Journal of Experimental Psychology: © 2016 American Psychological Association

Human Perception and Performance 0096-1523/16/$12.00 http://dx.doi.org/10.1037/xhp0000272

2016, Vol. 42, No. 12, 2084 –2102

Different (Key)Strokes for Different Folks: How Standard and Nonstandard

Typists Balance Fitts’ Law and Hick’s Law
Gordon D. Logan, Jana E. Ulrich, and Dakota R. B. Lindsey
Vanderbilt University

Fine motor skills like typing involve a mapping problem that trades Fitts’ law against Hick’s law. Eight
fingers have to be mapped onto 26 keys. Movement time increases with distance, so Fitts’ law is
optimized by recruiting more fingers. Choice difficulty increases with the number of alternatives, so
Hick’s law is optimized by recruiting fewer fingers. The effect of the number of alternatives decreases
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

with consistent practice, so skilled typists achieve a balance between Fitts’ law and Hick’s law through
This document is copyrighted by the American Psychological Association or one of its allied publishers.

learning. We tested this hypothesis by comparing standard typists who use the standard QWERTY
mapping consistently with nonstandard typists who use fewer fingers less consistently. Typing speed and
accuracy were lower for nonstandard typists, especially when visual guidance was reduced by removing
the letters from the keys or covering the keyboard. Regression analyses showed that accommodation to
Fitts’ law (number of fingers) and Hick’s law (consistency) predicted typing speed and accuracy. We
measured the automaticity of typing in both groups, testing for hierarchical control in 3 tasks: word
priming, which measures parallel activation of keystrokes, keyboard recall, which measures explicit
knowledge of letter locations, and hand cuing, which measures explicit knowledge of which hand types
which letter. Standard and nonstandard typists showed similar degrees of hierarchical control in all 3
tasks, suggesting that nonstandard typists type as automatically as standard typists, but their suboptimal
balance between Fitts’ law and Hick’s law limits their ability to type quickly and accurately.

Keywords: skill, automaticity, hierarchical control

We watch with amazement as the guitarist shreds the fretboard, amount of practice can change the distance between the keys.
the pianist tickles the ivories, and our fingers dance over the Hick’s law bends with practice, reducing the effect of number of
computer keyboard. The dazzling speed and effortless grace dis- choices (Logan, 1979). As it bends, more fingers can be recruited
guises the difficulty of the underlying choice process, which has to with less cost in RT. As more fingers are recruited, movement
map one of 10 fingers onto each of 120 –124 positions on the guitar distance decreases, which decreases the cost in movement time.
neck, 88 keys on the piano, and 60 –100 keys on the computer Performance becomes fast and fluent. Skill acquisition is further
keyboard. We suggest the choice is constrained by a tradeoff constrained by the principle of consistent mapping (Shiffrin &
between two fundamental laws of psychology: One is Fitts’ law, Schneider, 1977): Hick’s law only bends if the mapping between
which says that movement time increases linearly with the loga- fingers and targets is consistent throughout practice; departures
rithm of distance (for targets of equal size; Fitts, 1954). Fitts’ law from consistency impair the acquisition and expression of skill
may be optimized by recruiting all 10 fingers, which minimizes the (Logan, 1979).
distance between the nearest finger and the target. The other is This perspective leads us to predict that the expression of fine
Hick’s law, which says that response time (RT) increases linearly motor skills will depend on how effectively skilled practitioners
with the logarithm of the number of choices (for equiprobable balance the tradeoff between Fitts’ law and Hick’s law. Practitio-
choices; Hick, 1952; Hyman, 1953). Hick’s law may be optimized ners who use more fingers and use them more consistently should
by reducing the number of fingers to choose from to one. reach higher levels of performance than practitioners who use
To achieve the high levels of skill seen in guitarists, pianists, fewer fingers less consistently. The purpose of this article is to test
and typists, something has to give. Fitts’ law is immutable: No this prediction in skilled typewriting, comparing standard typists,
who use the standard QWERTY mapping consistently, to nonstan-
dard typists who depart from it by using fewer fingers or using
fingers inconsistently. Standard typists should type faster and more
This article was published Online First October 17, 2016. accurately than nonstandard typists. We also asked whether stan-
Gordon D. Logan, Jana E. Ulrich, and Dakota R. B. Lindsey, Depart- dard typists rely less on visual guidance (Snyder, Logan, & Yama-
ment of Psychology, Vanderbilt University. guchi, 2015) and type more automatically and hierarchically
This research was supported by Grant BCS 1257272. We are grateful to
(Crump & Logan, 2010b; Logan & Crump, 2009) than nonstan-
Kristy Snyder for help with programming and Yimin Qiu for help with the
dard typists.
video analysis. Programs, data, and analyses can be obtained at http://www
.psy.vanderbilt.edu/faculty/logan/#data. We focused on typewriting because it has become a nearly
Correspondence concerning this article should be addressed to Gordon universal skill since the dawn of the information age (Logan &
D. Logan, Department of Psychology, Vanderbilt University, Nashville, Crump, 2011). In a 2014 census, the United Nations found that
TN 37240. E-mail: gordon.logan@vanderbilt.edu 96% of people worldwide had cell phones and 43% of households
2084
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2085

worldwide had computers (79% in developed countries; http:// box was placed over typists’ hands and the keyboard, so neither
www.itu.int/en/ITU-D/Statistics/Pages/stat/default.aspx). Teach- hands nor letters were visible (Snyder et al., 2015; Tapp & Logan,
ers and students invest substantial amounts of time learning to 2011). We expect standard typists to be unaffected by this manip-
type, and our research has implications for how the time should be ulation. Nonstandard typists should get progressively worse across
invested. While formal training is common in American middle conditions, as more visual information is withheld.
schools, many typists acquire some skill earlier, hunting and
pecking on the keyboard of their family computer. Self-taught Automaticity and Hierarchical Control
skills may depart significantly from the standard mapping and may
compete with or dominate the standard mapping. Our research We asked whether nonstandard typing was less automatic than
measures the costs and benefits of nonstandard mapping, which standard typing. The balance between Fitts’ law and Hick’s law
should inform policy decisions about investments in formal train- limits the quality of performance, and that may limit the degree of
ing and remediation. automaticity typists may attain. Nonstandard typists with subopti-
mal balances may type less automatically than standard typists
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

with optimal balances. However, optimal performance is only one

Visual Guidance
This document is copyrighted by the American Psychological Association or one of its allied publishers.

criterion for automaticity (Logan, 1988; Moors & de Houwer,

We suggest that nonstandard typists rely more on visual guid- 2006). Nonstandard typists may type as effortlessly and autono-
ance than standard typists because of the nature of their skill and mously as standard typists even if their performance is suboptimal.
the way they acquire it. Everyone is a nonstandard typist to begin In real-world typing, autonomy and effort may be more important
with: Everyone starts with a hunt and peck strategy, searching the than optimal speed. Typists want to type without thinking about it
keyboard for each key and choosing a finger to strike it, possibly so they can focus on writing.
choosing the nearest to optimize Fitts’ law. At first, search and In typing, automaticity is expressed as hierarchical control.
finger choice are based on visual information gained from looking Unlike many of the cognitive tasks studied in the automaticity
at the keyboard. With practice, typists develop spatial representa- literature (Logan, 1988; Palmeri, 1997; Rickard, 1997; Shiffrin &
tions of the keyboard and their hands that they can access through Schneider, 1977; Siegler, 1987), the multiple steps in novice
kinesthesis and proprioception (Klatzky & Lederman, 2003; typing cannot be reduced to a single step by practice. Skilled
Rosenbaum, Meulenbroek, Vaughan, & Jansen, 2001) and choose typists still have to serially order the letters in the word, find the
the nearest finger without vision (Crump & Logan, 2010c; Snyder keys to strike, and choose the fingers to strike them with, just like
et al., 2015). We suggest that nonstandard typists never abandon novices. The difference is that novices execute these steps in
visual information entirely. It is compatible with kinesthesis and working memory under top– down control while experts execute
proprioception and provides converging evidence about key and them in the motor system under hierarchical control (see Figure 1;
finger locations. Looking at the keyboard is compatible with their
typing habits: Modern typists spend much more time composing
text than copying it (Logan & Crump, 2011), so there is nothing to
draw their eyes away from the keyboard. The downside is that the
closest finger is not always the one with the standard mapping.
Practice with visual guidance may lead to suboptimal habits.
By contrast, training with standard mapping is intended to teach
people to “touch type” without looking at the keyboard, so stan-
dard typists should not rely on visual information. We assume that
standard typists remember the locations of the keys and the fingers
used to strike them, retrieving keystrokes (Rosenbaum et al., 2001)
without comparing locations of fingers and keys and choosing the
nearest. The memory representations are implicit and accessible
primarily through kinesthesis and proprioception (Crump & Lo-
gan, 2010c).
Standard mapping optimizes Fitts’ law, recruiting all eight fin-
gers, and Hick’s law, mapping fingers to keys consistently. Each
key is associated with just one finger. This consistency reduces
interference from competing fingers, which speeds retrieval and
increases accuracy (Logan, 1988; Shiffrin & Schneider, 1977). Figure 1. Computational analysis of typing under top– down and hierar-
However, the retrieved finger may not always be the nearest, so chical control, including the representations held in working memory and
looking at the keyboard may create competition and interference. the motor system and the information flow between the processes that
operate on them. A word activates its constituent letters, a serial order
Standard typists may be better off not looking at the keyboard.
process selects a letter to be typed, selects the key associated with the letter,
We tested reliance on vision by having typists type paragraphs
selects a finger to strike it with, and executes a movement. Under top–
while we manipulated visibility in three conditions: In the visible down control, all of these processes except the movement are done in
condition, typists typed on a normal keyboard with the letters and working memory. Under hierarchical control, all of these processes are
their hands visible. In the blank condition, typists typed on a done in the motor system. Under top– down control, working memory
keyboard with blank stickers attached to the keys, so their hands represents words, letters, keys, and fingers. Under hierarchical control,
were visible but the letters were not. In the covered condition, a working memory only represents words.
 2086 LOGAN, ULRICH, AND LINDSEY

also see Logan & Crump, 2011; Shaffer, 1976; Sternberg, Knoll, & able to type quickly and accurately, suggesting that their knowl-
Turock, 1990). edge of letter locations is encapsulated in the motor system (Sny-
We assume hierarchical control emerges with practice. Hunt- der et al., 2014). We ask whether standard and nonstandard typists
and-peck typing under top– down control creates associations be- show the same poor explicit knowledge of the keyboard.
tween words, letters, keys, and movements in the motor system, The hand-cuing task tests the encapsulation of knowledge about
which strengthen with practice (Rosenbaum et al., 2001). Eventu- the mapping of letters to fingers in the motor system, asking
ally, the associations become strong enough that the motor system whether typists have explicit knowledge of which hand types
can retrieve the sequence of keystrokes on its own, given only the which letter (Logan & Crump, 2009; Snyder & Logan, 2013; Tapp
word to be typed. At some point, typists trust motor memory & Logan, 2011). Typists type single words preceded by a whole
enough to abandon top– down control and let the motor system cue, which tells them to type the whole word, or a hand cue, which
control typing. At this point, typing is hierarchical. Typists do not tells them to type only the letters assigned to one hand (e.g., if the
have to think about their typing precisely because top– down cue is Left and the word is dock, type d and c). Skilled typists find
processes do not have to select and sequence keys, fingers, and typing with hand cues very difficult. It slows their RT and typing
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

movements. This frees up working memory to think other thoughts rate dramatically, suggesting that their higher-level processes do
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(see Figure 1). not know which hand types which letter. We ask whether standard
We assume that hierarchical control emerges with practice in and nonstandard typists show the same disruption from hand cues.
much the same way other skills and habits emerge with practice:
There is a period in which control is exclusively top down, Method
followed by a period in which control is a mixture of top– down
and hierarchical, depending on relative strengths and strategic
choices, and ending with a period in which control is primarily
Subjects
hierarchical (Anderson, 1987; William & Harter, 1899; Fitts & We chose a sample size of 48 typists (24 standard, 24 nonstan-
Posner, 1967; Keller, 1958; Logan, 1988; Rickard, 1997). The dard) based on power estimates: A previous sample of 800 typists
details of the transitions are important topics for future research. typing similar paragraphs yielded a mean of 68 and a SD of 18
We also assume that the choice between top– down and hierarchi- words per minute (WPM; Crump & Logan, 2013). A sample size
cal control is voluntary, so skilled typists may switch between of 48 gives a power of 0.47 to detect a 5 WPM difference and a
them at will. They can bring top– down processes online to stop power of 0.96 to detect a 10 WPM difference. We recruited typists
(Logan, 1982), interrupt (Yamaguchi, Logan, & Li, 2013), or from the general university population through a computerized
correct typing (Crump & Logan, 2013), to deal with unfamiliar subject pool. The recruitment advertisements described the stan-
materials (Yamaguchi & Logan, 2014b), changes in key positions dard mapping. One asked for typists who used the standard map-
(Yamaguchi & Logan, 2014a) or unfamiliar keyboards (Crump & ping. The other asked for typists who used nonconventional map-
Logan, 2010c; Yamaguchi & Logan, 2014b). pings. The advertisements appear in the Appendix. We classified
Abundant evidence suggests that skilled typists achieve hierar- typists as standard or nonstandard when they arrived at the labo-
chical control (Logan & Crump, 2011; Shaffer, 1976; Sternberg et ratory by showing them the standard mapping in a picture of a
al., 1990). We asked whether standard and nonstandard typists keyboard with the keys assigned to each finger in a different color
achieve the same degree of hierarchical control by comparing them (see Figure A1), and asking them whether or not they used those
on three tasks that test critical properties of hierarchical control: fingers for those keys. They were to report “standard” if they typed
word priming, keyboard recall, and hand cuing. The word-priming every letter with standard mapping and “nonstandard” if they
task tests a hallmark feature of hierarchical control, which is ability typed one or more letters with a different mapping. Based on this
of one higher-level unit in working memory to activate several report, we assigned 24 typists to the standard group and 24 to the
lower-level units in the motor system (Crump & Logan, 2010b; nonstandard group. Typists received course credit or pay for par-
also see Logan, 2003; Logan, Miller, & Strayer, 2011). Word ticipating in one 90 min session (standard: 12 credit, 12 paid;
priming measures whether words activate their constituent letters nonstandard: 13 credit, 11 paid). Their mean age, sex, and typing
in parallel. Typists are given a word to type and are probed with a experience are presented in Table 1.
single letter from the first, middle, or last position in the word or
from another word. Skilled typists respond faster to probes from
Apparatus and Stimuli
the primed word than probes from another word, indicating that the
prime activates its constituent letters in parallel. The priming effect The words and letters to be typed and echoes of typists’ re-
decreases across position in the word, reflecting a gradient of sponses were displayed on a flat screen computer monitor (BenQ
activation. We ask whether standard and nonstandard typists show XL2411Z) controlled by a personal computer (ASUS M32BF).
the same within-word priming effect and the same gradient of Responses were collected on two standard computer keyboards
activation. (ASUS model KB73211). The keyboards and keys were black. We
The keyboard recall task tests the encapsulation of knowledge covered the keys on one keyboard with blank stickers (all black).
about the locations of letters on the keyboard in lower-level motor We covered the keys on the other keyboard with stickers depicting
processes, which is another hallmark of hierarchical control (Liu, the letters (white letters on a black background) so the blank,
Crump, & Logan, 2010; Snyder, Ashitaka, Shimada, Ulrich, & visible, and covered keyboards would feel the same. We covered
Logan, 2014). Typists are given a blank keyboard and are asked to the hands and keyboard with the top of a box of printer paper with
fill in the letters. Their accuracy reflects their explicit knowledge one side cut out, measuring 10.8 ⫻ 27.9 ⫻ 44.5 cm. All procedures
of the keyboard. Skilled typists do poorly on this task despite being were programmed in LIVECODE (http://livecode.com). The ma-
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2087

terial to be typed and the response echoes appeared as black Full-alphabet sentence. The 10 sentences were presented one
characters in a 24.1 ⫻ 19.7 cm light gray window centered on a at a time in random order, centered in the top half of the viewing
black screen. Instructions for all computerized tasks were pre- window. Typists’ responses were echoed below. On each trial,
sented in Helvetica font. Instructions were in 18 point font for the typists clicked on a “Begin Typing” box with a mouse, typed the
sentence and paragraph tasks, 20 point font for the priming task, sentence, and then clicked on an “End Typing” box. Video was
and 22 point font for the hand cuing task. The stimuli in the recorded continuously throughout the 10 sentences. The mouse
full-alphabet sentence task and the paragraphs in the speed tests movements provided visual markers of the beginning and end of
were presented in 18 point font. The stimuli in the word priming each sentence.
and hand cuing tasks were presented in 40 point font. Keyboard recall. In the keyboard recall task, typists were
The full-alphabet sentence task used 10 sentences, which appear presented with a blank keyboard on a 21.6 ⫻ 14 cm piece of paper
in the Appendix. Each sentence contained each letter of the alpha- (see Figure A2) and were asked to fill in the keys with the
bet at least once. The average length was 12.50 (SD ⫽ 2.46) words appropriate letters in 80 s. They changed pens every 20 s to
and 67.70 (SD ⫽ 9.55) characters, including spaces and punctua- provide coarse timing data, beginning with an orange pen, then
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tion. Many letters appeared more than once. Hand and finger changed to pink, green, and red pens. We counted the number of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

movements were recorded at 60 frames/sec on a video camera letters recalled in correct positions and the number of letters
(Canon VIXIA HF R500), which was mounted on a tripod (Ravelli recalled in incorrect positions in each 20 s epoch.
APLT2) behind the flat screen display aimed at the keyboard and Paragraph typing. Typists typed three paragraphs under
fingers. The keyboard and fingers filled 50.4% of the image. We three visual conditions (visible, blank, and covered). The order of
recorded performance on all tasks but only analyzed video for the visual conditions and paragraphs were counterbalanced separately
full-alphabet sentence task. across the 48 subjects, with eight subjects receiving each of the six
The paragraph typing task used three paragraphs expressing the possible orders of conditions or paragraphs. We planned to coun-
many merits of border collies (see Appendix) and are similar to terbalance standard and nonstandard groups, but after realizing that
other paragraphs we have used in our research expressing other many typists who reported using standard mapping actually used
merits of border collies (see Crump & Logan, 2013). Paragraphs nonstandard mappings, we focused on counterbalancing over the
1–3 were 106, 113, and 107 words in length (471, 517, and 490 whole group. Counterbalancing was not perfect within standard
characters), and included 5, 6, and 5 sentences. and nonstandard groups, but departures from perfect counterbal-
The word-priming task used 280 five-letter words, presented in ancing were not large.
the Appendix. All of the letters within each word were unique. The paragraphs to be typed were presented one at a time in the
Mean Kuèera–Francis word frequency was 63.43 (SD ⫽ 241.35) top half of the viewing window. Responses were echoed below.
per million (MRC Psycholinguistic Database; Coltheart, 1981; The experimenter set up the visibility condition, changing key-
http://www.psych.rl.ac.uk/). boards and placing the box to cover the keyboard and hands as
The hand-cuing task used three prime words (Left, Right, and necessary, and then left the typist to complete the paragraph. When
Both) and 48 four-letter target words, which are presented in the typists finished, they clicked “Next” with a mouse and opened the
Appendix. For the cues, the first letter was uppercase and the rest door to let the experimenter in to set up the next condition.
were lowercase. For the target words, all letters were uppercase. Word priming. The word priming, hand cuing, and word/
There were six examples of eight kinds of target words: two kinds nonword tasks were run in counterbalanced order. Over the 48
of unimanual words, in which all the letters were typed in one hand typists, eight had each of the six possible orders of conditions.
or the other (RRRR, LLLL) and six kinds of bimanual words, in Within standard and nonstandard groups, counterbalancing was
which two letters were typed with one hand and two letters were not perfect but it was close.
typed with the other (LLRR, RRLL, LRLR, RLRL, LRRL, The word-priming task involved 280 trials in which each word
RLLR). The mean Kuèera–Francis word frequency was 90.35 was presented once on a go trial or a probe trial. Each trial began
(SD ⫽ 172.35). with a 500 ms fixation cross in the center of the viewing window.
It was replaced by the prime word, which was displayed for 250
ms and then extinguished. On go trials (140 total), the screen
Procedure
remained blank for 1,000 ms, whereupon a go signal appeared (a
The 90 min experimental session included the following events. row of asterisks) instructing the typist to type the prime word as
First, typists gave informed consent in writing and filled out a quickly and accurately as possible and type the space bar to end the
typing survey asking them about their age, sex, and experience (10 trial. On probe trials (140 of total), the screen remained blank for
min). Then they performed the keyboard recall task (2 min). After 500 ms, whereupon a single capital letter appeared. Typists were
keyboard recall, we verbally confirmed their typing style, and took instructed to type the single letter as quickly and accurately as
them to a smaller room for computer testing. There, they did the possible and type the space bar to end the trial. The words or letters
full-alphabet sentence task (5 min), typed paragraphs with the they typed were echoed on the screen below the probe. There were
keyboard visible, blank, and covered (10 min), and then performed four probe types, three probing the first, middle, and last letter of
word priming, hand cuing, and word/nonword tasks (10 –15 min the prime word and one probing a letter from another word. Each
for each task). Finally, they returned to the reception room, had probe type occurred equally often. Words were assigned to go and
their hands traced for size measurements, were debriefed, com- probe trials randomly, with the same random assignment for all
pensated, and thanked for their participation (5 min). Throughout typists. The order of go and probe trials was randomized separately
the session, instructions and breaks took 10 –15 min. for each typist.
 2088 LOGAN, ULRICH, AND LINDSEY

Hand cuing. The hand-cuing task involved 144 trials, in nonstandard typists gave very similar estimates of their typing
which 48 unique words were presented three times, once with the speed and accuracy, suggesting that nonstandard typists are not
“whole” cue (Both) and once with each of the “hand” cues (Left aware that their style of typing may limit their performance.
and Right). The order was randomized for each typist. Each trial We assessed the number of fingers typists used and the consis-
began with a cue presented above the center of the viewing tency with which they assigned fingers to keys by analyzing videos
window, which remained on the screen throughout the trial. One of the typists typing the 10 full-alphabet sentences. Forty-four
second after cue onset, a fixation cross appeared below the cue for typists completed all 10 sentences. Four completed only nine
500 ms, which was replaced by the word to be typed. Typists were because of technical problems. We played the videos in slow
told to type the letters of the word appropriate to the cue and end motion or frame by frame to determine which finger struck which
their response by typing the space bar. The letters they typed were key for each of the (551 letter keystrokes ⫻ 48 typists ⬇) 26,448
echoed on the screen below the word to be typed. For unilateral keystrokes. The second author scored all of the keystrokes and her
words typed all in the forbidden hand, typists simply typed the results are reported in this article. We had a second person score
space bar. 5,432 keystrokes from five standard and five nonstandard typ-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Word/nonword. We included a 300-trial word/nonword task ists. She agreed with the second author on 5,403 keystrokes
This document is copyrighted by the American Psychological Association or one of its allied publishers.

in which typists typed equal numbers of four- and five-letter (99.5%).

words, pronounceable nonwords, and unpronounceable nonwords. The videos showed that standard and nonstandard typists bal-
We forgot to include RT measurement in the program, and accu- anced the tradeoff between Fitts’ law and Hick’s law differently.
racy was near ceiling, so we lost the most informative data and will We assessed typists’ adaptation to Fitts’ law by counting the
not discuss the task further. number of fingers each typist used across the sentences, including
a finger in the count if it was used at least once. The frequency
Results distribution for numbers of fingers used across all 48 typists is
presented in Figure 2 (top panel). As expected, standard typists
Standard Versus Nonstandard Typists accommodated better, typing with significantly more fingers than
nonstandard typists (7.67 vs. 5.92; see Table 1). Figure 3 plots the
The mean responses to the typing questionnaire for each group number of standard and nonstandard typists using each of the eight
are presented in Table 1. The groups were equivalent in age and fingers. Nonstandard typists focus on the index and middle fingers
hours spent typing each day. All typists grew up with access to near the center of the keyboard. Standard typists also recruit ring
computers in their homes, all had formal training in typing, and all and little fingers.
but one (a standard typist) currently owned a computer. We mea- We assessed adaptation to Hick’s law by calculating the con-
sured hand size in 16 standard and 19 nonstandard typists, and sistency with which typists mapped fingers onto keys. We mea-
found no difference (though the 16 men had larger hands than the sured consistency by counting the number of fingers each typist
19 women; 8.7 vs. 7.8 cm, t(33) ⫽ 4.89, p ⫽ .00003, MSE ⫽ used to type each letter across repetitions and sentences. The
0.176, JZS Bayes factor ⫽ 687.196(A)). The main difference number of repetitions of letters across sentences ranged from 10 (q,
between groups was that nonstandard typists began typing a year x) to 70 (e) and averaged 21. We defined consistent typing as using
and a half earlier and consequently had typed for more years. Thus, the same single finger to type a letter on each repetition. Using
nonstandard typists may be more likely to have ingrained self- another finger even once counted as inconsistent. The frequency
taught habits before formal instruction. Interestingly, standard and distribution for number of keys struck consistently across all 48

Table 1
Mean (SD in Brackets) Measures of Experience, Self Estimates of Typing Speed and Accuracy,
Middle Finger Length, Number of Fingers Used, and Number of Keys Struck Consistently for
Reported Standard and Reported Nonstandard Typists

Reported Reported JZS Bayes

Measure standard nonstandard t(46) factor

Years of age 20.08 (2.89) 20.92 (3.65) .878 2.542 (N)

Females 16/24 12/24a
ⴱ
Years spent typing 10.48 (3.09) 12.42 (3.23) 2.125 1.736 (A)
Age learned 9.67 (2.47) 8.15 (2.31) 2.207ⴱ 1.993 (A)
Weeks of formal training 27.46 (39.32) 22.58 (21.21) .535 3.096 (N)
Hours of typing per day 3.98 (1.61) 5.25 (3.35) 1.677 1.121 (N)
Self estimated words per minute 68.60 (15.56) 70.23 (18.95) .325 3.333 (N)
Self estimated accuracy 88.13 (7.08) 85.82 (9.11) .982 2.350 (N)
Middle finger length in cmb 8.19 (.55) 8.28 (.79) .376c 2.903 (N)
Number of fingers used 7.67 (.56) 5.92 (1.64) 4.944ⴱ 1562.319 (A)
Number of keys struck consistently 22.71 (5.08) 13.13 (4.84) 6.693ⴱ 358898.9 (A)
Note. For JZS Bayes factors, N ⫽ favors null hypothesis; A ⫽ favors alternative hypothesis.
a
One typist identified as “nonbinary.” b Length measures were obtained for 16 standard typists and 19
nonstandard typists. c Degrees of freedom ⫽ 33.
ⴱ
p ⬍ .05.
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2089

Standard Nonstandard

20

Number of Typists 15

10
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

5
This document is copyrighted by the American Psychological Association or one of its allied publishers.

0
1 2 3 4 5 6 7 8
Number of Fingers Used to Type

Standard Nonstandard
9

7
Number of Typists

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
Number of Keys Struck Consistently
Figure 2. Full-alphabet sentence typing: Frequency distributions of number of fingers used to type (top) and
number of keys struck consistently (bottom) across all 48 typists. Black bars represent standard typists. Gray bars
represent nonstandard typists.

typists is plotted in Figure 2 (bottom panel). We also counted the The numbers for each group are plotted on different images of the
number of fingers used to type each key, plotting the mean keyboard in Figure 5. The top panels represent the number of
distributions for standard and nonstandard typists in Figure 4. We typists typing each key typed consistently with any finger, using
calculated the entropy in fingers assigned to each key (entropy ⫽ standard or nonstandard mapping. Most standard typists typed
⌺-p·log2(p)). Across typists, the number of keys struck consis- most keys consistently with some deviations at the edges of the
tently correlated ⫺.689 with the mean number of fingers per key keyboard. Nonstandard typists typed less consistently with some-
and ⫺.982 with entropy, so we focused our analysis on the number what greater consistency at the edges. The bottom panels represent
of keys struck consistently. the number of typists typing each key consistently using nonstan-
As expected, standard typists accommodated better to Hick’s dard fingers. Few standard typists typed consistently with nonstan-
law, striking significantly more keys consistently than nonstandard dard fingers. The few who did tended to type letters at the edge of
typists (22.71 vs. 12.92; see Table 1). We counted the number of the keyboard consistently, using their ring fingers instead of their
standard and nonstandard typists typing each letter consistently. little fingers. Many nonstandard typists typed consistently with
 2090 LOGAN, ULRICH, AND LINDSEY

ity: visible, blank, and covered) mixed ANOVAs with mapping

between subjects factor and keyboard visibility within subjects.
Number of Typists Using Finger

20 Over conditions, standard typists were faster (79.99 vs. 65.63

WPM), F(1,46) ⫽ 7.79, p ⫽ .008, MSE ⫽ 952.24, ␩p2 ⫽ 0.15, and
more accurate (93.57 vs. 83.22%), F(1,46) ⫽ 14.51, p ⫽ .0004,
15 MSE ⫽ 272.60, ␩p2 ⫽ 0.24, than nonstandard typists. Over groups,
reducing keyboard visibility reduced speed, F(2,92) ⫽ 7.87, p ⫽
Standard .0007, MSE ⫽ 54.83, ␩p2 ⫽ 0.15, and accuracy, F(2,92) ⫽ 14.26,
10
Nonstandard p ⫽ .000004, MSE ⫽ 128.96, ␩p2 ⫽ 0.24. However, these main
effects were modulated by strong interactions between group and
5
visibility: for WPM, F(2,92) ⫽ 5.34, p ⫽ .006, MSE ⫽ 54.83,
␩p2 ⫽ 0.10; for percent correct, F(2,92) ⫽ 11.00, p ⫽ .00005,
MSE ⫽ 128.96, ␩p2 ⫽ 0.19. Keyboard visibility mattered only for
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

0 nonstandard typists. Standard typists were unaffected but nonstan-

This document is copyrighted by the American Psychological Association or one of its allied publishers.

LL LR LM LI RI RM RR RL dard typists got progressively worse as keyboard visibility was

Finger reduced, consistent with the hypothesis that nonstandard typists
rely on visual guidance more than standard typists.
Figure 3. Full-alphabet sentence typing: Number of standard and non- Standard typists were faster and more accurate than nonstandard
standard typists using each finger at least once in typing 10 full-alphabet
typists in all visibility conditions, but the differences were signif-
sentences.
icant only for the blank (WPM: t(46) ⫽ 3.25, p ⬍ .01, MSE ⫽
5.42, JZS Bayes factor [BF; Rouder, Speckman, Sun, Morey, &
nonstandard fingers, especially near the edges of the keyboard, Iverson, 2009] ⫽ 16.72(A); Accuracy: t(46) ⫽ 3.41, p ⫽ .001,
where letters would be typed with ring and middle fingers. MSE ⫽ 2.22, JZS BF ⫽ 24.57(A)) and covered conditions (WPM:
We confirmed typists’ self reports of their typing style with the t(46) ⫽ 2.68, p ⫽ .010, MSE ⫽ 6.24, JZS BF ⫽ 7.92(A);
video analysis, counting typists as standard if they used all eight Accuracy: t(46) ⫽ 3.61, p ⫽ .0008, MSE ⫽ 6.18, JZS BF ⫽
fingers consistently and nonstandard if they used fewer fingers or 24.57(A)). The differences in the visible conditions were neither
struck keys less consistently. By this criterion, all 24 typists who significant nor clearly null (WPM: t(46) ⫽ 1.93, p ⫽ .059, MSE ⫽
claimed to use a nonstandard mapping actually used one. Surpris- 4.50, JZS BF ⫽ 1.28(A); Accuracy: t(46) ⫽ 1.08, p ⫽ .2858,
ingly, only 10 of the 24 typists who claimed to use the standard MSE ⫽ 1.05, JZS BF ⫽ 2.16(N)). However, the 10 typists who
mapping actually used it. The 14 “nonstandard standard” typists actually used the standard mapping were faster than the 24 non-
fell between actual standard and reported nonstandard typists in standard subjects in the visible condition (84.96 vs. 71.72 WPM),
number of fingers used (8.00, 7.43, and 5.92 for actual standard, t(33) ⫽ 1.08, p ⫽ .044, MSE ⫽ 1.05, JZS BF ⫽ 1.71(A).
“nonstandard standard,” and reported nonstandard typists, respec- Dimensional analysis. We complemented the group analysis
tively) and the number of keys struck consistently (25.70, 20.57, with a dimensional analysis that focused on the measures that
and 12.92, respectively). The data for actual nonstandard, nonstan- defined group membership: the number of fingers used to type,
dard standard, and reported nonstandard typists are presented which reflects typists’ accommodation to Fitts’ law, and the num-
separately for paragraph typing, word priming, keyboard recall, ber of keys struck consistently, which reflects typists’ accommo-
and hand cuing in Figure A3. dation to Hick’s law. Figure 7 plots typing speed and accuracy
Because of the surprisingly small number of actual standard
typists, we analyzed the data in two ways. First, we did the group
analyses we planned initially, comparing typists who reported
25
using standard mapping with typists who reported using nonstan-
dard mappings using analysis of variance (ANOVA) and t tests.
Mean Number of Le ers

Second, we did dimensional analyses, correlating the number of 20

fingers used and the number of keys struck consistently with speed
and accuracy, using regression. The two analyses generally led to 15
the same conclusions.
NS
10 S
Keyboard Visibility and Typing Speed and Accuracy
Group analysis. We assessed typing speed and accuracy by 5
having typists type paragraphs with the keyboard visible, blank, or
covered. We calculated WPM by dividing the number of key-
0
strokes in the paragraph by the time between the first and last 0 1 2 3 4
keystrokes in minutes to get keystrokes per minute, and dividing
Number of Fingers Per Le er
by 5 to get WPM. The mean WPM and percentage of words typed
correctly for each group are plotted as a function of condition in Figure 4. Number of letters typed using 1– 4 fingers for standard and
Figure 6. No data were excluded. The data were analyzed in 2 nonstandard typists. Fewer fingers indicate greater consistency. Error bars
(mapping group: standard vs. nonstandard) ⫻ 3 (keyboard visibil- are SEMs.
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2091

Standard Typists All Fingers Nonstandard Typists All Fingers

24 19 18 22 21 17 21 19 21 23 20 12 6 7 10 1 4 7 5 14
12 21 20 22 24 21 23 21 22 14 10 13 19 11 11 17 12 9
20 20 21 24 21 24 24 15 12 17 15 11 21 22

Standard Typists Nonstandard Fingers Nonstandard Typists Nonstandard Fingers

6 1 0 0 0 1 2 0 0 6 19 5 2 1 0 1 2 2 4 13
0 1 1 0 0 0 0 0 1 8 4 7 0 0 2 1 3 5
2 4 11 0 5 0 0 12 8 17 0 7 1 1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 5. Full-alphabet sentence typing: Number of standard (left panel) and nonstandard (right panel) typists
striking each key consistently. Top panels include all keys struck consistently with any finger. Bottom panels
show keys struck consistently with “nonstandard” fingers that do not correspond to the standard mapping.

against these measures, with separate points for each typist in each Parallel activation of keystrokes is evident in two effects: One is
visibility condition (visible, blank, and covered). Data points in the a gradient of activation across the word, producing an increase in
rightmost column in each panel (8 fingers, 26 letters) represent RT across probe position. The other is greater activation for letters
typists who actually type with standard mapping. All the other within the word, producing longer RTs to probes that are not part
points reflect nonstandard typists. The data show considerable of the word (Crump & Logan, 2010b; also see Logan, 2003; Logan
variability among nonstandard typists. Each panel shows positive et al., 2011). We assessed these effects with planned orthogonal
trends, suggesting that better adaptation to Fitts’ law (fingers) and contrasts based on a 2 (mapping group: standard, nonstandard) ⫻
Hick’s law (consistency) produces better performance. 4 (probe position: first, middle, last, and other) mixed ANOVA on
Table 2 presents correlations between fingers and consistency the probe RTs and accuracies and found evidence supporting
and speed and accuracy in each visibility condition. All of the parallel activation. The ANOVA showed significant main effects
correlations were significant except for fingers in the visible con- of mapping group, F(1, 46) ⫽ 13.09, p ⫽ .0007, MSE ⫽ 27548.88,
dition. The correlations were much larger when the keyboard was ␩p2 ⫽ 0.22, and probe position, F(3, 138) ⫽ 148.46, p ⫽ .05 ⫻
covered, reflecting lower accuracy. Accuracy was especially bad 10⫺14, MSE ⫽ 1439.77, ␩p2 ⫽ 0.76. The interaction between them
in typists who used fewer than eight fingers or struck fewer than 20 was not significant, F(3, 138) ⫽ 2.33, p ⫽ .077, MSE ⫽ 1439.77,
keys consistently. ␩p2 ⫽ 0.05, but we proceeded with the planned comparisons any-
We dissected the ANOVA interactions between group and vis- way.
ibility in mixed maximum-likelihood regression analyses with We assessed the gradient of activation with a contrast that
fingers and consistency as between-subjects predictors and key- compared the linear trend in RTs to within word probes across
board visibility (visible, blank, and covered) as within-subjects position in the word, ultimately comparing the first position with
predictors. One analysis predicted speed; the other predicted ac-
curacy. Summary tables for both analyses are presented in Table 3.
Number of fingers—the Fitts’s law adaptation—affected both NS WPM S WPM NS ACC S ACC
speed and accuracy. Number of fingers interacted significantly
90 100
with visibility for both speed and accuracy, suggesting that visual
guidance is especially important for typists who type with fewer 95
fingers. Number of keys struck consistently—the Hick’s law ad- 80
Words Per Minute

90
Percent Correct

aptation—affected both speed and accuracy. Consistency inter-

acted with visibility in speed but not in accuracy. Using few 70 85

fingers inconsistently produces slower typing, especially when 80

visibility is limited. 60 75

Word Priming 70
50
Group analysis. The word-priming task is the first of three to 65
compare the extent of hierarchical processing in standard and 40 60
nonstandard typists, assessing parallel activation of the keystrokes Visible Blank Covered
in a word. We excluded trials with errors (9.13% of the data).
Keyboard
Mean RTs for correct responses to probes are plotted as a function
of probe position (first, middle, or last letter) for each group in the Figure 6. Typing speed (words per minute) and accuracy (percent cor-
top panel of Figure 8. RTs to control probes from other words are rect) for standard (S) and nonstandard (NS) typists as a function of
plotted as horizontal lines for comparison. keyboard visibility. Error bars are SEMs.
 2092 LOGAN, ULRICH, AND LINDSEY

140 100

120
A 90
B
80

Words Per Minute

100 70

Percent Correct
60
80
Visible 50 Visible
60 Blank Blank
40
Covered Covered
40 30

20
20
10

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Number of Fingers Used to Type Number of Fingers Used to Type
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

140
C 100
D
This document is copyrighted by the American Psychological Association or one of its allied publishers.

90
120
80
Words Per Minute

100 70

Percent Correct
60
80
Visible 50 Visible
60 Blank Blank
40
Covered Covered
40 30

20
20
10

0 0
0 5 10 15 20 25 0 5 10 15 20 25
Number of Le: ers Typed Consistently Number of Le: ers Typed Consistently

Figure 7. Dimensional analysis of typing speed and accuracy: The top panels show adaptation to Fitts’ law:
words per minute (Panel A) and accuracy (Panel B) as a function of the number of fingers used to Type 10
full-alphabet sentences. The bottom panels show adaptation to Hick’s law: words per minute (Panel C) and
accuracy (Panel D) as a function of the number of letters typed consistently.

the last. The contrast was significant overall, F(1, 138) ⫽ 340.31, 4.80, p ⫽ .003, MSE ⫽ 1439.77, suggesting less priming in
p ⬍ .1 ⫻ 10⫺20, and significant for both standard, F(1, 138) ⫽ nonstandard typists.
267.03, and nonstandard typists, F(1, 138) ⫽ 422.47, all ps ⬍ .1 ⫻ The bottom panel of Figure 8 shows the results expressed as
10⫺20, all MSEs ⫽ 1439.77. A contrast that compared the gradient priming scores. Consistent with the contrast analysis, both groups
in standard and nonstandard typists was significant, F(1, 138) ⫽ show priming and gradients of priming, but the gradient is some-
8.87, p ⫽ .003, MSE ⫽ 1439.77, suggesting a steeper gradient in what steeper for nonstandard typists. Priming is the same for the
nonstandard typists. two groups for the first letter but becomes smaller for subsequent
We assessed the amount of priming with a contrast that com- letters for nonstandard typists. The interpretation of these differ-
pared the average RT for the three within-word probes with RT for
other-word probes. The contrast was significant overall, F(1,
138) ⫽ 70.92, and significant for both standard, F(1, 138) ⫽ Table 3
99.43, and nonstandard typists, F(1, 138) ⫽ 47.23, all ps ⬍ .1 ⫻ Regression Analysis of Typing Speed (Words Per Minute) and
10⫺9, all MSEs ⫽ 1439.77. A contrast that compared priming in Accuracy (Percent Correct) With Number of Fingers Used to
standard and nonstandard typists was significant, F(1, 138) ⫽ Type (Fingers), Number of Keys Struck Consistently
(Consistency), and Keyboard Visibility (Visibility) as Predictors
Table 2 Error Degrees of Freedom ⫽ 44
Correlations of Typing Speed (Words Per Minute) and Accuracy
Words per minute Percent correct
(Percent Correct) With Number of Fingers Used to Type and
With Consistency of Finger Use in the Paragraph Typing Task Effect df F Cohen f df F Cohen f
ⴱ ⴱ
Visibility (V) 2 4.08 0.36 2 4.18 .36
Keyboard Fingers (F) 1 3.82ⴱ 0.24 1 13.15ⴱ .50
Predictor Measure Visible Blank Covered Consistency (C) 1 5.01ⴱ 0.29 1 7.36ⴱ .36
F⫻C 1 0.11 0.00 1 2.27 .16
ⴱ
Number of fingers Words per minute .283 .460 .514ⴱ V⫻F 2 5.10ⴱ 0.41 2 8.48ⴱ .56
Percent correct .189 .542ⴱ .709ⴱ V⫻C 2 3.40ⴱ 0.32 2 1.60 .16
Consistency Words per minute .343ⴱ .469ⴱ .437ⴱ V⫻F⫻C 2 3.18 0.30 2 1.19 .09
Percent correct .335ⴱ .503ⴱ .504ⴱ
Note. Error degrees of freedom (df) ⫽ 44.
ⴱ ⴱ
p ⬍ .05. p ⬍ .05.
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2093

900 For IKSI, t(46) ⫽ 1.76, p ⫽ .085, MSE ⫽ 15.42, JSZ BF ⫽

1.004(N); for accuracy, t(46) ⫽ 0.71, p ⫽ .484, MSE ⫽ 5.42, JSZ
BF ⫽ 2.841(N).
Mean Response Time in ms

800 Dimensional analysis. We performed a maximum-likelihood

regression analysis, predicting probe RT with probe position,
fingers, and consistency as predictors. The summary table for the
NS Primed analysis appears in Table 4. The only significant effects were
700
NS Control position, reflecting priming, and fingers, reflecting Fitts’ law:
S Primed Typists who use fewer fingers have longer distances to move to
S Control type the probe letter.
600

Keyboard Recall
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

500
Group analysis. The keyboard recall task assesses encapsu-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

First Middle Last lation of knowledge about the locations of letters on the keyboard
Probe Posi on in Word in lower-level processes (Snyder et al., 2014). We counted the
number of letters recalled in correct and incorrect (error) positions
200
in each 20 s epoch (identified by ink color). When typists wrote a
letter twice (57 of 1,248 total responses), we counted the first
response. The first response was clear when the two responses
Other RT - Within Word RT in ms

150 occurred in different epochs. When the two responses occurred in

the same epoch, we counted them as a single error if they were
both incorrect placements, and we counted them as .5 correct and
100 .5 error if one was correct and the other was an error. We tried
NS other scoring procedures, always counting correct and error place-
S ments in the same epoch as correct, always counting them as
50 Neutral incorrect, or counting them as both correct and incorrect. The
results were very similar and the pattern of significant effects was
the same. The mean number of letters recalled correctly and
0
incorrectly (error), counting correct and error responses in the
same epoch as .5, is plotted for 20 s epoch in the recall period for
First Middle Last each group in Figure 9. Correct and error data were analyzed in
-50
Probe Posi on in Word separate 2 (mapping group: standard, nonstandard) ⫻ 4 (recall
epoch: 1– 4) mixed ANOVAs. No data were excluded from anal-
Figure 8. Word priming: Top panel: Mean response time (RT) to single ysis.
letter probes as a function of their position in the prime word for standard Both standard and nonstandard typists showed incomplete and
(S) and nonstandard (NS) typists. Control trials represent letter probes from inaccurate explicit knowledge of the keyboard. Standard typists
other words. Bottom panel: Priming scores (control RT—within word RT)
recalled 17.1 letters correctly, placed 4.4 in the wrong location,
as a function of their position in the word for standard and nonstandard
and omitted 4.5. Nonstandard typists correctly recalled 14.6, in-
typists. Error bars are Fisher’s Least Significant Difference for p ⬍ .05
calculated from the highest order interaction. correctly placed 6.6, and omitted 4.8. Correct recall was highest in
the first 20 s and declined after that, F(3, 138) ⫽ 93.95, p ⫽ .4 ⫻
10⫺32, MSE ⫽ 8.75, ␩p2 ⫽ 0.67, suggesting that typists ran out of
ences between groups must be tempered by the lack of a signifi- knowledge, not time. Standard typists placed more letters in cor-
cant interaction between groups and probe position, on which these rect positions than nonstandard typists, but the difference was not
contrasts are based.
A 2 (mapping group: standard, nonstandard) ⫻ 4 (probe posi-
tion: first, middle, last, and other) mixed ANOVA on accuracy Table 4
showed no significant differences. Accuracy was high for all probe Summary Table for Regression Analysis of Word
positions in both groups (standard ⫽ 95.8, 95.0, 93.7, and 96.2% Priming Effects
for first, middle, last, and other, respectively; nonstandard ⫽ 94.8,
Effect df F Cohen f̂
95.1, 94.3, and 95.0%, respectively).
ⴱ
On go trials, in which typists typed the whole word, standard Position (P) 3 88.02 2.33
typists had shorter RTs than nonstandard typists (487 vs. 585 ms), Fingers (F) 1 8.02ⴱ .38
Consistency (C) 1 1.47 .10
t(46) ⫽ 2.84, p ⫽ .007, MSE ⫽ 10.42, JSZ BF ⫽ 6.678(A), F⫻C 1 .03 .00
consistent with the difference in probe RTs, perhaps because P⫻F 3 1.73 .21
nonstandard typists used fewer fingers and so had to move greater P⫻C 3 1.15 .10
distances. Standard typists had shorter interkeystroke intervals P⫻F⫻C 3 2.5 .31
(IKSIs: 134 vs. 149 ms) and higher accuracy (87.38 vs. 86.13%) Note. Error degree of freedom (df) ⫽ 44 for each effect.
ⴱ
than nonstandard typists, but neither difference was significant. p ⬍ .05.
 2094 LOGAN, ULRICH, AND LINDSEY

12 letters (Logan & Crump, 2009; Snyder & Logan, 2013; Tapp &
Logan, 2011). Two typists in the nonstandard group were excluded
10 for not following instructions (responding appropriately to the
hand cue). We excluded data from trials with errors (see Table 4).
Number Recalled

8
Mean RTs to the first keystroke and IKSIs for subsequent key-
strokes are plotted in Figure 10 as a function of cuing condition.
Correct NS Accuracy data are presented in Table 5.
6
Correct S Both standard and nonstandard typists showed evidence of
Error NS encapsulation: Hand cue trials, which required explicit monitoring
4 of the hands and fingers, produced longer RTs and IKSIs than
Error S
whole cues, which did not require monitoring. For RT, the differ-
2 ences were 446 and 440 ms for standard and nonstandard typists,
respectively. For IKSI, the differences were 130 and 124 ms,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

0 respectively. These are large disruptions compared to baseline RT

This document is copyrighted by the American Psychological Association or one of its allied publishers.

1 2 3 4 and IKSI, suggesting that knowledge of key to finger mapping is

20 Second Epoch not directly available to conscious awareness (Logan & Crump,
2009; Snyder & Logan, 2013; Tapp & Logan, 2011).
Figure 9. Keyboard recall: Mean number of letters recalled in correct These conclusions were supported by a 2 (mapping group:
(correct) and incorrect (error) positions by standard (S) and nonstandard
standard vs. nonstandard) ⫻ 2 (cue type: hand vs. whole) ⫻ 2
(NS) typists as a function of epoch in the 80 second recall period. Error
bars are SEMs.

1600

significant, F(1, 46) ⫽ 3.72, p ⫽ .060, MSE ⫽ 5.22, ␩p2

⫽ 0.08. 1400
Mean Response Time in ms

Error recall did not vary across epochs, F(3, 138) ⫽ 1.40, p ⫽
.246, MSE ⫽ 3.198, ␩p2 ⫽ 0.03. Standard typists placed signifi- 1200

cantly fewer letters in incorrect positions than nonstandard typists, 1000

F(1, 46) ⫽ 4.84, p ⫽ .033, MSE ⫽ 3.26, ␩p2 ⫽ 0.09). The poorer
performance for nonstandard typists was surprising given the 800

evidence that they rely more on visual guidance. Apparently, 600

S
looking at the keys does not guarantee memory for their locations. NS
For both groups, the low accuracy and high error and omission 400

rates in keyboard recall contrast sharply with their high accuracy 200
typing the same letters on the sentence typing task, which taps both
implicit and explicit knowledge. Standard and nonstandard typists 0
Bimanual Unimanual Bimanual Unimanual
typed 91.8 and 89.9% of the words correctly. If the letters are
Whole Hand
typed independently, then the probability of typing the whole word
correctly is the product of the probabilities of typing each letter Cuing Condi on
correctly. If the probabilities are equal, then the probability of 350
Mean Interkeystroke Interval in ms

typing an N-letter word correctly is the probability of typing a

letter correctly raised to the Nth power: P(word correct) ⫽ P(letter 300
correct)N. Thus, the probability of typing letters correctly can be
estimated as the Nth root of the probability of typing words 250

correctly: P(letter correct) ⫽ P(word correct)1/N. The words in the

200
sentences averaged 4.51 letters in length, so accuracy per letter is
the 4.51th root of accuracy per word. For standard and nonstandard 150 S
typists, accuracy per letter was 98.1 and 97.7%, respectively. Both
NS
values are much higher than keyboard recall accuracy (65.8 and 100
56.2%, respectively), suggesting that both groups rely on implicit
knowledge to support accurate typing. Explicit knowledge of the 50

keyboard is not sufficient in either group.

0
Dimensional analysis. We performed least-squares regres- Bimanual Unimanual Bimanual Unimanual
sion analyses predicting correct and error recall (separately) Whole Hand
summed over recall epoch from fingers and consistency. No effect Cuing Condi on
was significant in either analysis.
Figure 10. Hand cuing: Mean response time (top) and interkeystroke
Hand Cuing interval (bottom) for standard (S) and nonstandard (NS) typists as a
function of cue type (whole, hand) and word type (unimanual, bimanual).
Group analysis. The hand-cuing task assesses encapsulation Error bars are Fisher’s Least Significant Difference for p ⬍ .05 calculated
of knowledge about the hands and fingers used to type specific from the highest order interaction.
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2095

Table 5 with group may result from differences in consistency of mapping

Mean Accuracy (Percent Correct) for the Hand-Cuing Task as a fingers onto keys between groups. Nonstandard typists may type
Function of Group, Cue Type, and Word Type some unimanual words with two hands, making monitoring more
difficult.
Whole cue Hand cue An ANOVA with the same structure on accuracy showed a
Typing style Bimanual Unimanual Bimanual Unimanual significant main effect of cue type, F(1, 44) ⫽ 24.16, p ⫽ .00001,
MSE ⫽ 51.79, ␩p2 ⫽ 0.35, and a significant interaction between
Standard 89.93 90.97 84.90 88.19
Nonstandard 92.55 89.39 85.92 82.95
mapping group and word type, F(1, 44) ⫽ 7.84, p ⫽ .0076, MSE ⫽
40.09, ␩p2 ⫽ 0.15. Standard typists did better on unimanual words
than bimanual words; nonstandard subjects did the opposite. No
other effects were significant.
(word type: unimanual vs. bimanual) mixed ANOVAs on the RTs.
Dimensional analysis. We performed separate maximum-
The RT ANOVA showed significant main effects of cue type, F(1,
likelihood regression analyses on RT, IKSI, and accuracy using
44) ⫽ 217.13, p ⫽ .01 ⫻ 10⫺16, MSE ⫽ 41598.81, ␩p2 ⫽ 0.83, and
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

hand cue, word type, fingers, and consistency as predictors. Sum-

word type, F(1, 44) ⫽ 17.58, p ⫽ .0001, MSE ⫽ 15992.03, ␩p2 ⫽
This document is copyrighted by the American Psychological Association or one of its allied publishers.

mary tables for the analyses appear in Table 6. The effect of hand
0.29, and a significant interaction between them, F(1, 44) ⫽ 18.23,
cue was significant in all three analyses and the interaction be-
p ⫽ .0001, MSE ⫽ 17509.32, ␩p2 ⫽ 0.29. Neither the main effect
tween hand cue and word type were significant in the RT and IKSI
of group nor any of its interactions were significant: Fs(1,44) ⫽
analyses, reflecting encapsulation of knowledge in lower level
0.09, 0.04, 1.74, and 3.18 for the main effect, Group ⫻ Cue type,
processes. The interaction between fingers and consistency was
Group ⫻ Word type, and Group ⫻ Cue type ⫻ Word type,
significant in the RT and accuracy analyses, reflecting the cost of
respectively, MSEs ⫽ 26783.53, 41598.81, 15992.03, and
a poorer balance between Fitts’ law and Hick’s law. The remaining
17509.32, respectively, ps ⫽ .77, .84, .19, and .08, respectively,
significant interactions involving fingers and consistency were
␩p2s ⫽ 0.00, 0.00, 0.04, and 0.07, respectively.
significant in only one of the analyses.
An ANOVA with the same structure on IKSIs showed a significant
main effect of cue type, F(1, 44) ⫽ 96.75, p ⫽ .01 ⫻ 10⫺11, MSE ⫽
Postscript
462.45, ␩p2 ⫽ 0.69, and significant interactions between cue type and
word type, F(1, 44) ⫽ 36.42, p ⫽ .0000002, MSE ⫽ 2348.53, While this article was under review, Feit, Weir, and Oulasvirta
␩p2 ⫽ 0.45, word type and group, F(1, 44) ⫽ 4.20, p ⫽ .046, MSE ⫽ (2016) published an article in the proceedings of a computer
3737.93, ␩p2 ⫽ 0.09, and cue type, word type, and group, F(1, 44) ⫽ science conference (CHI ’16) that also compared standard and
5.36, p ⫽ .025, MSE ⫽ 2348.53, ␩p2 ⫽ 0.11. The interactions reflect nonstandard typists. They used motion capture software to analyze
the relative difficulty of typing unimanual words and the relative ease movements while typists typed visually displayed sentences on
of monitoring them. Unimanual words require hand repetitions, which visible keyboards with the letters marked on the keys. Their study
are difficult, while bimanual words allow hand alternations, which are was conducted in Finland and sampled both Finnish and English
easy (Salthouse, 1986), so unimanual words are slower than bimanual speakers, testing them in the language in which they preferred to
words on trials with whole cues. By contrast, hand repetitions may type. Their sample size was smaller (N ⫽ 30 vs. 48) and their
make monitoring easier, so unimanual words are monitored more typists were slower on average (58.5 vs. 76.0 WPM), but their
quickly than bimanual ones on trials with hand cues. The interactions results are very similar to ours (or rather, ours are similar to theirs).

Table 6
Summary Tables for Regression Analyses on Hand Cuing Data

Interkeystroke
Response time interval Percent correct
Effect Error df F Cohen f̂ F Cohen f̂ F Cohen f̂
ⴱ ⴱ ⴱ
Hand cue (H) 45 70.03 1.23 61.15 1.14 6.23 .34
Word (W) 45 2.69 .19 .00 .00 .11 .00
Finger (F) 42 .83 .00 1.23 .07 1.10 .05
Consistency (C) 42 1.91 .14 .23 .00 3.06 .21
H⫻W 45 4.46ⴱ .27 26.23ⴱ .74 1.64 .12
F⫻C 42 12.06ⴱ .49 1.47 .10 8.01ⴱ .39
H⫻F 126 .09 .00 2.81 .20 .43 .00
W⫻C 126 1.13 .05 8.95ⴱ .42 5.45ⴱ .31
H⫻C 126 .05 .00 4.86ⴱ .29 .85 .00
W⫻F 126 .07 .00 .23 .00 .38 .00
H⫻F⫻C 126 .00 .00 3.19 .22 1.17 .06
W⫻F⫻C 126 2.88 .20 1.11 .05 .73 .00
H⫻W⫻F 126 .42 .00 .49 .00 1.05 .03
H⫻W⫻C 126 2.22 .16 16.57ⴱ .58 .35 .00
H⫻W⫻F⫻C 126 11.50ⴱ .48 1.93 .14 1.55 .11
Note. Each effect has 1 degree of freedom (df).
ⴱ
p ⬍ .05.
 2096 LOGAN, ULRICH, AND LINDSEY

Like us, Feit et al. (2016) found no difference in typing speed suggesting parallel activation of keystrokes in both groups (Crump
between their 13 standard typists and their 17 nonstandard typists & Logan, 2010b). The gradient was somewhat steeper in nonstan-
(57.8 vs. 58.9 WPM, respectively), while typing on a visible key- dard typists, suggesting a narrower range of parallel activation.
board, t(28) ⫽ 0.221, p ⫽ .827, MSE ⫽ 4.978, JZS BF ⫽ 2.832(N). Neither group had enough explicit knowledge of the keyboard to
Their sample size limited their power (power to detect a 10 WPM support their demonstrated ability to type accurately, suggesting
difference was only .56) but the Bayes Factor favors the null hypoth- that their knowledge of letter to key mapping is mostly implicit in
esis. the motor system (Snyder et al., 2014). Standard and nonstandard
Motion capture analysis showed that standard typists used more typists both slowed dramatically when typing with hand cues,
fingers than nonstandard typists and used them more consistently, suggesting their knowledge of the mapping of keys to movements
but only 3 of the 13 standard typists used the standard mapping was also implicit in the motor system (Logan & Crump, 2009;
perfectly consistently. Regression analyses showed that typing Snyder & Logan, 2013; Tapp & Logan, 2011). Thus, nonstandard
speed was predicted by the consistency of mapping fingers onto typists seem to have the same degree of hierarchical control as
keys, measured as entropy, suggesting that fast typing depends on standard typists (Logan & Crump, 2011; Shaffer, 1976; Sternberg
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

adaptation to Hick’s law. Typing speed was also predicted by et al., 1990). Both seem able to type without thinking about letters,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

measures of movement distance, consistent with Fitts’ law. Feit et keys, and movements, having handed that off to the motor system.
al. did not find significant correlations with number of fingers used It is interesting that nonstandard typists achieved the same
for typing, testing typists with the keyboard visible. We found no degree of hierarchical control as standard typists with suboptimal
significant correlation between fingers and typing speed when the adaptations to Fitts’ law and Hick’s law. It runs counter to the
keyboard was visible, though our correlations were significant in common intuition that higher degrees of automaticity are associ-
the blank and covered keyboard conditions (see Table 2). ated with better performance (Cohen, Dunbar, & McClelland,
Interestingly, Feit et al. measured typists’ gaze direction while 1990; Logan, 1988; MacLeod & Dunbar, 1988; Rickard, 1997;
typing and found that nonstandard typists spent much more time Shiffrin & Schneider, 1977; Siegler, 1987). Typists may value
looking at the keyboard than standard typists (41 vs. 20%, respec- hierarchical control because it reduces the load on working mem-
tively). This corroborates our finding that restricting keyboard ory (compare the left and right panels of Figure 1), and that may
visibility selectively impairs nonstandard typists. Overall, the re- motivate a shift from top– down to hierarchical control before
sults agree quite closely, demonstrating that the basic effects performance is optimal. We assume that typists adopt hierarchical
survive independent replication. control when the memories in the motor system are reliable
enough to support typing without top– down control. Memories
Discussion strengthen with practice, and may become reliable before they are
optimal (Crump & Logan, 2010a). Memories for nonstandard
Balancing Fitts’ Law and Hick’s Law mappings also improve with practice (Yamaguchi & Logan,
2014a). They may become reliable even if they are suboptimal. We
We began with the hypothesis that fine motor skills involve suspect our nonstandard typists performed worse because of sub-
balancing a tradeoff between Fitts’ law and Hick’s law. Recruiting optimal mapping, not lack of practice. They started typing a year
more fingers reduces movement time but complicates the choice and a half before the standard typists and spent the same time on
between fingers. Choice time can be reduced by consistent prac- computers each day (see Table 1).
tice. Thus, we predicted higher levels of speed and accuracy in Typing performance is difficult to optimize because it requires
typists who use more fingers more consistently. The standard achieving a balance between Fitts’ law and Hick’s law. Different
mapping of fingers to keys minimizes distance and maximizes balances impose different constraints on performance, and typists
consistency, so we predicted better performance in typists who must adapt to the balance they choose. Balances that recruit more
used the standard mapping than in typists who used nonstandard fingers consistently will lead to better performance, but every
mappings. The results confirmed both predictions. Standard typists balance can be improved with practice, headed toward its own
used more fingers more consistently and typed faster and more local maximum. Many paths to local maxima may produce motor
accurately than nonstandard typists, especially when visibility of memories that are reliable enough to support typing without top–
the keyboard was degraded. Typing speed and accuracy correlated down control (Rosenbaum et al., 2001).
with the number of fingers used and the number of keys struck
consistently. Thus, skilled performance depends on the balance
between Fitts’ law and Hick’s law. We suggest the balance deter- Visual Guidance
mines the highest level of performance a typist can attain. Subop-
timal balances lead to lower performance at asymptote. Our results show that the cost of nonstandard typing is
increased reliance on vision (also see Feit et al., 2016). We
suspect that nonstandard typists retain vestiges of the hunt and
Automaticity and Hierarchical Control
peck strategy, sometimes choosing the finger closest to the
We asked whether the better performance of standard typists target key. Visual or spatial guidance may be helpful in resolv-
was accompanied by a higher degree of automaticity, expressed as ing interference. When retrieval suggests two candidates, prox-
greater hierarchical control. We tested hierarchical control in three imity to the target may break the tie, allowing a faster choice
tasks and found the same qualitative and quantitative effects for than retrieval by itself. However, the nearest finger is not
standard and nonstandard typists in each task. They showed similar always the one assigned in standard mapping, so using vision to
gradients of activation across position in the word-priming task, resolve present interference may strengthen memories that
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2097

cause more interference in the future. Visual guidance may general population. We were surprised to find that only 10 of 24
subvert adaptation to Hick’s law. typists who reported using the standard mapping actually used it.
Increased reliance on visual guidance may have other costs. The Altogether, 79% of typists in our sample were nonstandard. Feit et
eyes cannot be on the keyboard and the screen at the same time. Thus, al. (2016) found 90% were nonstandard. It would be interesting to
nonstandard typists would derive fewer benefits from information on know how these numbers compare to the general population.
the screen as they type. For example, typists monitor the screen for Perhaps standard typists are a rare breed.
explicit error detection (Logan & Crump, 2010; Snyder et al., 2015).
Nonstandard typists might miss more errors. When people write Beyond Typing
compositions, they often read the text as they type it to check for Typing is one example of a broad range of fine motor skills that pit
errors of grammar, meaning, and style. Attention to the keyboard Fitts’ law against Hick’s law. Playing musical instruments, operating
would disrupt reading and impair error detection. Attention to the equipment, and interfacing with machines all require mapping fingers
keyboard might produce more specific interference when writing onto positions. The range of distances and the nature of the patterns
about space or spatial relations (Logan, 1994). However, these effects vary from one skill to another, but Fitts’ law and Hick’s law hold
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

might be small. All but two of our typists typed on laptops, where the nevertheless. The mapping is more complicated with musical instru-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

screen is close to the keyboard, making switching between them ments than with typing. Different fingers may play the same note in
easier and potentially reducing the costs. Whether such costs exist is different contexts (chords, melodies), but the same finger often plays
an important question for future research. the same note in a given context. The content may vary between
skills, but learning and performance are governed by the same prin-
Remedial Training for Nonstandard Typists? ciples: consistency within context facilitates encoding and retrieval
(Logan, 1988; Logan & Etherton, 1994; Shiffrin & Schneider, 1977).
Should nonstandard typists invest in remedial training to learn Thus, our results with typing should generalize to a broad range of
the standard mapping? Feit et al. (2016) found no difference in fine motor skills: People who balance Fitts’ law and Hick’s law better
typing speed between standard and nonstandard typists (58 vs. 59 will perform better.
WPM, respectively). In the present study, with faster typists, the Fine motor skills may be limited by other laws besides Fitts’
differences between standard and nonstandard typists were clearer, and Hick’s. Music must be played in rhythm, and that adds
but they were relatively small. When the keyboard was visible, temporal constraints to the retrieval process (Palmer & Pfor-
nonstandard typists averaged 72 WPM and our one two-finger dresher, 2003; Pfordresher, Palmer, & Jungers, 2007). Hands
typist managed 60 WPM. That is fast enough to qualify as a must be coordinated (Heijink & Meulenbroek, 2002). Emotion
professional typist. It may be fast enough for many typing situa- must be conveyed (Juslin, 2000). Improvising requires choosing
tions, where top speed is not so critical. An unpublished study in the notes to play as well as choosing the fingers to play them
our laboratory found that skilled typists who typed at 78 WPM on with (Johnson-Laird, 2002). Playing in a group requires coor-
a speed test slowed to 45 WPM when composing texts. Modern dinating pitch and rhythm with others (Zamm, Pfordresher, &
typists spend much of their time composing (Logan & Crump, Palmer, 2015). Each skill involves its own constraints and its
2011), so the suboptimal top speed of nonstandard typists may not own tradeoffs between fundamental psychological laws. Much
matter much. From this perspective, remedial training on the can be learned from identifying the laws that apply, learning
standard mapping may not be worth the investment. Typists may how they trade off against each other, and learning how skilled
care more about typing hierarchically than typing quickly. Hier- practitioners balance the tradeoff. We hope we have shown this
archical control lets them type without thinking so they can think approach is productive in skilled typing.
about higher level goals.
Should schools invest in teaching typing earlier? Our nonstan- References
dard typists started typing at an earlier age than our standard
typists, so their habits may have been more strongly ingrained Anderson, J. R. (1987). Skill acquisition: Compilation of weak-method
before they began formal training. In debriefing, we asked our problem situations. Psychological Review, 94, 192–210. http://dx.doi
.org/10.1037/0033-295X.94.2.192
nonstandard typists why they persisted with nonstandard mapping
Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the control of
despite having formal training on the standard mapping. Of the 22 automatic processes: A parallel distributed processing account of the
who responded, 10 said their own style was faster or easier and 12 Stroop effect. Psychological Review, 97, 322–361. http://dx.doi.org/10
said it was more comfortable or convenient. Thus, they had little .1037/0033-295X.97.3.332
incentive to acquire high levels of skill with standard mapping. Coltheart, M. (1981). The MRC psycholinguistic database. Quarterly Jour-
Earlier training on standard mapping may prevent the development nal of Experimental Psychology, 33, 497–505. http://dx.doi.org/10.1080/
of suboptimal habits, but suboptimal habits may be good enough. 14640748108400805
The benefits of earlier training may not be large enough outweigh Crump, M. J. C., & Logan, G. D. (2010a). Episodic contributions to
the costs the typist and the educational system would have to pay. sequential control: Learning from a typist’s touch. Journal of Experi-
These speculations rest on how representative our standard and mental Psychology: Human Perception and Performance, 36, 662– 672.
http://dx.doi.org/10.1037/a0018390
nonstandard typists were of the general population. Our nonstan-
Crump, M. J. C., & Logan, G. D. (2010b). Hierarchical control and skilled
dard typists may have been especially good. They were recruited typing: Evidence for word-level control over the execution of individual
from a competitive university environment and may have re- keystrokes. Journal of Experimental Psychology: Learning, Memory,
sponded to our advertisement to show off their high level of skill. and Cognition, 36, 1369 –1380. http://dx.doi.org/10.1037/a0020696
Feit et al. (2016) found similar results in a sample from a broader Crump, M. J. C., & Logan, G. D. (2010c). Warning: This keyboard will
population. It would be interesting to know the levels of skill in the deconstruct—The role of the keyboard in skilled typewriting. Psycho-
 2098 LOGAN, ULRICH, AND LINDSEY

nomic Bulletin & Review, 17, 394 –399. http://dx.doi.org/10.3758/PBR Logan, G. D., & Crump, M. J. C. (2010). Cognitive illusions of authorship
.17.3.394 reveal hierarchical error detection in skilled typists. Science, 330, 683–
Crump, M. J. C., & Logan, G. D. (2013). Prevention and correction in 686. http://dx.doi.org/10.1126/science.1190483
post-error performance: An ounce of prevention, a pound of cure. Logan, G. D., & Crump, M. J. C. (2011). Hierarchical control of cognitive
Journal of Experimental Psychology: General, 142, 692–709. http://dx processes: The case for skilled typewriting. In B. H. Ross (Ed.), The
.doi.org/10.1037/a0030014 psychology of learning and motivation (Vol. 54, pp. 1–27). Burlington,
Feit, A. M., Weir, D., & Oulasvirta, A. (2016). How we type: Movement MA: Academic Press. http://dx.doi.org/10.1016/B978-0-12-385527-5
strategies and performance in everyday typing. In CHI ’16: Proceedings .00001-2
of the 2016 CHI Conference on Human Factors in Computing Systems Logan, G. D., & Etherton, J. L. (1994). What is learned during automati-
(pp. 4262– 4273). New York, NY: ACM. http://dx.doi.org/10.1145/ zation? The role of attention in constructing an instance. Journal of
2858036.2858233 Experimental Psychology: Learning, Memory, and Cognition, 20, 1022–
Fitts, P. M. (1954). The information capacity of the human motor system 1050. http://dx.doi.org/10.1037/0278-7393.20.5.1022
in controlling the amplitude of movement. Journal of Experimental Logan, G. D., Miller, A. E., & Strayer, D. L. (2011). Electrophysiological
Psychology, 47, 381–391. http://dx.doi.org/10.1037/h0055392 evidence for parallel response selection in skilled typists. Psychological
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Fitts, P. M., & Posner, M. I. (1967). Human performance. Oxford, Eng- Science, 22, 54 –56. http://dx.doi.org/10.1177/0956797610390382
This document is copyrighted by the American Psychological Association or one of its allied publishers.

land: Brooks/Cole. MacLeod, C. M., & Dunbar, K. (1988). Training and Stroop-like interfer-
Heijink, H., & Meulenbroek, R. G. J. (2002). On the complexity of ence: Evidence for a continuum of automaticity. Journal of Experimen-
classical guitar playing: Functional adaptations to task constraints. Jour- tal Psychology: Learning, Memory and Cognition, 14, 126 –135. http://
nal of Motor Behavior, 34, 339 –351. http://dx.doi.org/10.1080/ dx.doi.org/10.1037/0278-7393.14.1.126
00222890209601952 Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and
Hick, W. E. (1952). On the rate of gain of information. The Quarterly conceptual analysis. Psychological Bulletin, 132, 297–326. http://dx.doi
Journal of Experimental Psychology, 4, 11–26. http://dx.doi.org/10 .org/10.1037/0033-2909.132.2.297
.1080/17470215208416600 Palmer, C., & Pfordresher, P. Q. (2003). Incremental planning in sequence
Hyman, R. (1953). Stimulus information as a determinant of reaction time. production. Psychological Review, 110, 683–712. http://dx.doi.org/10
Journal of Experimental Psychology, 45, 188 –196. http://dx.doi.org/10 .1037/0033-295X.110.4.683
.1037/h0056940 Palmeri, T. J. (1997). Exemplar similarity and the development of
Johnson-Laird, P. N. (2002). How jazz musicians improvise. Music Per- automaticity. Journal of Experimental Psychology: Learning, Mem-
ception, 19, 415– 442. http://dx.doi.org/10.1525/mp.2002.19.3.415 ory and Cognition, 23, 324 –354. http://dx.doi.org/10.1037/0278-
Juslin, P. N. (2000). Cue utilization in communication of emotion in music 7393.23.2.324
performance: Relating performance to perception. Journal of Experi- Pfordresher, P. Q., Palmer, C., & Jungers, M. K. (2007). Speed, accuracy,
mental Psychology: Human Perception and Performance, 26, 1797– and serial order in sequence production. Cognitive Science, 31, 63–98.
1812. http://dx.doi.org/10.1037/0096-1523.26.6.1797 http://dx.doi.org/10.1080/03640210709336985
Keller, F. S. (1958). The phantom plateau. Journal of the Experimental Rickard, T. C. (1997). Bending the power law: A CMPL theory of strategy
Analysis of Behavior, 1, 1–13. http://dx.doi.org/10.1901/jeab.1958.1-1 shifts and the automatization of cognitive skills. Journal of Experimental
Klatzky, R. L., & Lederman, S. J. (2003). Representing spatial location and Psychology: General, 126, 288 –311. http://dx.doi.org/10.1037/0096-
layout from sparse kinesthetic contacts. Journal of Experimental Psy- 3445.126.3.288
chology: Human Perception and Performance, 29, 310 –325. http://dx Rosenbaum, D. A., Meulenbroek, R. J., Vaughan, J., & Jansen, C. (2001).
.doi.org/10.1037/0096-1523.29.2.310 Posture-based motion planning: Applications to grasping. Psychological
Liu, X., Crump, M. J. C., & Logan, G. D. (2010). Do you know where your Review, 108, 709 –734.
fingers have been? Explicit knowledge of the spatial layout of the Rouder, J. N., Speckman, P. L., Sun, D., Morey, R. D., & Iverson, G.
keyboard in skilled typists. Memory & Cognition, 38, 474 – 484. http:// (2009). Bayesian t tests for accepting and rejecting the null hypothesis.
dx.doi.org/10.3758/MC.38.4.474 Psychonomic Bulletin & Review, 16, 225–237. http://dx.doi.org/10
Logan, G. D. (1979). On the use of a concurrent memory load to measure .3758/PBR.16.2.225
attention and automaticity. Journal of Experimental Psychology: Human Salthouse, T. A. (1986). Perceptual, cognitive, and motoric aspects of
Perception and Performance, 5, 189 –207. http://dx.doi.org/10.1037/ transcription typing. Psychological Bulletin, 99, 303–319. http://dx.doi
0096-1523.5.2.189 .org/10.1037/0033-2909.99.3.303
Logan, G. D. (1982). On the ability to inhibit complex actions: A stop- Shaffer, L. H. (1976). Intention and performance. Psychological Review,
signal study of typewriting. Journal of Experimental Psychology: Hu- 83, 375–393. http://dx.doi.org/10.1037/0033-295X.83.5.375
man Perception and Performance, 8, 778 –792. http://dx.doi.org/10 Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human
.1037/0096-1523.8.6.778 information processing: II. Perceptual learning, automatic attending, and
Logan, G. D. (1988). Toward an instance theory of automatization. Psy- a general theory. Psychological Review, 84, 127–190. http://dx.doi.org/
chological Review, 95, 492–527. http://dx.doi.org/10.1037/0033-295X 10.1037/0033-295X.84.2.127
.95.4.492 Siegler, R. S. (1987). The perils of averaging data over strategies: An
Logan, G. D. (1994). Spatial attention and the apprehension of spatial rela- example from children’s addition. Journal of Experimental Psychol-
tions. Journal of Experimental Psychology: Human Perception and Perfor- ogy: General, 116, 250 –264. http://dx.doi.org/10.1037/0096-3445
mance, 20, 1015–1036. http://dx.doi.org/10.1037/0096-1523.20.5.1015 .116.3.250
Logan, G. D. (2003). Simon-type effects: Chronometric evidence for Snyder, K. M., Ashitaka, Y., Shimada, H., Ulrich, J. E., & Logan, G. D.
keypress schemata in typewriting. Journal of Experimental Psychology: (2014). What skilled typists don’t know about the QWERTY keyboard.
Human Perception and Performance, 29, 741–757. http://dx.doi.org/10 Attention, Perception, & Psychophysics, 76, 162–171. http://dx.doi.org/
.1037/0096-1523.29.4.741 10.3758/s13414-013-0548-4
Logan, G. D., & Crump, M. J. C. (2009). The left hand doesn’t know what Snyder, K. M., & Logan, G. D. (2013). Monitoring-induced disruption in
the right hand is doing: The disruptive effects of attention to the hands skilled typewriting. Journal of Experimental Psychology: Human Per-
in skilled typewriting. Psychological Science, 20, 1296 –1300. http://dx ception and Performance, 39, 1409 –1420. http://dx.doi.org/10.1037/
.doi.org/10.1111/j.1467-9280.2009.02442.x a0031007
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2099

Snyder, K. M., Logan, G. D., & Yamaguchi, M. (2015). Watch what you tion of typing skill. Journal of Experimental Psychology: Learning,
type: The role of visual feedback from the screen and hands in skilled Memory, and Cognition, 40, 1713–1732. http://dx.doi.org/10.1037/
typewriting. Attention, Perception, & Psychophysics, 77, 282–292. xlm0000026
http://dx.doi.org/10.3758/s13414-014-0756-6 Yamaguchi, M., & Logan, G. D. (2014b). Pushing typists back on the
Sternberg, S., Knoll, R. L., & Turock, D. L. (1990). Hierarchical control in learning curve: Revealing chunking in skilled typewriting. Journal of
the execution of action sequences: Tests of two invariance properties. In Experimental Psychology: Human Perception and Performance, 40,
M. Jeannerod (Ed.), Attention and performance XIII (pp. 3–55). Hills-
592– 612. http://dx.doi.org/10.1037/a0033809
dale, NJ: Erlbaum.
Yamaguchi, M., Logan, G. D., & Li, V. (2013). Multiple bottlenecks in
Tapp, K. M., & Logan, G. D. (2011). Attention to the hands disrupts skilled
hierarchical control of action sequences: What does “response selection”
typewriting: The role of vision in producing the disruption. Attention,
Perception, & Psychophysics, 73, 2379 –2383. http://dx.doi.org/10 select in skilled typewriting? Journal of Experimental Psychology: Hu-
.3758/s13414-011-0208-5 man Perception and Performance, 39, 1059 –1084. http://dx.doi.org/10
William, L. B., & Harter, N. (1899). Studies of the telegraphic language. .1037/a0030431
The acquisition of a hierarchy of habits. Psychological Review, 6, Zamm, A., Pfordresher, P. Q., & Palmer, C. (2015). Temporal coordination
345–375. http://dx.doi.org/10.1037/h0073117 in joint music performance: Effects of endogenous rhythms and auditory
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Yamaguchi, M., & Logan, G. D. (2014a). Pushing typists back on the feedback. Experimental Brain Research, 233, 607– 615. http://dx.doi
This document is copyrighted by the American Psychological Association or one of its allied publishers.

learning curve: Contributions of multiple linguistic units in the acquisi- .org/10.1007/s00221-014-4140-5

Appendix
Subject Recruitment Advertisements on SONA System

Standard Typing Study perform a series of typing tasks in which you are to type a word
or letter as quickly and as accurately as you can. This study
ⴱⴱⴱⴱⴱⴱⴱⴱⴱ
PLEASE ONLY SIGN UP FOR THIS STUDY IF YOU requires subjects to be capable of touch-typing using 10 fingers
MEET THE TYPING STYLE REQUIREMENTS. IF YOU DO at the conventional finger placement on the keyboard. The
NOT MEET THE TYPING STYLE REQUIREMENTS BELOW, finger assignments in the conventional finger placement are as
SIGN UP FOR “Paid Study - Nonstandard Typing Style” (A follows: left pinky: QAZ, left ring: WSX, left middle: EDC, left
SEPARATE STUDY). If you have any questions about the index: RFVTGB, right index: YHNUJM, right middle: IK, right
requirements, please email the researcher.ⴱⴱⴱⴱⴱⴱⴱⴱⴱ You will ring: OL, right pinky: P

Figure A1. Display of standard mapping used to confirm whether typists use the standard mapping (depicted)
or a nonstandard mapping. See the online article for the color version of this figure.

(Appendix continues)
 2100 LOGAN, ULRICH, AND LINDSEY

Figure A2. Blank keyboard used for the keyboard recall task.

Nonstandard Typing Study The July sun caused a fragment of black pine wax to ooze on the
ⴱⴱⴱⴱⴱⴱⴱⴱⴱ
velvet quilt.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

PLEASE ONLY SIGN UP FOR THIS STUDY IF YOU

This document is copyrighted by the American Psychological Association or one of its allied publishers.

MEET THE TYPING STYLE REQUIREMENTS. IF YOU DO

NOT MEET THE TYPING STYLE REQUIREMENTS BELOW, Paragraphs for Typing Speed and Accuracy Tests
SIGN UP FOR “Paid Study - Standard Typing Style” (A SEPARATE
STUDY). If you have any questions about the requirements, please 1. It is difficult to know how man ever managed large flocks
email the researcher.ⴱⴱⴱⴱⴱⴱⴱⴱⴱ You will perform a series of typing of sheep on the rough and hilly terrain of these areas
tasks in which you are to type a word or letter as quickly and as without the help of these wonderful dogs. The strains that
accurately as you can. This study requires subjects that type using a proved most adept at the specialized type of work required
NONCONVENTIONAL finger placement on the keyboard. The fin- were highly prized and selectively bred from. This pro-
ger assignments in the CONVENTIONAL finger placement are as duced the sort of collie we know today. From looking at
follows: left pinky: QAZ, left ring: WSX, left middle: EDC, left very old photographs, it is remarkable how little they have
index: RFVTGB, right index: YHNUJM, right middle: IK, right ring: changed in the last hundred years or so. It proves that the
OL, right pinky: P ⴱⴱⴱⴱWe are looking for participants that do not type
early flockmasters knew well the type of dog that was built
using the finger to key mapping listed aboveⴱⴱⴱⴱ
on the correct lines for the job it was intended to do.

Full-Alphabet Sentences 2. One other sphere where border collies are most successful
is in search and rescue. Dog handlers are required to go out
The quick brown fox jumps over the lazy dog.
and look for missing climbers and walkers. A lot of these
Six javelins thrown by the quick savages whizzed 40 paces people get lost in areas where sheep are grazed. Border
beyond the mark. collies have to range well ahead of the handlers to cover
the maximum amount of ground, so they must be tested for
The public was amazed to view the quickness and dexterity of their trustworthiness with sheep before training starts. It
the juggler. does show what an adaptable breed the border collie is, in
that it can be taught to ignore an animal that it has been
We quickly seized the black axle and just saved it from going
specifically bred to herd. Border collies are becoming more
past him.
and more popular for this purpose.
A mad boxer shot a quick, gloved jab to the jaw of his dizzy
opponent. 3. A border collie from the correct source can be a charming
pet. However, dogs bred from a strong working line can
Whenever the black fox jumped the squirrel gazed suspi- become very frustrated and destructive if they find them-
ciously. selves in an environment where there is nothing for them to
do. The job is the border collie’s main reason for living.
The job requires extra pluck and zeal from every young wage
The desperate need to work is slightly diluted in certain
earner.
lines of border collies that are bred for the show ring. It is
While making deep excavations we found some quaint bronze important to remember that, although a border collie is
jewelry. usually quite happy to be a loving pet, he will need plenty
of exercise, and preferably some occupation for his very
Six big juicy steaks sizzled in a pan as five workmen left the quarry. able brain.

(Appendix continues)
 FITTS’ LAW VERSUS HICK’S LAW IN SKILLED TYPING 2101

Paragraph Speed Paragraph Accuracy

90
A 100
B
95
80

Words per Minute

90

Percent Correct
85
70

NS/NS 80 NS/NS
60 NS/S NS/S
75
S/S S/S
70
50
65

40 60
Visible Blank Covered Visible Blank Covered
Keyboard Keyboard

Word Priming Keyboard Recall

C D
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

900 14
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Mean Response Time in ms

12

800

Number Recalled
10
NS/NS Primed Correct NS/NS
8
NS/NS Control Correct NS/S
700
NS/S Primed 6 Correct S/S
NS/S Control Error NS/NS
S/S Primed 4 Error NS/S
600
S/S Control Error S/S
2

500 0
First Middle Last 1 2 3 4
Probe Posi; on in Word 20 sec Epoch

Hand Cuing Hand Cuing

1600
E 350
F
Mean Interkeystroke Interval in ms
Mean Response Time in ms

1400 300
1200
250
1000
200
800
NS/NS 150 NS/NS
600
NS/S NS/S
100
400 S/S S/S

200 50

0 0
Bimanual Unimanual Bimanual Unimanual Bimanual Unimanual Bimanual Unimanual
Whole Hand Whole Hand
Cuing Condi=on Cuing Condi?on

Figure A3. Mean performance by actual standard (S/S; N ⫽ 10), nonstandard standard (NS/S; N ⫽ 14), and
actual nonstandard (NS/NS; N ⫽ 24) typists on paragraph typing (Panels A, B), word priming (Panel C),
keyboard recall (Panel D), and hand cuing (Panels E, F). The NS data in previous figures are NS/NS data here.
The S data were a mixture of the NS/S and S/S data here. Error bars are SEMs.

Words for Word Priming Task flyer, foamy, forge, forty, frail, frame, fudge, ghost, gipsy, gland,
Abide, about adobe, adore, agent agile, agony, alter, amber, glare, glide, glint, gloat, glove, graft, grape, grasp, grate, groan,
amour, amuse, angel, anvil, arise, arose, aside aspen awful, baron, guide, harsh, hasty, haunt, heady, heart, hinge, hotel, house, hyena,
baton, beach, beard, begin, begun, blame blank, block, blush, ideal, image, inert, irony, jerky, joint, juicy, jumpy, knelt, knife,
bonus, brave, brawl, bread, bride, brush, brute, budge, built, bulge, laden, laugh, learn, light, liken, liner, liver, lodge, lofty, logic,
bunch, buxom, cadet, cameo, canoe, caper, chafe, chalk, cheat, lover, lower lucid, lucky lymph, lynch, magic, maple, maybe,
chief, chunk, churn, cigar, claim, clean, cobra, comet, comic, midst, milky, mirth, molar, money, month, munch, musty, north,
corps, crash, crate, crawl, crazy, crush, daunt, deity, delta, devil, noted, novel, nudge, orbit, other, paste, pasty, patch, pause, piano,
dimly, ditch, douse, drape, dream, dwarf, empty, envoy, extra, pilot, pique, plant, plate, poker, polka, porch, prime, prize, pulse,
exult, fairy, field, fiend, filth, first, flame, flesh, float, flung, flush, punch, ranch, range, rangy, reach, react, relax, relic, rider, ridge,

(Appendix continues)
 2102 LOGAN, ULRICH, AND LINDSEY

rifle, rival, roast, rough, rusty, sable, sandy, saucy, scald, scamp, LRLR: both, clap dock, form, sick, wish
serum, shack, shaft, shaky, shank, shave, shirt, shock, shone,
snore, shown, since, sixth, sixty, slain, slink, smack, smear, smite, LRRL: find, glow, role, song, suit, tune
snare, sneak, snort, solar, space, spark, spent, spoil, spoke, spray,
spurt, squat, stain, stake, stale, stalk, stamp, stand, stare, stern, RLLR: harm, item, mash, navy, park, peak
stick, story, stung, surge, swarm, swine, swing, syrup, thick, thine,
think, thong, threw, throb, thumb, thump, tough, trace, track, trade, RLRL: hair, idle, laid, land, melt, pant
train, twirl, unfit, untie, vague, value, vocal, vomit, waste, watch,
RRLL: hide, just, live, mice node, poet
waxen, weigh, whack, whisk, whole, width, wield, winch, windy,
women, would, wreck, wrest, write, wrong, wrote, yearn. RRRR: holy, jump, link, milk, plum, pony

Words for Hand Cuing Task

Received January 31, 2016
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

LLLL: brag, cage, date, fast, gear, star

Revision received May 2, 2016
This document is copyrighted by the American Psychological Association or one of its allied publishers.

LLRR: bail, crop, echo, grin, sail, walk Accepted May 13, 2016 䡲