabelle Reymen. 2019. Configuring collective digital-
technology usage in dynamic and complex design practices. Research Policy 48 (8): 103696.
https://doi.org/10.1016/j.respol.2018.10.020.
Volk, Rebekka, Julian Stengel, and Frank Schultmann. 2014. Building Information Modeling (BIM)
for existing buildings—Literature review and future needs. Automation in Construction 38: 109–
127. https://doi.org/10.1016/j.autcon.2013.10.023.
Won, P.H. 2001. The comparison between visual thinking using computer and conventional media
in the concept generation stages of design. Automation in Construction 10 (3): 319–325. https://
doi.org/10.1016/S0926-5805(00)00048-0.
Chapter 7
Design Thinking and the Digital
Ecosystem

Abstract Because design thinking is contextual, insomuch as it varies depending on

the tools being used and the environments that support it, there is a need to understand
the cognitive impacts of any new platforms that are developed. The focus of this
chapter is design thinking in digital, interactive and collective platforms. Drawing
on past research, models and theories, it proposes two conceptual frameworks for
facilitating interactive and collective design thinking in a digital ecosystem. The new
frameworks of Digital Design Thinking (DDT) processes and functionalities are then
used to examine six examples of collective and interactive platforms. This chapter
contributes to a better understanding of creative, collaborative design thinking using
digital tools and within digital environments.

7.1 Introduction

The rapid evolution of communication and networked technologies is changing our

world, redefining the ways we create, access and trade digital content. This digital-
isation of everyday life also challenges traditional approaches to design creativity,
collaboration and culture. This chapter introduces and explores the concept of Digital
Design Thinking (DDT), which extends conventional concepts of design thinking to
encompass both digital environments and tools as well as collaborative approaches
to innovation and creativity. The specific environments and tools considered in this
chapter are interactive and collective platforms. The relationship between these plat-
forms and DDT also encompasses a consideration of Collective Intelligence (CI).
These three themes—DDT, CI and design platforms that operate in the digital design
ecology—are briefly introduced in this first section.
A technical definition of DDT is that it is a “non-typological and non-
deterministic” approach, which emphasises the use of “discrete and differentiated”
components to develop, explore and test design propositions (Oxman 2006b, p. 262).
This definition contains traces of both computational design’s heuristic and algo-
rithmic modes and classical design’s cognitive and behavioural approaches. More
recently, however, the definition of DDT has been expanded to accommodate the
types of social creativity that are enabled by digital networks. In conventional

© Springer Nature Switzerland AG 2020 165

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_7
166 7 Design Thinking and the Digital Ecosystem

design thinking some well-known strategies for social creativity include participatory
design (Luck 2018; Smith and Iversen 2018), co-creation (Nonaka and Konno 1998;
Prahalad and Ramaswamy 2004) and co-design (Huybrechts et al. 2017; Mitchell
et al. 2016). In the contemporary world, however, social design interactions are
increasingly occurring online using specialised digital design platforms. This is why
a more recent definition of DDT describes it as a cognitive process of “collective
creativity” that is supported by digital platforms (Nakakoji et al. 2000). The present
chapter adopts this expanded definition of DDT and investigates it in the context of
the world’s growing digital ecology.
The concept of a digital ecology is central to some of the most influential business
models of the last two decades. Its core components include the commodification
of digital processes, support for collective and interactive platforms (Markus and
Loebbecke 2013) and the use of distributed supply-chains to create a “sharing econ-
omy” (Richter et al. 2017). The technologies that enable these business models—
mobile computing, social media, “internet of things” and big data—are now so ubiq-
uitous that their collective agency is almost taken for granted. The concept of a digital
ecosystem reflects this omnipresence, describing a connected network where digital
actors (customers, partners and providers), their activities and online organisations,
evolve to sustain production (Elia et al. 2020).
The technologies that enable the digital ecosystem, and which also support DDT,
encompass artefacts, infrastructures and platforms (Nambisan et al. 2017). The
artefacts are the products, both physical and virtual, produced or traded within
the ecology. Some of the key characteristics of these artefacts are that they are
malleable, flexible or adaptable. The infrastructure is the set of digital systems that
allow communication and collaboration. These can range from cloud computing to
physical, networked makerspaces. The platforms are the services and architectures
that support or merge with the infrastructure and are the foundations for the DDT
environment. In combination, the artefacts, infrastructures and platforms sustain the
digital economy and are core to its capacity for innovation (Nambisan et al. 2017;
Tiwana et al. 2010; Verstegen et al. 2019).
The focus of this chapter is interactive and collective platforms, which are the parts
of the digital ecology that support collaboration. Collaboration in the digital ecology
is significant because it has been scaled up and linked to CI and innovation (Alag
2008; Halpin 2008; Heylighen 1999; Lévy 2010). CI is core to “distributed partici-
patory design” and its more common counterpart, “crowdsourcing” (Farshchian and
Divitini 1999). The design industry is an important part of the digital ecology and it
is not surprising that design research should be interested in CI as a component of
DDT.
Pierre Lévy defines CI as “the capacity of human collectives to engage in intel-
lectual cooperation in order to create, innovate and invent” (Lévy 2010, p. 71). More
than just a simple aggregation of individuals, in the digital ecology CI combines
the social network with the design environment itself (Halpin 2008). CI often uses
dynamic, user-created and responsive systems (e.g. “Web 2.0”) to improve social
network services (Lopez Flores et al. 2015b) and support collaborative innovation
networks (Gloor 2006). Like artificial intelligence and open-platform architecture,
7.1 Introduction 167

CI is frequently associated with innovation (Hüsig and Kohn 2011; Leon 2009; Lopez
Flores et al. 2015a, b; Wang et al. 2002).
This introduction to DDT, CI and collaborative and interactive platforms, sets
the scene for the present chapter. As discussed previously in Part I of this book,
design thinking in “digital” (scripting) environments differs from design thinking in
“traditional” (pen-and-paper) environments (Oxman 2006a, b; Lee et al. 2013, 2014a;
Reffat 2006). Design thinking in immersive, collaborative, shared and responsive
environments (the digital ecosystem) is also dissimilar to its more traditional co-
located, face-to-face version. This is why the present chapter commences by adapting
an analytical framework for understanding DDT and creative, collaborative design.
The framework encompasses key DDT characteristics identified in past research on
digital collaboration and CI. A second framework is then developed to investigate
DDT functionalities in interactive and collective platforms. Finally, both frameworks
are used to compare six examples of interactive art and media platforms. The chapter
concludes with a discussion of the implications of these examples for DDT and
further applications of the frameworks.

7.2 A Digital Design Thinking (DDT) Framework

This section presents two conceptual frameworks for understanding and analysing
DDT in the digital ecosystem. The first—adapted and expanded versions of digital
design models in the literature—is used for identifying DDT processes. The second
identifies DDT functionalities in interactive and collective platforms.

7.2.1 Types of DDT Processes

As a precursor to proposing a framework for DDT processes, it is useful to consider

past research that theorises relationships between digital environments, design and
CI. Five past theories are reviewed in this section, all with different intentions, but
which cover dimensions of DDT. The five theories are from Oxman (2006b), Burdick
and Willis (2011), Fischer et al. (2005), Balestrini et al. (2017) and Alag (2008).
Firstly, Oxman’s (2006b) “integrated compound” model combines DDT processes
of formation, generation, evaluation and performance, describing interactions and
information flows in an integrated digital system. For the purpose of improving
form generation, she identifies the performance-based process as the best (Oxman
2007). However, despite setting up these parameters, Oxman’s five types of DDT
models (Table 3.7) are closer in spirit to those of computational designers, than the
current group of collaborative DDT models which are more concerned with shared
design processes in distributed, ubiquitous and mobile environments. Thus, Oxman’s
framework is valuable, but it does not accommodate the CI processes or the digital
ecosystem that are so important today.
168 7 Design Thinking and the Digital Ecosystem

Burdick and Willis (2011) propose a framework that, while not explicitly
concerned with DDT, offers a epistemic understanding of its strategies and processes.
They identify three design thinking models to support new modes of learning: (i)
interpretive (performative and rhetorical), (ii) situated (networked and contingent)
and (iii) user-oriented. Their framework, despite being intended for new-media
education and digital-humanities research, accommodates the idea that CI—by way
of the Internet and social networking—can emerge through the open contribution and
participation of a large group (Alag 2008; Elia et al. 2020; Lévy 1997; Malone et al.
2010). While emergent CI is not the same as the “wisdom of crowds” (Surowiecki
2005), both operate through decentralised user engagement. The difference is that
CI produces aggregative knowledge through increased interactivity, whereas the
“wisdom of crowds” is often used to describe a reductive group problem-solving
process (Lee and Chang 2010).
In order to differentiate between types of social creativity Fischer et al. (2005)
propose five models: (i) the “fish-scale” model of overlapping flexible collaboration;
(ii) the structural model based on communities (“of practice” or “of interest”); (iii) the
defined objective model, which is supported by distributed cognition; (iv) the seeding,
evolutionary growth and reseeding (SER) process model; and (v) the meta-design
model. Across these collective cognitive models, CI is one of the main mechanisms
that supports interactions between users Fischer et al. (2005). Also highlight the
distributed nature of interactions and the need for open and transparent access to
systems.
Balestrini et al. (2017) identify six cyclic steps (identification, framing, design,
deployment, orchestration and outcome) for exploring public engagement on
community issues. This model is of interest because Satnam Alag (2008) highlights
a similar cyclic CI framework, where one user influences others through the produc-
tion of reviews, ratings, recommendations or blogs. Alag (2008) also proposes three
ways of using CI: (i) allowing interaction; (ii) aggregating contributions; and (iii)
leveraging models to develop personalised content.
Using Alag’s (2008) work as a foundation, and informed by the other models
described previously in this section, it is possible to adapt his interactive and
collective components to propose a DDT process model (Fig. 7.1).
• The interactive side of the DDT model supports learning about each user’s connec-
tions, transferences and contributions in CI platforms. Its detailed content encom-
passes concepts of individual creativity discussed in Part I of this book, and also
has overlap with Oxman’s digital design models.
• The collective side of the DDT model is aligned to the Design Team Cognition
(DTC) model for collaborative design (introduced in Chap. 5). It also embodies
the main theories used for social creativity and creative interactions between an
individual and society (Fischer et al. 2005).
A core sequence in this DDT and CI process framework is interacting, collecting and
leveraging (I-C-L) (Fig. 7.1).
7.2 A Digital Design Thinking (DDT) Framework 169

(iii) Leveraging CollecƟve

Personalised content Intelligence

(i) InteracƟng (ii) CollecƟng

User InteracƟve ColleccƟve
DDT model DDT model
User

Contribute and interact Contribute and interact

Content

Fig. 7.1 A cyclic I-C-L (Interacting, Collecting and Leveraging) process in digital platforms,
adapted from “three components to harnessing collective intelligence” (Alag 2008, p. 11)

Table 7.1 Digital Design Thinking (DDT) processes identified in past research
Study Elia et al. (2020) Elia and Margherita (2018) Stelzle et al. (2017)
Type Digital collaboration Problem-solving Collective design
DDT process • Conceptualising • Problem identification and • Project initiation
• Creating conceptualisation • Co-briefing
• Deciding • Problem analysis and study • Co-designing
• Inspiring • Problem synthesis and modelling • Professional
• Networking • Solutions proposition and design
• Recommending definition • Ranking voting
• Sharing • Solutions prototyping and Test • Integration
• Suggesting • Solution implementation • Approving
• Transferring • Solution maintenance • Formal assessment

In addition to the CI processes that support DDT, three further processes are
significant in the present context: digital collaboration, problem-solving and collec-
tive design (CD) (Table 7.1). For the first of these, Elia et al. (2020) identify the
importance of digital collaboration in their CI model of the digital entrepreneurship
ecosystem. They divide the process of digital collaboration into ten steps, flows or
actions—conceptualising, creating, deciding, inspiring, networking, recommending,
sharing, suggesting and transferring. For the second of the three processes, the role
of CI in problem-solving, Elia and Margherita (2018) identify seven steps in a CI
system to support collective problem analysis and solution generation. These steps are
problem identification, analysis, synthesis, solution proposition, prototyping, imple-
mentation and maintenance. They are similar to those in the famous five-stage design
thinking model: empathise, define, ideate, prototype and test. Each step requires an
appropriate application and leads to the production of solutions to complex or non-
linear problems. For example, a “sematic aggregator” for problem formulation and
conceptualisation leads to a shared problem definition. Then a “parameter graph
generator” develops the taxonomy of factors and variables for problem analysis. In
this way, a problem visualiser, solution matrix, solution builder, application monitor
and performance assessment are all components of the CI system. For the third
170 7 Design Thinking and the Digital Ecosystem

Asynchronous global

Asynchronous local

Synchronous global

Synchronous local

Content
Ex
tri

ra
ns

rc
ic
In

hy
cr
tri

ow
ns

User
d
ic
In

h
tri

ie
ra
ns

rc
ic

hy
cr
ow
d

Fig. 7.2 Three functionalities of interactive and collective platforms

process, CD, Stelzle et al. (2017) develop a platform identifying co-design and co-
decision processes. They focus on the importance of convergent decision-making
after divergent ideation, along with key types of decisions in CD. The decision types
involve prioritising/ranking, solution selection, criteria/value setting and stakeholder
setting.
These three DDT processes, supported and promoted by digital collaboration and
CI platforms, facilitate individual and social creativity. Collectively, these processes
are all dependent on the functionalities of digital platforms. Which is why four DDT
processes—(i) I-C-L (Fig. 7.2), (ii) digital collaboration, (iii) problem-solving and
(iv) collective design—must be accommodated in the framework. The following
section defines the key functionalities of interactive and collective platforms which
make up the second framework.

7.2.2 Key Functionalities of Interactive and Collective

Platforms

The platforms that currently support DDT in design have often arisen from advances
in Computer-Aided Design (CAD) that have enabled collaborative and synchronous
designing in real-time virtual environments (Reffat 2006). There are, however, other
7.2 A Digital Design Thinking (DDT) Framework 171

types of platforms which are also significant for understanding DDT. For example,
Morschheuser et al. (2017) identify “gamification” and “crowdsourcing” as drivers
for particular interactive, collective platforms. Both are related to crowd-rating (by
way of collective homogeneous inputs) and crowd-creating (by way of collective
heterogeneous inputs). The former is more concerned with immersive or contextual
interaction (collective interaction), while the latter emphasises emergent, collective
activities (interactive collectivity). An alternative example of a design platform is Lee
and Chang’s (2010) genetic algorithm for CI that uses both (i) interactive behaviours
between designers and a design environment and (ii) collective, effective responses
from clients and users.
Interactive and collective platforms developed to support CI have three compa-
rable abilities: (i) increased intelligence, (ii) increased sense and (iii) collective
evaluation.
• “Increased intelligence”, like the wisdom of crowds, refers to the emergence of CI
through the volume of participants making contributions to a synergistic solution,
like user-generated content (for example, Wikipedia or YouTube). That is, CI can
be regarded as socially extended knowledge.
• “Increased sense” refers to the capacity to expand data collection using digital
platforms and devices. For example, a smartphone may have embedded sensors
such as an accelerometer, electronic compass, gyroscope and image sensor. Many
Location-Based Services (LBSs) and Social Networking Services (SNSs) on
smartphones interact with unrestricted information created by unlimited numbers
of users using a diverse range of sensors.
• “Collective evaluation” captures the decision-support functions provided by
crowd reviews, ratings and recommendations (for example, Tripadvisor and
Amazon). CI in these examples supports the generation of a collective decision-
making process (Maher et al. 2011), while collective problem-solving requires
suitable platforms to coordinate the actions of collective users (Heylighen 1999).
The characteristics of CI-supported platforms have been researched in the past
using the “5W1H” framework, which asks “who, what, where, when, why and how”
(Dey 2000; Ha et al. 2006; Lee et al. 2014b; Schilit et al. 1994). Malone et al.
(2010), for example, develop a gene-table of CI systems categorised using the over-
arching design questions: what, who, why and how. The who question traces two
genes, crowd and hierarchy (distributed and structured organisations), to identify the
nature of the collective undertaking. The why relates to motivations or incentives
that drive the activity. These could include financial gain, social capital or emotional
satisfaction. The what interrogates two basic genes that determine actors conduct:
create and decide. Finally, how has two genes arising from the answers to what.
How–Create identifies processes of collection, contestation and collaboration. How–
Decide deals with group decision processes (voting and consensus) and individual
decisions (social network). Huang et al. (2017, p. 2) extend the genome framework
to include two additional categories—when and where—for analysing “mobile and
situated crowdsourcing systems”.
The combined properties of interactive and collective digital platforms for design
thinking can be explored using three types of functionalities, each of which reflect
172 7 Design Thinking and the Digital Ecosystem

aspects of the “5W1H” genome framework. The first is user-based functionality,

associated with the who and why categories. The next is content-based functionality,
which is developed from the what and how questions. Finally, location and time-based
functionality responds to where and when themes. These CI platform attributes can
be mapped into a comprehensive set of DDT functionalities (Table 7.2). Each of the
three categories and their sub attributes are described in more detail hereafter.
The first category, user-based functionality, has four attributes: extrinsic hier-
archy, extrinsic crowd, intrinsic hierarchy and intrinsic crowd. These attributes are
concerned with the people who contribute to content creation in an interactive and
collective platform and why they do it. Other than normal users, the people behind
user-generated content are commonly classified into four types: power users, content
providers, developers and service providers (Belimpasakis et al. 2010). The relations
between the different users can be conceptualised as interactions between Csikszent-
mihalyi’s three systems: person, field and domain (Csikszentmihalyi 1988, 1994,

Table 7.2 Key functionalities of interactive and collective platforms developed from CI attributes
Malone et al. (2010) Huang et al. (2017) DDT functionality
and its attributes
Who • Crowd Who • Crowd User-based
• Hierarchy (or • Requester/Contributor • Extrinsic
management) • Practitioner/Crowd hierarchy
Why • Money Why • Extrinsic • Extrinsic crowd
• Love • Intrinsic • Intrinsic
• Glory hierarchy
• Intrinsic crowd
What • Create What • Environment-centric Content-based
• Decide • People-centric • Collaborative
• Service-centric creation
How-create • Collection How • Participatory • Collaborative
How-decide • Collaboration • Opportunistic decision
• Group decision • Collective
• Individual creation
• Collective
decision
decision
• Opportunistic
creation
• Opportunistic
decision
* Malone et al. identify only four Where • Local Location and
gene categories above • Anywhere time-based
When • Time-bound • Synchronous
• Anytime local
• Synchronous
global
• Asynchronous
local
• Asynchronous
global
7.2 A Digital Design Thinking (DDT) Framework 173

1997). Setting aside these types of users and their dynamic operations, interac-
tive and collective platforms deal with two user structures: crowd and hierarchy.
A crowd is a collection of people who are widely distributed, with roles that are not
known in advance, but have an implicit purpose or function (Malone et al. 2010).
In contrast, a hierarchy is a collection of users who are organised by explicit roles
or functions. In addition to these user structures, user-based functionality deals with
the motivations of participations (why). Huang et al. (2017) suggest differentiating
extrinsic and intrinsic motivations although combinations of the two are potentially
more common. Extrinsic motivations, for example, relate to external rewards such
as wealth, power or fame. Intrinsic ones are independent of other consequences and
arguably, self-motivated or voluntary participants may be the most important for a
platform. Furthermore, growth in interactive and collective activities occurs in the
shift from extrinsic hierarchy to intrinsic crowd, leading to potentially unlimited
relationships between thousands of users.
The second category of DDT functionality, content-based functionality has six
attributes: collaborative creation, collaborative decision, collective creation, collec-
tive decision, opportunistic creation and opportunistic decision. These attributes are
used to understand what is occurring and how. Interactive and collective platforms
allow and respond to each user’s personal interactivity, sharing, collaboration, user-
generated content and artistic content. For example, users participating in an SNS
blogging platform produce a variety of content that is shared with others. Further-
more, a Mixed Reality (MR) platform is usually developed as an open-source Appli-
cation Programming Interface (API) for power users to easily generate and share
content. These activities are associated with creation, and the response from other
users is the need for evaluation and decision. When the processes and products
of creation are expressed or externalised, social evaluation and social appreciation
(rewards and acknowledgements) may result, being the catalyst for further social
creativity (Fischer et al. 2005). Examples of the creation and decision attributes
are “crowd-creating” and “crowd-rating”, respectively (Morschheuser et al. 2017).
Both creation and decision attributes can be employed as part of three different
approaches: collaborative, collective and opportunistic. Collaborative approaches
are used in hierarchies and collective in crowds. These first two attributes are related
to “participatory” contributions by users, whereas non-participatory or opportunistic
contributions are typically generated by sensors or settings in mobile devices (Huang
et al. 2017). For example, LBSs or health-monitoring applications of mobile devices
generate unforeseen, pervasive content, using location tracking and wearable sensors.
The last category of DDT functionality is location and time-based. A capacity
to work anywhere (local verses global) and anytime (asynchronous or synchronous)
is a defining feature of interactive and collective platforms for design. Thus, this
category has four attributes: synchronous local, synchronous global, asynchronous
local and asynchronous global. Collectively they explain where and when a function
or activity is occurring. For example, considering location-based functionality, the
primary division is between local and global, although Huang et al. (2017) suggest
an alternative, “local” and “anywhere”. The former is defined as “place-specific”,
whereas the latter is not. Location-aware technologies—Global Positioning Systems
174 7 Design Thinking and the Digital Ecosystem

(GPS) or Wi-fi Positioning Systems (WPS)—allow for the creation, management

and sharing of data from the local to the global. Interactive and collective plat-
forms usually support this type of mobility-enabling functionality. The second part
of this category differentiates asynchronous from synchronous activities. Although an
alternative framing of these attributes could consider “time-limited” and “non-time-
limited” functionalities, from a design perspective the primary issues are commu-
nication and co-working which is why the former pairing is more appropriate. In a
DDT platform “co-creation” may require synchronisation (Nonaka and Konno 1998)
or it could occur in serial or parallel (Fischer et al. 2005). For this reason, most digital
design platforms accommodate both asynchronous and synchronous interaction. In
summary, location and time-based functionality increase along a continuum from
synchronous local to asynchronous global.
Figure 7.2 illustrates the three categories of DDT functionalities of interactive and
collective platforms. It also identifies their attribute values and maps them into a 3D
(x, y, z dimensions) model for quantitative and qualitative analysis and assessment.
The DDT functionalities framework (Fig. 7.2) also identifies three optimal or
increasingly functional directions for interactive and collective experience: the z-
axis increases to intrinsic crowd (user); x-axis to opportunistic decision (content);
and y-axis to asynchronisation (location and time). The functionality (F) of a digital
platform can, therefore, be quantified as follows:

F = aU + bC + cL (7.1)

In this formula, U is the user-based functionality, C is the content-based func-

tionality and T is the location and time-based functionality. The second functionality
(content) ranges from 1 to 6, with the other two, from 1 to 4. In order to normalise
or weight each functionality, the formula has three constants (a, b, c) as correction
factors, which can be measured using case studies and evaluations. Nonetheless, for a
simple assessment of a balanced-use platform, “1” can be used as a default constant.
As an example of how this would work (Fig. 7.2), consider a hypothetical “bal-
anced” interactive and collective platform that can accommodate collective decision-
making (C = 4), asynchronous local interaction (L = 3) and an extrinsic crowd (U =
2). The combined DDT functionality (F = 9) reflects the way these attributes shape
the platform’s cognitive impacts. Using this method, the functionality of digital plat-
forms can be quantitatively compared. The following section presents a series of
examples of mobile and situated platforms, which are examined using the DDT
process and functionality frameworks. The primary examples are of interactive art
and media platforms, although some of these also support design processes.
7.3 Detailed Examples and Analysis 175

7.3 Detailed Examples and Analysis

Interactive, collective platforms, from BIM to social-media blogging sites, provide

the foundations or the architecture for groups to produce innovative or creative
outcomes. The two DDT frameworks presented in this chapter propose ways of under-
standing platform processes and comparing functionalities. While these frameworks
might appear to be optimised for specific design platforms, the rise of pervasive and
ubiquitous computing means they are also useful for investigating many other types
of collective, interactive systems. Today, microprocessors and sensors are increas-
ingly embedded in everyday objects, buildings and environments, engaging users
in a collective, interactive process. This capacity for interaction, conscious or not,
is central to many new creative platforms developed for the arts. Before consid-
ering specific media art platforms, it is important to differentiate between several
aspects of Tangible User Interfaces (TUIs) and draw some distinctions between CI
in experiential art and in the design process.
Using Hornecker and Buur’s (2006) classification of social and spatial interac-
tion, it is possible to differentiate between four types of platform interfaces: tangible
manipulation, spatial interaction, embodied facilitation and expressive representa-
tion. The first of these refers to physical or tactile interfaces, the second to those
that are site-specific and the third to collective interfaces in a physical space. The
last encompasses interfaces that provide a clear and responsive connection between
an action and an outcome. These categories are not conventional ones in design
research. The closest example of a collective tangible interaction in design might
be the creation of a physical model by a team, something that is becoming rare in
design practice today. In the arts, however, collective interaction between the general
public, artists and a platform can constitute the entirety of a creative output. Or to
put this another way, in interactive media arts, the process is the product (Ascott
1969). Being engaged in this collective, dynamic and creative process can influence
the cognitive behaviours of individuals, groups and even whole communities. Thus,
interactive artworks provide some informative examples of interactive, collective
platforms that engage with CI.
The following sections describe and analyse two types of interactive platforms—
mobile and situated—and their processes and functionalities. The mobile platforms
engage participants in an on-going creative process that exists in the continuum
between real and digital worlds. Mobile platforms are often used by artists to
engage audiences with their work using a Mobile Augment Reality (MAR) plat-
form, typically supported by smartphones, Personal Digital Assistants (PDAs) or
tablet PCs. The participants are not a passive audience in these examples, they are
actors, composers and performers, self-organising their behaviours in response to the
environment. The work Differential Life Integral City by Tesoc Hah is an example
of interactive art using a mobile platform for its collective interface (Fig. 7.3). In
contrast, situated platforms rely on two-way simultaneous communication supported
by ubiquitous sensor networks. They extract real-time data from the “thick air” of
information in the world (Velikov et al. 2012) and provide new, collective opportu-
nities to interact with, and create in public space (Yoon 2008). Open Columns by
176 7 Design Thinking and the Digital Ecosystem

a. DiīerenƟal life Integral city b. Open Columns

Fig. 7.3 Differential Life Integral City (left) by Tesoc Hah and Open Columns (right) by Omar
Khan. Two examples of interactive and collective platforms. Incheon International Digital Art
Festival, South Korea, 2010

Omar Khan is an interactive artwork that uses a situated platform to gather data from
participants and respond to their presence (Fig. 7.3). These two examples, along with
four more, are investigated and assessed in the following sections.
Throughout the following discussion of mobile and situated platforms, reference
is made to the DDT framework’s four types of processes (I-C-L, digital collabora-
tion, problem-solving and collective design) and three categories of functionalities
(user-based, content-based and location and time-based). Using these frameworks,
the chapter characterises and compares the emergent creative and collaborative
characteristics of six interactive digital platforms.

7.3.1 Mobile Platforms

The smartphone is arguably the most powerful platform for ubiquitous computing
available today (Henrysson and Ollila 2004; Abowd et al. 2005). Smartphones
support a myriad of interactive and collective behaviours, creating a pervasive Mixed
Reality (MR) (Stapleton and Rolland 2010). In combination, the mobility and ubiq-
uity of the smartphone’s MR environment support basic DDT functionality, providing
a platform for creative interactive processes. The smartphone is also a TUI, which
gives “physical form to digital information” and to social interaction (Ullmer and
Ishii 2000). For all of these reasons, it is ideal for MAR (Lee and Kim 2011; Lee
et al. 2013, 2019).
In media arts, the smartphone is regularly used to blur the boundaries between art
and computer science, artist and viewer. As an example, Tesoc Hah’s 2010 Differen-
tial Life Integral City presents a collective virtual environment that supports a creative
7.3 Detailed Examples and Analysis 177

on-going process (Fig. 7.3a). Visitors to the work input their lifestyle information
into the platform using a smartphone, and the platform simultaneously generates and
continuously transforms virtual residences in response to each participant’s needs.
This process not only shapes individual housing units, but collectively develops a
dynamic virtual city in response to the crowd’s requirements. While the audience can
participate in the platform anytime and from anywhere, the exhibition itself remains
spatially situated. Furthermore, while people can actively shape its content, they
cannot completely determine its outcome.
Differential Life Integral City can be analysed using the frameworks presented
previously in this chapter. In terms of its functionalities, its user category is intrinsic
crowd, its content is collective creation and its location and time are asynchronous
global. This leads to an unweighed functionality value of 11 (4 + 3 + 4). The
DDT process supported by this platform is a simple three step I-C-L (Interacting,
Collecting and Leveraging). The audience interacts with the digital platform through
input of their information and the platform collects and simulates their data in virtual
space. Platforms of this type make the link between tangible and intangible culture
more interactive and playful (Marques and Borba 2017). Simulated visualisation
based on participation can also be understood as “embodied design thinking” which
is situated and generated through socio-cultural experience (Diethelm 2019; Jerald
2015). Digital platforms using MAR technologies like this, support direct access to
digital media, projecting a collective experience.

7.3.1.1 Mobile Augmented Reality (MAR) Platforms

Mixed Reality (MR) refers to real-time interaction that spans the boundaries between
the real and the virtual. Rather than having fixed properties, MR can be conceptualised
as a continuum ranging from Augmented Reality (AR) to Augmented Virtuality
(Milgram and Kishino 1994). MR platforms use mobile ubiquitous computing to
create interactive relations between people, objects and locations, and then overlay
computer-generated and real visualisations on the physical world. Such platforms
visually connect the physical environment with digital information, as well as with
people, through the social network. An example of an MR project is SLARiPS
(Second Life Augmented Reality in Physical Space), which set out to extract Second
Life avatars from the virtual world and make them visible and responsive in the real
world (Stadon 2009).
MAR, which combines ubiquitous computing and AR (Liberati 2016), is often
supported by cloud-based servers (Chatzopoulos et al. 2017). An example of a
smartphone-based MAR is EYEPLY (Hurwitz and Jeffs 2009), which provides
personalised marketing and promotion of products and services to individual users.
EYEPLY operates in real time at stadiums, parks, conventions or shopping malls.
Because it shares controllable contents already developed from existing third-party
data, its content-based functionality is limited. EYEPLY ’s functionalities can be cate-
gorised as follows. For the user category, extrinsic crowd; for the content category,
collaborative creation; and for location and time, synchronous local. Therefore,
178 7 Design Thinking and the Digital Ecosystem

its functionality value is 4 (2 + 1 + 1), which is low for an interactive, collabo-

rative platform. In addition to its functionality, its DDT process consists of three
steps, interacting, collecting and leveraging, meaning it uses a basic I-C-L process
(Fig. 7.1).
Santana et al. (2017) developed a MAR platform using an LBS enabling visual
analysis of temporal data series and 3D building models. This type of MAR plat-
form uses AR to visualise and simulate virtual models at a particular location in
the real world. Like most location-based MAR platforms, it has limited functionali-
ties: extrinsic crowd, collaborative creation and synchronous local (F = 4). Despite
this, AR combining physical and virtual interaction can be a powerful mechanism
for engagement and decision-making (Allen et al. 2011; Gordon and Manosevitch
2010).
An example of a more recent MAR platform is Urban CoBuilder, which uses
a multiplayer, outdoor urban simulation to support rule-based emergent planning
(Imottesjo and Kain 2018). Using the collective results of individual design and
planning decisions, Urban CoBuilder enables process simulation through multi-
stakeholder inclusion as well as rule-based simulation using a gaming engine. Its
capacity for interactive and collective visualisation seeks to support increased social
participation and communication between experts and the community in the DDT
process. Thus, the Urban CoBuilder platform serves collaborative design and collec-
tive decision-making. The functional characteristics of Urban CoBuilder are as
follows: intrinsic crowd, collective decision and asynchronous local. Its function-
ality value is 11 (4 + 4 + 3). Although this location-based MAR platform involves
both hierarchy and crowd users, the creation of digital content is only undertaken
by the extrinsic hierarchy with the collaborative work occurring separately. Further-
more, collective decision by the crowd does not directly support “co-creation” but
is limited to suggesting ideas for designers (hierarchy). Thus, the DDT process is
closer to digital collaboration—focusing on deciding, inspiring, networking, recom-
mending and sharing—and only partly supports problem-solving limited to solution
prototyping and testing.

7.3.2 Situated Platforms

As the previous section notes, while computers in the contemporary world have
almost “disappeared”, computation is everywhere (Weiser 1991). The concept of a
“tangible bit” encapsulates integration of different types of interactive surfaces and
ambient media into objects, rendering it both invisible and ever-present (Ishii and
Ullmer 1997). Situated platforms use ubiquitous computing (Weiser 1991) to support
spatial interaction and embodied facilitation (Hornecker and Buur 2006). These
platforms empower collective interactions and context-awareness in physical space,
sustaining opportunistic creation and decision functionalities. They recognise users’
behaviours and needs and then respond with suitable, personalised services (Hara
et al. 2002; Lee et al. 2014b; MacDorman et al. 2004; Nakauchi et al. 2003). In this
7.3 Detailed Examples and Analysis 179

way, “embodied interaction” in a situated platform leads to “embodied cognition”,

arising through the crowd’s unconscious or intuitive interactions in space (Dourish
2004). This phenomenon, the leveraging of cognition through interaction, is central
to technologically mediated social participation (Fortin et al. 2014b).
As a simple example of a situated platform, the interactive artwork Open Columns
by Omar Khan uses carbon dioxide (CO2 ) sensors in an exhibition space to record
the physical outputs of participants and reflect this in the installation (Fig. 7.3b). The
installation, made up of a mesh of columns, changes in response to the number of
people in the space and their interactions. The functionalities of this platform can
be categorised as intrinsic crowd, opportunistic creation and synchronous local. The
combined functionality value is 10 (4 + 5 + 1). In comparison with the mobile
platform supporting Differential Life Integral City, this situated platform provides
opportunistic creation, pervasively generated by sensors and a self-organising system
(columns) within the space of the exhibition. Because Open Columns lacks a decision
point, its DDT process conforms to the most common model for interactive artworks,
I-C-L.

7.3.2.1 Interactive Media (IM) Platforms

The location of an IM platform in a public space is an example of a situated platform

that can blur and merge physical and digital spaces. This type of platform has, in
the recent past, largely relied on the use of ubiquitous computing, sensor networks,
robotics, middleware and agent-based software (Cook and Das 2007). In an IM
installation, spatial interactions are embedded and situated in a physical space, and
consequently users need to physically interact with the platform. Moreover, inter-
actions are not restricted to touching or moving objects, but can require physically
moving around an area (Hornecker and Buur 2006; O’Neill 2008). As such, IM
displays in urban space often reflect collective experience or movement.
To sustain social interactions in an IM platform, Hespanhol and Tomitsch (2012)
suggest that three factors are important. First, active participation, second, well-
located sensor locations and finally, close alignment of medium and message. These
points are especially relevant when considering the performative aspects of IM plat-
forms. As participants become aware of their collective presence they may begin
to perform to an audience of their peers. Hespanhol and Tomitsch (2012) warn that
performative settings “may pose cognitive challenges for the audience but also allow
for more open expressive co-experiences to emerge” (p. 40). The most common
approach to IM platforms is to employ large “situated displays” in public spaces
to facilitate in-situ and real-time responses. Some examples of these include Opin-
ionizer (Brignull and Rogers 2003), CityWall (Morrison et al. 2008), Discussions in
Space (Schroeter and Foth 2009), Sapporo World Window (Choi and Seeburger 2011),
SMSlingshot (Fischer and Hornecker 2012), Smart Citizen Sentiment Dashboard
(Behrens et al. 2014) and City-Share (Ludwig et al. 2017).
Ludvigsen’s (2005) interactive media installation iFloor is an IM platform that
explores social interaction in a public building. Ludvigsen argues that an effective IM
platform must have a shared representation, support collective responses and possess
180 7 Design Thinking and the Digital Ecosystem

social interfaces. Ludvigsen’s iFloor is an interactive floor surface that provides

library users with a new way of finding materials. This situated platform allows users
to post questions to the digital floor using text messages or email, and then review
questions using a shared cursor. The cursor is operated pervasively by the physical
position of users who also receive visual feedback of their movements. The combined
physical activity and shared interface facilitate situated engagement. This game-like
platform provides previously unforeseen opportunities for collaborative exploration.
Collectively, the functionalities of iFloor are intrinsic crowd, opportunistic creation
and synchronous local, leading to a functionality value of 10 (4 + 5 + 1). Its DDT
process is digital collaboration including creating, deciding, inspiring, networking,
recommending and sharing. It is, however, limited to face-to-face communication in
a fixed location.
Fortin et al.’s (2014a, b) paired media installation, Mégaphone and Speakers’
Corner, is focussed on the act of public speaking and its reception. In Speakers’
Corner a person’s voice is captured by a microphone and sent to a voice recogni-
tion system. The data collected and processed in this way is then projected on the
Mégaphone platform’s interfaces: loudspeakers, a set of responsive stage lights and
media façades. In this work, the public not only contribute to creating the content
(through “public” speaking) but also receive other’s content, visually and aurally,
stimulating social interaction. The functionalities of this platform are intrinsic crowd,
collective creation and synchronous local, and its functionality value is 8 (4 + 3 +
1). Compared to iFloor’s embodied interaction, the Speakers’ Corner and Méga-
phone situated platform involves direct creation (the initial voice input), although
the platform automatically transfers this into various formats. Collectively, its DDT
process is not a simple I-C-L, but rather digital collaboration addressing creation,
inspiration, networking, sharing and transferring. However, without the deciding and
recommending steps, its collaborative potential is relatively limited.

7.4 Discussion

The relatively recent convergence of wireless technologies and sensor networks has
enabled new types of everyday experience in the digital ecosystem. In this ecosystem
local interactions and self-organising behaviours enable individuals to seamlessly
interact with the environment and each other. Despite this, our understanding of
the design thinking process and its products in this ecosystem is still in its infancy.
Thus, this chapter adapts and develops two frameworks to systematically investigate
the properties and functionalities of digital platforms. Using these frameworks, the
previous sections in this chapter analysed two types of interactive and collective
platforms, mobile and situated. The results of this analysis are summarised in Table
7.3.
The DDT processes in the six cases are limited to just two types: I-C-L and digital
collaboration. The mobile and situated platforms examined in this chapter do not
use “co-decision”, which is one of the most important CI processes. In most of the
7.4 Discussion 181

Table 7.3 DDT process and functionality of interactive and collective platforms analysed in the
case study (F: functionality value)
Platform Case Process Functionality F
Mobile Differential life Integral I-C-L Intrinsic crowd, 11
city Collective creation,
Asynchronous global
EYEPLY I-C-L Extrinsic crowd, 4
Collaborative creation,
Synchronous local
Urban CoBuilder Digital collaboration Intrinsic crowd, 11
Collective decision,
Asynchronous local
Situated Open Columns I-C-L Intrinsic crowd, 10
Opportunistic creation,
Synchronous local
IFloor Digital collaboration Intrinsic crowd, 10
Opportunistic creation,
Synchronous local
Mégaphone and Speakers’ Digital collaboration Intrinsic crowd, 8
Corner Collective creation,
Synchronous local

platforms, a hierarchy is the foundation for the operations of the crowd or collective,
limiting the capacity for true creative and collective emergence. Without capacity
for collective, convergent thought processes, users’ engagement and immersion are
constrained. DDT is an interactive and collective process like I-C-L, but it evolves
through collaborative and collective decision-making. By incorporating the third
or fourth types of DDT processes (Table 7.1) in a platform, more interactive and
collective activities will arise in the ecosystem.
The Collective Design (CD) process, in particular, could offer an optimal model
of DDT in the future. CD, by way of the cloud or online sharing platforms, chal-
lenges traditional design thinking (Özkil 2017), accommodating participatory design
(Barcellini et al. 2015; Smith and Iversen 2018), collaborative product development
(Lin et al. 2010; Wang and Zhang 2010) and computer-supported cooperation (Papan-
gelis et al. 2019; Schrott and Glückler 2004). CD, however, has different knowl-
edge production and communication processes, through loose or informal design
networks (Özkil 2017). Whereas collaborative design is a controlled formal process,
CD proposes an informal, flexible collaborative process (Huang et al. 2010). CD’s
benefits are that it is intrinsically collaborative, part of on-going process of devel-
oping and sharing design knowledge across a body of physically distributed, but
similarly motivated people. This feature of CD processes has parallels with the prop-
erties of SMMs and DMMs (Chap. 5). This also signals that the DDT framework
can accommodate the DTC model in Chap. 5, highlighting the communication and
information processing of individual, collective and distributed MMs.
182 7 Design Thinking and the Digital Ecosystem

In terms of functionality, the mobile platforms examined in this chapter tend

to address asynchronous interaction (Table 7.3). In contrast, the situated platforms
naturally support synchronous local functionality. Acknowledging the spatial and
temporal limitations of IM installations, future media platforms should consider three
additional functionalities: synchronous global, asynchronous local and asynchronous
global. With the expansion of social networking, future platforms should have access
to these functions. In addition, socially extended cognition is mediated by interaction
with the digital platforms examined in this chapter, which are in turn updated to reflect
users’ collective authorising actions (Bruns 2008). Consequently, this heterogeneous,
collective interaction often emerges through the use of hand-held devices and tangible
interaction. In this way, embodied interaction in situated platforms can be extended to
supporting asynchronous global activities. Indeed, as larger numbers of users engage
in CI, user experience may be further amplified through inclusion of three additional
functionalities: (i) intrinsic crowd, (ii) opportunistic decision and (iii) asynchronous
global.
The functionality (F) values in Table 7.3 enable the quantitative comparison of
DDT platforms, although the measures adopt a simple default position, using “1” for
all constants. Furthermore, the results indicate that three functionalities are common
in interactive and collective platforms. For example, the functionality values of Differ-
ential Life Integral City and Urban CoBuilder are identical, but their content and
location and time-based functionalities are different. In addition, the use of oppor-
tunistic creation in the two situated platforms results in a similar functionality value
(F = 10), which is the highest of the two mobile platforms. Therefore, this measure
offers a promising quantitative mechanism for reviewing platform DDT functionality.
A further factor to consider when assessing mobile or situated platforms is there
are now a growing number of examples of mobile-situated platforms. For example,
SMSlingshot (Fischer and Hornecker 2012), Smart Citizen Sentiment Dashboard
(Behrens et al. 2014), mood.cloud (Scolere et al. 2016) and City-Share (Ludwig
et al. 2017) are all integrated mobile-situated platforms. These combined platforms
may be catalysts for further opportunities in the digital ecosystem.

7.5 Conclusion

With the advent of mobile and ubiquitous computing, our everyday experience is
getting closer to that illustrated in Langton’s (1995) Artificial Life. For example, using
embodied interaction, seamless responsive behaviours are exhibited in interactive art
and media platforms. In most cases, the goal of the art is to develop performative and
pervasive interactions, based on direct or indirect communication between people
and the environment. These participatory behaviours can be enhanced, and their
creative outcomes improved, through incorporation of advances in three types of
functionality: (i) user, (ii) content and (iii) location and time. A heightened under-
standing of DDT process types is also a precursor to improved outcomes. The four
DDT processes identified in this chapter can be regarded as stages in the evolution of
7.5 Conclusion 183

DDT and CI, ranging from the more primitive I-C-L to the more advanced collective
design process.
The analytic approach proposed in this chapter facilitates the systematic explo-
ration of both DDT processes and functionalities, as can be seen in the six example
platforms. The functionality measure (F) provides a way of comparing digital plat-
forms and evaluating their potential for supporting CI, DMM and TMM. The two
DDT frameworks in the chapter do, however, require further development and testing
in different contexts to understand their limits and opportunities, for example, to be
adopted in and advance professional or commercial applications.
This chapter, the last of Part II Collaboration, addresses interaction collabora-
tion and collectivity in design thinking. It also identifies the way social interac-
tion can support social creativity and CI. In Part III, these themes are expanded to
consider cultural aspects of design thinking, which necessarily include those relating
to communication and language in design.

References

Abowd, G.D., L. Iftode, and H. Mitchell. 2005. Guest editors’ introduction: The smart phone—A
first platform for pervasive computing. Pervasive Computing, IEEE 4 (2): 18–19.
Alag, S. 2008. Collective Intelligence in Action. Shelter Island, NY: Manning Publications.
Allen, M., H. Regenbrecht, and M. Abbott. 2011. Smart-phone augmented reality for public
participation in urban planning. Paper presented at the Proceedings of the 23rd Australian
Computer-Human Interaction Conference, Canberra, Australia,
Ascott, Roy. 1969. Behaviourables and futuribles. Control 5: 3.
Balestrini, Mara, Yvonne Rogers, Carolyn Hassan, Javi Creus, Martha King, and Paul Marshall.
2017. A city in common: A framework to orchestrate large-scale citizen engagement around urban
issues. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems.
Denver, Colorado, USA: Association for Computing Machinery.
Barcellini, Flore, Lorène Prost, and Marianne Cerf. 2015. Designers’ and users’ roles in participatory
design: What is actually co-designed by participants? Applied Ergonomics 50: 31–40. https://doi.
org/10.1016/j.apergo.2015.02.005.
Behrens, Moritz, Nina Valkanova, Ava Fatah gen. Schieck, and Duncan P. Brumby. 2014. Smart
citizen sentiment dashboard: A case study into media architectural interfaces. In Proceedings
of The International Symposium on Pervasive Displays. Copenhagen, Denmark: Association for
Computing Machinery.
Belimpasakis, P., P. Selonen, and Y. You. 2010. Bringing user-generated content from Internet
services to mobile augmented reality clients. In 2010 Cloud-Mobile Convergence for Virtual
Reality Workshop (CMCVR 2010) Proceedings.
Brignull, H., and Y. Rogers. 2003. Enticing people to interact with large public displays in public
spaces. In INTERACT. Amsterdam: IOS Press.
Bruns, Axel. 2008. Blogs, Wikipedia, Second Life, and Beyond: From Production to Produsage.
Peter Lang.
Burdick, Anne, and Holly Willis. 2011. Digital learning, digital scholarship and design thinking.
Design Studies 32 (6): 546–556. https://doi.org/10.1016/j.destud.2011.07.005.
Chatzopoulos, D., C. Bermejo, Z. Huang, and P. Hui. 2017. Mobile augmented reality survey: From
where we are to where we go. IEEE Access 5: 6917–6950. https://doi.org/10.1109/ACCESS.
2017.2698164.
184 7 Design Thinking and the Digital Ecosystem

Choi, J.H., and J. Seeburger. 2011. Sapporo world window Urban interaction through public
and private screens. In 2011 IEEE International Conference on Pervasive Computing and
Communications Workshops (PERCOM Workshops).
Cook, Diane J., and Sajal K. Das. 2007. How smart are our environments? An updated look at the
state of the art. Pervasive and Mobile Computing 3 (2): 53–73. https://doi.org/10.1016/j.pmcj.
2006.12.001.
Csikszentmihalyi, M. 1988. Society, culture, and person: A systems view of creativity. In The Nature
of Creativity, ed. R.J. Sternberg, 325–339. Cambridge, UK: Cambridge University Press.
Csikszentmihalyi, M. 1994. The domain of creativity. In Changing the World: A Framework for the
Study of Creativity, ed. D.H. Feldman, M. Csikszentmihalyi, and H. Gardner, 135–158. Westport,
CT: Praeger.
Csikszentmihalyi, M. 1997. Creativity: Flow and the Psychology of Discovery and Invention. New
York: HarperCollins.
Dey, Anind Kumar. 2000. Providing architectural support for building context-aware applications.
GA, USA: Georgia Institute of Technology Atlanta.
Diethelm, Jerry. 2019. Embodied Design Thinking. She Ji: The Journal of Design, Economics, and
Innovation 5 (1): 44–54. https://doi.org/10.1016/j.sheji.2019.02.001.
Dourish, Paul. 2004. Where The Action Is: The Foundations of Embodied Interaction. Cambridge:
MIT Press.
Elia, Gianluca, and Alessandro Margherita. 2018. Can we solve wicked problems? A conceptual
framework and a collective intelligence system to support problem analysis and solution design
for complex social issues. Technological Forecasting and Social Change 133: 279–286. https://
doi.org/10.1016/j.techfore.2018.03.010.
Elia, Gianluca, Alessandro Margherita, and Giuseppina Passiante. 2020. Digital entrepreneurship
ecosystem: How digital technologies and collective intelligence are reshaping the entrepreneurial
process. Technological Forecasting and Social Change 150: 119791. https://doi.org/10.1016/j.
techfore.2019.119791.
Farshchian, Babak A., and Monica Divitini. 1999. Using email and WWW in a distributed partici-
patory design project. ACM SIGGROUP Bulletin 20 (1): 10–15. https://doi.org/10.1145/327556.
327602.
Fischer, Patrick Tobias, and Eva Hornecker. 2012. Urban HCI: Spatial aspects in the design of shared
encounters for media facades. In Proceedings of the SIGCHI Conference on Human Factors in
Computing Systems. Austin, Texas, USA: Association for Computing Machinery.
Fischer, Gerhard, Elisa Giaccardi, Hal Eden, Masanori Sugimoto, and Yunwen Ye. 2005. Beyond
binary choices: Integrating individual and social creativity. International Journal of Human-
Computer Studies 63 (4–5): 482–512.
Lopez Flores, René, Jean Pierre Belaud, Stéphane Negny, and Jean Marc Le Lann. 2015a. Open
computer aided innovation to promote innovation in process engineering. Chemical Engineering
Research and Design 103: 90–107. https://doi.org/10.1016/j.cherd.2015.08.015.
Flores, Lopez, Stéphane Negny. René, Jean Pierre Belaud, and Jean-Marc. Le Lann. 2015b. Collec-
tive intelligence to solve creative problems in conceptual design phase. Procedia Engineering
131: 850–860. https://doi.org/10.1016/j.proeng.2015.12.394.
Fortin, Claude, Kate Hennessy, and Hughes Sweeney. 2014a. The “Making of” mégaphone, an
interactive “Speakers” Corner” and digitally-augmented agora in public space’. In Proceedings
of The International Symposium on Pervasive Displays. Copenhagen, Denmark: Association for
Computing Machinery.
Fortin, Claude, Carman Neustaedter, and Kate Hennessy. 2014b. The appropriation of a digital
“speakers” corner: lessons learned from the deployment of mégaphone. In Proceedings of the
2014 conference on Designing interactive systems. Vancouver, BC, Canada: Association for
Computing Machinery.
Gloor, P.A. 2006. Swarm Creativity: Competitive Advantage through Collaborative Innovation
Networks. Oxford, MA: Oxford University Press.
References 185

Gordon, Eric, and Edith Manosevitch. 2010. Augmented deliberation: Merging physical and virtual
interaction to engage communities in urban planning. New Media & Society 13 (1): 75–95. https://
doi.org/10.1177/1461444810365315.
Ha, Tae, Ji. Jung, and Oh. Sung. 2006. Method to analyze user behavior in home environ-
ment. Personal and Ubiquitous Computing 10 (2): 110–121. https://doi.org/10.1007/s00779-
005-0016-9.
Halpin, H. 2008. Foundations of a philosophy of collective intelligence. In AISB 2008 Convention:
Communication, Interaction and Social Intelligence, 1st-4th April 2008, University of Aberdeen,
eds. Frank Guerin, and Wamberto Weber Vasconcelos, 12–19. Computing & Philosophy.
Hara, K., T. Omori, and R. Ueno. 2002. Detection of unusual human behavior in intelligent house.
Paper presented at the Proceedings of the 2002 12th IEEE Workshop on Neural Networks for
Signal Processing, 2002
Henrysson, Anders, and Mark Ollila. 2004. UMAR: Ubiquitous Mobile Augmented Reality. Paper
presented at the Proceedings of the 3rd International Conference on Mobile and ubiquitous
multimedia, College Park, Maryland, USA, October 27–29, 2004
Hespanhol, Luke, and Martin Tomitsch. 2012. Designing for collective participation with media
installations in public spaces. In Proceedings of the 4th Media Architecture Biennale Conference:
Participation. Aarhus, Denmark: Association for Computing Machinery.
Heylighen, Francis. 1999. Collective intelligence and its implementation on the web: Algorithms to
develop a collective mental map. Computational and Mathematical Organization Theory 5 (3):
253–280. https://doi.org/10.1023/a:1009690407292.
Hornecker, Eva, and Jacob Buur. 2006. Getting a grip on tangible interaction: A framework on phys-
ical space and social interaction. Paper presented at the Proceedings of the SIGCHI Conference
on Human Factors in Computing Systems, Montréal, Québec, Canada,
Huang, Hong-Zhong., Xu. Huan-Wei, and Xu. Zu. 2010. Petri Net-based coordination component
for collaborative design. Concurrent Engineering 18 (3): 199–205. https://doi.org/10.1177/106
3293X10379765.
Huang, Yun, Alain Shema, and Huichuan Xia. 2017. A proposed genome of mobile and situated
crowdsourcing and its design implications for encouraging contributions. International Journal
of Human-Computer Studies 102: 69–80. https://doi.org/10.1016/j.ijhcs.2016.08.004.
Hurwitz, A., and A. Jeffs. 2009. EYEPLY: Baseball proof of concept—Mobile augmentation for
entertainment and shopping venues. Paper presented at the. IEEE International Symposium
on Mixed and Augmented Reality—Arts, Media and Humanities, 2009. ISMAR-AMH 2009,
Orlando, USA, 19–22 October 2009
Hüsig, Stefan, and Stefan Kohn. 2011. “Open CAI 2.0”—Computer aided innovation in the era of
open innovation and Web 2.0. Computers in Industry 62 (4):407–413. https://doi.org/10.1016/j.
compind.2010.12.003.
Huybrechts, Liesbeth, Henric Benesch, and Jon Geib. 2017. Institutioning: Participatory design,
co-design and the public realm. CoDesign 13 (3): 148–159. https://doi.org/10.1080/15710882.
2017.1355006.
Imottesjo, Hyekyung, and Jaan-Henrik. Kain. 2018. The Urban CoBuilder—A mobile augmented
reality tool for crowd-sourced simulation of emergent urban development patterns: Requirements,
prototyping and assessment. Computers, Environment and Urban Systems 71: 120–130. https://
doi.org/10.1016/j.compenvurbsys.2018.05.003.
Ishii, Hiroshi, and Brygg Ullmer. 1997. Tangible bits: Towards seamless interfaces between people,
bits and atoms. Paper presented at the Proceedings of the ACM SIGCHI Conference on Human
factors in computing systems, Atlanta, Georgia, USA.
Jerald, Jason. 2015. The VR Book: Human-Centered Design for Virtual Reality. San Rafael, CA:
Association for Computing Machinery and Morgan & Claypool.
Langton, Christopher G. 1995. Artificial Life: An Overview. Cambridge: MIT Press.
Lee, Ji-Hyun, and Ming-Luen Chang. 2010. Stimulating designers’ creativity based on a creative
evolutionary system and collective intelligence in product design. International Journal of
Industrial Ergonomics 40 (3): 295–305. https://doi.org/10.1016/j.ergon.2009.11.001.
186 7 Design Thinking and the Digital Ecosystem

Lee, Ju Hyun, and Mi Jeong Kim. 2011. A Context immersion of mixed reality at a new stage.
In Circuit Bending, Breaking and Mending 16th International Conference on Computer-Aided
Architectural Design Research in Asia (CAADRIA) 2011, ed. C.M. Herr, N. Gu, S. Roudavski,
and M.A. Schnabel, 199–208. Newcastle, Australia: CAADRIA.
Lee, Ju Hyun, Mi Jeong Kim, and Mary Lou Maher. 2013. Designing for interactive and collective
mobile creativity. In Proceedings of the 9th ACM Conference on Creativity and Cognition, ed.
E.Y.L. Do, S. Dow, J. Ox, S. Smith, K. Nishimoto, and C.T. Tan, 345–348. New York, NY: ACM.
Lee, Ju Hyun, Ning Gu, and Anthony Williams. 2014a. Parametric design strategies for the gener-
ation of creative designs. International Journal of Architectural Computing 12 (3): 263–282.
https://doi.org/10.1260/1478-0771.12.3.263.
Lee, Ju Hyun, Hyunsoo Lee, Mi Jeong Kim, Xiangyu Wang, and Peter E. D. Love. 2014b. Context-
aware inference in ubiquitous residential environments. Computers in Industry 65 (1): 148–157.
https://doi.org/10.1016/j.compind.2013.08.005.
Lee, Hochul, Jaehun Lee, Young Choon Lee, Hyuck Han, and Sooyong Kang. 2019. Mobile collab-
orative computing on the fly. Pervasive and Mobile Computing 58: 101027. https://doi.org/10.
1016/j.pmcj.2019.05.008.
Leon, Noel. 2009. The future of computer-aided innovation. Computers in Industry 60 (8): 539–550.
https://doi.org/10.1016/j.compind.2009.05.010.
Lévy, Pierre. 1997. Collective Intelligence: Mankind’s Emerging World in Cyberspace, trans. Robert
Bononno. Cambridge, MA: Perseus Books.
Lévy, Pierre. 2010. From social computing to reflexive collective intelligence: The IEML research
program. Information Sciences 180 (1): 71–94. https://doi.org/10.1016/j.ins.2009.08.001.
Liberati, Nicola. 2016. Augmented reality and ubiquitous computing: The hidden potentialities of
augmented reality. AI & Society 31 (1): 17–28. https://doi.org/10.1007/s00146-014-0543-x.
Lin, Shiyong, Roger J. Narayan, and Yuan-Shin. Lee. 2010. Hybrid client–server architecture and
control techniques for collaborative product development using haptic interfaces. Computers in
Industry 61 (1): 83–96. https://doi.org/10.1016/j.compind.2009.07.004.
Luck, Rachael. 2018. Participatory design in architectural practice: Changing practices in future
making in uncertain times. Design Studies 59: 139–157. https://doi.org/10.1016/j.destud.2018.
10.003.
Ludvigsen, Martin. 2005. Designing for social use in public places—A conceptual framework of
social interaction. In 2005 International Conference on Designing Pleasurable Products and
Interfaces (DPPI 2005). Eindhoven, The Netherlands.
Ludwig, Thomas, Christoph Kotthaus, Christian Reuter, S.ören van Dongen, and Volkmar Pipek.
2017. Situated crowdsourcing during disasters: Managing the tasks of spontaneous volunteers
through public displays. International Journal of Human-Computer Studies 102: 103–121. https://
doi.org/10.1016/j.ijhcs.2016.09.008.
MacDorman, K., H. Nobuta, T. Minato, and H. Ishiguro. 2004. Memory-based recognition of human
behavior based on sensory data of high dimensionality. Paper presented at the 2004 IEEE/RSJ
International Conference on Intelligent Robots and Systems, 2004. (IROS 2004), 28 September
2 October 2004
Maher, Mary Lou, Mercedes Paulini, and Paul Murty. 2011. Scaling up: From individual design to
collaborative design to collective design. In Design Computing and Cognition ’10, ed. John S.
Gero. Dordrecht: Springer Netherlands.
Malone, T.W., R.J. Laubacher, and C. Dellarocas. 2010. The collective intelligence genome. MIT
Sloan Management Review 51 (3): 21–31.
Markus, M. Lynne., and Claudia Loebbecke. 2013. Commoditized digital processes and business
community platforms: New opportunities and challenges for digital business strategies. MIS
Quarterly 37 (2): 649–653.
Marques, L.énia, and Carla Borba. 2017. Co-creating the city: Digital technology and creative
tourism. Tourism Management Perspectives 24: 86–93. https://doi.org/10.1016/j.tmp.2017.
07.007.
References 187

Milgram, Paul, and Fumio Kishino. 1994. A taxonomy of mixed reality visual displays. IEICE
Transactions on Information and Systems 77 (12): 1321–1329.
Mitchell, Val, Tracy Ross, Andrew May, Ruth Sims, and Christopher Parker. 2016. Empirical
investigation of the impact of using co-design methods when generating proposals for sustainable
travel solutions. CoDesign 12 (4): 205–220. https://doi.org/10.1080/15710882.2015.1091894.
Morrison, A., G. Jacucci, and P. Peltonen. 2008. CityWall: Limitations of a multi-touch environment.
Paper presented at the Public and Private Displays Workshop (PPD 08), Naples, Italy, 31 May
Morschheuser, Benedikt, Juho Hamari, Jonna Koivisto, and Alexander Maedche. 2017. Gamified
crowdsourcing: Conceptualization, literature review, and future agenda. International Journal of
Human-Computer Studies 106: 26–43. https://doi.org/10.1016/j.ijhcs.2017.04.005.
Nakakoji, K., Y. Yamamoto, and M. Ohira. 2000. Computational support for collective creativity.
Knowledge-Based Systems 13 (7–8): 451–458.
Nakauchi, Y., K. Noguchi, P. Somwong, T. Matsubara, and A. Namatame. 2003. Vivid room: human
intention detection and activity support environment for ubiquitous autonomy. Paper presented at
the 2003 IEEE/RSJ International Conference ons Intelligent Robots and Systems, 2003. (IROS
2003), 27–31 October 2003
Nambisan, Satish, Kalle Lyytinen, Ann Majchrzak, and Michael Song. 2017. Digital innovation
management: Reinventing innovation management research in a digital world. MIS Quarterly 41
(1): 223–238.
Nonaka, Ikujiro, and Noboru Konno. 1998. The concept of “Ba”: Building a foundation for
knowledge creation. California Management Review 40 (3): 40–54. https://doi.org/10.2307/411
65942.
O’Neill, Shaleph. 2008. Interactive Media: The Semiotics of Embodied Interaction. London:
Springer Science & Business Media.
Oxman, Rivka. 2006a. Digital design thinking: in the new design is the new pedagogy. In
Proceedings of the 11th Conference on Computer-Aided Architectural Design Research in Asia
(CAADRIA), ed. A. Kaga, and R. Naka, 37–46. Kumamoto, Japan: School of Architecture and
Civil Engineering; Kumamoto University.
Oxman, Rivka. 2006b. Theory and design in the first digital age. Design Studies 27 (3): 229–265.
https://doi.org/10.1016/j.destud.2005.11.002.
Oxman, Rivka. 2007. A Performance-based model in digital design: PER-FORMATIVE—design
beyond aesthetic. Architectural Engineering and Design Management 3 (3): 169–180. https://
doi.org/10.1080/17452007.2007.9684640.
Özkil, Ali Gürcan. 2017. Collective design in 3D printing: A large scale empirical study of designs,
designers and evolution. Design Studies 51: 66–89. https://doi.org/10.1016/j.destud.2017.04.004.
Papangelis, Konstantinos, Domenico Potena, Waleed W. Smari, Emanuele Storti, and Wu. Keqin.
2019. Advanced technologies and systems for collaboration and computer supported cooperative
work. Future Generation Computer Systems 95: 764–774. https://doi.org/10.1016/j.future.2019.
02.041.
Prahalad, C.K., and Venkat Ramaswamy. 2004. Co-creating unique value with customers. Strategy
& Leadership 32 (3): 4–9. https://doi.org/10.1108/10878570410699249.
Reffat, R.M. 2006. Computing in architectural design: Reflections and an approach to new
generations of CAAD. ITcon 11: 655–668.
Richter, Chris, Sascha Kraus, Alexander Brem, Susanne Durst, and Clemens Giselbrecht. 2017.
Digital entrepreneurship: Innovative business models for the sharing economy. Creativity and
innovation management 26 (3): 300–310. https://doi.org/10.1111/caim.12227.
Santana, José Miguel., Jochen Wendel, Agustín Trujillo, José Pablo. Suárez, Alexander Simons,
and Andreas Koch. 2017. Multimodal location based services—Semantic 3D City data as virtual
and augmented reality. In Progress in Location-Based Services 2016, ed. Georg Gartner and
Haosheng Huang, 329–353. Cham: Springer International Publishing.
Schilit, B., N. Adams, and R. Want. 1994. Context-aware computing applications. Paper presented
at the Workshop on Mobile Computing Systems and Applications, 1994, 8–9 December 1994
188 7 Design Thinking and the Digital Ecosystem

Schroeter, Ronald, and Marcus Foth. 2009. Discussions in space. In Proceedings of the 21st Annual
Conference of the Australian Computer-Human Interaction Special Interest Group: Design: Open
24/7. Melbourne, Australia: Association for Computing Machinery.
Schrott, Gregor, and Johannes Glückler. 2004. What makes mobile computer supported cooper-
ative work mobile? Towards a better understanding of cooperative mobile interactions. Inter-
national Journal of Human-Computer Studies 60 (5): 737–752. https://doi.org/10.1016/j.ijhcs.
2003.11.006.
Scolere, Leah M., Eric P.S. Baumer, Lindsay Reynolds, and Geri Gay. 2016. Building mood, building
community: Usage patterns of an interactive art installation. In Proceedings of the 19th Inter-
national Conference on Supporting Group Work. Sanibel Island, Florida, USA: Association for
Computing Machinery.
Smith, Rachel Charlotte, and Ole Sejer Iversen. 2018. Participatory design for sustainable social
change. Design Studies 59: 9–36. https://doi.org/10.1016/j.destud.2018.05.005.
Stadon, J. 2009. Project SLARiPS: An investigation of mediated mixed reality. Paper presented at
the 2009 IEEE International Symposium on Mixed and Augmented Reality—Arts, Media and
Humanities, 19–22 October.
Stapleton, Christopher, and Jannick Rolland. 2010. Mixing realities at Ismar 2009: Scary and
wondrous. Computer Graphics and Applications, IEEE 30 (3): 89–95.
Stelzle, Benjamin, Anja Jannack, and Jörg Rainer. Noennig. 2017. Co-design and co-decision:
Decision making on collaborative design platforms. Procedia Computer Science 112: 2435–2444.
https://doi.org/10.1016/j.procs.2017.08.095.
Surowiecki, J. 2005. The Wisdom of Crowds. New York: Anchor.
Tiwana, Amrit, Benn Konsynski, and Ashley A. Bush. 2010. Research commentary—Platform
evolution: Coevolution of platform architecture, governance, and environmental dynamics.
Information Systems Research 21 (4): 675–687. https://doi.org/10.1287/isre.1100.0323.
Ullmer, B., and H. Ishii. 2000. Emerging frameworks for tangible user interfaces. IBM Systems
Journal 39 (3.4): 915–931. https://doi.org/10.1147/sj.393.0915.
Velikov, Kathy, Geoffrey Thün, and Colin Ripley. 2012. Thick air. Journal of Architectural
Education 65 (2): 69–79. https://doi.org/10.1111/j.1531-314X.2011.01199.x.
Verstegen, Luuk, Wybo Houkes, and Isabelle Reymen. 2019. Configuring collective digital-
technology usage in dynamic and complex design practices. Research Policy 48 (8): 103696.
https://doi.org/10.1016/j.respol.2018.10.020.
Wang, Hongwei, and Heming Zhang. 2010. A distributed and interactive system to integrated
design and simulation for collaborative product development. Robotics and Computer-Integrated
Manufacturing 26 (6): 778–789. https://doi.org/10.1016/j.rcim.2010.05.005.
Wang, Lihui, Weiming Shen, Helen Xie, Joseph Neelamkavil, and Ajit Pardasani. 2002. Collab-
orative conceptual design—State of the art and future trends. Computer-Aided Design 34 (13):
981–996. https://doi.org/10.1016/S0010-4485(01)00157-9.
Weiser, Mark. 1991. The computer for the 21st century. Scientific American 265 (3): 94–104.
Yoon, Meejin J. 2008. Public works. Journal of Architectural Education 61 (4): 59–68. https://doi.
org/10.1111/j.1531-314X.2008.00188.x.
Part III
Culture
Chapter 8
Design Thinking Across Borders

Abstract This chapter investigates cross-national aspects of design thinking, with

a focus on three themes: design cognition, complexity and spatial language. The
chapter uses a new method, a dual-coding system (design cognition and language) for
protocol data combined with linkography. This system formally captures cognitive
and linguistic characteristics of the design process. This method is applied to analyse
parametric design processes undertaken by a set of Australian and Swedish architects.
Through this analysis, and the discussion of its results, the chapter contributes to the
in-depth exploration of multiple aspects of cross-national design and cultural design
thinking.

8.1 Introduction

Globalisation is a process where organisations and systems operate across national

boundaries to support productivity or to achieve power, success or influence. Glob-
alisation typically entails integration of resources, flows of production and trans-
portation networks (James 2009). Trade of raw materials, technology or expertise,
has historically been a driver of globalisation. The exchange or sale of design exper-
tise is today recognised as being central to the global “creative economy”. Florida
(2002) famously predicted that the creative economy, and the “creative class” behind
it, could eventually rise to challenge traditional economic and social capital models,
built on resources or manufacturing. While it is arguable whether this has occurred,
it is clear that design practice has become increasingly global in its aspirations and
operations.
Contemporary design teams typically include members from many countries, with
different linguistic and cultural backgrounds. This situation has become so preva-
lent that the challenges of culturally and geographically distributed teams are now
recognised as a core issue in design practice and thinking (Man 2014). Regardless of
whether these teams are working in the same building, or in different cities, multi-
cultural and multi-lingual teams pose both a challenge and an opportunity for the
global creative economy. These challenges and opportunities are at the heart of the
present chapter, which is about linguistic and cultural dimensions in design cognition.

© Springer Nature Switzerland AG 2020 191

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_8
192 8 Design Thinking Across Borders

Language, be it verbal, visual or gestural, is central to the development of Shared,

Team and Distributed Mental Models (SMMs, TMMs and DMMs, respectively).
Language is also culturally and socially coded, which has implications for both
knowledge transmission and reception. As such, the native language (the spoken
or written word) is just one of many factors affecting communication in teams.
Individual religious beliefs, cultural practices, social attitudes and learnt behaviours
all potentially complicate the capacity to share mental constructs. In a simple sense,
homogeneity in teams might be the best starting point if the focus is solely on
productivity and a linear process, because it limits complicating factors associated
with communication (Horwitz and Horwitz 2007). Conversely, and more importantly
for design, heterogeneity in teams is often linked to innovation (Chen 2018) and
multi-cultural diversity is a factor in team creativity (Cox and Stacy 1991; Jackson
1991; Santandreu Calonge and Safiullin 2015; Shalley et al. 2004). On balance, the
evidence in favour of the benefits of a diverse design team outweighs any detriments.
Despite understanding the benefits of diversity in teams, just including designers
with different backgrounds does not guarantee creative thinking. The Mental Models
(MMs) of team members must be properly shared to be effective (Rosenman et al.
2007). Chen (2018), for example, argues that bicultural communication, which
combines two cultures’ original features into an integrated solution, is a necessary
prerequisite for innovation in heterogenous teams. Whether this is true or not, such
arguments about diversity and creativity, and the need for TMMs, SMMs or DMMs,
must all eventually consider the relationship between language and cognition.
Language, as a system, is both a reflection of the way we think and of our socio-
cultural differences and values (Boroditsky 2001; Gleitman and Papafragou 2005).
Linguists, philosophers, anthropologists and psychologists have long been interested
in the relationship between language and thought (Lee et al. 2016). Structuralist
thinkers, for example, argue that language is fundamentally universal in its cognitive
intent and logic (Chomsky 1965), whereas post-structuralists maintain that cognition
must precede language development (Pinker 1994). There is also disagreement about
the nature of cognitive in communication, with some researchers emphasising the
importance of visual imagery and models (Arnheim 1997; Paivio 1971, 1983; Ware
2008), while a counterargument maintains that word order and information structure
are more important (Hörnig et al. (2006). In both cases, evidence confirms that
language influences conceptual development, which in turn shapes the capacity to
develop and communicate MMs (Bowerman 1996; Levinson 1996, 2003). In design
teams, this capacity turns out to be especially reliant on spatial reasoning.
Tenbrink and Ragni (2012) identify three stages of spatial thinking in the commu-
nication of MMs: comprehension, description and validation. In the comprehension
stage, the given conditions or controls are integrated into a unified MM using people’s
knowledge of the semantics of spatial expression. In the description stage, the MM
is defined and examined to reveal implicit relations. Finally, in the validation stage,
if there is an alternative model of the original conditions or controls the designer
returns to the start to review it. Alternatively, the team can arrive at an agreed MM
and apply it.
8.1 Introduction 193

Because MMs are “abstractions and represent only the essential parts” of a process
or system, they tend to be “more abstract than visual models” and thereby priori-
tise “qualitative relations instead of metric information” (Tenbrink and Ragni 2012,
p. 281). Thus, the tripartite comprehension-description-validation model is itself an
example of a MM that is reliant on spatial reasoning. It has both a linear progression
and an in-built loop returning to the start if needed. There is also an assumption in this
model that MMs use spatial reasoning for communication and this improves their
effectiveness, because spatial reasoning is relatively universal. The evidence for this
perspective is not, however, so compelling. There are indicators that the MMs which
some teams presume are shared, are not. Misunderstandings of this type are often
linked to linguistic terms or concepts and they may have general consequences for
problem-solving (Munnich et al. 2001). But is this also true of design? Are the MMs
of designers linguistically constrained? To address these two questions, this chapter
conducts a parallel investigation of both cognitive and linguistic characteristics in
the design process.
To undertake this investigation a method must be available to record and analyse
cognitive and linguistic characteristics in design. As a useful point of reference
for this, Paivio’s dual coding theory identifies three types of cognitive processing:
representational, referential and associative (Clark and Paivio 1987; Paivio 1983,
1991). Representational processing refers to the direct activation of verbal or non-
verbal characterisations. Referential processing deals with the activation of the verbal
system by the non-verbal system or vice versa. Associative processing is the activation
of representations within the same verbal or non-verbal system. As discussed previ-
ously (Chap. 5), both visual and verbal representations not only support the develop-
ment of individuals’ design cognition and distributed MMs, but also facilitate design
communication for SMMs. In addition, several components of the Design Team
Cognition (DTC) model (Fig. 5.1) directly support design teams. Spatial language
is crucial to cognitive representation and communication in design teams. Further-
more, individual fluency in verbal representation has an impact on both the practical
operations of the team, and the transformation of MMs to SMMs. As such, “verbal
literacy” is important in design because language facilitates bridging between indi-
vidual knowledge and experience and that of the team (Dong 2005; Jara 2014). Past
research into communication (Burleson and Caplan 1998) also identifies the role
played by an individual’s social information-processing capacity in the development
of “team-related cognition” (DeChurch and Mesmer-Magnus 2010).
This brief overview sets the context for the present chapter which examines
multiple approaches to understanding cultural design thinking in terms of (i) design
cognition, (ii) cognitive and syntactical complexities and (iii) spatial language. The
multi-aspect methodology developed in this chapter is used to ensure that the inves-
tigation of design cognition considers language, and the investigation of language
takes into account cognitive factors. Thus, a dual-coding system, which captures
both design cognition and spatial language, is used to analyse protocol data derived
from design experiments. It allows for comparisons to be constructed between the
actions and thought processes of designers and the ways they describe these actions.
The protocol data is derived from two Australian design protocols and two Swedish
194 8 Design Thinking Across Borders

design protocols developed under the same experimental conditions. The protocol
data is quantitatively measured, mathematically analysed and then visualised using
linkography (see Chap. 4).

8.2 Research Procedure

By observing groups of designers working separately on the same task and in the same
environment, a direct comparison can be made between the ways different designers
or groups think and work. For this chapter two Australian designers (Au1 and Au2 in
Chap. 2) and two Swedish (Sw1 and Sw2) undertook the same task (a concept design
for a high-rise building), using the same tool, Grasshopper, and while “thinking
aloud” during their design sessions. All participants had experience in parametric
design including successfully completing at least one major architectural design
project using the graphical algorithm editor. All designers were asked to undertake
the task in approximately one hour. However, taking into account varying needs
of processing time and debugging, they were allowed to finish earlier or continue
overtime. The research procedure is the same as Study II in Chap. 2.
All think-aloud vocalisations for this study were in English regardless of the
native language of the participant. Thus, the interpretation of the actions has an
innate English bias, but the common language supports a more consistent coding
and analysis process. This also responds to the limitations identified by Boroditsky
(2001) when using and comparing native languages. Importantly, both English and
Swedish languages share some similar characteristics, but they also have differences.
For example, both languages use the standard word order—Subject-Verb-Object
(SVO),—while Swedish also follows the Verb-second (V2) order, as does the German
language (Lee et al. 2016). The composition, “there + verb”, is common in Swedish,
whereas “there + to be” is more common in English. In the case of this study, there
might be further differences in the use of spatial language between two groups of
designers. However, this study doesn’t aim to capture this level of cultural or linguistic
differentiation, but rather to highlight whether or not the spatial language reasoning
of each group can be identified through the spatial language coding scheme. The
analysis of the results of the cross-national protocols is reported in three stages:
design cognition, complexity and spatial language.

8.3 Design Cognition

8.3.1 Coding Results

Table 8.1 reports the general characteristics of the four design protocols. Three partic-
ipants in the experiment completed the design within one hour, while Au2 took longer.
8.3 Design Cognition 195

Table 8.1 General results of the four design protocols

Designer Design time Num. of segments Coded segments Average time of segments
(s)
Au1 47 min 54.7 s 220 208 (94.5%) 13.0
Au2 1 h 27 min 8 s 319 277 (86.8%) 16.5
Sw1 1 h 4 min 22 s 360 347 (96.4%) 10.7
Sw2 1 h 3 min 12 s 303 278 (91.7%) 12.5
Mean 1 h 5 min 42 s 300.5 277.5 (92.3%) 13.2
SD 16 min 9.9 s 58.8 56.7 2.4

The average duration of each coded segment was 13.2 s and the average value of the
number of segments was 300.5. Sw1’s protocol has the highest number of encoded
segments (360), and the shortest average time of segments (10.7 s). This result indi-
cates that Sw1 produced clearer cognitive activities and developed them more quickly
than the others. In contrast, Au2 has the smallest percentage of the encoded segments
(86.8%) and therefore took more time to complete a cognitive activity. Au2 also
produced more unnecessary (non-encoded) and trouble-shooting activities, including
finding algorithms (rules) and arranging rules. On average, 92.3% of segments were
encoded using the design cognition coding scheme.
Table 8.2 reports the coding coverage (%) of each cognitive subclass. The most
dominant activity coded in this study is A-Rule (writing an algorithm), the second
most is E-Geometry (visually evaluating the outcome of a rule) and the third is E-Rule
(evaluating the rule itself). On average, these three activities account for over 50%
of the codes, a result that is similar to the percentage of coding results for Study II
in Chap. 2. The algorithmic representation is the preferred medium in the physical
level of parametric design, where it is regularly evaluated in the 3D view as well
as in the scripting view. This switching pattern closely corresponds to the creative
micro-processes (see Chap. 3) and co-evolutionary processes between design spaces
and modes (see Chap. 5).
Despite some similarities, there are clear differences between the results for the
participants’ cognitive activities. For example, Au2 and Sw2 each produce a larger
number of geometric activities, in essence, because they both drew shapes in the
3D view and then imported them into the algorithm editor. Both also constantly re-
evaluated their algorithmic rules (E-Rule), which accounts for over 16% of their time,
and were more limited in generating behaviours (G-Generation). Of interest is the
fact that the Australian designers addressed rule-creation activities more frequently,
whereas the Swedish designers developed more parameters than the comparative
group. Au1 and Au2 often undertook perceptual activities, while Sw1 and Sw2
tended to evaluate existing parameters (E-Parameter). The sample is too small to
draw conclusions about these trends, but the coding scheme identifies these clear
differences. In addition, each national group followed a similar pattern of algorithmic
representations, which might be shaped by their past experience and skills. That
is, while this protocol study is limited to identifying cross-national differences in
196 8 Design Thinking Across Borders

Table 8.2 The percentage of coding results

Level Category Subclass Au1 Au2 Sw1 Sw2 Mean SD
Physical Geometry G-Geometry 2.0 6.6 0.0 4.2 3.2 2.8
G-Change 0.5 9.3 0.7 0.4 2.7 4.4
Algorithm A-Parameter 4.9 2.6 6.0 6.1 4.9 1.6
A-Change 10.0 1.7 10.2 8.0 7.5 4.0
Parameter
A-Rule 24.7 20.5 17.4 16.9 19.9 3.6
A-Change 6.0 4.3 8.9 5.2 6.1 2.0
Rule
A-Reference 0.0 0.0 1.4 0.0 0.4 0.7
Perceptual Geometry P-Geometry 2.8 1.3 0.1 0.4 1.1 1.2
Algorithm P-Algorithm 3.2 2.9 1.0 2.2 2.3 1.0
Conceptual Problem-finding F-Initial Goal 3.0 0.9 0.6 4.2 2.2 1.7
F-Geometry 7.0 8.7 5.9 6.0 6.9 1.3
Sub Goal
F-Algorithm 4.1 4.0 1.8 7.1 4.3 2.2
Sub goal
Solution-generating G-Generation 4.0 2.0 3.9 2.5 3.1 1.0
Solution-evaluating E-Geometry 19.4 18.8 22.2 11.2 18.0 4.7
E-Parameter 0.8 0.0 9.6 9.0 4.9 5.2
E-Rule 7.6 16.3 10.2 16.5 12.7 4.5
E-Reference 0.0 0.0 0.0 0.0 0.0 0.0
Sum 100 100 100 100 100 100

parametric design, it also captures cognitive similarity and disparity between divisible
groups.

8.3.2 Linkographic Analysis

As discussed previously in Chaps. 4 and 5, linkography is typically used for

measuring “productivity” in teamwork (Costa and Sobek 2004; Goldschmidt 1990,
1995). After developing linkographs of the four design sessions, their link indexes
are calculated. The link indexes of Au1 (2.75) and Sw2 (2.76) are slightly higher
than the others (Au2, 2.39 and Sw1, 2.62) in their language groups. That is, the
volume of linking activities in Au1’s and Sw2’s design protocols indicate that the
interconnectedness of their thoughts is greater than the others. Goldschmidt and
Tatsa (2005) argue that there is a correlation between such high link-index values
and productivity. Furthermore, moves that are particularly rich in links also serve as a
sound, and arguably the best, indicator of productivity. In particular, “link-intensive”
8.3 Design Cognition 197

or “Critical Move” (CM) links are clearly related to four distinct moves in para-
metric design: introducing geometric/algorithmic ideas, creating algorithms (as a
unit), modification activities and evaluation activities (see Chap. 4). These moves
result in CM (forelinks) that earn their designation.
Table 8.3 shows the CMs which have more than five (CM5 ), six (CM6 ) and seven
links (CM7 ). Au1’s protocol develops the highest proportions of CMs over the total
number of moves, while Sw1 produced the lowest. For example, the CM5 percentage
of the total number of moves is 28.6% for Au1 and 19.7% for Sw1. If we accept
the rationale of Goldschmidt (1995), Kan and Gero (2008), the link indexes and the
figures in Table 8.3 (the percentages of CMs) confirm that both Au1 and Sw2 are
more productive than the others in their language groups. Interestingly, Sw1’s design
activities (Table 8.2), which featured the highest number of moves, didn’t contribute
to productivity in terms of idea generation. These linkographic indexes are useful for
identifying individual “productivity” in design, while less so for revealing similarity
and disparity between the two groups of designers.
A critical aspect of parametric design is the algorithmic component or set of
components in an algorithm unit. Thus, the number of algorithmic units in a protocol
is a possible indicator of productivity in parametric design. Figure 8.1 illustrates algo-
rithmic scripts showing clear graphical links between algorithmic units (or compo-
nents). Unexpectedly, the two less productive designers (Au2 and Sw1) developed
more algorithmic units than the more productive designers. That is, the number of
algorithmic units generated in a parametric design session may be not related to the
generation of ideas or to productive design thinking. Alternatively, a better way to
analyse graphical algorithms may be topological analysis using graph theory.

Table 8.3 Critical moves with more than five (CM5 ), six (CM6 ) and seven links (CM7 )
Designer Link CM5 (CM5 %) CM6 (CM6 %) CM7 (CM7 %)
Au1 Forelinks 44 (20.0%) 40 (18.2%) 35 (15.9%)
Backlinks 19 (8.6%) 10 (4.6%) 9 (4.1%)
Total 63 (28.6%) 50 (22.7%) 44 (20.0%)
Au2 Forelinks 51 (16.0%) 39 (12.2%) 34 (10.7%)
Backlinks 24 (7.5%) 15 (4.7%) 9 (2.8%)
Total 75 (23.5%) 54 (16.9%) 43 (13.5%)
Sw1 Forelinks 51 (14.2%) 44 (12.2%) 41 (11.4%)
Backlinks 20 (5.6%) 7 (1.9%) 5 (1.4%)
Total 71 (19.7%) 51 (14.1%) 46 (12.8%)
Sw2 Forelinks 48 (15.8%) 43 (14.2%) 36 (11.9%)
Backlinks 31 (10.2%) 20 (6.6%) 13 (4.3%)
Total 79 (26.0%) 63 (20.8%) 49 (16.2%)
198 8 Design Thinking Across Borders

Au1

Au2

Sw1

Sw2

Fig. 8.1 Four designers’ algorithmic scripts consisting of algorithmic units and links
8.4 Cognitive and Syntactical Complexities 199

8.4 Cognitive and Syntactical Complexities

8.4.1 Cognitive Complexity

8.4.1.1 Content Complexity

Table 8.4 presents the entropy and content complexity (H C ) values of the four design
protocols. The H C —being the summary of entropy values of all subclasses—of Sw2’s
protocol is the highest (3.5048) and Au2’s is the lowest (3.2540). Interestingly, the
ranking of H C values of the four protocols is identical to the order of their link
indexes. That is, H C may provide an alternative index for design productivity (idea
generation), which does not always result in a constructive outcome. One reason for
this is that a designer may be distracted by too many ideas in a limited timeframe.
For example, Sw2 generated many ideas but barely completed his design in the
one-hour session. Thus, Sw2’s H C value indicates that his design process is very
productive, even though it isn’t necessarily very timely or efficient. In addition, H C
is dependent on the development of categories (subclasses) of a coding scheme as
discussed in Chap. 4. Nonetheless, H C provides a simpler quantification of individual
design cognition than structural complexity, which requires the more time-consuming

Table 8.4 Entropy and content complexity (H C ) values of the four design protocols
Subclass Au1 Au2 Sw1 Sw2 Mean SD
G-Geometry 0.1129 0.2588 – 0.1921 0.1879 0.0730
G-Change 0.0382 0.3187 0.0501 0.0319 0.1097 0.1395
A-Parameter 0.2132 0.1369 0.2435 0.2461 0.2099 0.0509
A-Change Parameter 0.3322 0.0999 0.3359 0.2915 0.2649 0.1118
A-Rule 0.4983 0.4687 0.4390 0.4335 0.4599 0.0299
A-Change Rule 0.2435 0.1952 0.3106 0.2218 0.2428 0.0493
A-Reference – – 0.0862 – 0.0862 –
P-Geometry 0.1444 0.0814 0.0100 0.0319 0.0669 0.0597
P-Algorithm 0.1589 0.1481 0.0664 0.1211 0.1236 0.0413
F-Initial Goal 0.1518 0.0612 0.0443 0.1921 0.1124 0.0711
F-Geometry Sub Goal 0.2686 0.3065 0.2409 0.2435 0.2649 0.0304
F-Algorithm Sub goal 0.1889 0.1858 0.1043 0.2709 0.1875 0.0680
G-Generation 0.1858 0.1129 0.1825 0.1330 0.1536 0.0363
E-Geometry 0.459 0.4533 0.4820 0.3537 0.4370 0.0569
E-Parameter 0.0557 – 0.3246 0.3127 0.2310 0.1519
E-Rule 0.2826 0.4266 0.3359 0.4289 0.3685 0.0718
E-Reference – – – – – –
HC 3.3339 3.2540 3.2564 3.5048 3.3373 0.1177
200 8 Design Thinking Across Borders

development of a linkograph. Furthermore, since the entropy values of subclasses are

based on the cognitive coding results, they may be able to capture cognitive similarity
and disparity between design groups.

8.4.1.2 Structural Complexity

Entropy values developed from a linkograph can used to measure structural

complexity (H S ). Table 8.5 shows the results of entropy calculations derived from
the design protocols. In all protocols, backlink entropy, which measures the opportu-
nities arising from enhancements or responses (Kan and Gero 2008), is higher than
forelink entropy. To normalise the entropy values developed from different lengths
of design protocols, Table 8.6 records the entropy per move of the four design proto-
cols. The total cumulative entropy per move is regarded as the H S of each protocol.
Interestingly, the H S values parallel those of the percentages of CMs (Table 8.3).
The entropies per move in Au1’s protocol is higher than the others (Table 8.6).
This implies that Au1 has a higher level of H S (0.4635) and the resulting linkograph
has relatively greater information clustering and abstraction. Conversely, the lowest
value of H S in Sw1’s protocol (0.3053) suggests a relative reduction in cognitive
complexity. Sw1’s protocol has the shortest average duration of moves with the
highest number of segments that may be related to higher productivity. However,
the lowest value of entropy reveals that Sw1’s transition between design activities
in the process is fast but doesn’t produce a meaningful CM in terms of information
clustering and abstraction. Both Sw1’s and Au2’s protocols are relatively weaker
in cognitive terms, as a result of lower opportunities relating to idea generation
(forelink), enhancements or responses (backlink) and cohesiveness or incubation
(horizonlink).

Table 8.5 Entropy of the four design protocols

Forelink total H Backlink total H Horizonlink total H Cumulative total
Au1 35.23 52.56 14.17 101.96
Au2 32.70 50.67 15.93 99.30
Sw1 35.22 57.64 17.05 109.91
Sw2 33.50 55.14 17.95 106.59

Table 8.6 Entropy per move and structural complexity (H S ) of the four design protocols
Forelink H Backlink H Horizonlink H HS
Au1 0.1601 0.2389 0.0644 0.4635
Au2 0.1025 0.1588 0.0499 0.3113
Sw1 0.0978 0.1601 0.0474 0.3053
Sw2 0.1106 0.1820 0.0592 0.3518
8.4 Cognitive and Syntactical Complexities 201

This data reveals that H S is also potentially a good indicator of productivity. In

addition, while H C corresponds with the link index of each protocol, H S corresponds
to the result of CMs. That is, both complexities can be used to explore “productiv-
ity”, but the latter (H S ) is more related to the support for creativity, because quality
outcomes and creativity rely on ideas in CMs (Goldschmidt and Tatsa 2005). Thus,
H S might be more useful for comparing individual creative behaviours in a set of
design protocols.

Fig. 8.2 Decile growths of four designers’ linkographs

202 8 Design Thinking Across Borders

Au1 Au2 Sw1 Sw2

0.2
0.18 0.30
0.16
0.2
Forelink Backlink
0.25
Entropy per total move

Entropy per total move

0.1
0.14
0.1
0.12 0.20
0.10
0.1
0.15
0.08
0.1

0.1
0.06 0.10

0.04
0.05
0.02

0.00 0.00
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Decile time Decile time

0.10 0.5
0.50
0.09 Horizontalink 0.45
0.5 CumulaƟve
Entropy per total move

Entropy per total move

0.08 0.4
0.40
0.07 0.35
0.4
0.06 0.30
0.3
0.05 0.3
0.25
0.04 0.2
0.20
0.03 0.2
0.15
0.02 0.10
0.1
0.01 0.1
0.05
0.00 0.00
0.0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10

Decile time Decile time

Fig. 8.3 Changes of entropy per total moves

Decile growth plots of the linkograph for each protocol (Fig. 8.2) can be used to
describe the changes of entropy per total moves over time (Fig. 8.3), which shows
individual differences in the set of protocols and their unique patterns. For example,
in Fig. 8.3 Au1’s data dominates the results for forelink, backlink and horizonlink
entropies, and Sw2’s results are second. This is reflected in the final cumulative value
per move (H S ) column in Table 8.6, but the figures of the forelink and horizonlink
entropy capture some additional features of individual cognitive complexity.
Au1’s results for forelink entropy increase until the ninth decile time, while Au2’s
and Sw2’s entropies show only a small amount of change after the sixth and the
seventh deciles, respectively (Fig. 8.3). In the horizonlink entropy results, Au1’s
protocol develops a pyramidal peak at the fifth decile, while the others are almost
flat after the fourth. The horizonlink entropy result is related to the shape of each
linkograph, and in particular to any large “chunks” (the dense pyramidal zones) in
the graph. For example, adopting the problem-forwarding generative strategy that
develops designs in a step-by-step manner, Au1’s initial idea is divided into two
design goals, and the resultant linkographs develop into two large chunks (Fig. 8.2).
Thus, Au1’s horizonlink entropy has its highest value at the fifth decile where the
linkograph of the first chunk is completed (see Fig. 8.3). In contrast, Sw1’s entropy
values at the third or fourth decile are higher than the others, but they are almost
flat or even reduce thereafter. This unique pattern may be caused by too many small
chucks, meaning a lower level of information clustering in Sw1’s linkograph. While
the interpretation of the results of this small sample size is necessarily limited, the
8.4 Cognitive and Syntactical Complexities 203

measurement of entropies over time is useful for revealing features of individual

design processes as well as for visualising cognitive complexity in design protocols.

8.4.2 Syntactical Complexity

In the fields of psychology, education, communication and linguistics, cognitive

research uses verbal and/or written data derived from experiments to rate the rela-
tive complexity of participants (Benet-Martínez et al. 2006; O’Keefe and Sypher
1981). To quantitatively calculate the linguistic complexity embodied in transcrip-
tions of design protocols, this chapter uses syntactic complexity. Ortega (2003)
defines syntactic complexity as the range of forms uncovered in language production
and the degree of sophistication of these forms. Syntactic complexity is often used
to describe linguistic maturity or competency and many measures of it have been
developed in linguistics, including length of sentence, length of T-unit (thought-unit),
length of clause, clauses per T-unit (C/T ) and dependant clauses per clause. A T-unit
(or minimal terminable unit) is a unit of “thought”, consisting of one main clause with
all subordinate clausal and non-clausal elements embedded in it (Hunt 1970). The
number of clauses per T-unit (or per utterance) is accepted as an index of linguistic
complexity (Cheung and Kemper 1992). As an example, Table 8.7 presents part of a
design protocol transcript that indicates the number of clauses and T-units. Based on
these values, the T-unit complexity ratio (C/T ) for this fragment of the transcript can
be calculated as 2.33 (7/3). The higher the value of this ratio, the higher the syntactic
complexity of the description.
Lu (2010) categorises fourteen measures of syntactic complexity into five types:
length of production unit, sentence complexity, subordination, coordination and
specific structures. Among the complexity measures, T-unit complexity ratio allows
for linguistic comparisons of relative complexity in grammatical constructions. This
measure is especially useful in the present context because designers’ utterances are
so variable in length in the design process, that a comparable value is required to
measure differences.

Table 8.7 Example of a design protocol showing the number of clauses and T-units
Transcript Clauses T-units
“So the brief is for the design of a high-rise building … its a conceptual design 2 1
of a high-rise building that will have two main areas, hotel and office. …
The first thing I’ll do is I’ll put the site constraints in, which is a maximum 4 1
floor area of fifty by fifty metres. …
So I’ll start by putting in a box.” 1 1
Total 7 3
C/T = 2.33
204 8 Design Thinking Across Borders

Table 8.8 Syntactic complexity of each protocol

Num of segments Num of clauses (C) Num of T-units (T ) Syntactic complexity
(C/T )
Au1 152.0 382.0 (2.51) 211.0 (1.39) 1.81
Au2 105.0 141.0 (1.34) 112.0 (1.07) 1.26
Sw1 192.0 116.0 (0.60) 73.0 (0.38) 1.59
Sw2 144.0 235.0 (1.63) 136.0 (0.94) 1.73
Mean 148.3 218.5 (1.47) 133.0 (0.90) 1.60
SD 35.7 120.4 (0.80) 58.1 (0.40) 0.20
(…) = value per segment

The linguistic analysis of cross-national protocol data in this chapter commences

by examining the syntactic complexity (C/T ) of each protocol (Table 8.8). The
syntactic analysis involves 30-min protocols developed from the four design
protocols.
Au1 produced the most complex protocol syntactically (1.81), while Au2 had
the lowest (1.26). In the cognitive coding, Au1 followed a higher proportion of
“expert” design processes, resulting in the problem-forwarding generative strategy
and a higher level of cognitive complexity. In contrast, Au2 presented a more “novice”
approach, often stopping to solve a problem as it emerged. The two Swedish language
designers also demonstrated quite different patterns of results. Sw1 produced the
smallest number of clause (C) and T-unit (T ) codes, while Sw2 produced slightly
higher figures. In addition, there are more trouble-shooting activities in Sw2’s proto-
cols, while Sw1 developed the shortest value in the average time of moves. That
is, Sw1 changed design activities relatively quickly but used only a limited volume
of “thinking aloud”. This might be a limitation of the concurrent verbalisation tech-
nique adopted in the study, but the shorter moves resulted in a lower level of cognitive
complexity in the previous section. Conversely, Sw1 produced an average value of
the syntactic complexity (C/T ). Collectively, the figures of C/T in Table 8.8 are
similar to the levels of H S in Table 8.6. That is, a designer’s syntactic complexity
is related to their cognitive complexity, despite being developed from completely
different coding schemes.

8.5 Spatial Language

As earlier sections in this book and chapter reveal, “spatial language” is central to
the development of design cognition, MMs and their shared, team or distributed
counterparts. In a transcript, explanations of objects, and their absolute or relative
relationships can be used to analyse the depth and intricacy of the spatial language
being used. Four types of spatial language—spatial relation, spatial object, spatial
8.5 Spatial Language 205

Table 8.9 Spatial language coding scheme

Category Subclass Description
Spatial relation L-relation Local relation (e.g. the usage of projective terms (left, right,
G-relation front, behind, above))
D-pronoun Global relation (e.g. the usage of compass-based terms such as
north and south)
Demonstrative pronouns (e.g. the use of here, there)
Spatial object 1-object Describe one object
2-object Describe two objects
M-object Describe multiple objects
Spatial direction Direct Express a direction, in which an object is located
Indirect e.g. the usage of “beside”
Syntactic format L-st-R Locatum—Spatial term—Relatum
st-R-L Spatial term—Relatum—Locatum
st-L Spatial term—Locatum

direction and syntactic format—are captured by the spatial language coding scheme
in Table 8.9.
The first category, spatial relation, characterises the spatial terms in each protocol
used to describe the connection between locatum (a position of something) and
relatum(s) (which it is relative too). This category accommodates the axial structure
of the reference object and its contact or support with respect to a surface (Munnich
et al. 2001). It also includes functional relationships between entities based on spatial
descriptions (Coventry and Garrod 2004) and the tendency to use spatial relational
terms that are sufficient and determinate enough to qualitatively define the spatial
relationship (Tenbrink and Ragni 2012). The spatial relation category also classifies
spatial terms into three subclasses: L-relation (local relation), G-relation (global rela-
tion) and D-pronouns (demonstrative pronouns). L-relation and G-relation highlight
the use of prepositional phrases to describe location, while D-pronouns (“there” and
“here”) replace the longer or more detailed descriptions of place or time.
The spatial object category in the coding identifies the number of objects involved
in each description. The description of an object and its relationship to other objects
is an indicator of the linguistic purpose, complexity and expertise.
Past linguistic research identifies that the third category, spatial direction, is signif-
icant because projective terms (left, right, front, behind, above, below) can have
very different meanings in different languages (Tenbrink and Ragni 2012). Spatio-
temporal configurations and spatial distinctions are both examples where linguistic
differences are exhibited (van der Zee and Slack 2003).
The final category, syntactic format, focuses on the way speakers frame their
spatial descriptions syntactically. Hörnig et al. (2006) argue that word order and
information structure are an important part of the mental reasoning process. Speakers
typically adopt syntactic formats that are appropriate for both the information struc-
ture chosen and for their descriptions (Tenbrink and Ragni 2012). For Tenbrink and
Ragni (2012) a locatum is necessarily described in terms of both a relatum and the
206 8 Design Thinking Across Borders

Table 8.10 Spatial language coding results

Spatial relation Spatial object Spatial Syntactic
direction format
L-relation D-pronoun 1-object 2-object Direct L-st-R
Au1 27 (0.24) 10 (0.05) 3 (0.01) 3 (0.01) 1 (0.00) 24 (0.11)
Au2 7 (0.03) 3 (0.03) 0 0 1 (0.01) 3 (0.03)
Sw1 7 (0.10) 6 (0.08) 0 0 0 (0.00) 1 (0.01)
Sw2 7 (0.05) 18 (0.13) 0 1 (0.01) 1 (0.01) 16 (0.12)
Mean 12 (0.08) 9.25 (0.07) 1.50 (0.01) 1.41 (0.01) 0.75 (0.01) 11.00
(0.07)
SD 10 (0.03) 6.50 (0.05) 0.75 (0.00) 1.00 (0.01) 0.50 (0.00) 10.92
(0.06)
(…) = normalised value calculated by the frequency per T-unit

particular spatial term that defines the nature of the relationship. For example, the
sentence, “a bedroom is to the right of the entrance”, is encoded as “L-sr-R” because
it consists of a locatum (bedroom), a spatial term (to the right of) and a relatum
(entrance). Such a coding reveals the linguistic structure of a decision or proposal,
as well as its general grammatical format.
The coding results for spatial language are presented in Table 8.10. In the spatial
format category, the designers only used “L-st-R”. Furthermore, no designers used
compass-based terms describing global relations (G-relation) and indirect direct-
ness. For the spatial object category, the linguistic analysis identifies only a small
number of objects that are in spatial relationships. The two native Swedish speakers
used higher frequencies of demonstrative pronouns, 6 and 18, respectively, than the
two native English speakers. Au1, the more “expert” designer, used spatial relation,
spatial object and syntactic format terms and clauses more often than Au2, the more
“novice”. While these outcomes generally support the syntactic complexity values
of the four protocols, these results may also reflect linguistic or cognitive differ-
ences. In order to examine this aspect of concurrent verbalisations and designers’
different think-aloud skills, the Frequency (F) divided by T-unit normalises the
data. These normalised values (F/T ) confirm the possibility that the cognitive design
process of a designer may be related to the linguistic features of his or her cognitive
representations.

8.6 Conclusion

This chapter has examined multiple approaches to exploring cultural design thinking
in cross-national design protocols. Its purpose was to explore the relationship between
language and cognition in design, a relationship which is significant for the formation
and communication of MMs, SMMs, TMMs and DMMs.
8.6 Conclusion 207

The first major finding of the chapter is methodological. This chapter demon-
strates that a dual-coding system (design cognition and spatial language) enables
the capturing of (i) design cognition, (ii) cognitive and syntactical complexities and
(iii) spatial language in a design environment. With a small sample size, the study
is not able to compare the design thinking processes of the two sets of designers in
a generalisable way, but the results confirm that the mixed analytic approach can be
effective for exploring diverse aspects of the design process and design cognition.
A potentially more important outcome of the chapter is that its analysis suggests
that cognitive and linguistic indexes—including link index, the percentage of CM,
Hc, Hs and syntactical complexity—are co-related with each other. These indexes are
potential indicators of creative, collaborative and cultural aspects of design thinking.
In particular, cognitive analysis using Hc may offer a valuable method for under-
standing similarities and differences in the cognitive characteristics of different
groups of designers. Linguistic analysis using the spatial language coding scheme can
capture cross-national differences in spatial representation. However, a conceptual
design-form generation task would not be appropriate for future research, because it
may only contain limited spatial relationship content. That’s why linguistic research
often emphasises the importance of spatial descriptive tasks, like “3-term series
tasks”.
Finally, this chapter demonstrates that the CMs and structural complexity
measures developed through linkography can be employed to capture the individual
differences of design productivity as well as creativity in a set of protocols. The
development of linkography, however, is still a significant challenge and a time-
consuming process. As an alternative, this chapter shows that Hc might be used for
measuring both individual and group characteristics in the design process. The next
chapter of Part III Culture in this book, continues the exploration of multi-cultural
design thinking in terms of the language of design.

References

Arnheim, R. 1997. Visual Thinking. Los Angeles: University of California Press.

Benet-Martínez, Verónica, Fiona Lee, and Janxin Leu. 2006. Biculturalism and cognitive
complexity: expertise in cultural representations. Journal of Cross-Cultural Psychology 37 (4):
386–407. https://doi.org/10.1177/0022022106288476.
Boroditsky, Lera. 2001. Does language shape thought?: Mandarin and English speakers’ conceptions
of time. Cognitive Psychology 43 (1): 1–22. https://doi.org/10.1006/cogp.2001.0748.
Bowerman, M. 1996. Which way to the present? Cross-linguistic differences in thinking about time.
In Rethinking Linguistic Relativity, ed. J. Gumperz and S. Levinson, 145–176. Cambridge, MA:
Cambridge University Press.
Burleson, B.R., and S.E. Caplan. 1998. Cognitive complexity. In Communication and personality:
Trait perspectives, eds. J.C. McCroskey, J.A. Daly, M.M. Martin, and M.J. Beatty, 233–286.
Cresskill, NJ: Hampton Press.
Chen, Chong-Wen. 2018. New product styles and concepts in the bicultural context. The Design
Journal 21 (6): 771–787. https://doi.org/10.1080/14606925.2018.1516496.
208 8 Design Thinking Across Borders

Cheung, Hintat, and Susan Kemper. 1992. Competing complexity metrics and adults’ production
of complex sentences. Applied Psycholinguistics 13 (01): 53–76. https://doi.org/10.1017/S01427
16400005427.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Clark, James M., and Allan Paivio. 1987. A Dual Coding Perspective on Encoding Processes. In
Imagery and Related Mnemonic Processes: Theories, Individual Differences, and Applications,
eds. Mark A. McDaniel, and Michael Pressley, 5–33. New York, NY: Springer New York.
Costa, Ramon, and Durward K Sobek. 2004. How process affects performance: an analysis of student
design productivity. In ASME 2004 International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference. Salt Lake City, Utah, USA: American
Society of Mechanical Engineers.
Coventry, K.R., and S.C. Garrod. 2004. Saying, Seeing And Acting: The Psychological Semantics
of Spatial Prepositions. Hove and New YorK: Psychology Press.
Cox, Taylor H., and Blake Stacy. 1991. Managing cultural diversity: Implications for organizational
competitiveness. The Executive 5 (3): 45–56.
DeChurch, Leslie A., and Jessica R. Mesmer-Magnus. 2010. The cognitive underpinnings of effec-
tive teamwork: A meta-analysis. Journal of Applied Psychology 95 (1): 32–53. https://doi.org/
10.1037/a0017328.
Dong, Andy. 2005. The latent semantic approach to studying design team communication. Design
Studies 26 (5): 445–461.
Florida, R. 2002. The Rise of the Creative Class: And How It’s Transforming Work, Leisure,
Community and Everyday Life. New York: Basic Books.
Gleitman, Lila, and Anna Papafragou. 2005. Language and thought. In Cambridge Handbook of
Thinking and Reasoning, ed. Keith J. Holyoak and Robert G. Morrison, 633–661. New York:
Cambridge University Press.
Goldschmidt, Gabriela. 1990. Linkography: Assessing design productivity. In Cyberbetics and
System ‘90, World Scientific, ed. R. Trappl, 291–298. Singapore: World Scientific.
Goldschmidt, Gabriela. 1995. The designer as a team of one. Design Studies 16 (2): 189–209.
https://doi.org/10.1016/0142-694X(94)00009-3.
Goldschmidt, Gabriela, and Dan Tatsa. 2005. How good are good ideas? Correlates of design
creativity. Design Studies 26 (6): 593–611. https://doi.org/10.1016/j.destud.2005.02.004.
Hörnig, Robin, Klaus Oberauer, and Andrea Weidenfeld. 2006. Between reasoning. Quarterly
Journal of Experimental Psychology 59 (10): 1805–1825. https://doi.org/10.1080/174702105
00416151.
Horwitz, Sujin K., and Irwin B. Horwitz. 2007. The effects of team diversity on team outcomes:
A meta-analytic review of team demography. Journal of Management 33 (6): 987–1015. https://
doi.org/10.1177/0149206307308587.
Hunt, Kellogg W. 1970. Syntactic maturity in schoolchildren and adults. Monographs of the Society
for Research in Child Development 35 (1): iii–67. https://doi.org/10.2307/1165818.
Jackson, Susan E. 1991. Team composition in organizational settings: Issues in managing an increas-
ingly diverse work force. In Group Process and Productivity, 138–173. Thousand Oaks, CA, US:
Sage Publications, Inc.
James, Harold. 2009. The Creation and Destruction of Value: The Globalization Cycle. Cambridge,
MA: Harvard University Press.
Jara, Cynthia. 2014. Verbal literacy in the design process: Enthusiasm and reservation. In
ARCC/EAAE 2014 International Conference on Architectural Research Conference. Honolulu,
USA: University of Hawai‘i at Manoa.
Kan, Jeff W.T., and John S. Gero. 2008. Acquiring information from linkography in protocol studies
of designing. Design Studies 29 (4): 315–337. https://doi.org/10.1016/j.destud.2008.03.001.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2016. The language of design: Spatial cognition
and spatial language in parametric design. International Journal of Architectural Computing 14
(3): 277–288. https://doi.org/10.1177/1478077116663350.
References 209

Levinson, S.C. 1996. Frames of reference and Molyneux’s question: Crosslinguistic evidence. In
Language and Space, ed. P. Bloom and M. Peterson, 109–169. Cambridge, MA: MIT Press.
Levinson, S.C. 2003. Space in Language and Cognition: Explorations in Cognitive Diversity.
Cambridge University Press.
Lu, Xiaofei. 2010. Automatic analysis of syntactic complexity in second language writing.
International Journal of Corpus Linguistics 15 (4): 474–496.
Man, J. 2014. Design Teamwork in Distributed Intercultural Teams: Competition, Collaboration,
Cooperation. Eindhoven: Technische Universiteit Eindhoven.
Munnich, Edward, Barbara Landau, and Barbara Anne Dosher. 2001. Spatial language and spatial
representation: A cross-linguistic comparison. Cognition 81 (3): 171–208. https://doi.org/10.
1016/S0010-0277(01)00127-5.
O’Keefe, Daniel J., and Howard E. Sypher. 1981. Cognitive complexity measures and the relation-
ship of cogntive complexity to communication. Human Communication Research 8 (1): 72–92.
https://doi.org/10.1111/j.1468-2958.1981.tb00657.x.
Ortega, Lourdes. 2003. Syntactic complexity measures and their relationship to L2 Proficiency: A
research synthesis of college-level L2 writing. Applied Linguistics 24 (4): 492–518. https://doi.
org/10.1093/applin/24.4.492.
Paivio, Allan. 1971. Imagery and Verbal Processes. Oxford: Holt, Rinehart & Winston.
Paivio, Allan. 1983. Mental Representations: A Dual Coding Approach. Oxford: Oxford University
Press.
Paivio, Allan. 1991. Dual coding theory: Retrospect and current status. Canadian Journal of
Psychology/Revue canadienne de psychologie 45 (3): 255–287. https://doi.org/10.1037/h00
84295.
Pinker, S. 1994. The Language Instinct: The New Science of Language and Mind. New York:
William Morrow and Company.
Rosenman, M.A., G. Smith, M.L. Maher, L. Ding, and D. Marchant. 2007. Multidisciplinary collab-
orative design in virtual environments. Automation in Construction 16 (1): 37–44. https://doi.org/
10.1016/j.autcon.2005.10.007.
Santandreu Calonge, David, and Askhat F. Safiullin. 2015. Can culturally, disciplinarily and educa-
tionally diverse (D3) teams function and be creative? A case study in a Korean university.
Educational Studies 41 (4): 369–392. https://doi.org/10.1080/03055698.2015.1018871.
Shalley, Christina E., Jing Zhou, and Greg R. Oldham. 2004. The effects of personal and contextual
characteristics on creativity: Where should we go from here? Journal of Management 30 (6):
933–958. https://doi.org/10.1016/j.jm.2004.06.007.
Tenbrink, Thora, and Marco Ragni. 2012. Linguistic principles for spatial relational reasoning. In
Spatial Cognition VIII, ed. Cyrill Stachniss, Kerstin Schill, and David Uttal, 279-298. Lecture
Notes in Computer Science. Berlin Heidelberg: Springer.
van der Zee, E., and J.M. Slack. 2003. Representing Direction in Language and Space. Oxford
University Press.
Ware, C. 2008. Visual Thinking for Design. Burlington, MA: Morgan Kaufmann.
Chapter 9
The Language of Design Thinking

Abstract This chapter investigates the relationship between language and cognition
in design. This is a critical topic for supporting effective multi-national design teams,
and it also illuminates assumptions about design’s capacity to function as a type of
universal language. The chapter reports on the results of a design experiment where
23 participants from three culturally and linguistically different groups completed the
same design task. The protocol analysis reveals the characteristics of “design think-
ing” and “design language” across the groups. The comparative analysis confirms
that cognitive allocation is related to the production of information categories and
spatial linguistic principles. It also indicates that there are observable differences in
design cognition, design information and spatial language between Australian and
Asian designers. This research develops fundamental knowledge about how different
cultural and linguistic groups understand, communicate and undertake design.

9.1 Introduction

Throughout history, a major driver of globalisation was trade in resources, from one
area with a surfeit (of materials, expertise or capacity) to another with a relative lack.
In effect, trade thrives where there is an inequity of access to resources. As a result of
this, the value proposition of globalisation is predicated on the existence of diversity
at a local level. Indeed, contrary to concerns about potential mass homogenisation
of consumer habits or values, a common side effect of globalisation is a revival of
interest in local produce, practices and experiences (Luna and Forquer Gupta 2001).
Even the obvious signs of a global economy, including the proliferation of brand
names or restaurant chains across international borders, often mask the fact that
the products or menus they offer vary from place to place to accommodate local
conditions. Values, social structures and behavioural characteristics differ across
nations and cultures, and globalisation does not necessarily erase these. People from
different countries have different value orientations and the capacity to merge global
and local conditions is arguably a necessity for survival in the modern world (Man
2014).

© Springer Nature Switzerland AG 2020 211

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_9
212 9 The Language of Design Thinking

In terms of cultural behaviours in a globalised world, it has been proposed that

western and eastern countries have much greater potential to work collaboratively
than they are currently (Molina et al. 2005). This is especially true in the case of multi-
national design teams, where a mixture of local and global knowledge is regularly
linked to enhanced productivity and creativity. There is, however, on-going debate
about the relationship between language and cognition in general, and specifically
in design teams. Although linguistic determinism (the view that language shapes
thought) has often been criticised, research into the use of language to describe space
and time has grown in significance, developing important insights about communi-
cation (Boroditsky et al. 2011; Corballis 2018; Tenbrink 2011). Nonetheless, thus far
relatively little is known about the relationship between design language and design
thinking.
Linguistic studies on spatial reasoning suggest that language, along with its
intrinsic cultural and value structures, has a clear impact on how knowledge about
space and form is developed, assessed and transmitted (Herskovits 1986; Tenbrink
and Ragni 2012; van der Zee and Slack 2003). Despite such findings, equivalent infor-
mation about design thinking and design language remains scarce. Our own research
on these topics (Lee et al. 2016), for example, draws connections between spatial
language use and productive and entropic cognitive processes. In order to expand
our knowledge base about this topic, this chapter compares cognitive design activ-
ities across three culturally and linguistically diverse groups. Its goal is to develop
an understanding of the way the characteristics of visual and verbal representa-
tions correlate to design cognition. This research necessarily considers the impact
of language on design communication by way of information categories (Suwa and
Tversky 1997) and spatio-linguistic principles (Tenbrink and Ragni 2012).
This chapter uses protocol analysis to investigate recordings of a design process
and then encode and classify various activities using a specific system (Chai and
Xiao 2012; Cross et al. 1996; Gero and Neill 1998; Lee et al. 2015). This system is a
triple-coding scheme of design cognition, design information and spatial language.
It captures design thinking activities (design cognition) and design language prop-
erties (“design information” and “spatial language”) in the design process. Twenty-
three protocols are examined in this chapter, selected from each of the three
different cultural and linguistic groups. This volume of participants and protocols
is statistically significant for this method, providing results that are reliable and
generalisable.
The chapter commences by presenting a conceptual model for analysing the rela-
tionship between design thinking and design language. After describing the model,
the chapter briefly outlines its methodological procedures. Thereafter the chapter
reports the results of the protocol analysis of the design experiment, including a
series of descriptive analyses, correlations and ANOVA results. It concludes with a
discussion about the implications of the results.
9.2 Design Thinking and Language Model 213

9.2 Design Thinking and Language Model

This chapter uses data derived from an empirical study to explore the relationship
between design thinking and design language during design communication. In order
to do this, a conceptual model is developed to study the language of design across
two parallel domains, thereby providing a new way of investigating how they are
related (Fig. 9.1). The core of the method is a triple-perspective-coding scheme
for protocol analysis. The first perspective analyses design activities to measure
the cognitive allocations of designers. The next two perspectives examine the words
used by designers while they are engaged in the design process. The second considers
instances of “design information” which are central to the way design is represented
(design representation). The third addresses the use of “spatial language” (spatial
representation). Consequently, the design thinking and language model (Fig. 9.1) is
suggested from the coding categories that were developed from past research (Suwa
et al. 1998; Suwa and Tversky 1997; Tenbrink and Ragni 2012) and are described
previously in this book (Chaps. 2, 5 and 8, respectively).
The first component of the model, design cognition, captures five categories:
representation, perception, function, evaluation and goal setting. These categories
are selectively adapted from Suwa et al.’s coding scheme (1998). Representation
activities consist of three subclasses: R-drawing, R-writing and R-label. R-drawing
is used to capture designer’s drawing activities, while R-writing and R-label encapsu-
late writing texts and making labels, respectively. The perception category comprises
focussing on, or paying close attention to, something. It has three subclasses, P-
drawing, P-brief , and P-writing, which relate to a designer paying attention to draw-
ings, a design brief or a given site (functional or performative needs), and its written
content, respectively. The function, evaluation and goal-setting categories also have
subclasses to code each cognitive allocation in the design process.

Design cogniƟon
 RepresentaƟon  EvaluaƟon
 PercepƟon  Goal seƫng
Design Thinking
CogniƟve acƟviƟes
 FuncƟon

The Language
of Design

Design informaƟon SpaƟal language

 Property  Object
Design Language
 FuncƟonality  LocalisaƟon
CogniƟve representaƟons
 Directness

Fig. 9.1 A design thinking and language model

214 9 The Language of Design Thinking

Design language is explored in the second component of the model, which

considers both “design information” and “spatial language” in visual and verbal
(cognitive) representations. The “design information” code highlights two categories,
property and functionality, which are based on Suwa and Tversky’s coding scheme
(1997) (Table 5.4). The property category consists of five subclasses: P-space, P-
object, P-shape, P-size and P-enclosure. The first four subclasses are adopted from
the emergent properties in Suwa and Tversky’s development, while the last captures
the depiction of an enclosure (e.g. walls, doors, windows, etc.). The functionality
category also consists of five subclasses, F-role, F-features, F-view, F-light and
F-circulation, which relate to designers’ consideration of practical roles, abstract
features and their interactions, views, lights and circulation, respectively.
The last component of the model, “spatial language” consists of three categories—
object, localisation and directness—which address spatial representations charac-
terised by the tendency to employ spatio-relational terms. The spatial language coding
scheme (see Table 8.9) draws on the work of Tenbrink and Ragni (2012). The object
category identifies the number of objects in each segment. Since the objects refer
to representations of spatial relations by way of position, the first category captures
the fundamental property of spatial representation. The localisation category has
three subclasses: L-projective, L-absolute and L-demonstrative. L-projective relates
to projective expressions often used in spatial-reasoning tasks, while L-absolute
identifies the absolute reference-reasoning system. Directions and spatial distinc-
tions can be used to identify the different linguistic expressions used when referring
to spatio-temporal configurations (van der Zee and Slack 2003). Thus, the directness
coding category consists of two subclasses, D-direct and D-indirect, to investigate
both direct and indirect spatial relationships.

9.3 Methodological Procedure

9.3.1 Design Experiment

The 23 participants in the design experiment were masters-level architecture students

studying in Australia, but drawn from multiple countries of origin. The participants
were divided into three culturally and linguistically different groups: Group A, native
English speakers; Group B from Asian countries or regions for which English is
an official or common language; Group C, from Asian countries or regions for
which English is not an official language. All participants had proficient language
skills to conduct general verbal and written communication in English. The students
were recruited from the University of Newcastle, University of South Australia,
Deakin University and RMIT University. The experimental procedures and encoding
processes are identical to the ones for Study I in Chap. 2.
All of the participants were provided with a design brief and a 45-min timeframe
and asked to produce a number of design sketches, diagrams and drawings. They
9.3 Methodological Procedure 215

were also asked to verbalise their thoughts in English as they carried out design
activities. The think-aloud verbalisation included a request to describe what they
were looking at, thinking, doing and feeling. Participants were allowed to cease the
experiment at any time. Each design experiment was video recorded for analysis.
The design brief was for a house on a vacant suburban block, to accommodate
the needs of a mixed-generational family (see Box 2.1). Designers were asked to
deliver sketches of design stages/components of the mixed-generational housing
design, including initial conceptual design, site layout(s) of the mixed-generational
house, plan(s) and elevation(s) and/or section(s). Design drawings produced during
the experiment were analysed in conjunction with the transcript to reveal the design
thinking processes.

9.3.2 Analysis Procedure

For the initial analysis, a descriptive review of the coding results was conducted to
identify and describe the basic features of the protocol data. Then bivariate (Pearson)
correlations were run using IBM SPSS Statistics (vers. 25) to provide a structure for
defining the relationships between the categories of the three schemes. Finally, one-
way analysis of variance (ANOVA) with Scheffé post hoc tests were conducted
to examine if there are any statistically significant differences in terms of design
cognition, design information and spatial language among the three cultural groups.
The Scheffe tests allow comprehensive comparisons involving contrasts of more
than two means at a time (Stevens 1999). Before the ANOVA, normality tests are
used to determine if each variable in a set of protocol data has a normalised distribu-
tion. In summary, the Pearson correlations are used to investigate the relationships
between the three sets of results encoded with the triple-coding systems for design
cognition, design information and spatial language. The ANOVA is used to compare
the cognitive and linguistic characteristics among the three cultural groups (as one
control and two experiment groups).

9.4 Results

9.4.1 Coding Results

The 23 design protocols collected for analysis consist of eight from Group A (control
group), eight from Group B (experiment group) and seven from Group C (experiment
group). The complete group comprises 12 females and 11 males. Table 9.1 shows the
properties of the design protocols. The average time per segment is 17.73 s, while
the average time duration is 42 min 19 s to complete the conceptual design task.
Because some participants took slightly longer or shorter times to complete the task,
216 9 The Language of Design Thinking

Table 9.1 Design protocols analysed across the three groups

Group Designer code Gender Time duration Num. of Segments Time per segment (s)
A A1 Female 44 min 5 s 216 12.24
A2 Male 42 min 56 s 134 19.22
A11 Female 44 min 45 s 170 15.80
A15 Male 41 min 21 s 120 20.68
A23 Female 44 min 33 s 173 15.45
A24 Male 38 min 18 s 121 18.99
A28 Female 43 min 4 s 143 18.07
A29 Male 44 min 16 s 145 18.32
B B3 Male 43 min 20 s 182 14.29
B5 Male 44 min 41 s 140 19.15
B6 Female 46 min 27 s 153 18.22
B13 Female 46 min 52 s 174 16.16
B20 Female 44 min 2 s 135 19.57
B21 Female 46 min 8 s 149 18.58
B26 Male 37 min 35 s 128 17.61
B38 Female 41 min 1 s 123 20.01
C C8 Female 46 min 20 s 172 16.16
C19 Male 44 min 4 s 151 17.51
C31 Female 32 min 5 s 120 16.04
C36 Male 43 min 0 s 143 18.04
C37 Male 42 min 20 s 131 19.39
C39 Male 31 min 34 s 93 20.36
C40 Female 40 min 33 s 135 18.03
Mean 42 min 19 s 145.70 17.73
SD 4 min 6 s 26.41 2.06

the results are normalised for comparison. The coding results of “design cognition”
are then weighted by time duration of each code, while the coding results of “design
information” and “spatial language” use the average frequencies per five minutes of
each protocol.

9.4.1.1 Design Cognition

Table 9.2 presents the percentage of the frequency weighted by time duration of
each code of Group A’s protocols. The table also presents the content complexity
(H C ) value of each protocol. On average, representation activities account for 50.4%
(R-drawing: 43.6%, R-writing: 3.6%, R-label: 3.2%); perception activities account
 9.4 Results

Table 9.2 The coding results (the percentage of time duration) of Group A’s protocols (R: Representation, P: Perception, F: Function, E: Evaluation, G: Goal
setting)
Subclass A1 A2 A11 A15 A23 A24 A28 A29 Mean SD
R-drawing 30.0 58.5 43.3 50.1 41.3 43.6 39.8 42.0 43.6 8.21
R-writing 7.5 2.3 3.4 6.2 2.7 0.2 6.7 0.0 3.6 2.89
R-label 4.3 4.0 13.2 2.9 0.2 0.0 0.4 0.5 3.2 4.41
P-drawing 3.3 14.2 14.4 12.5 15.8 17.6 12.9 21.3 14.0 5.18
P-brief 10.0 6.8 3.0 10.8 14.9 11.5 10.2 15.3 10.3 4.03
P-writing 0.5 0.0 0.0 1.7 0.2 0.0 7.6 0.0 1.3 2.63
F-interaction 27.4 5.7 6.8 3.8 9.3 11.1 16.9 9.0 11.3 7.63
F-reaction 3.2 0.0 1.2 0.0 1.4 0.0 0.0 0.0 0.7 1.16
E-drawing 6.9 3.3 11.0 8.4 9.1 12.7 3.0 8.3 7.8 3.39
E-writing 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.1 0.25
E-label 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00
G-initial 2.7 2.4 2.7 0.6 2.1 2.4 0.2 2.1 1.9 0.96
G-sub 3.2 0.8 0.2 1.5 1.3 0.4 1.5 0.6 1.2 0.95
G-repeat 1.0 2.1 0.0 1.4 1.7 0.4 0.9 0.9 1.1 0.68
Sum 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 –
HC 2.8322 2.1269 2.5280 2.4189 2.5513 2.2630 2.5770 2.2866 2.4480 0.2222
217
218 9 The Language of Design Thinking

for 25.6% (P-drawing: 14.0%, P-brief : 10.3%, P-writing: 1.3%); function activities
account for 12.0% (F-interaction: 11.3%, F-reaction: 0.7%); evaluation activities
account for 7.9% (E-drawing: 7.8%, E-writing: 0.1%); and goal-setting activities
account for 4.1% (G-initial: 1.9%, G-sub: 1.2%, G-repeat: 1.1%). The average H C
value is 2.4480.
The coding results for each Group also identify the different design strategies
used by participants. For example, A1 documented formal and configurational ideas
in writing at an early stage of her design session and often revisited the brief during
the session to set up new goals. She used graphic devices (arrow symbols) to capture
interactions between people and design elements. In this way, her protocol produces
the highest percentages of R-writing (7.5%), F-interaction (27.4%) and F-reaction
(3.2%) in Group A. Conversely, A1 also generated the lowest percentage of R-
drawing (30.0%) and her design session developed the highset H C value (2.8322) in
Group A. In contrast, A2, despite being a productive designer, offered only limited
verbalisations while drawing. He repeatedly examined massing options and then
focused on the production of a detailed plan during the session. Thus, his protocol
produces the highest percentage of R-drawing (58.5%) and lower percentages of
F-interaction (5.7%) and E-drawing (3.3%). Collectively his protocol results in the
lowest H C value (2.4480) in Group A. These are just a few examples of distinct
cognitive design strategies in the results.
As identified in Chap. 2, three strategies are common in sketch-based design cogni-
tive activities. Evidence of the first of these, the textual-goal forwarding strategy, is
seen in three protocols in Group A (A1, A2 and A23). This strategy involves devel-
oping a clear, written summary of goals at the beginning of a design session, and then
continuing to refer to these throughout for assessing the designs against the textual-
goals. The second strategy, graphical-goal forwarding is found in the work of A28
and A29. They initially developed a set of conceptual sketches closely relating to
their goal-setting activities and then subsequently generated designs based on these
graphical-goals. The third strategy, drawing reflection, is seen in the work of A11
and A24 who each developed their ideas through graphic exploration and subse-
quent contemplation of these sketches. Both designers drew over existing drawings
after attending to (P-drawing) or evaluating (E-drawing) them. These approaches
explain the relatively larger proportions of E-drawing in the two protocols (11.0%,
12.7%, respectively). There is also evidence in the work of A23 of drawing reflection
behaviours, where she combines both textual-goal forwarding and drawing reflection.
Table 9.3 presents the results of Group B’s protocols and associated H C values. On
average, representation activities account for 56.2%; perception for 23.6%; function
for 10.74%; evaluation for 4.3%; and goal setting for 5.1% of the overall activities.
Compared to Group A’s results, Group B tended to produce more representation
and goal-setting activities, but less perception, function and evaluation activities.
Although design activities in the problem space, like goal setting, have a positive
influence on cognitive complexity (see Chaps. 4 and 8), the average H C value of
Group B (2.3974) is lower than Group A. This may indicate that Group B’s design
processes are less complex or productive than Group A’s.
 9.4 Results

Table 9.3 The coding results (the percentage of time duration) of Group B’s protocols (R: Representation, P: Perception, F: Function, E: Evaluation, G: Goal
setting)
Subclass B3 B5 B6 B13 B20 B21 B26 B38 Mean SD
R-drawing 42.5 46.5 58.6 48.6 58.9 41.0 35.5 39.7 46.4 8.60
R-writing 6.1 4.0 4.9 5.7 0.0 2.9 2.3 1.8 3.5 2.10
R-label 10.9 4.7 3.6 2.8 9.0 6.9 5.4 7.2 6.3 2.74
P-drawing 12.1 16.4 14.2 16.9 17.0 17.0 20.1 22.0 17.0 3.09
P-brief 3.4 8.3 2.2 4.9 3.9 8.6 16.7 1.7 6.2 4.95
P-writing 2.4 0.0 0.0 0.0 0.0 0.5 0.0 0.6 0.4 0.83
F-interaction 10.6 10.6 10.8 3.9 3.2 10.7 9.4 20.8 10.0 5.38
F-reaction 0.9 0.5 0.0 1.1 1.2 0.7 0.0 1.5 0.7 0.55
E-drawing 3.4 4.5 5.7 8.5 3.2 3.4 5.6 0.2 4.3 2.42
E-writing 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00
E-label 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00
G-initial 1.3 3.4 0.0 3.3 1.2 5.7 4.1 2.3 2.7 1.83
G-sub 6.4 1.2 0.0 3.8 0.8 1.5 0.0 0.7 1.8 2.21
G-repeat 0.0 0.0 0.0 0.4 1.6 0.8 0.9 1.6 0.7 0.68
Sum 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 –
HC 2.6885 2.4577 1.9410 2.4625 2.0017 2.6620 2.5831 2.3823 2.3974 0.2834
219
220 9 The Language of Design Thinking

Considering individual cognitive design strategies, B3’s and B5’s protocols

demonstrate a textual-goal forwarding strategy, while B38’s uses a graphical-goal-
forwarding strategy. Two of the members in Group B (B13 and B21) employ both
textual-goal and graphical-goal-forwarding strategies in their design sessions. The
H C values of these five designers confirm that their design processes are relatively
higher in complexity than the others in this group. In contrast, designers B6 and B20
develop their designs by responding to the given design brief or site, without devel-
oping a goal-setting activity in advance. This results in a relatively larger proportion of
R-drawing (58.6% and 58.9%) and smaller amounts of goal-setting activities among
Group B. Thus, data for both B6 and B20 shapes the lowest H C values in Group B,
1.9410 and 2.0017. In contrast, B26’s data shows a lack of a clear design strategy.
Despite conducting a detailed site analysis, drawing site sections and formulating
a written list of requirements he only produced a minimal plan layout. Thus, his
protocol has the lowest amount of R-drawing (35.5%), the highest amount of P-brief
(16.7%) and the second highest amount of P-drawing (20.1%) in Group B. One of
the most significant observations about Group B is that several designers tried to
remember their experiences growing up in native countries or regions, such as Hong
Kong and Malaysia. While doing this they would often physically close their eyes or
look upward, exhibiting the “memory recall or retrieval” behaviour. This behaviour
is not apparent in Group A’s protocols and may be worthy of further investigation.
The coding results for Group C’s protocols and H C values are in Table 9.4. On
average, representation activities account for 52.5% (perception 33.9%; function for
8.0%; evaluation for 3.4%; and goal setting for 2.1%) of the overall time spent on
activities. The results indicate that Group C’s designers tended to undertake more
perception activities than the other groups, with only limited evaluation and goal
setting. The average H C value of Group C (2.2311) is lower than that of the other
two groups. Thus, based on these measures, Group C’s design processes are the
least complex and productive of the three. Interestingly, when solving the design
task, most of the designers in this group also exhibited overt memory and recall
activities, evoking their own experience of multi-generational living in their native
countries or regions. Group C also spent a significant amount of time calculating
dimensions, sizes or volumes. That is, they tended to employ what might be called
an “engineering” approach to design thinking. Because of this pattern across Group
C, it could be speculated that their educational or cultural experiences (or potentially
both) have led to this approach.
Looking more closely at the cognitive patterns in Group C, C8 and C40 stand out
from the rest because their protocols emphasise graphical-goal-setting activities at the
start of their sessions. In contrast, the other designers in the Group C do not conduct
explicit goal-setting activities. In addition to the graphical-goal-forwarding strategy,
C8 and C40 also adopt a drawing reflection strategy and produce relatively higher
proportions of P-drawing (29.1%, 31.3%, respectively) and relatively lower propor-
tions of R-drawing (30.9%, 30.5%, respectively). Both designers’ protocols develop
relatively high H C values in their group, further differentiating them. Conversely,
C39 spent the majority of the session developing a detailed plan layout with only
limited verbalisation. He then completed the session by evaluating and extending
 9.4 Results

Table 9.4 The coding results (the percentage of time duration) of Group C’s protocols (R: Representation, P: Perception, F: Function, E: Evaluation, G: Goal
setting)
Subclass C8 C19 C31 C36 C37 C39 C40 Mean SD
R-drawing 30.9 58.8 53.8 38.4 32.2 61.7 30.5 43.8 13.87
R-writing 2.3 4.2 1.8 0.7 9.1 0.4 3.3 3.1 2.96
R-label 6.7 6.1 2.4 12.3 5.3 0.0 6.8 5.7 3.86
P-drawing 29.1 21.3 20.7 26.8 15.6 15.8 31.3 22.9 6.27
P-brief 16.6 3.8 5.7 10.5 17.6 14.0 8.2 10.9 5.35
P-writing 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.0 0.09
F-interaction 5.1 3.1 8.6 7.3 11.2 3.0 15.4 7.7 4.53
F-reaction 0.0 0.6 0.0 0.8 0.6 0.0 0.0 0.3 0.36
E-drawing 8.5 2.2 2.5 0.6 5.3 1.4 3.6 3.4 2.70
E-writing 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00
E-label 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.00
G-initial 0.7 0.0 2.6 1.5 0.7 2.1 0.6 1.2 0.93
G-sub 0.0 0.0 1.9 0.4 2.3 0.0 0.0 0.7 1.00
G-repeat 0.0 0.0 0.0 0.6 0.0 1.5 0.0 0.3 0.53
Sum 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 –
HC 2.4297 1.864 2.1035 2.3454 2.7229 1.7253 2.4267 2.2311 0.3511
221
222 9 The Language of Design Thinking

the design with a perspective and two sections. His design approach results in the
highest percentage of R-drawing (61.7%) and the relatively lower percentages of F-
interaction (3.0%) and E-drawing (1.4%) across Group C. Collectively, C39’s design
session results in the lowest H C value (1.7253) in the experiments.
Although each designer in Group C uses slightly different cognitive strategies
(see Table 9.4) the data uncovers similar distributions and weights across the codes,
reflecting some common tendencies. Most of Group C’s designers, for example, only
developed “micro” goals, that do not direct the overall design in an integrated way.
Such micro-goals tend to arise from specific requirements of the design brief, rather
than holistic ones. By prioritising micro-goals, the primary drivers for the design
are limited to local or isolated configurational solutions to practical issues (like
separation, mobility, privacy and accessibility). Thus, these designers continuously
developed secondary solutions by reflecting on the capacity of their proposals (in
existing drawings) to meet specific aspects of the brief. This leads to multiple P-
drawing and P-brief activities as a precursor to finalising the design deliverables.

Group A
Coding coverage (%)

60.0

RepresentaƟon
40.0 PercepƟon
FuncƟon
20.0 EvaluaƟon
Goal seƫng
0.0
A1 A2 A11 A15 A23 A24 A28 A29 Average
Designer

Group B
Coding coverage (%)

60.0

RepresentaƟon
40.0
PercepƟon
FuncƟon
20.0 EvaluaƟon
Goal seƫng
0.0
B3 B5 B6 B13 B20 B21 B26 B38 Average
Designer

Group C
Coding coverage (%)

60.0

RepresentaƟon
40.0 PercepƟon
FuncƟon
20.0 EvaluaƟon
Goal seƫng
0.0
C8 C19 C31 C36 C37 C39 C40 Average
Designer

Fig. 9.2 The percentage of coding results of each cognitive category across the three groups
9.4 Results 223

Figure 9.2 presents graphs of the three groups’ coding results for “design cogni-
tion”. These graphs display the proportion (%) of the frequency, weighted by time
duration, of each category. The most dominant category is representation, the next is
perception and the last is function. While the graphs for each group suggest a similar
order of cognitive allocation, they also demonstrate different patterns in the data. For
example, the graphs show that the rise in representation activities often corresponds
to a fall in perception activities. Group A’s designers tended to produce more function
and evaluation activities than the other groups, while Group B’s designers tended
to exhibit more representation and goal-setting activities than the other groups. The
graphs also indicate that Group C’s designers tended to produce more perception
activities and less evaluation and goal-setting activities, than the other two groups.

9.4.1.2 Design Information

Figure 9.3 graphs the three groups’ coding results for “design information”, the first
component of “design language”. The graphs describe the average frequencies of
each category for each five-minute interval. Across the three groups, most designers
produce more terms and clauses associated with the property category than func-
tionality. Although the graphs show individual differences, on average Group A’s
designers produce more “design information” (average 20.1 for five minutes) than
the other two groups. In contrast, Group C’s designers produce the least (average
9.3).
Although Group C’s designers may be more limited in their capacity for verbal
representation in English, Group B’s designers display a similarly low level of verbal
representation. This might mean that English proficiency isn’t the only factor shaping
the result. It is also possible that different design styles and strategies have an impact
on cognitive allocations. For example, the two designers (A1 and A28) who are the
most productive in terms of “design information”, adopted clear design strategies:
textual-goal forwarding and graphical-goal forwarding, respectively. These classical
or well-organised design strategies may be related to the advantageous use of design
language. In contrast, designer C39 has the lowest frequency in the coding results
of “design information”. He also developed the highest percentage of representation
category (69.1%) in Fig. 9.2. This could be the reason for his low values for function
and evaluation categories in “design cognition”, which also result in the lowest
frequencies in both “design information” (Fig. 9.3) and “spatial language” (Fig. 9.4).

9.4.1.3 Spatial Language

Tables 9.5, 9.6 and 9.7 report the three groups’ coding results for “spatial language”,
the second component of “design language”. These results describe the average
frequencies of each subclass for each five-minute interval.
On average, the designers tend to describe either one object or multiple objects,
but only rarely two objects. This result is somewhat contrary to expectations, as the
224 9 The Language of Design Thinking

Property FuncƟonality
Group A
30.0
Frequency

20.0

10.0

0.0
A1 A2 A11 A15 A23 A24 A28 A29 Average
Designer

Property FuncƟonality Group B

30.0
Frequency

20.0

10.0

0.0
B3 B5 B6 B13 B20 B21 B26 B38 Average
Designer

Property FuncƟonality Group C

30.0
Frequency

20.0

10.0

0.0
C8 C19 C31 C36 C37 C39 C40 Average
Designer

Fig. 9.3 The frequencies of each category for five minutes (design information) across the three
groups

PercepƟon EvaluaƟon Goal Seƫng

Percentage

Percentage

Percentage

A B C A B C A B C

Fig. 9.4 Box and whisker plots of the coding results for perception, evaluation and goal setting
across Groups A, B and C
9.4 Results 225

Table 9.5 The coding results of Group A’s protocols for spatial language (average frequencies per
5 min)
Category Subclass A1 A2 A11 A15 A23 A24 A28 A29 Mean SD
Object O-one 2.2 0.0 0.8 0.1 0.3 0.3 2.0 0.1 0.7 0.87
O-two 0.6 0.0 0.1 0.0 0.0 0.3 0.0 0.0 0.1 0.20
O-multiple 1.5 0.0 0.7 0.1 0.2 0.1 1.2 0.2 0.5 0.55
Localisation L-projective 3.3 3.3 4.2 2.2 6.8 2.4 4.4 3.6 3.8 1.47
L-absolute 2.6 0.6 0.4 0.0 2.4 1.3 3.8 0.9 1.5 1.31
L-demonstrative 18.2 3.5 19.1 0.6 14.5 11.2 6.4 12.4 10.7 6.72
Directness D-direct 3.9 2.1 2.7 1.2 7.3 2.1 6.7 2.6 3.6 2.26
D-indirect 2.0 1.7 0.3 1.0 1.7 1.3 1.5 1.9 1.4 0.56
Sum 34.2 11.2 28.4 5.2 33.2 18.9 26.0 21.8 22.4 10.27

Table 9.6 The coding results of Group B’s protocols for spatial language (average frequencies per
5 min)
Category Subclass B3 B5 B6 B13 B20 B21 B26 B38 Mean SD
Object O-one 2.0 0.3 1.1 0.0 0.0 0.0 0.1 0.6 0.5 0.70
O-two 0.0 0.0 0.0 0.0 0.1 0.0 0.1 0.0 0.0 0.06
O-multiple 1.5 0.0 0.0 0.0 0.0 0.2 0.8 0.0 0.3 0.55
Localisation L-projective 2.1 2.0 1.3 1.3 1.1 1.1 2.7 3.2 1.8 0.78
L-absolute 0.1 1.1 0.5 0.0 0.0 1.7 1.3 1.2 0.8 0.68
L-demonstrative 8.0 2.5 6.1 16.9 5.8 3.5 6.4 4.3 6.7 4.48
Directness D-direct 0.9 2.0 1.3 0.6 0.8 2.4 1.6 3.2 1.6 0.88
D-indirect 1.3 1.1 0.5 0.6 0.3 0.4 2.4 1.2 1.0 0.67
Sum 15.8 9.1 10.9 19.4 8.2 9.3 15.4 13.7 12.7 3.99

Table 9.7 The coding results of Group C’s protocols for spatial language (average frequencies per
5 min)
Category Subclass C8 C19 C31 C36 C37 C39 C40 Mean SD
Object O-one 0.0 0.1 0.6 0.1 0.6 0.0 0.2 0.2 0.26
O-two 0.1 0.0 0.3 0.0 0.0 0.0 0.1 0.1 0.12
O-multiple 0.2 0.2 0.8 0.5 0.4 0.0 0.1 0.3 0.26
Localisation L-projective 1.7 0.1 1.6 1.9 1.7 1.4 1.4 1.4 0.59
L-absolute 0.5 0.1 0.6 0.0 0.7 0.0 0.2 0.3 0.30
L-demonstrative 13.6 5.0 12.5 1.4 3.0 0.2 1.5 5.3 5.51
Directness D-direct 1.4 0.2 1.9 0.2 1.4 0.5 0.5 0.9 0.67
D-indirect 0.9 0.0 0.3 1.6 0.9 1.0 1.1 0.8 0.53
Sum 18.5 5.8 18.5 5.7 8.6 3.0 5.2 9.3
226 9 The Language of Design Thinking

design task deals with space, enclosure and function, which might be assumed to have
singular, binary and multiple components. It is, however, a complex design task, and
not a relational-reasoning task that specifically considers pairs (Tenbrink and Ragni
2012). In the data the most dominant code is L-demonstrative (using a demonstrative
pronoun system), the second is L-projective (a projective expression using projective
terms) and the third is D-direct (expressing a direct relationship between objects).
While designers only rarely described two objects, the production of subclasses in
the object category is limited across the three groups of designers. Consequently, the
most dominant category is localisation, the second is directness and the third is object.
In summary, Group A’s designers produced more “spatial language” (average 22.4
for five minutes) than the other two groups. Group A’s designers naturally used more
projective terms and demonstrative pronouns and produce more direct relationships
than the other groups.
Table 9.5 shows the coding results of Group A’s protocols for “spatial language”,
based on the average frequencies for five minutes. Two designers, A1 and A23,
produced the most frequent spatial-language activities, 34.2 and 33.2, respectively.
A23 only produced an average number of codes in “design information”, but she
produced the highest number of terms and clauses associated with localisation (23.7)
and directness (9.0) in the spatial-language coding. These results may reflect her
practice of evaluating drawings and generating ideas by reflecting on her drawings.
In contrast, A15, who produced the lowest frequency overall in “design information”
coding, also developed the lowest frequency in the production of “spatial language”
(in total, 5.2).
The coding results of Group B’s protocols for “spatial language” are in Table 9.6.
B13 produced the highest frequency of terms and clauses associated with “spatial
language” (19.4), but this result may be traced to the frequent use of L-demonstrative
(demonstrative pronoun system). Other than this specific code, her spatial language
result is very limited. The two designers (B3 and B26), who produced the highest
numbers of activities in “design information”, also generated relatively high numbers
of terms and clauses in the spatial-language coding, 15.8 and 15.4, respectively. In
contrast, B20 who had the least frequency overall in “design information” coding
also had the least frequency in the use of “spatial language” (8.2).
Table 9.7 shows the coding results of Group C’s protocols for “spatial language”
based on the average frequencies for five-minute intervals. C8’s data contain the
highest frequency overall for the use of terms and clauses associated with spatial
language, although she develops only an average number of codes in the “design
information” results (Fig. 9.3). Her design style focussed on graphical-goal settings
at a micro-level for more than 17 min in the experiment. She also repeatedly reviewed
the design brief and site information. The combination of these factors contributes to
the overall result. C31, one of the most productive designers in the “design informa-
tion” coding, and C8 produced an equally high level of “spatial language” (18.5). In
contrast, C39, who was less productive in the “design information” coding, develops
the least activities in spatial-language coding (3.0).
An interesting observation arising from the data is that the designers in each
group responsible for producing the lowest frequency of “design information” (A15,
9.4 Results 227

B20, and C39) also develop the lowest frequency in each group of “spatial language”.
This implies that the production of “design information” is strongly related to the use
of “spatial language”. Conversely, the highest frequency use of terms and clauses
associated with “spatial language” tend to be associated with the frequent use of
L-demonstrative. Thus, higher frequencies of “spatial language” are often unre-
lated to higher frequency of “design information”. That is, the production of L-
demonstrative alone does not develop a meaningful design language. The next section
conducts a detailed investigation of the relationships between design cognition,
design information and spatial language.

9.4.2 Correlation

Correlation analyses (Table 9.8) using Pearson correlation coefficients (r) and signifi-
cance values (p) for the number of cases (n = 23) provide an indicator of the strengths
and directions of the relationships between variables (Brace et al. 2012). First, the
representation category in “design cognition” has an inverse relationship with the
functionality and directness categories (p < 0.05). This suggests that representation
activities typically associated with the production of functionality and directness, are
frequently negatively related and are moderately affected (r = −0.472 and −0.521,
respectively). The function category in “design cognition” has a significant, positive
relationship with all categories in “design information” and two categories (object
and directness) in “spatial language”. This suggests that designers who produce
more function-related activities tend to use more terms and clauses associated with
“design information” and “spatial language”. The evaluation category in “design
cognition” is also a predictor of the functionality and localisation categories in
“design information” and “spatial language” (0.419 and 0.637, respectively). That
is, the evaluation-based “design thinking” is closely related to both components of
“design language”.
Both the property and functionality categories in “design information” have signif-
icant, positive correlations with all categories in “spatial language”. The property
category has a particularly strong, positive relationship with the object category
(r = 0.859, p < 0.01) and the functionality category also has a strong, positive
relationship with the directness category (r = 0.778, p < 0.01). In summary, the
data suggest that cognitive activities are related to designers’ considerations of both
design information and spatial language. Specifically, the more function and evalu-
ation activities a designer employs, the higher the frequency of the use of terms and
clauses related to design information and spatial language. In a multi-cultural and
multi-linguistic context, this finding suggests that design thinking may be influenced
by the different usages of design language shaped by cultural and linguistic back-
grounds and differences (through their impact on “design information” and “spatial
language”).
 228

Table 9.8 Correlations between the coding results of design cognition, design information, and spatial language (n = 23)
Design cognition Design information Spatial language
Representation Perception Function Evaluation Goal setting Property Functionality Object Localisation
Perception −0.616**
Function −0.519* −0.208
Evaluation −0.297 −0.017 −0.077
Goal setting −0.067 −0.396 0.239 0.007
Property −0.308 −0.229 0.604** 0.131 0.231
Functionality −0.472* −0.182 0.577** 0.419* 0.379 0.671**
Object −0.185 −0.371 0.661** 0.003 0.251 0.859** 0.544**
Localisation −0.358 −0.180 0.326 0.637** 0.269 0.606** 0.632** 0.439*
Directness −0.521* 0.041 0.527** 0.211 0.248 0.664** 0.778** 0.401 0.576**
** p < 0.01, *p < 0.05
9 The Language of Design Thinking
9.4 Results 229

FuncƟonality LocalisaƟon Directness

Frequency

Frequency

Frequency
A B C A B C A B C

Fig. 9.5 Box and whisker plots of the coding results for functionality, localisation and directness
across Groups A, B and C

9.4.3 Comparison

The second important result from the experimental data highlights the differences
across the three cultural groups in terms of design cognition, design information and
spatial language. Significant differences are observed in the coding results of six out
of nine categories. Figure 9.4 shows the “box and whisker” plots of the coding results
of three “design cognition” categories (the percentage of the frequency weighted by
time duration) across the three cultural groups, while Fig. 9.5 illustrates the coding
results for functionality (from “design information”), localisation and directness
categories (from “spatial language”), describing average frequencies for five-minute
intervals. The box and whisker plots indicate that there are differences in the cognitive
allocation as well as in the production of “design information” and “spatial language”
in each cultural group.
Table 9.9 shows the results of one-way ANOVA with post hoc Scheffé test for
the eight selected subclasses where significant differences were identified among
groups. Group C’s designers, for example, more frequently attended to drawings
(P-drawing) than Group A’s designers, while Group A’s designers more frequently
evaluated existing drawings (‘E-drawing) than Group C’s designers. Consequently,
the group of Australian designers produced more terms relating to shape (e.g. place,
room, space, area and site) than the other two groups of Asian designers. Group
A’s designers also produced more descriptions of abstract features or concepts and
interactions than Group C’s designers. The Australian designers also considered
circulation more explicitly than the other two groups.
In terms of spatial language, three subclasses show significant differences across
the groups. The Australian designers describe spatial relations more frequently using
projective terms than the other two groups, which is supported by the largest F ratio
and the smallest P value (F = 11.632, P = 0.000). Group A data also shows a more
frequent production of absolute reference-reasoning systems and expresses more
direct relationships between objects than Group C. Thus, the results of ANOVA
for the eight selected subclasses confirm there are significant differences in “design
cognition” and “spatial language” between the three cultural groups.
230 9 The Language of Design Thinking

Table 9.9 The results of one-way ANOVA with Scheffé post hoc test for eight selected subclasses
Category Group Mean SD F Sig Scheffé a, b
P-drawing A 14.000 5.177 6.253** 0.008 A<C
B 16.963 3.090
C 22.943 6.262
E-drawing A 7.838 3.394 5.046* 0.017 A>C
B 4.313 2.419
C 3.443 2.700
P-shape A 5.550 3.673 7.393** 0.004 A > B, C
B 3.938 2.605
C 2.986 2.706
F-features A 3.613 1.596 6.484** 0.007 A>C
B 2.713 1.438
C 1.186 0.584
F-circulation A 1.400 0.693 6.506** 0.007 A > B, C
B 0.638 0.421
C 0.529 0.359
L-projective A 3.775 1.443 11.632** 0.000 A > B, C
B 1.850 0.789
C 1.400 0.600
L-absolute A 1.500 1.308 3.574* 0.047 A>C
B 0.738 0.670
C 0.300 0.294
D-direct A 3.575 2.251 6.832** 0.005 A>C
B 1.600 0.893
C 0.871 0.682
** p< 0.01, *p < 0.05
a Uses Harmonic Mean Sample Size = 7.636
b The group sizes are unequal and because of this the harmonic mean of the group sizes is used.

Type I error levels are not guaranteed

9.5 Discussion

Past research (Gerrig and Banaji 1994; Gleitman and Papafragou 2005) claims that
language, as a system, is both a reflection of the way we think and of our socio-
cultural differences and values. Importantly, linguistic systems may shape our thought
processes, and cognitive processes may be linguistically constrained. It could be
assumed then, that this is a bidirectional effect between design cognition and spatial
language (Lee et al. 2016; Lee et al. 2019). Thus, this chapter investigates the influ-
ence of “design language” on “design thinking” and vice versa. Our examination,
furthermore, explores cultural and linguistic differences in the design process.
9.5 Discussion 231

The results of correlations and ANOVA in this chapter confirm that there is a
clear relationship between design and language. There are also clear differences
between culturally and linguistically different groups in terms of design cognition,
design information and spatial language. However, Pinker (1994) provides empir-
ical evidence that there are thought processes at work long before people develop
any language skills. All individuals initially possess a “universal mentalese” (Pinker
1994). In parallel, Arnheim (1969) provides evidence of visual reasoning indepen-
dent from language. Visual perception could, therefore, be an intelligent act with
no relation to language. Thus, the arguments in this chapter may be open to criti-
cisms of linguistic determinism and bias. Nonetheless, this chapter finds that “spatial
language” is strongly correlated with “design information” in design language, oper-
ating complementarily in terms of cognitive representations. These also correlate with
representation, function and evaluation categories in design cognition, which are crit-
ical to design thinking. That is, both spatial representation (or linguistic abstraction)
and the use of design information relate to cognitive allocations. As such, surely
design language is closely linked to design thinking. This finding is reflected in the
results of previous linguistic studies (Munnich and Landau 2003; Munnich et al.
2001; Tenbrink and Ragni 2012).
This chapter not only highlights the relationship between design and language,
but it also addresses the impact of culture and language on design. There are,
however, several limitations in our experimental procedures. For example, although
there are some differences in the level of English proficiency across the groups of
designers, all participants were asked to verbalise their thoughts in English. This
was more difficult for Group C’s designers. Thus, the interpretation of the differ-
ences observed between the groups might have some limits. However, the results of
our research have parallels with past results that suggest speakers of different native
languages might approach spatial problem-solving quite differently (Munnich and
Landau 2003). For example, Group A’s and B’s designers often developed textual-
goals or graphical-goals at the beginning of their design sessions, while Group C’s
designers rarely conducted obvious goal-setting activities, solving isolated micro-
problems in a more mechanistic way. The different design thinking styles evident
across the three cultural groups may be related to the rational and the creative phases
of the design process (Bashier 2014) or “rational-intuitive” and “intuitive-rational”
design approaches (Donoso et al. 2018). In addition, Group A’s designers tended to
develop “macro” design solutions while most of Group C’s designers were limited
to producing “local” design configurations.
Statistically, there are significant differences in perception, evaluation and goal-
setting categories among the three cultural groups. Group C’s designers emphasise
perception activities in design cognition (also subclass P-drawing), while Group A’s
designers produce more evaluation activities (also subclass E-drawing) than Group
C’s designers. Group B’s designers produce more goal-setting activities than Group
C’s designers. This result implies that a group of designers of similar cultural back-
ground can have their own style of design thinking (cognitive process and preference),
232 9 The Language of Design Thinking

which may produce similar patterns of cognitive activities within the group when
solving a design problem. Moreover, these may well differ from the design thinking
strategies used by other cultural groups.

9.6 Conclusion

This chapter has developed and investigated a rich set of design protocols from three
different cultural and linguistic groups. Protocol analysis has been used to formally
capture the characteristics of both design thinking and design language using a new
triple-coding system. The results of correlations and ANOVA support our assump-
tions regarding the existence of certain relationships between design and language
as well as revealing some cultural and linguistic differences in design. We argue that
a bidirectional relationship between design thinking and design language is not just
likely, but is perhaps a key element that needs to be understood for supporting a
global creative economy. The different characteristics observed between Australian
and Asian designers suggest that, even if design is a “universal language”, cogni-
tive allocation and representation are influenced by individual designers’ cultural
experiences and preferences.
The cognitive (content) complexity (H C ) introduced in Chap. 4 of this book
has been used in this chapter to quantify individual design cognitions. It facilitates
capturing the different levels of complexity and productivity in a set of design proto-
cols. Thus, this chapter confirms that cognitive complexity is a promising measure
for analysis of this type. Finally, both chapters of Part III contribute to a deeper
understanding of the relationship between design thinking and design language as
well as design processes in the multi-cultural context. The findings are especially
relevant to the growing numbers of international studios in both architectural practice
and design education. Future research could also consider collaborative settings with
participants across different cultural groups to enrich and deepen the investigation.

References

Arnheim, R. 1969. Visual Thinking. Berkeley, Los Angeles, London: University of California Press.
Bashier, Fathi. 2014. Reflections on architectural design education: The return of rationalism in the
studio. Frontiers of Architectural Research 3 (4): 424–430. https://doi.org/10.1016/j.foar.2014.
08.004.
Boroditsky, Lera, Orly Fuhrman, and Kelly McCormick. 2011. Do English and Mandarin speakers
think about time differently? Cognition 118 (1): 123–129. https://doi.org/10.1016/j.cognition.
2010.09.010.
Brace, Nicola, Richard Kemp, and Rosemary Snelgar. 2012. SPSS for Psychologists. Basingstoke,
UK: Palgrave Macmillan.
Chai, Kah-Hin., and Xin Xiao. 2012. Understanding design research: A bibliometric analysis of
Design Studies (1996–2010). Design Studies 33 (1): 24–43.
References 233

Corballis, Michael C. 2018. Space, time, and language. Cognitive Processing 19 (1): 89–92. https://
doi.org/10.1007/s10339-018-0878-1.
Cross, Nigel, Henri Christiaans, and Kees Dorst. 1996. Analysing Design Activity. New York: Willy.
Donoso, Sergio, Pedro Mirauda, and Rubén Jacob. 2018. Some ideological considerations in the
Bauhaus for the development of didactic activities: The influence of the Montessori method, the
modernism and the gothic. Thinking Skills and Creativity 27: 167–176. https://doi.org/10.1016/
j.tsc.2018.02.007.
Gero, J.S., and Thomas Mc Neill. 1998. An approach to the analysis of design protocols. Design
Studies 19 (1): 21–61.
Gerrig, R. J., and M. R. Banaji. 1994. Language and thought. In Thinking and Problem Solving.
Handbook of Perception and Cognition, ed. R.J. Sternberg, 233–261. London: Academic Press.
Gleitman, Lila, and Anna Papafragou. 2005. Language and thought. In Cambridge Handbook of
Thinking and Reasoning, 633–661.
Herskovits, A. 1986. Language and Spatial Cognition: A Interdisciplinary Study of the Prepositions
in English. Cambridge, England: Cambridge University Press.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2015. Creativity and parametric design? Comparing
designer’s cognitive approaches with assessed levels of creativity. International Journal of Design
Creativity and Innovation 3 (2): 78–94. https://doi.org/10.1080/21650349.2014.931826.
Lee, Ju Hyun, Ning Gu, and Michael J. Ostwald. 2019. Cognitive and linguistic differences in
architectural design. Architectural Science Review 62 (3): 248–260. https://doi.org/10.1080/000
38628.2019.1606777.
Lee, Ju Hyun, Michael J. Ostwald, and Ning Gu. 2016. The language of design: Spatial cognition
and spatial language in parametric design. International Journal of Architectural Computing 14
(3): 277–288. https://doi.org/10.1177/1478077116663350.
Luna, D., and S. Forquer Gupta. 2001. An integrative framework for cross-cultural consumer
behavior. International Marketing Review 18 (1): 45–69.
Man, Jinfan. 2014. Design Teamwork in Distributed Intercultural Teams: Competition, Collabora-
tion, Cooperation. Eindhoven: Technische Universiteit Eindhoven.
Molina, A., J. Aca, and P. Wright. 2005. Global collaborative engineering environment for integrated
product development. International Journal of Computer Integrated Manufacturing 18 (8): 635–
651. https://doi.org/10.1080/09511920500324472.
Munnich, Edward, and Barbara Landau. 2003. The effects of spatial language on spatial represen-
tation: Setting some boundaries. In Language in Mind: Advances in the Study of Language and
Thought, ed. Dedre Getner, and Susan Goldin-Meadow, 113–155. Mit Press.
Munnich, Edward, Barbara Landau, and Barbara Anne Dosher. 2001. Spatial language and spatial
representation: A cross-linguistic comparison. Cognition 81 (3): 171–208.
Pinker, Steven. 1994. The Language Instinct: How the Mind Creates Language. New York: William
Morrow and Company.
Stevens, J.P. 1999. Intermediate Statistics: A Modern Approach, 2nd ed. Mahwah, NJ: Lawrence
Erlbaum Associates.
Suwa, Masaki, Terry Purcell, and John S. Gero. 1998. Macroscopic analysis of design processes
based on a scheme for coding designers’ cognitive actions. Design Studies 19 (4): 455–483.
Suwa, Masaki, and Barbara Tversky. 1997. What do architects and students perceive in their design
sketches? A protocol analysis. Design Studies 18 (4): 385–403. https://doi.org/10.1016/S0142-
694X(97)00008-2.
Tenbrink, Thora. 2011. Reference frames of space and time in language. Journal of Pragmatics 43
(3): 704–722. https://doi.org/10.1016/j.pragma.2010.06.020.
Tenbrink, Thora, and Marco Ragni. 2012. Linguistic principles for spatial relational reasoning. In
Cyrill Stachniss, Kerstin Schill, and David Uttal, ed. V.I.I.I. Spatial Cognition, 279–298. Lecture
Notes in Computer Science: Springer, Berlin Heidelberg.
van der Zee, E., and J.M. Slack. 2003. Representing Direction in Language and Space. Oxford
University Press.
 Part IV
Conclusion
Chapter 10
Conclusion: Three C’s of Design
Thinking

Abstract The final chapter in this book reflects on the findings, models and frame-
works presented previously. The first part, “creative design thinking” revisits the
results of two cognitive studies (Chap. 2) and the qualitative analysis and quantita-
tive measures of complexity developed for design thinking (Chaps. 3 and 4). The
second part, “collaborative design thinking”, considers the wider implications of
the Design Team Cognition (DTC) model and the Digital Design Thinking (DDT)
frameworks (Chaps. 5, 6 and 7). Finally, “cultural design thinking” returns to the
language of design and its consequences for spatial reasoning and communication
(Chaps. 8 and 9). This chapter summarises the book’s contribution to advances in
design thinking in terms of creativity, collaboration and culture.

10.1 Introduction

In the preface to this book, we employed a simple spatial analogy to introduce the
topic of design thinking. This classic descriptive trope—the problem of navigating
an unknown landscape—was used to develop an initial picture of the complexity of
design cognition. It is also an example of a Mental Model (MM), which was shared
with readers to foreshadow some key messages and themes in the book. For example,
in the MM we used spatio-relational reasoning to hint at the importance of both
problem spaces (“off-ramps”, “cul-de-sacs” and “dead-ends”) and solution spaces
(“freeways”, “bridges” and “tunnels”) while travelling to a destination. The different
types of paths taken by expert and novice explorers were also characterised in terms
of way-finding strategies. Several of the design strategies discussed in this book—
“forwarding”, “searching” and “reflecting”—encapsulate these paths and the sense
of exploration they entail. We also warned the readers that experts were more likely
to be open to the possibility that, not only isn’t the planned destination accessible, it
may not even be desirable. Design is not just about finding a solution to a problem, it
may require questioning the problem itself. Similarly, design is not necessarily about
optimal functionality, a solution that balances function, inspiration and innovation is
likely to be more effective.

© Springer Nature Switzerland AG 2020 237

J. H. Lee et al., Design Thinking: Creativity, Collaboration and Culture,
https://doi.org/10.1007/978-3-030-56558-9_10
238 10 Conclusion: Three C’s of Design Thinking

Like any MM, the information contained in our preface was abstract and simplistic.
Readers of the complete book will be aware that investigating design thinking requires
special methods and techniques. Using a detailed review of past theories, coupled
with new frameworks, experimental results and mathematical analysis, this book
constructs a deep and critical review of design thinking for the digital era. This
context, the “digital era” (and its counterparts, the “digital ecosystem” and “dig-
ital Anthropocene”), is significant because design collaboration and production are
increasingly digital. In the design fields of architecture and industrial design, BIM,
CAD and parametric design are fast becoming the primary platforms. Furthermore,
there are multiple digital design platforms in use in business, marketing and the arts.
In the contemporary world, design thinking has become so intricately tied to digital
contexts that there is an urgent need to develop an understanding of the relation-
ship between the two. This is especially important as a growing body of research
is now indicating that designers think and act differently when working in diverse
environments.
Within this general scope, this book identifies three themes—creativity, collabo-
ration and culture—that are crucial for the future of design thinking. These “three
C’s” are not just connected. A reciprocal relationship exists between them (Fig. 1.1).
Creativity is one of the most desired by-products of design thinking and arguably
the entire reason the design process has been identified as a unique and valuable
resource. Designers, however, rarely work alone and creativity is often tied to a
team’s capacity to develop and communicate MMs. Furthermore, team members
from different cultural and linguistic backgrounds promote divergent thinking and
collective intelligence, which recursively foster creativity. Thus, the “three C’s”
cannot be isolated from each other and all are important in contemporary design.
This book is the first to take a consistent approach to examining creativity, collab-
oration and culture in the context of the design process and product of the digital era.
Its purpose, however, is not just to describe or propose methodological advances in
cognitive research. Instead, this book introduces and then uses quantitative measures
of cognitive complexity (H C and H S ), to study patterns in design thinking associated
with the “three C’s”. Furthermore, within each of these themes, new design thinking
models for the digital era are developed. The three main ones in this book are the
“creative micro-process” model (Fig. 3.5), the DTC model (Fig. 5.1) and the “design
thinking and language” model (Fig. 9.1). Furthermore, while many cognitive studies
are only concerned with individual strategies and behaviours, the DTC model in this
book deals with team cognition.
This concluding chapter revisits the content of the book, including the results
of empirical protocol studies and the application of advanced methods, and then
reiterates their contributions and limitations.
10.2 Creative Design Thinking 239

10.2 Creative Design Thinking

Design thinking is valued for its capacity to generate innovative ideas and solu-
tions that have the potential to transform society and industry (Brown 2009; Martin
2009; Neumeier 2009). While design thinking is sometimes conflated with creative
thinking in cognitive research, there is a common alternative position which argues
that design thinking pursues innovation rather than creativity. This position arises
from the idea that the processes and products of design are measurable and manage-
able in terms of the levels of innovation they develop. In contrast, creativity has
often been viewed as elusive and subjective, being a mysterious territory in the realm
of complex human operations. As such, some people argue that creativity is not a
product of design thinking, because strategic and deliberate thinking leads to innova-
tion. This disagreement partially explains why much past research on design thinking
has been limited to examining explicit cognitive processes and products. Despite this
position, studies about creativity have been conducted using rigorous and scientific
methods for many years (Sternberg and Lubart 1999). Multiple formal models of,
and theories about creativity have been developed throughout the twentieth century.
Some of the most important of these are Wallas’s (1926) four-stage model of the
creative process, Gordon’s (1961) synectics theory, Guilford’s (1967) Structure of
Intellect (SI) theory, Hayes’ (1989) cognitive processes in creativity and Gero’s
(2000) computational model of innovative and creative design processes. This past
research confirms that creativity in design thinking is not uncharted territory, it can be
formally explored, mapped and measured. In this context, Part I of this book not only
provides an in-depth investigation into creative design thinking but also introduces
new measures of design cognition. Specifically, the results of a series of protocol
studies are used to reveal implicit processes and hidden cognitive representations in
the design process. The creativity implicit in a product is also examined using expert
panel assessment and a set of evaluation criteria. Thus, this book contributes to the
exploration of creative design thinking in both breadth and depth, neither of which
are common in design thinking research.
Chapter 2 uses the results of two protocol studies to identify individual design
strategies that support creativity. The first of these studies explores sketch-based
design activities and the second investigates parametric design activities. The
individual strategies identified in these cognitive studies contribute to improving
designers’ creative practices and are especially valuable for novice designers. Impor-
tantly, this chapter confirms that design thinking in a digital environment does differ
from design thinking in a more traditional, pen-and-paper or sketching environment.
While acknowledging limitations associated with the small sample sizes, and the
difficulty of developing intercoder-reliability measures for cognitive studies, Chap. 2
develops three forwarding (sketch-based) strategies and two generative (parametric)
strategies that are useful for understanding creative design thinking. To balance the
problems of conducting design studies in controlled contexts, both studies adopt
post-experimental interviews as retrospective protocols. While interview and obser-
vational (Cross 2011) studies may be superior for examining design practice for
240 10 Conclusion: Three C’s of Design Thinking

diverse “real-world” briefs, controlled experimental settings provide the best basis
for comparison and generalisation of results. Nevertheless, the results developed in
the first two studies are most informative for examining creative micro-processes in
a short-term design task. Advantageously, these are similar to the short “bursts” of
design activity, which occur in the quotidian work of a professional designer.
Chapter 3 provides an in-depth investigation of parametric design protocols,
capturing important micro-processes that support creativity. As the first stage of this,
the combined method of protocol analysis and expert panel assessment identifies
four parametric design activities—(i) changing parameters, (ii) perceiving geome-
tries, (iii) introducing algorithmic ideas, and (iv) evaluating geometries—that support
creative problem-solving processes. A close review of encoded segments in the proto-
cols reveals iterative cognitive patterns related to these activities. From this data, the
chapter develops a creative micro-process model (Fig. 3.5) for parametric design. The
importance of micro design activities and patterns has been identified in a number
of past studies of design cognition, while the evidence-based selection of key design
activities (the mapping process) is valuable for a focussed study on cognitive activ-
ities. Furthermore, the new understanding of sequential and/or cyclic design activi-
ties (both algorithmic and geometric activities) in the creative micro-process model
contributes to developing generative capacity in parametric design, as well as the
design thinking in the digital ecosystem.
While Chaps. 2 and 3 address cognitive activities that support creativity in design,
Chap. 4 develops two measures to quantify design thinking in terms of cognitive
complexity. In the design domain, complexity has been used as a criterion to assess
both the design product and the design problem. There has, however, been no clear
way to quantify complexity in the design process or in design cognition. Thus, Chap. 4
proposes two measures of cognitive complexity (H C and H S ), providing a significant
methodological contribution to research in this field. Both can be used to index
individual cognitive differences and facilitate comparisons between design processes.
The content complexity (H C ) measure is particularly useful, as demonstrated in
Part III, as it provides an alternative index for design productivity (idea generation).
Measuring information entropy for H C is based on the micro-categorisation of design
activities (i.e. subclasses), which is then correlated to other cognitive and linguistic
indexes. However, the verification of variables and categorisations for H C requires
further research. Nevertheless, Chap. 8 later indicates that H C corresponds with the
link index of each protocol, while structural complexity (H S ) corresponds to the
result of Critical Moves (CMs), which is more closely related to creativity.
In summary, each chapter in Part I of this book contributes to advances in creative
design thinking. The combined method of protocol analysis and expert panel assess-
ment in Chap. 2 is also applied in multiple chapters in this book and can be further
used for the mapping process of design process and product (Chap. 3) and the inter-
pretation of complexity measures (Chap. 4). Importantly, the two generative strate-
gies (Chap. 2) and the creative micro-process model (Chap. 3) are beneficial for
developing creative design thinking in parametric design.
10.3 Collaborative Design Thinking 241

10.3 Collaborative Design Thinking

While the first part of this book examines individual design thinking in relation to
creativity, Part II deals with design cognition and communication in teams. Team-
work protocols have been extensively explored in past design research (Goldschmidt
1995; Stempfle and Badke-Schaub 2002; Valkenburg and Dorst 1998). Rather than
revisiting this work, this book highlights the way individual designer’s work, as a
precursor to understanding the “collective” and “interactive” cognitive processes
that underpin the DTC model. In Part II, collaborative design thinking in the digital
ecosystem is examined using a series of technology-enabled collaborative platforms
as cases (e.g. BIM in Chap. 6; mobile, situated and collective digital platforms in
Chap. 7). In contrast to the other Parts of this book, which use data derived from
protocol studies, Part II contains a series of detailed reviews of theories of “team
cognition”, “cognitive representation”, “emerging technologies”, “Digital Design
Thinking (DDT) processes” and “DDT platforms”. Through this process, it uncovers
new or emerging trends in digital, networked design collaboration.
Chapter 5 presents a review of past research on “team cognition” and “cognitive
representation” and then distills key lessons from this about collaborative design.
This review highlights the significance of Transactive Memories (TMs), shared and
distributed MMs (SMMs and DMMs, respectively) and visual and verbal representa-
tions. These elements provide the foundations for an integrated team cognition model
for design collaboration. This DTC model positions TM as the system supporting
emergence and sharedness and distributed knowledge. While the model can be under-
stood as merging aspects of SMMs and DMMs, its purpose is to accommodate the
performative and creative aspects of collaborative design. To examine the model,
two of the four creative components—“design spaces” and “design modes”—are
explored using the protocols developed in Chap. 2. Two performative components
(“design information” and “spatial language”) of designers’ image-based and verbal
representations in MMs are also investigated through cross-national design proto-
cols. Although the capacity to verify the DTC model using this data is limited, it
demonstrates the importance of individuals’ creative design activities (distributed
knowledge) and their design communication processes (emergence and sharedness).
Thus, the DTC model provides a baseline for conducting future research on collab-
orative design thinking, social cognition, coordination and design management. The
knowledge contained in the DTC model also contributes to future team cognition
models and formal research approaches in the design domain. The “design thinking
and language” model proposed in Part III of this book is also founded on the way
the DTC model deals with different modes of design activities as well as linguistic
expression and spatial relational reasoning in design teams.
Chapter 6 introduces an advanced BIM knowledge framework for design collab-
oration. BIM is one of the most important platforms supporting design visualisation,
communication and collaboration in the Architecture, Engineering and Construc-
tion (AEC) sector. Unlike the other theories and models discussed in this book, the
BIM framework is a domain-specific knowledge construct. However, since BIM
242 10 Conclusion: Three C’s of Design Thinking

enables both interaction and collectivity (Verstegen et al. 2019), it is one of the most
practical examples of Computer Supported Cooperative Work (CSCW) tools. The
BIM framework in Chap. 6 provides a better understanding of the BIM microcosm
of the digital ecosystem, in which products, processes and people co-evolve. This
chapter describes a domain-specific example of digital design collaboration, while
Chap. 7 uses DDT frameworks to understand the mobile, situated and collective
digital platforms that surround us every day.
Chapter 7 proposes two DDT frameworks that represent an important step towards
improving our understanding of the relationship between design thinking, design
creativity and collaboration. The first framework encapsulates processes (I-C-L,
digital collaboration, problem-solving and collective design) and the second, func-
tionality (user-based, content-based and location and time-based). Both frameworks
are founded in the results of past research into Collective Intelligence (CI), digital
collaboration and Collective Design (CD). The DDT processes and functionalities
frameworks are also closely related to the DTC model’s emergence and sharedness
(SMMs) and distributed knowledge (DMMs). In Chap. 7 the DDT frameworks are
used for analysing and measuring the information processing and functional proper-
ties of examples of mobile, situated and collective platforms. For each platform, the
functionalities are quantified and their capacities and characteristics compared. This
enables the evaluation of these DDT platforms in terms of CI, CD, DMM and SMM.
Importantly, the two DDT frameworks in Chap. 7 identify the growth or evolution
trajectories of interactive and collective capacity in digital platforms. For example,
DDT processes evolve from the I-C-L to collective design to accommodate more
crowd-based activities, while DDT functionalities evolve in three possible dimen-
sions: from extrinsic hierarchy to intrinsic crowd, from collaborative creation to
opportunistic decision and from synchronous local to asynchronous global. Because
of this capacity to model developmental trajectories, the DDT frameworks can be
used to examine or shape the creation of a range of digital platforms including design
thinking tools and collaboration software.
The major frameworks presented in Part II—the DTC model (Chap. 5), BIM
knowledge framework (Chap. 6) and two DDT frameworks (Chap. 7)—support
improved collaborative design thinking through engaging social creativity in the
digital ecology. While the exploratory research in each chapter is limited to providing
models or frameworks, the reviews or past findings contribute to building a compre-
hensive knowledge base in the field of design collaboration. The topics covered
in Part II are expanded in Part III to accommodate the cultural aspects of design
thinking, in particular, “design language”.

10.4 Cultural Design Thinking

The cultural aspects of design thinking have become increasingly important in

contemporary design practice, with significant growth in geographically distributed,
multi-national teams (Man 2014). Part III addresses the relationship between design
10.4 Cultural Design Thinking 243

and language, which is significant because language shapes, and is in turn shaped
by, thought (Boroditsky 2001; Gleitman and Papafragou 2005). Moreover, all of the
design thinking topics addressed in this book—from design strategies and creative
processes to TMs, SMMs and DMMs—must consider linguistic factors. Given the
importance of this topic, it is surprising how little research has been conducted.
For this reason, the “design thinking and language” model proposed on Chap. 9 is
one of the most valuable outcomes of this book. Acknowledging the criticisms of
linguistic determinism (Chomsky 1965; Pinker 1994) and the parallel importance of
visual imagery (Arnheim 1997; Paivio 1971; Ware 2008), language both shapes and
shares our ideas in design environments. The two chapters in Part III examine the
cultural aspects of design thinking, developing and applying formal methodological
frameworks and using empirical data to support the process.
Chapter 8 examines multiple aspects of cross-national design thinking using
cognitive and linguistic indexes: link index, percentage of CM, H C , H S and syntac-
tical complexity. Although there are methodological limitations associated with the
sample size and experimental setting, this chapter presents multiple new measures
and assesses them in a comprehensive way. The cognitive and linguistic indexes are
valuable because they can be used to identify diverse aspects of creative, collabo-
rative and cultural design thinking. Specifically, the chapter reveals the relationship
between cognitive and linguistic characteristics in the design process, including the
correlation of quantitative indicators. The dual-coding system (“design cognition”
and “spatial language”), which enables in-depth investigation of the language of
design, is further elaborated in the following chapter.
Chapter 9 presents a research model for “design thinking and language”, which
is used for investigating the relationship between design and language. It highlights
three components of cultural design thinking (“design cognition”, “design informa-
tion” and “spatial language”) which are closely related to the DTC model. That is,
the “design thinking and language” model is not only used for investigating cultural
design thinking, but also facilitates the understanding of creativity and collaboration
in design thinking. The protocol study in this chapter (23 participants) rigorously
assesses and confirms that “design thinking” (in terms of cognitive allocation) is
related to “design language” (in terms of the use of information categories and spatial
linguistic principles). It concludes that different cultural and linguistic groups use
different cognitive methods or strategies to develop and represent design.
In traditional design research, the concept of “culture” is used to describe a distinct
value, belief and action (Strickfaden and Heylighen 2010). In contrast, in this book
it is concerned with the linguistic dimension of design thinking. The linguistic and
cognitive results presented in Part III are, however, explored in an individual exper-
imental setting, and future research should consider different team settings (collab-
oration, cooperation and competition) as well as “co-creation” protocols in a multi-
cultural team. Thus, the joint application of the DTC model and the “design thinking
and language” model will not just be valuable for future research, but such a combi-
nation may well be inevitable as multi-cultural teams continue to play a major role
in design. The “creativity” aspects of design thinking also need to be revisited to
elaborate both models. That is, the triangular model of the “three C’s” presented in
244 10 Conclusion: Three C’s of Design Thinking

this book provides a significant step towards transforming the traditional concept of
“design thinking” into an integrated notion of multiple correlative themes.

10.5 Conclusion

This book conducts the first integrated study of the creative, collaborative and cultural
aspects of design thinking. These three themes facilitate and promote diverse design
thinking in the digital ecosystem. The models and frameworks presented in this book
combine multiple aspects of design process and product, developing new insights
while also asking new questions. They contribute to the structuring and content
of future studies on individual and team design thinking, corresponding to their
cognitive and communicative properties.
Although this book highlights emerging design thinking arising from digital tech-
nologies and new networked platforms, traditional design thinking and its conven-
tional approaches are also comprehensively discussed in the literature reviews, as the
hybrid nature of the co-existence of the old and new will continue. The human opera-
tions are unchanged, but are supported by, or evolve with, the digital ecosystem. Thus,
it is acknowledged that this book is really a starting point of a longer journey which
cannot be accommodated in a single volume. Nonetheless, this book provides an
important map or guide, for people navigating their way around design thinking and
considering its applications in design practice, critical thinking and design research.

References

Arnheim, R. 1997. Visual Thinking. Los Angeles: University of California Press.

Boroditsky, Lera. 2001. Does language shape thought?: Mandarin and English speakers’ conceptions
of time. Cognitive Psychology 43 (1): 1–22. https://doi.org/10.1006/cogp.2001.0748.
Brown, Tim. 2009. Change by Design, How Design Thinking Transforms Organizations and Inspires
Innovation. New York: Harper Collins Publishers.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Cross, Nigel. 2011. Design Thinking: Understanding How Designers Think and Work. Oxford, New
York: Bloomsbury.
Gero, John S. 2000. Computational models of innovative and creative design processes. Technolog-
ical Forecasting and Social Change 64 (2): 183–196. https://doi.org/10.1016/S0040-1625(99)001
05-5.
Gleitman, Lila, and Anna Papafragou. 2005. Language and thought. In Cambridge Handbook of
Thinking and Reasoning, ed. Keith J. Holyoak and Robert G. Morrison, 633–661. New York:
Cambridge University Press.
Goldschmidt, Gabriela. 1995. The designer as a team of one. Design Studies 16 (2): 189–209.
https://doi.org/10.1016/0142-694X(94)00009-3.
Gordon, William J.J. 1961. Synectics: The Development of Creative Capacity. New York: Harper
and Row.
Guilford, Joy P. 1967. The Nature of Human Intelligence. New York: McGraw-Hill.
References 245

Hayes, John R. 1989. Cognitive processes in Creativity. In Handbook of creativity. Perspectives on

individual differences, ed. John A. Glover, Royce R. Ronning, and Cecil R. Reynolds, 135–145.
New York: Plenum Press.
Man, J. 2014. Design Teamwork in Distributed Intercultural Teams: Competition, Collaboration,
Cooperation. Eindhoven: Technische Universiteit Eindhoven.
Martin, Roger. 2009. The Design of Business: Why Design Thinking is the Next Competitive
Advantage. Boston, MA: Harvard Business Press.
Neumeier, Marty. 2009. The Designful Company: How to build a culture of nonstop innovation.
Berkeley, CA: New Riders.
Paivio, Allan. 1971. Imagery and Verbal Processes. Oxford: Holt, Rinehart & Winston.
Pinker, S. 1994. The Language Instinct: The New Science of Language and Mind. New York:
William Morrow and Company.
Stempfle, Joachim, and Petra Badke-Schaub. 2002. Thinking in design teams—An analysis of team
communication. Design Studies 23 (5): 473–496. https://doi.org/10.1016/S0142-694X(02)000
04-2.
Sternberg, Robert J., and Todd I. Lubart. 1999. The Concept of Creativity: Prospects and Paradigms.
In Handbook of Creativity, ed. Robert J. Sternberg, 3–15. Cambridge University Press.
Strickfaden, Megan, and Ann Heylighen. 2010. Cultural Capital: A Thesaurus for Teaching Design.
International Journal of Art & Design Education 29 (2): 121–133. https://doi.org/10.1111/j.1476-
8070.2010.01653.x.
Valkenburg, Rianne, and Kees Dorst. 1998. The reflective practice of design teams. Design Studies
19 (3): 249–271. https://doi.org/10.1016/S0142-694X(98)00011-8.
Verstegen, Luuk, Wybo Houkes, and Isabelle Reymen. 2019. Configuring collective digital-
technology usage in dynamic and complex design practices. Research Policy 48 (8): 103696.
https://doi.org/10.1016/j.respol.2018.10.020.
Wallas, Graham. 1926. The Art of Thought. New York: Harcourt Brace.
Ware, C. 2008. Visual Thinking for Design. Burlington, MA: Morgan Kaufmann.