Bronfenbrenner, U. (1977). Toward an experimental ecology of human development.

American
Psychologist. 513-531. Retrieved from http://jmsw.org/hdf/facultystaff/Tudge/
Bronfenbrenner1977.pdf (10 p.)
Toward an Experimental Ecology
of HumanDevelopment

URIE BRONFENBRENNER Cornell University

ABSTRACT: A broader approach to research in hu- in research, but often with indifference to or open
j man development is proposed that focuses on the pro- rejection of rigor. In its more extreme manifesta-
\ gressive accommodation, throughout the life span, tions, this trend has taken the form of excluding
between the growing human organism and the changing the scientists themselves from the research process.
environments in which it actually lives and grows. For example, one major foundation has recently
\ The latter include not only the immediate settings
stated as its new policy that, henceforth, grants
containing the developing person but also the larger
for research will be awarded only to persons who
social contexts, both formal and informal, in which
these settings are embedded. In terms of method, the are themselves the victims of social injusticeA
approach emphasizes the use of rigorousj^d^igned Other, less radical expressions of this trend in-1
exp_erjments, both naturalistic and contrived, beginning volve reliance on existential approaches in which 1
in the early stages of the research process. The chang- "experience" takes the place of observation and I
ing relation between person and environment is con- analysis is foregone in favor of a more personalized I
ceived in systems terms. These systems properties and direct "understanding" gained through inti- \
are set forth in a series of propositions, each illus- mate involvement in the field situation. More, N.
trated by concrete research examples. common, and more scientifically defensible, is an /"
emphasis on naturalistic observation, but with the /
This article delineates certain scientific limitations stipulation that it be unguided by any hypotheses i
in prevailing approaches to research on human de- formulated in advance and uncontaminated by V
velopment and suggests broader perspectives in structured experimental designs imposed prior to /
theory, method, and substance. The point of de- data collection.
parture for this undertaking is the view that, espe-
cially in recent decades, research in human de- This article represents a synthesis and further develop-
ment of ideas originally presented by the author in two
velopment has pursued a divided course, with each addresses at successive annual meetings of the American
direction tangential to genuine scientific progress. Psychological Association. The first was a presidential
To corrupt a contemporary metaphor, we risk be- address to the Division of Personality and Social Psy-
chology in 1974; the second was an invited Master Lec-
ing caught between a rock and a soft place. The ture in 197S.
rock is rigor, and the soft place relevance. As I The article grew out of work carried out by the author
have argued elsewhere (Bronfenbrenner, 1974; as a Belding Fellow of the Foundation for Child Develop-
ment. Appreciation is expressed to the Foundation and
Note 1), the emphasis on rigor has led to experi- its staff, in particular to Orville Brim and Heidi Sigal.
ments that are elegantly designed but often lim- The author is also indebted to the following colleagues for
ited in scope. This limitation derives from the their constructive criticisms of earlier drafts of the manu-
script: Irwin Altman, Melvin Kohn, Eleanor Maccoby,
fact that many of these experiments involve situa- Rudolf Moos, John Weisz, and Sheldon White.
tions that are unfamiliar, artificial, and short-lived Requests for reprints should be sent to Urie Bronfen-
and that call for unusual behaviors that are difficult brenner, Department of Human Development and Family
Studies, Cornell University, Martha Van Rensselaer Hall,
to generalize to other settings. From this per- Ithaca, New York 148S3.
1
spective, it can be said that much of contemporary In a recent survey of all studies in child development
developmental psychology is the science of the (N = 902) published between 1972 and 1974 in three
prominent research journals (Child Development, Devel-
strange behavior of children in strange situations opmental Psychology, Journal of Genetic Psychology),
with strange adults for the briefest possible periods Larson (Note 2) found that 76% of all the investigations
of time.* had employed the experimental laboratory paradigm; the
next highest category was research using pencil-and-paper
Partially in reaction to such shortcomings, other techniques (17%); observational studies were in lowest
workers have stressed the need for social relevance place (8%).

AMERICAN PSYCHOLOGIST • JULY 1977 • 513

 The orientation proposed here rejects both the pects of the environment beyond the immediate
implied dichotomy between rigor and relevance situation containing the subject. .._-^
and the assumed incompatibility between the re- Specifically, in this essay, I propose first an
quirements of research in naturalistic situations expansion and then a_jcp_n_yergence Of both the
and the applicability of structured experiments at naturalistic and the experimental approaches—
an early stage in the scientific process. Specifi- more precisely, an expansion and convergence in
cally, it rejects as spurious the argument that, be- the theoretical conceptions of the environment that
cause naturalistic observation preceded experimen- underlie each of them. I refer to this evolving sci-
tation in both the physical and biological sciences, /entific perspective as the ecology o] human devel-
this progression is necessarily the strategy of opment. The major dimension-Tof this perspective
choice in the study of human behavior and devel- ''are outlined below.
opment. Such an interpretation mistakes a his-
torical sequence for a causal one and represents
Terms oj Reference
yet another instance of the logical pitfalls inherent
in the ever-seductive post hoc, propter hoc infer- Let us begin with some definitions of focus, con-
ence. In my view, 20th-century science possesses text, and method.
research strategies that, had they been available
to the 19th-century naturalists, would have en- DEFINITION 1. The ecology of human develop-
abled them to leapfrog years of painstaking, ex- ment is the scientific study of the progressive, mu-
haustive description in arriving at a formulation of tual accommodation, throughout the life span, be-
tween a growing human organism and the changing
biographical principles and laws. This is not to
imply that taxonomy is not an essential scientific immediate environments in which it lives, as this
task but only to assert that a phase of comprehen- process is affected by relations obtaining within
sive observation, recording, and classification may and between these immediate settings, as well as
not be a necessary condition for making progress the larger social contexts, both formal and infor-
in the understanding of process, and that the early mal, in which the settings are embedded.
application of experimental paradigms may in fact The conception of the environment implicit in
lead to more appropriate taxonomies for achieving the foregoing definition is considerably broader
the requisite work of systematic description. and more differentiated than that found in psy-
There is yet another restriction unnecessarily chology in general and in developmental psychol-
imposed on the strategy of naturalistic observation, ogy in particular. Specifically:
particularly as applied to the human case by
its principal advocates—the ethologists (Jones, DEFINITION 2. The ecological environment is
1972; McGrew, 1972) and the psychological ecol- conceived topologically as a nested arrangement of
ogists of the Kansas school (Barker & Schoggen, structures, each contained within the next. (For
1973; Barker & Wright, 1954). Both groups have the purpose of describing these successive levels, I
shall employ a terminology adapted from Brim
/ adapted to the study of human behavior a model
[1975].)
1 originally developed for the observation of sub-
1. A microsystem is the complex of relations be-
\human species. Implicit in this model is a con- [ tween the developing person and environment in an
icept of the environment that may be quite ade- I immediate setting containing that person (e.g.,
quate for the study of behavior in animals but that home, school, workplace, etc.). A>.S£liing is de-
fs hardly sufficient for the human case. Specifi- i fined as a place with particular physical features
cally, it is limited to the immediate, concrete set- f in which the participants engage in particular ac-
ting containing the living creature and focuses on tivities in particular roles (e.g., daughter, parent,
the observation of the behavior of one or, at most, teacher, employee, etc.) for particular periods of
two beings at a time in only one setting. As I time. The factors of place, time, physical fea-
shall argue below, the understanding of human de- tures, activity, participant, and role constitute the
velopment demands going beyond the direct ob- elements of a setting.
servation of behavior on the part of one or two In psychological research, especially in the lab-
persons in the same place; it requires examination oratory, these elements are often given short shrift.
of multiperson systems of interaction not limited In particular, roles other than those of experi-
to a single setting and must take into account as- menter and subject that might in fact be operative

514 • JULY 1977 • AMERICAN PSYCHOLOGIST

 for the participants are disregarded, and behavior practice in everyday life. To give a formal
is examined primarily in terms of process (e.g., definition:
modes of interaction, reinforcement schedules, re-
sponse rates) rather than content (e.g.,_thfi-*iature 4. Ajm&crnsyntfm refers to the overarching in-
and purpose of the task). So that this substantive stitutional patterns of the culture or subculture,
aspect is not overlooked,T^tts»-the_tejm activity such as the economic, social, educational, legal, and
rather than behavior to identify this essential fea- political systems, of which micro-, meso-, and exo-
ture of the microsystem. systems are the concrete manifestations. Macro-
systems are conceived and examined not only in
2. A mesosystem comprises the interrelations structural terms but as carriers of information and
/ among major settings containing the developing ideology that, both explicitly and implicitly, endow
*-. person at a particular point in his or her life. meaning and motivation to particular agencies, so-
Thus, for an American 12-year-old, the mesosys- cial networks, roles, activities, and their interrela-
tem typically encompasses interactions among fam- tions. What place or priority children and those
4ly, school, and peer group; for some children, it responsible for their care have in such macrosys-
might also include church, camp, or workplace, al- tems is of special importance in determining how a
though the last would be less common in the child and his or her caretakers are treated and in-
United States than in some other societies. In teract with each other in different types of settings.
sum, stated succinctly, a mesosystem is a system of
microsystems. Especially in its formal properties, the foregoing
3. An exosystem is an extension of the meso- conception of the environment, as well as the dy-
system embracing other specific social structures, namic relation between person and situation im-
both formal and informal, that do not themselves plied in the definition of the ecology of human de-
contain the developing person but impinge upon or velopment, draws heavily on the theories of Kurt
encompass the immediate settings in which that Lewin (1935, 1936, 1948, 1951). Indeed, thisN
person is found, and thereby influence, delimit, or article may be viewed as an attempt to provide ]
even determine what goes on there. These struc- psychological and sociological substance"to Lewin's^x
tures include the major institutions of the society, brilliantly conceived topological territories.
both deliberately structured and spontaneously Having outlined' the structure of the ecological
evolving, as they operate at a concrete local level. environment, we are in a position to examine a
They encompass, among other structures, the world construct often alluded to in recent discussions of
of work, the neighborhood, the mass media, agen- developmental research—ecological validity^ Al-
cies of government (local, state, and national), the though this term, as yet, Tias no accepted defini-
distribution of goods and services, communication tion, one can infer from discussions of the topic a
and transportation facilities, and informal social common underlying conception: An investigation^
networks. is regarded as ecologically valid if it is carried out
in a naturalistic setting and involves objects and
A macrosystem differs in a fundamental way activities from everyday life. Although originally '
from the preceding forms in that it refers not to attracted to this notion, upon reflection I have
the specific contexts affecting the life of a par- come to view it not only as too simplistic but
ticular person but to general prototypes, existing as scientifically^ unsound on several counts. First.
in the culture or subculture, that set the pattern while I agree wholeheartedly with the desirability
for the structures and activities occurring at the of extending research activities beyond the labora-
concrete level. Thus, within a given society, one tory, I question the seemingly automatic grant of
school classroom looks and functions much like scientific legitimacy to a research effort merely
another. The same holds true for other settings on the basis of its being conducted in a real-life
and institutions, both informal and formal. It is situation. Even more arbitrary, however, is the
as if all were constructed from the same_Jjlue- converse implication that any investigation carried
Rrints. These "blueprints"are the_maciasystems. out in a nonnaturalistic setting is necessarily eco-
Some actually exist in explicit form as recorded logically invalid, and thereby scientifically suspect
• laws, regulations, and rules. But most macrosys- on purely a priori grounds. Surely, this is to pre-
tems are informal and implicit—carried, often un- judge the issue. Moreover, the term ecological
wittingly, in the minds of the society's members validity as it is currently used has no logical rela-
as ideology made manifest through custom and tion to the classical definition of validity—namely,

AMERICAN PSYCHOLOGIST • JULY 1977 • 515

 the extent to which a research procedure measures valid or invalid on a priori ground*y Thus, de-
'what it is supposed to measure. Indeed, there is pending on the problem, the laboratory may be
a basic conflict in the theoretical assumptions un- an altogether appropriate setting for an investiga-
derlying the two definitions. In the classical con- tion and certain real-life environments may be
ception, validity is ultimately determined by the highly inappropriate. Suppose, for example, one
nature of the problem under investigation. In is interested in studying the interaction between
contrast, ecological validity, as presently defined, mother and child when the child is placed in a
is apparently determined once and for all by the strange and unfamiliar situation. Clearly the lab-
setting in which the study is being conducted, oratory approximates this condition far better than
without regard to the question being investigated. the home. Conversely, if the focus of inquiry is
Surely, in any research endeavor this last consid- the modal pattern of parent-child activity pre-
eration must be the most decisive in assessing vailing in the family, observations confined to the
validity of whatever kind. laboratory can be misleading. As I have docu-
At the same time, implicit in current concerns mented elsewhere in greater detail (Bronfenbren-
with ecological validity is another principle that ner, in press), patterns of parent-child interaction
can no longer be disregarded in the light of avail- in the laboratory are substantially and systemati-
able evidence. This is the proposition that the cally different than those in the home. Specifi-
properties of the environmental context in which cally, so far as young children are concerned, the
research is carried out influence the processes that results indicate that the strangeness of the labora-
take place within that context and thereby affect tory situation tends to increase anxiety and other
the interpretation and generalizability of the re- negative feeling states and to decrease manifesta-
search findings. tions of social competence (Lamb, 1976b; Ross,
I have therefore sought to formulate a definition Kagan, Zelazo, & Kotelchuck, 1975; Lamb, Note
of ecological validity that takes both of these prin- 3). Possibly in response to this reaction of the
ciples into account. Once this task became clear, child, parents tend to exhibit more positive inter-
it was not difficult to achieve. All that was re- i actions toward their children in the laboratory
quired was a logical extension of the classical ' than in the home (Schlieper, 1975; Shalock, 1956;
definition of validity. As traditionally formulated, Belsky, Note 4). In addition, Lamb (1976b;
this definition is limited in focus, applying only Note 3) reported that the tendency of the infant
to the measurement procedures employed in re- at home to display more affiliative behaviors (e.g.,
search operations. The definition of ecological looking, smiling, reaching, vocalizing) toward the
validity proposed here expands the scope of the father than the mother was reversed in the labora-
original concept to include the environmental con- tory. Moreover, consistent with the arguments of
text in which the research is conducted. Sroufe (1970) and Tulkin (1972) that the lab-
oratory is especially likely to be an anxiety-arous-
DEFINITION 3. Ecological validity refers to the ing situation for lower-class families, Lamb found
extent to which the environment experienced by socioeconomic differences in father-infant inter-
the subjectzJitt- a scientific investigation has the action favoring the middle class in the laboratory,
properties it is supposed or assumed to have by the 'whereas such differences had not been present in
investigator. ~~ I the home.
Two features of the foregoing definition deserve Again, the fact that research results obtained in
special comment. Fjxsi, the relevant features of the laboratory differ from those observed in the
/•the environment include not only its objective home cannot be interpreted as evidence for the
/ properties but also the way in which it is perceived superiority of one setting over the other, except
^-by the research subjects. This stipulation takes in relation to a specific research question. At the
cognizance of perhaps the only proposition in so- yvery least, such differences serve to illuminate the
cial science that approaches the status of an im- \special properties of the laboratory as an ecologi-
mutable law—W. I. Thomas's inexorable dictum: cal context. More importantly, they illustrate the
"If men define situations as real, they are real in as-yet-unexploited power of the laboratory as an
their consequences" (Thomas & Thomas, 1928, p. , ecological contrast for highlighting the distinctive
572). \features of other types of settings as they affect
Second, note that Definition 3 does not desig- behavior and development. From this point of
/ nate any particular kind of research setting as view, an ecological orientation increases rather

516 • JULY 1977 • AMERICAN PSYCHOLOGIST

 than reduces opportunities for laboratory research trol other sources of influence either by random
by pointing to new knowledge that can be achieved assignment (contrived experiment) or by matching
through close and continuing articulation between (natural experiment).
laboratory and field.
At a more general level, the comparison of I deliberately eschew the term quasi-experiment,
results obtained in laboratory and real-life settings typically employed in the research literaiure, be-
provides an illustration of the basic strategy cause it suggests a lower level of methodological
through which ecological validity can be demon- rigor, an implication I regard as unwarranted on \
strated or found wanting. As in the case of the strictly scientific grounds. As I shall endeavor to
definition of the concept, the method represents an show, there are instances in which a design ex-
I extension of the procedures employed for investi- ploiting an experiment of nature provides a more
gating validity in its classical form. Essentially, critical contrast, insures greater objectivity, and
the process is one of establishing construct validity permits more precise and theoretically significant
(Cronbach & Meehl, 19S5), in this instance by inferences—in short, is more elegant and consti-
testing the ecological theory underlying the re- tutes "harder" science—than the best possible con-
search operations—that is, the assumptions being trived experiment addressed to the same research
made about the nature and generalizability of the, question.
environment in which the research is being con- In other respects, of course, the definition has
ducted. For example, when a laboratory study a familiar ring. In keeping with the commitment
is regarded as representative of behavior elsewhere, to rigor affirmed at the outset, the main body of
evidence must be provided of an empirical relation the definition is a restatement of the basic logic
to similar activities in the other setting—in other of the experimental method. What may be chal-
(words, validation against an external ecological lenged about this formulation is not the procedure
I criterion, with the possibility of systematic diver- advocated but the timing and the target of its ap-
\gence explicitly taken into account. It should be plication. Specifically, I am proposing that ex- ,
recognized, moreover, that such divergence may periments be employed in the very first phases of I
take the form not merely of differences in average scientific inquiry, not for the usual objective of -*
response, but in the total pattern oj relationships, testing hypotheses (although this device is used as
/and in the underlying processes that they are pre- a means to an end) but for heuristic purposes— V
l^sumed to reflect. Some examples of substantial namely, to analyze systematically the nature of
shifts in pattern and process from one ecological the existing accommodation between the person-
context to another are cited further on in this and the. surrounding milieu.
article. The need for early experimentation derives from
/The foregoing discussion of ecological validity the nature of the problem under investigation.
leads directly to the principal methodological the- The "accommodation" or "fit" between person and
""sis of this exposition. As should be true of any environment is not an easy phenomenon to recog-
scientific endeavor, decisions on research design are nize. Here, looking is usually not enough. As
dictated by theoretical considerations. Thus, in Goethe wrote with his poet's prescience: "Was ist
;the present instance, given the complex conception das Schwerste von allem? Was dir das Leichteste
of the ecological environment in terms of inter- diinket, mit den Augen zu sehen, was vor den
I dependent, nested systems, the question arises as Augen dir liegt." (What is the most difficult of
\ to how these interdependencies can be investigated all? That which seems to you the easiest, to see
empirically. I shall argue that a strategy espe- with one's eyes what is lying before them.)
If looking is not enough, what is one to do?
cially well_suited,,fQr this purpose, from the earliest
How can the observer quicken his or her sensi-
stages of research forward, is an ecological experi-
tivity to the critical features of the observed? The
ment, defined as follows:
answer to this question was given me more than
DEFINITION 4. An ecological experiment is an 30 years ago, long before I was ready to appre-
effort to investigate the progressive accommodation ciate it, by my first mentor in graduate school,
between the growing human organism and its en- Walter Fenno Dearborn. In his quiet, crisp New
vironment through a systematic contrast between England accent, he once remarked: "Bronfenbren-
two or more environmental systems or their struc- ner, if you want to understand something, try to
tural components, with a careful attempt to con- change it." And whether one studies change by

AMERICAN PSYCHOLOGIST • JULY 1977 • 517

 deliberately altering conditions in a contrived ex- single variable at a time and attempts to "con-
periment or by systematically exploiting an "ex- trol out" all others, in ecological research the in-
periment of nature," the scientific purpose and vestigator seeks to "control in" as many theoreti-
effect are the same: To maximize one's sensitivity cally relevant ecological contrasts as possible
to phenomena through the juxtaposition of the within the constraints of practical feasibility and
similar but different constitutes the core of the rigorous experimental design. For only in this
experimental method and creates its magnifying way can one assess the generalizability of a phe-
.power. nomenon beyond a specific ecological situation and,
The case presented here for early and continuing equally significant from a developmental perspec-
application of experimental paradigms should not tive, identify the processes of mutual accommoda-
be misinterpreted as an argument against the tion between a growing organism and its changing
use of other methods, such as ethnographic de- surround. For instance, in studying socialization
scription, naturalistic observation, case studies, I strategies, one might do well to stratify the sample
field surveys, etc. Such strategies can provide i not only, as is commonly done, by social class, but
invaluable scientific information and insights. The \also by family structure and/or child-care setting
point being made is a positive one—namely, that (home versus day care). Such stratification in
the experiment plays a critical role in ecological terms of two or more ecological dimensions serves
investigation not only for testing hypotheses he scientifically useful function of providing a
but, at prior stages, for detecting and analyz- lystematically differentiated and thereby poten-
ing systems properties within the immediate set- ,iially sensitive grid that makes possible the detec-
ting and beyond. The special suitability of the Jtion and description of patterns of organism-en-
experiment for this purpose is highlighted by an vironment interactions across a range of ecological
adaptation of Dearborn's dictum to the ecological '-contexts. Moreover, given the extraordinary ca-
realm: // you wish to understand the relation be- pacity of the species homo sapiens to adapt to its
tween the developing person and some aspect of milieu, these patterns are more likely to be com-
his or her environment, try to budge the one, and plex than simple. To corrupt, only slightly, the
see what happens to the other. Implicit in this terminology of experimental design: In ecological
1
injunction is the recognition that the relation be- research, the principal main effects are likely to be/
tween person and environment has the properties interactions^ ~
of a system with a momentum of its own; the This brings us to the final and most challenging
only way to discover the nature of this inertia and requirement of a research model for investigating
its interdependencies is to try to disturb the exist- the ecology of human development: Namely, en-
ing balance. vironmental^ structures, and the processes taking
It is from this perspective that the primary pur- place within and between them, musT be viewed as
pose of the ecological experiment becomes not hy- interdependent and must be analyzed in systems
pothesis. testing but discovery—the identification terms. The specification of these interdependen-
of those systems properties and processes that af- cies constitutes a major task of the proposed ap-
fect, and are affected by, the behavior and de- proach. The rest of this article represents a be-
velopment of the human being. Moreover, if the ginning effort in this direction in the form of a
objective is the identification of systems proper- series of propositions outlining the requirements
ties, then it is essential that such systems proper- of an ecological model for research at each of the
ties not be excluded from the research design be- four successive levels stipulated in the conceptual
fore the fact by restricting observation to only one Ciramework of the environment. Each proposition^
setting, one variable, and one subject at a time. is accompanied by one or more examples of con- \
Human environments and—even more so—the ca- crete investigations—actual when available, hypo- I
pacities of human beings to adapt and restructure thetical when not—to illustrate the given require-/
these environments are so complex in their basic ment, either by demonstration or default.
organization that they are not likely to be cap- The reference to illustration by default reflects
tured, let alone comprehended, through simplistic the fact that for reasons already indicated, well-
unidimensional research models that make no pro- designed, ecological experiments are, as yet, not
vision for assessing ecological structure and varia- easy to find. In an effort to alter this state of
tion. Accordingly, in contrast to the classical affairs, I was fortunate in enlisting the support of
laboratory experiment in which one focuses on a the Foundation for Child Development (FCD) in

518 • JULY 1977 • AMERICAN PSYCHOLOGIST

 initiating a small-scale program of research grants Klaus, 197S; Hales, Note 5). Taking as their
and career development awards in the ecology of point of departure observations on animals reveal-
human development. The aim of the program is ing complex, species-specific patterns of mother-
to encourage scientific work and training in the neonate interaction immediately after delivery
systematic study of "the behavior and develop- (Rheingold, 1963), the investigators undertook to
ment of children, and those who care for them, in explore this phenomenon in the human case. Not-
the enduring environments in which they live." ing that prevailing hospital practices resulted in
A number of ecological experiments cited in this minimal opportunities for contact between mother
article were supported by grants from the FCD and newborn, the researchers modified the estab-
program.2 lished procedures so as to permit mothers to have
their naked infants with them for about an hour
Properties of the Microsystem shortly after delivery and for several hours daily
thereafter. Randomly assigned control groups ex-
RECIPROCITY perienced the usual routine in American hospitals
—"a glance at their baby shortly after birth, a
It is a sign of some progress that the first systems short visit six to 12 hours after birth for identifi-
property to which I call attention is one that cation purposes, and then 20- to 30-minute visits
many readers will recognize and applaud. In the for feeding every four hours during the day"
classical, psychological research model, whether in (Kennell et al., 1974, p. 173).
the laboratory or in the field, there were, and often
The reported results of these experiments strain
still are, only two parties—an experimenter, iden-
the credulity of the reader. One month after the
tified solely, and apparently still acceptably, as E,
brief extended contact at birth, the mothers in the
and another person equally informatively described
experimental group were more attentive and af-
as 5, the subject. The term subject is apt, for it
fectionate toward their babies and more solicitous
reflects the fact that with few exceptions, the
about their welfare (Klaus et al., 1972). Not only
process operating between E and 5 has been viewed
were these differences still in evidence at the end
as unidirectional; the experimenter presents the
of the 1st year, but 2 years later the mothers, in
- stimulus, and the subject gives the response.
speaking to their children, used significantly more
Nowadays, we all know that the process goes both
questions, adjectives, and words per proposition
ways. In more formal terms:
and fewer commands and content words than did
PROPOSITION 1. In contrast to the traditional, the control mothers.
unidirectional research model typically employed Finally, the most recent experiment in the series
in the laboratory, an ecological experiment must (Hales, Note 5) not only provides a much-needed
allow for reciprocal processes; that is, not only the replication of the initial studies in a larger sample
effect of A on B, but also the effect of B on A. (N = 60) but does so in a different cultural con-
This is the requirement of reciprocity. text (Guatemala) and with a more rigorous ex-
While the thesis that most behavior in social perimental design that permits pinning down the
heretofore unresolved issue of whether there exists1^
situations is reciprocal is generally accepted in
a critical period of susceptibility to extended con-'
principle, it is often disregarded in practice. As a tact between mother and infant. Hales clarified
striking case in point, we may consider a series of this issue by introducing two early-contact groups:
ingenious ecological experiments and follow-up one limited to 45 minutes immediately after de-
studies conducted by a group of investigators from livery and the other to an equal interval but be-
the Department of Pediatrics at Case Western Re- ginning 12 hours after the infant's birth. The
serve University (Kennell et al., 1974; Klaus, results were unequivocal. Only the mothers in the"^
Kennell, Plumb, & Zuehlke, 1970; Klaus et al., immediate contact group were affected. /
1972; Ringler, Kennell, Jarvella, Navojosky, &
RECOGNIZING THE FUNCTIONAL SOCIAL SYSTEM

2
Information about the program may be obtained by From an ecological perspective, even more re-
writing to Joyce Brainard, Administrative Aide, Program markable than the dramatic results reported in
on the Ecology of Human Development, Department of
Human Development and Family Studies, Cornell Uni- this series of experiments are the data they omit.
versity, Ithaca, New York 148S3. In none of the papers cited is there a single word

AMERICAN PSYCHOLOGIST • JULY 1977 • 519

 about the behavior of the infants, and all of the BEYOND THE DYAD
experimental effects are attributed entirely to the
mothers. Thus the investigators refer repeatedly The Case Western Reserve University experiments
A to a "maternal sensitive period" or "a special at- reflect the influence of the traditional laboratory
^ tachment period existing in the human mother" paradigm in still one other respect; they are lim-
(Klaus et al., 1972, p. 463; Kennell et al., 1974, ited to a two-person model. As previously noted,
p. 173). The principle of reciprocity, of course, the classical psychological experiment allows for
raises the question of whether the distinctive be- only two participants: E and S. Even in those
havior of the mothers in the experimental group researches that take into account the activities of
might not have occurred, at least in part, as a more than two persons in differing roles, the be-
response to a sequence of activities initiated by havior of each is usually analyzed separately and
(the developing infant and reciprocated by the interpreted as an independent effect. As a case in
mother in a progressively evolving pattern of social point, we may consider recent work on father-
interaction. Regrettably, the possibility remains infant interaction.3 Much of this research treats
unexplored. In keeping with the classical experi- the behavior of the father, and any reaction it
mental model, the focus of scientific attention in may evoke in the child, in exclusively class-theo-
these studies was limited to the subjects of the retical terms (Lewin, 1935) as attributable entirely
research, who, in this instance, were not the chil- to the father, without regard to the possibility that
dren but the mothers. The omission is all the both the father's action and the child's responses
more remarkable given the fact that the infants may be influenced by the mother—her presence or
j were always present in the research situation and, absence and the possible effect of her behavior on
what is more, that all of the mothers' behavior the interaction of the father with the child. I refer
xbeing observed was directed toward them.
/to this kind of indirect influence as a second-order
Taken as a whole, this series of experiments on effect. To state the issue in prepositional form:
the effects of early, extended mother-infant con- (PROPOSITION 3. In contrast to the conventional
tact provides an excellent illustration of several tdyadic research model, which is limited to assessing
( defining properties of an ecological research model, \the direct effect of two agents on each other, the
both by demonstration and default. On the one \ design oj an ecological experiment must take into
hand, the work constitutes a clear instance of eco- I account the existence in the setting of systems that
logically valid experimentation focused directly on I include more than two persons (N + 2 systems).
developmental processes. Moreover, it presents an I Such larger systems must be analyzed in terms oj
example par excellence of how experimental in- J all possible subsystems (i.e., dyads, triads, etc.)
tervention can bring to light critical features of an \ and the potential second- and higher order effects
\iecological process hardly likely to be identified associated with them.
\through straightforward naturalistic observation in -<*

\he unaltered, existing setting. On the other hand, It will be observed that this proposition repre-
If the research represents a striking case of failure sents, in effect, an extension and further specifica-
\to take into account the total social system actu- tion of Proposition 2 as applied to a system in-
jally functioning in the given situation. volving more than two persons. To illustrate the
'T This dramatic lacuna in an otherwise impres- application of the principle, let us turn to three
sive series of studies gives rise to the next propo- recent studies of parent-child interaction that, ex-
^-sition. plicitly or implicitly, employed a three-person
model. Parke (1976) and his co-workers observed
^ PROPOSITION 2. An ecological experiment re- both parents with their newborns in a hospital
quires recognition of the social system actually setting to determine what effect each parent had
operative in the research setting. This system will on the other's interactions with the infant. In
typically involve all of the participants present, not each case,
excluding the experimenter. This is the require-
The presence of the spouse significantly altered the be-
ment oj recognizing the totality of the functional havior of the other parent, specifically, both father and
social system in the setting. mother expressed more positive affect (smiling) toward

This proposition becomes increasingly important

as one moves on to a consideration of systems in- 8
For a comprehensive review of this literature, see Lamb
volving more than two persons. (1975, 1976c).

520 • JULY 1977 • AMERICAN PSYCHOLOGIST

 their infant and showed a negative level of expectation reveals the role of the parent as a source of
when the other parent was present . . . . These results
indicate that parent-infant interaction patterns are modi-
security .for the child and, in terms of a three-
fied by the presence of another adult; in turn, the impli- person model, as a catalyst for the child's inter-
cation is that we have assumed prematurely that parent- action with the environment, including other, un-
infant interaction can be understood by our sole focus on
the parent-infant dyad alone. (Parke, 1976, pp. 33-34)
familiar persons. Thus, in all the "strange-situa-
tion" experiments, the mother's presence in the
Support for Parke's conclusion comes from a laboratory reduces the child's anxiety and resist-
study by Pederson (Note 6), in which the second- ance to the "stranger." Indeed, especially when
order effect is somewhat more remote but equally, the experiments are carried out in the home (e.g., \
if not more, consequential. This investigator ex- Lamb, 1976b; Note 3), infants in the company of '
amined the influence of the husband-wife relation- their parents look and smile at the stranger more
ship (assessed through interview) on mother-in- often than at their mothers.
fant interaction in a feeding context (as observed The mother-father-child triad is of course not\
in the home). His results are summarized as the only three-person system of developmental im- ,
follows: portance within a family. Other common com-
binations include two siblings and a parent; par-
f The husband-wife relationship was linked to the mother-
infant unit. When the father was supportive of the ent, child, and grandparent, aunt, or uncle, etc. I
', mother . . . she was more effective in feeding the baby have been able to find only one study of the effect
• . . . . High tension and conflict in the marriage was asso- i of the impending arrival of a second child on the
-V dated with more inept feeding on the part of the mother.
' (Pederson, Note 6, p. 6) parental treatment of the first, that done by a
prescient leader in the field over a quarter of a
• Pederson also found that the developmental century ago (Baldwin, 1947). Other triadic com-
status of the infant, as measured on the Brazelton binations in the family apparently remain wholly
\ scale, was inversely related to the degree of tension unexplored and hence constitute- a promising eco-
and conflict in the marriage. Consistent with the logical domain for developmental research.
present Proposition 1, he notes appropriately that The application of a three-person model to a
the causal direction could go both ways.4 developmental context outside the home is likewise
Pederson's results indicate that this second- a rarity. There does exist one elegant study,
order effect can have inhibitory as well as facili- however, documenting a second-order effect in a
tative impact. Indeed, Lamb (1976a) suggests, classroom setting. Seaver (1973) ingeniously ex-
/• on the basis of experimental findings, that as the ploited an "experiment of nature" to investigate
infant gets older (i.e., 18 months) the presence of the controversial phenomenon of induced teacher
the second parent may reduce rather than increase expectancies (Rosenthal & Jacobson, 1968). Sea-
parent-child interaction. The experiment, how- ver examined differences in the academic achieve-
ever, was carried out in the laboratory. As pre- ment of elementapy-school pupils with older sib-
viously noted, a number of comparative studies lings who had had the same teacher and had per-
(including one by Lamb) have shown that both formed either exceptionally well or exceptionally
parents and children behave rather differently in poorly. Children taught by teachers who had not
laboratory and real-life situations; hence it would instructed the older siblings served as controls. In
be important to replicate Lamb's experiment in a contrast to earlier studies, which had produced
home setting. inconsistent, weak, or questionable effects, the re>
When interpreted in an ecological perspective, suits of Seaver's natural experiment gave substan-J
however, the results of laboratory studies pro- tial support to the teacherjexpectancy hypothesis:
vide an important complement to research carried As Seaver himself acknowledged, however, it was
out in real-life environments. For example, if not clear who was the mediator of the observed
the laboratory is viewed as what it almost in- effect. Were the teacher's expectations changed be-
variably is for a young child—namely, a "strange cause of her prior experience with the older sibling,
situation" (Ainsworth & Bell, 1970)—it clearly or did the younger sibling evoke a different re-
sponse from the teacher because of the younger
child's expectations created by the older sibling
*The reciprocal interaction between the marital and or by the parents (based on their previous ac-
the parent-child dyads in a three-person system is demon- quaintance with the teacher), or both? The re-
strated even more dramatically in Hetherington's (Note 7)
comparative study of divorced versus two-parent families. maining ambiguity in interpretation testifies to the

AMERICAN PSYCHOLOGIST • JULY 1977 • 521

 importance of analyzing subsystems and higher members of their families—were also exposed to
order effects as stipulated in Proposition 3. traffic noise and, in all likelihood, affected by it.
The involvement of parents as intermediaries in If so, the possibility remains that the impairment
a process already involving two siblings and a of the children's auditory discrimination and ver-
teacher would of course escalate the system from '1 bal skills might have come about not only as a
a triad to a quintet, or, more generally, an TV + 3 j function of their own difficulties in hearing or
system. To my knowledge, no studies utilizing 1 sustaining attention in a noisy environment, but
such a model have been carried out within a | also because others around them were similarly af-
single setting, despite the fact that the modal Ifected and engaged less frequently in conversa-
American family with two parents and two chil- tions, in reading aloud, or in correcting their chil-
dren constitutes a readily available example. The dren's verbal utterances. No data are available
wide prevalence of this structure raises the ques- to demonstrate or disconfirm the existence of such
tion of the optimal size and form of systems for a second-order effect, but relevant information
fostering human development. could Tiave readily been obtained had the other
The evidence cited above suggests that as one participants in the setting been included in the
moves beyond the dyad, the resulting structures research design.
may offer possibilities for greater stability, mutual Similar considerations apply to research on the
assistance, complementarity, spelling each other effects of television. Almost all investigations in
off, and reinforcement, both directly and indirectly this area have been concerned with the direct im-
through third parties. Although the power of an pact of the program viewed by the child on his or
+ 3 system within a single context such as the her knowledge, attitudes, and behavior; indirect
home or school remains unknown, the paradigm influences through the modification of patterns of
can be applied to some researches that have been family life have scarcely been mentioned, let alone
carried out in multiple settings. Before turning investigated. In a review of research literature
/^to a consideration of this topic, however, we must bearing on this issue, Garbarino (1975) was able
\ take note of yet another source of higher order to identify only one investigation that dealt with
effects. the question explicitly and systematically. In a
field survey, Maccoby (1951) found that 78% of
INDIRECT IMPACT OF PHYSICAL FACTORS the respondents indicated no conversation occurred
during viewing, except at specified times such as
Environmental influences on development are of commercials, and that 60% reported that no ac-
course not limited to human beings. However, in tivity was engaged in while watching. On the
keeping with -the classic two-element research basis of her findings, Maccoby concluded:
model, these influences are usually thought of as
The television atmosphere in most households is one of
acting directly on the subject; the possibility of quiet absorption on the part of family members who are
- ' higher order effects operating indirectly has been present. The nature of the family social life during a
overlooked. The following are two examples. program could be described as "parallel" rather than in-
teractive, and the set does seem quite clearly to dominate
The first is provided by an elegant ecological family life when it is on. (p. 428)
study of the influence of apartment noise on hu-
man development (Cohen, Glass, & Singer, 1973). It is noteworthy that Maccoby's study was pub-
lished a quarter of a century ago and that, appar-
The investigators found that children living on the
lower floors of 3 2-story buildings near noisy traffic ently, no further research has been done on the
showed greater impairment of auditory discrimina- problem since that time. With the rapid growth
of television, and the television culture, in the
tion and reading achievement than a matched
. sample living in higher floor apartments. Cohen intervening years, the impact of the medium on
et al. viewed their study as a real-life counterpart family life has, in all probability, become both
to laboratory experiments demonstrating degrada- more pervasive and profound. The question of
tion of task performance as a direct aftereffect of how any resulting change in family patterns has,
exposure to noise. The two situations are not in turn, affected the behavior and development of
analogous, however, since the real-life setting in- children (i.e., the second-order effect) remains
cluded other persons besides the children selected completely unexplored.
as the subjects of the study. Moreover, these These and related studies lead to the following
other persons—the children's parents and other proposition:

522 • JULY 1977 • AMERICAN PSYCHOLOGIST