Patterns of Dominance and Imitation
in an Infant Peer Group
A. E. Russon and B. E. Waite
Department of Psychology, Glendon College, York University, Toronto,
Ontario
The present study explored whether model choice in infant peer imitation is related to
peer social dominance. Twelve 11-1Cmonth old infants in a stable infant daycare group
were videotaped systematically over a two-week period, providing six 20-minute focai
individual samples of free social activity each. Dominance-related peer encounters and
peer imitation were identified in the videotaped samples, then coded in behavioral
terms. Both rights of possession and agonism influenced dominance. Group dominance
structures of high, mid, and low rank subgroups were identiiled. Dominance was related
to peer imitation: Infants preferred higher rank over lower rank models. Possible processes
underlying this pattern are discussed. Results show that model choice is one way
In which peer experience could influence development; they also highlight the interconnectedness
rather than the independence of individual dimensions of infant peer
experience.
KEY WORDS: Social dominance; Imitation; Infant peer relationships; Infant peers;
Peers; Infant daycare.
E
xperience with peers during infancy, especially the sort of extensive
experience available through daycare, is now acknowledged
as an important contributor to social development (Apolloni and
Cooke 1975; Belsky and Steinberg 1978; Rubenstein and Howes
1979). Effects of infant daycare as compared with home rearing include
enhancement of social development and modification of social orientation.
Apolloni and Cooke suggested that peer imitation may be one mechanism
involved. Peer imitation could have an impact as early as one year of age,
when it becomes common (e.g., Brenner and Mueller 1982; Russon 1985).
Understanding the patterns in infant peer imitation may be important to
understanding the nature of its impact on social development.
A considerable amount is now known about patterns in the actions infants
select to imitate in their peers (e.g., Brenner and Mueller 1982; Bridges
Received June 20, 1989; revised May 10, 1990.
Address reprint requests to: A. E. Russon, Ph.D., Department of Psychology, Glendon College,
York University, 2275 Bayview Avenue, Toronto, Ontario, Canada M4N 3M6.
Ethology and Sociobiology 13: 55-73 (1991)
0 Elsevier Science Publishing Co., Inc., 1991
655 Avenue of the Americas, New York, NY 10010
0162-3095/91/$03.50
56 A. E. Russon and B. E. Waite
1933; McCabe and Uzgiris 1983; Mueller and DeStefano 1973; Mueller and
Lucas 1975). Like others, they prefer to imitate actions representing levels
of cognitive functioning similar to or slightly superior to their own. This is
not the only basis for patterning in imitation. Imitation is also a fundamentally
social process which is experienced as prosocial (Thelen and Kirkland
1976). In fact, Yando, Seitz, and Zigler (1978) have proposed that a major
social or interpersonal function exists for imitation in addition to the cognitive
or knowledge acquisition function traditionally studied. As yet, there
has been little empirical research on the social function of imitation (Nadel
1986).
Social factors, particularly model choice, would be expected to play a
particularly central role in imitation during infancy, when actions and their
actors are as yet poorly differentiated (Aronfreed 1969; Nadel 1986; Yando,
Seitz, and Zigler 1978; Uzgiris 1981). Factors likely to influence model choice
in infants are established interpersonal relationships with the model. With
peers, both dominance and affiliative or “friendship” relationships have
been shown to relate to model selection in older children (Hartrup and Coates
1970; Uzgiris 1981; Yando, Seitz, and Zigler 1978). In some form, both have
been identified in stable infant peer groups (Howes 1983; Strayer and Trudel
1984). In the current project we studied a stable infant daycare group in
order to determine whether and/or how one of these types of peer relationship,
dominance, is related to model choice in infant peer imitation.
THEORETICAL CONTEXT
There are no systematic empirical data on the interpersonal factors influencing
model choice in infant peers, although patterns can be predicted using
findings from other contexts. Experimental studies using adult models have
demonstrated that infants can and do differentiate models on the basis of
established relationships. Attachment to the model increases the likelihood
that an infant will imitate as well as attempt to imitate unusual actions
(McCabe and Uzgiris 1983; Yando, Seitz, and Zigler 1978). Suggestions
about the relationship factors related to model choice in peers come from
studies of older children and nonhuman primates. These have focused on
two types of relationship, friendship and dominance.
Friendships are affiliative relationships characterized, in mature form,
by mutuality, cooperation, and interpersonal sensitivity (Smollar and Youniss
1982). Imitation could serve in building, maintaining, and strengthening
friendship bonds because of its general effect of facilitating social relations.
It has been shown to serve as a means of establishing communication and
of strengthening attachment bonds (Nadel 1986). However, based on currently
available evidence, infant peer friendships are much more primitive
than the matures ones defined-they reduce to mutual preference, enjoyment,
reciprocity, and complementarity-and even these occur infrequently
Dominance and Imitation in Infant Peers 57
(Howes 1983). No empirical studies have yet explicitly investigated associations
between friendship and imitation in infant peers.
In contrast, the conceptual and methodological bases for studying infant
peer dominance are well established, and infant peer dominance relationships
appear to occur commonly (Bakeman and Brownlee 1982; Missakian
1980; Strayer and Trudel 1984). Consequently, although affiliative bonds are
expected to be important contributors to patterns in imitation, we focused
attention for this study on dominance.
The logic behind the expectation that imitation will be linked to dominance
is most elegantly articulated in Charlesworth’s (1988) presentation
of a general framework for studying the ontogeny of resource acquisition.
He considers resource acquisition to be a fundamental goal of behavior which
commonly leads to competition. There are a variety of behavioral approaches
to competing, of which Charlesworth has identified five: aggression,
manipulation, deception, intimidation, and cooperation. Both ontogeny
and ecology are important determinants of which approaches are used.
Of these approaches, aggression and therefore dominance are likely to
be common in stable infant peer groups. The peer group context per se is
likely to be intensely competitive because peers’ equality concentrates competition
on a limited number of resources. Because aggression appears when
resources are scarce, it is likely to emerge as a characteristic approach to
peer competition. This is even more likely in infant peers: Aggressive encounters
become common in infant peer groups from the end of the first
year, and because of infants’ limited capabilities, very few alternatives are
likely to have developed. When aggression is a central approach to resource
competition, dyadic dominance relations and group dominance hierarchies
commonly result (Silk 1987). They are considered to have adaptive value in
replacing disruptive social conflict with rules of priority as means of relating
competition (Charlesworth 1988). Relations showing dominance qualities
have been identified in infant peer groups by the end of the first year (Bakeman
and Brownlee 1982; Missakian 1980; Strayer and Trudel, 1984).
In stable groups, competition tends to become increasingly complex as
close interconnections are established among group members. One direction
that increasing complexity quickly takes in the peer context is towards generating
alternative, conditionally employed approaches to replace aggression
as at means of resource competition. Aggression is socially disruptive and
potentially dangerous. Individuals also learn that with some competitors,
especially more dominant ones, aggression is an ineffective means of competing.
The range of alternative approaches which can appear in infancy is
narrow because of infants’ limited behavioral and cognitive capacities; however,
two are possible-cooperation and imitation. Both emerge as prominent
themes in infant peer encounters near the end of the first year (e.g.,
Brenner and Mueller 1982; Mueller and Lucas 1975; Russon 1985).
Imitation could be very effective as an alternative means of competition
in the context of established dominance relations. Imitating dominant in-
58 A. E. Russon and B. E. Waite
dividuals could serve as a means of acquiring the dominant’s knowledge and
skills about competing effectively. Dominant individuals by definition succeed
often in competitions over resources; they therefore know where valuable
resources are and have the skills to acquire and retain them. In line
with this analysis, it has been shown that dominance draws attention (Abramovitch
1976) and that child and nonhuman primate imitators prefer models
of high dominance rank (e.g., Strayer 1981; Walters and Seyfarth 1987).
Competency principles further lead to the suggestion that imitators may
prefer models of similar or only slightly higher rank, because they tend to
be optimally more competent (Missakian 1980; Nadel 1986; Wishart 1986;
Yando, Seitz, and Zigler 1978). Because of its socially ingratiating qualities,
imitation could also be used as a means of gaining tolerance, acceptance,
or even power from higher status individuals, all of which could lead to
increased access to resources (e.g., Abramovitch and Grusec 1978; LaFreniere
and Charlesworth 1983; Nadel 1986; Walters & Seyfarth, 1987).
There is some evidence that the dominance-imitation link may hold in infant
peers, from a study of an infant chimpanzee dyad: In all incidents of peer
imitation, the model was the dominant partner (Russon, in press).
Although empirical research has shown that some form of dominance
is common in infant peer groups, its nature is disputed. Since debates range
over assessment, structure, and even basic definition, overviews of relevant
arguments and our own position are provided.
Dominance relationships are defined basically for dyads. Although
many formal definitions have been proposed (Walters and Seyfarth 1987),
the central concept of dominance is always asymmetry in outcomes of social
encounters-outcomes are regularly more advantageous for one partner than
the other (e.g., Daudelin 1980; Deag 1977; LaFreniere and Charlesworth
1983; Noe, de Waal, and van Hooff 1980). The term “relationship,” as used
here, refers to relatively enduring social patterns shown in stable consistencies
in partners’ encounters (Hinde 1979). The dominant is the partner
who consistently takes precedence, wins, or controls and limits the behavior
of the other; the latter is the subordinate. Within stable social groups of
many species, patterns commonly emerge among the dyadic dominance relationships
which generate dominance structures organizing the whole
group.
The most common basis for assessing dominance in infant peers has
been the process of agonism in dyads (Bakeman and Brownlee 1982; Missakian
1980; Strayer and Trudel 1984). However, the use of a strictly agonistic
basis derives from research on species such as chickens and cercopithecines,
and it is now known that the processes behind asymmetric
outcomes vary considerably between species (Walters and Seyfarth 1987).
Recent work on humans and apes by LaFreniere and Charlesworth (1983),
Noe, de Waal, and Van Hooff (1980), and Silk (1987) found that assessments
based narrowly on agonism did not adequately predict the asymmetric outcome
patterns seen in these species. In these species, asymmetry in out-
Dominance and Imitation in Infant Peers 59
comes can be determined by multiple processes rather than the single one
of dyadic agonism; these processes can be relatively independent but then
interact or compete to determine asymmetric outcomes. Examples include
coalitions and principles of rights of possession. Second, dominance supposedly
reduces detrimental social conflict. Thus, even if dominance relationships
are initially formed via agonism, once they have been established,
outcomes should increasingly be determined by other means, such as signals
of dominance status (LaFreniere and Charlesworth 1983). On both counts,
the basis for assessing established dominance structures in human groups
should go beyond agonism.
It is generally agreed that dominance structures at the group level can
be inferred from the patterning of the dyadic dominance relationships in the
group. Linear, transitive dominance hierarchies where each individual can
be assigned a unique ordered rank tend to result when dominance is determined
by agonism (Noe, de Waal, and van Hooff 1980; Silk 1987). Traditionally,
researchers have tried to fit linear, transitive hierarchies to their
data. However, if multiple processes interact with agonistic success to determine
asymmetric outcomes, such models are unlikely to be appropriate
(Noe, de Waal, and van Hooff 1980; Silk 1987). A related issue is that of
individual versus grouped ranks. If linear hierarchies do not hold then individual
ranks may not either. In fact, researchers studying very young
human children and large groups of male chimpanzees, where multiple processes
and thus nonlinearity may apply, have found that individuals ranks
do not fit their data well (Abramovitch 1976; Walters and Seyfarth 1987;
Missakian 1980). They have adopted subgroup rather than individual ranks
as more accurately representing the dominance structures in these groups.
Determining which models best reflect the dominance patterns in human
infant peer groups is clearly a prerequisite to linking peer dominance and
imitation.
The present study was designed to assess whether dominance relationships
are related to model choice in infant peers. We planned an observational
study of spontaneously occurring imitation and dominance structures
in a stable infant daycare group, where imitation would be facilitated and
relationships already formed. Goals of the study were to establish the nature
of group dominance structures, and then to assess links between peer model
choice and model dominance rank.
METHODS
Subjects and Setting
Subjects were full-time members of the Senior Infants group at the York
University Childcare Centre. Infants were predominantly children of faculty,
staff, and students at the university, coming from well-educated families and
60 A. E. Russon and B. E. Waite
diverse cultural backgrounds. Inclusion in the study was based on written
parental permission plus regular attendance at daycare during data collection
and for at least two months prior to that time. One potential subject was
excluded due to parental refusal, three others due to irregular attendance.
The final focal group consisted of 12 of the 16 infants in the group (6 male,
6 female) aged 11-16 months (mean = 13.2 mo, sd = 1.83).
The study was conducted in the infants’ normal daycare environment.
This consisted of a three-room complex with two connecting playrooms and
a sleeping room, an adjoining private playground, and a shared gymnasium.
The regular social milieu consisted of their infant classmates, four permanent
qualified daycare workers (three female, one male) and part-time parent and
student assistants. Parents, students, and other children visited the class
periodically. Three members of the research team attended the class for a
week prior to data collection for the purpose of habituation. During data
collection, when they were not involved in research activities, they integrated
their activities with the daycare routine by assisting with caregiving
or playing with infants. The researchers did thereby become part of the
daycare social structure, and may have influenced infants’ social interactions.
However, infants actively solicited assistance from any adult present
and became distressed if it was not forthcoming, so that participating seemed
less disruptive than remaining aloof. Based on videotaped data, there were
no obvious ways in which the researchers interacted differently with the
infants than other adults.
Data Collection
Raw data consisted of videotaped samples of infants’ free activity during
the normal daycare routine. To coordinate with the infants’ daily schedule,
we taped mainly from 9:30 to 11:30 AM and 1:30 to 3:30 PM on weekdays.
Data were structured as six 20-minute focal individual samples per subject
taken over two weeks in May, 1986. Samples for each subject were balanced
across different periods of the day, settings, and the two-week sampling
period. We collected adjunct data from daycare staff and parents on each
infant’s date of birth, sex, and dates of entering both daycare and the Senior
Infants class. We also kept a daily log, on an ad lib basis, to note incidents
which might influence patterns in peer social activities (e.g., absence, late
arrival, sickness, stress, special events).
Data Coding
Other than adjunct information, data consisted of coded descriptions of dominance-related
and imitative peer encounters identified in the videotaped
samples. Criteria used to separate events were either change in content
(partner or theme) and/or a break of at least 10 seconds between the end of
one social action and the start of another. We excluded peer encounters
Dominance and Imitation in Infant Peers 61
where adults intervened because this commonly influenced outcomes. Although
adults normally involve themselves in infants’ peer encounters, infant
peer research has concentrated almost exclusively on encounters free of
adult influence because of interest in how infants work out peer encounters
independently (Bakeman and Brownlee 1982; Russon 1985; Russon, Waite,
and Rochester, 1990). Because we were similarly interested in patterns that
infants themselves generated and maintained, we likewise limited the current
study to peer encounters free of adult intervention. For discussions of patterns
found when adults do interfere, see Russon, Waite, and Rochester,
1990).
Infant peer imitation. The focus in identifying imitation was on infants’
attempts to perform the same behavior as a model, regardless of the apparent
motive or underlying mental process (Russon 1985). The formal criteria we
adopted were those typical of imitation research (Nadel 1986). Imitation was
identified when one infant I (imitator) observed another infant M’s (model)
behavior, and, because of.the observation, I’s behavior was more similar to
M’s observed behavior than it would have been without the observation.
For immediate imitation, observational criteria were: 1) I must have received
information from M’s action, 2) the copier’s behavior followed the model’s
closely in time (maximum 10 sets delay), and 3) at least some features of
the modeled actions were clearly reproduced. Interrater agreement was required
to accept imperfect or partial copies. For delayed imitation, identification
relied on the precision of the copy, particularly details of the modeled
act; again, interrater agreement was required.
We located all imitative events involving the focal child as either model
or imitator. We succumbed to the lure of perfection offered by videotape
and attempted to generate an exhaustive catalog of focal imitative events.
Three judges first established high interrater reliabilities on training samples.
At least two judges then coded each sample to identify all imitative events
and to generate a detailed behavioral transcript of each, with the second
judge revising the transcript produced by the tirst. The identities of model
and imitator and the nature of the action copied were coded for each event.
Changes generated by the second judge’s revision included accepting or
rejecting events that the first identified as questionable, adding detail to
descriptions and adding missed events. Serious disagreements were discussed
with the principal investigator (AR). Differences between the first
and second judges’ work consisted mainly of omissions, with the second
invariably tinding more imitative events than the first. Overt disagreements
were very rare. They involved splitting versus grouping and questionable
events; none involved the identities of imitator or model or the action copied.
Infant peer social dominance. In line with earlier discussions and our focus
on established structures, we focused dominance assessments on outcomes
and thus included nonagonistic events. Borrowing from other work under-
62 A. E. Russon and B. E. Waite
taken from a similar viewpoint (Daudelin 1980; LaFreniere and Charlesworth
1983), we included the following types of encounters: disputed and undisputed
object/place/partner possession encounters, aggressive and submissive
gestures, pursuits, requests for contact, and disturbing a partner’s activities.
The dominant in each encounter is the winner in possession
encounters, or the partner who performs aggressive gestures, pursues, disturbs
a partner’s activities, or provides contact. The subordinate is the loser,
or the partner who takes possession only once items are abandoned, performs
submissive gestures, or solicits contact.
Two judges working independently performed coding. They established
high levels of interrater reliabilities in training sessions, then checked agreement
periodically during formal coding on a randomly selected 10% of samples.
Based on comparison of rates of agreement versus disagreement (A/
(A + D)), the two judges achieved 80%- 100% agreement on identification of
events, 100% on identification of participants, and 80%-100% on outcomes.
The maximum number of disagreements identified in any one sample was
2; lower rates of agreement therefore reflect a low frequency of incidents
for the sample. The judges invariably identified the same incidents as dominance-related.
Disagreements most commonly concerned whether or where
to split or group in prolonged episodes (e.g., repeated bouts of possession
struggles interspersed with pauses). Disagreements would not have affected
the dominance hierarchy we established in any of the cases examined.
We identified dyadic dominance relationships when a dyad’s encounters
showed an outcome pattern in favor of one partner. The criteria used were
more stringent than simple excess of wins over losses; following LaFreniere
and Charlesworth (1983) we required that one partner dominate in at least
two-thirds of that dyad’s encounters with an excess of at least two wins.
RESULTS AND DISCUSSION
Since interpretation of dominance patterns is necessary for assessing links
between dominance and imitation. results and discussion are combined.
DOMINANCE RELATIONSHIPS AND THE DOMINANCE
HIERARCHY
Videotape samples provided 391 clear dominance-related peer encounters.
Their distribution is shown in Table 1.
For the group dominance structure, we first attempted to generate a traditional
linear hierarchy with individual ranks using standard procedures (see
LaFreniere and Charlesworth 1983). The dominance matrix and the derived
hierarchy are shown in Table 2. Individual cells show the number of dyadic
encounters where the row partner dominated the column partner. Domi-
Dominance and Imitation in Infant Peers 63
Table 1. Distribution of Dominance-Related Encounters
Type Frequency Behavioral Content
Possessing items
Aggression
Assertion
Submission
AggressionlAssertion-
Submission
Comfort
Total
245
37
24
18
21
1
391
Possession struggles, uncontested
takes, abandoned item takes (objects,
171; places, 23; persons, 51)
Physical aggression (hit, poke, pull,
push, scratch, bite)
Persisting with goals despite obstruction
(maintain hold or position, remove
partner’s grasp, bump peer, interrupt
peer)
Alter activity because of peer (detour
away from peer, withdraw, pause
own activity when peer arrives, cry
for help)
Encounters involving exchanges of
these (A-hit, B-cry; A-push, B-cry
and leave)
Offer comfort to a peer (pat)
Table 2. Dyadic Dominance Matrix
Rank
Dominant
Subordinate
Class Individual (Age/Sex) AX EN CR PL KT CT LE WJ SM ER KA ST Totals
High 2
High 2
High 2
High 4
Mid 5
Mid 6
Mid 7
Mid 8.5
Mid 8.5
Low 10
Low 11
Low 12
(14%/M)
($
(l4%M)
(l5%F)
(12%,F)
(lOC&F)
(ZiF)
(13::,F)
(GYM)
(%I,
(lzf,
(ll%)
6
9
3
2
0
2
1
0
2
4
0
7 14 11 2 0 4 3 2 9 4
4 1 2 0 1 1 1 3 0
3 10 5 9 2 12 6 8 0
2 10 10 1 4 0 1 12 5
1 5 3, 11 14 1 3 0 1
0 7 2 7 2 8 0 6 2
0 0 0 6 0 7 I 1 2
0 1 0 4 1 3 2 3 2
0 1 1 3 1 0 1 5 1
2 3 6 0 1 1 2 2 I1
1 0 2 1 2 3 0 1 1
0 2 1 I 0 2 2 0 0 0
1
2
2
1
3
5
3
4
3
2
1
57
21
66
49
44
39
22
21
16
32
16
8
Totals 29 16 47 37 41 26 36 37 19 48 28 27 391
Age is given in months, as of the start of sample (5/20/86).
64 A. E. Russon and B. E. Waite
nance relationships could be calculated directly for almost one-half of the
dyads (30/66-46%). For the remaining dyads, either win-loss differences or
cell frequencies were too small to identify dominance relationships. When
the win-loss difference did not meet asymmetry criteria but the total frequency
of dominance-related encounters was above five, we labeled the
relationship “tied for dominance” (10/66); when the frequency of encounters
was under five, “indeterminate for dominance” (26/66).
Linearity and rigidity for this hierarchy are 97% (one reversal in 30
calculable dyadic dominance relationships) and 68%, respectively. Linearity
and rigidity refer to the percentage of dyadic relationships and dyadic encounters
outcomes, respectively, that conform to a linear transitive model
(LaFrenier and Charlesworth, 1983). Calculations are
Linearity = 100 x
total calculable dyads-relationship reversals
total calculable dyads
Rigidity = 100 x
total incidents-incident reversals
total incidents
Despite the fact that these levels of linearity and rigidity are similar to those
from related studies (see Table 3), accepting a linear dominance model is
problematic for our group. The linear solution is not well-defined because
it is not unique: Several infants (e.g., EN, SM) could equally well have
occupied very different ranks. In search of an explanation, we compared
our results with those from the related studies. Reanalysis using LaFreniere
and Charlesworth’s criteria was possible for two studies that published their
dominance matrices (Missakian 1980; Strayer and Trudel 1984).
We found a larger number of ties and indeterminate dyadic dominance
relationships than these two studies, based on a raw sample similar to Strayer
and Trudel’s (two hr/subject) but much smaller than Missakian’s. Comparisons
of Strayer and Trudel’s results with Missakian’s suggest only small
improvements in definition with more extensive sampling (4% more calculTable
3. Comparisons with Dominance Assessments in Related Studies
Study
Ages
(mo)
Missakian
Strayer and Trudel
Russon and Waite
x = 10.3
(6-20)
r( = 18
(sd = 6.2)
X = 13.2
(sd = 1.8)
Bakeman and Brownles(***) F; = 18.4
(12-24)
Relationships* Linearity
ObsiDyad Dyads Calc-Tie-Indet
(No.) (No.) (%) a* b** Rigidit
18.72 120 54-7-39 100 96 86
7.69 36 50-l l-39 100 90 79
5.79 66 46-15-39 97 91 68
3.49 55 ?_?_? ? 83 69
* Using LaFrenier and Charlesworth’s (1983) criteria for determining dominance relationships.
** Using wins-losses > 0 to determine dominance relationships, as in Bakeman and Brownies (1982), Missakian (19801, Stray1
and Trudel (1984).
*** Bakeman and Brownlee’s dominance matrix not published: calculations are those published in their report.
Dominance and Imitation in Infant Peers 65
able dominance relationships, 4% fewer ties). Our sample generated fewer
dominance-related events per dyad, a lower percentage of calculable dominance
relationships and a higher percentage of ties than Strayer and Trudel’s
or Missakian’s, despite our wider range of dominance-related encounters
and a raw sample similar to Strayer and Trudel’s. The first two differences
are reasonable, however, because our group was slightly younger than
Strayer and Trudel’s; peer conflict is commonly lower in the first than in
the second year (Missakian 1980; Russon 1985), and relationships differentiate
with increasing age (Howes 1983; Strayer and Trudel 1984). Our
higher rate of ties could be due to the narrower age range in our group. The
closer infants are in age, the more likely they will be competitive equals.
Our sample generated more observations per dyad than Bakeman and
Brownlee’s with a similar raw sample, but this is probably because their
dominance assessments were based only on object possession conflicts, a
subset of agonistic encounters. Nonetheless, our dominance matrix was very
similar in linearity and rigidity to theirs.
We therefore accepted our dominance pattern as a valid representation
of our group rather than as a methodological artifact. Dyadic dominance
relationships appeared to exist; some infants showed observable differences
in peer interactive behavior consistent with their partner’s dominance rank
(e.g., Observation 1). A group dominance structure was deemed present
because there were group level patterns among the dyadic dominance re-
lationships.
Observation I. PL was sitting and manipulating a Fischer-Price toy hanging
on the side of a toy box. Two other infants, ER, then CR, tried to intrude
on her use of this toy.
When ER reached for her toy, PL looked at him abruptly, grabbed his
reaching hand and tried to bite him twice. One of the daycare workers (ED),
intervened by warning her not to bite. ER managed to pull away and leave.
PL glared and grabbed after him.
As ER left, CR approached and reached for PL’s same toy. PL seized
his hand, pushed it away twice, and vocalized “aaaah” in protest (calling
for adult help, Hay 1985; Maccoby 1983). CR persisted in using the toy. A
tussle ensued (ED again intervened verbally), until finally PL turned back
to the toy and grudgingly (frowning) let CR share it. CR gradually edged
her out so that finally she left altogether, leaving him with exclusive use.
However, because of the ties and the relatively low linearity and rigidity,
we tested whether alternative models offered a better tit. Dominance models
based on multiple competing principles are consistent with deviations from
linearity like those we found. Bakeman and Brownlee (1982) and Noe, de
Waal, and van Hooff (1980) found patterns similar to ours: substantial numbers
of indeterminate and tied dyadic relationships, producing low rigidity.
Both concluded that this pattern reflected two relatively independent principles,
agonistic success and indisputable “rights” of possession, that interacted
to determine asymmetric outcomes. Rights commonly pertain to
66 A. E. Russon and B. E. Waite
use of space, objects, and social partners. Assertion of these rights does not
imply conflict, even when subordinates assert them over dominants.
There was evidence that a principle of rights of possession influenced
asymmetric outcomes in our infant group. Many of the reversals involved
temporary ownership; that is, subordinates won in competitions with otherwise
more dominant partners when attempting to retain an item which they
currently possessed. Two rare reversals involving permanent ownership occurred
in disputes over clothing or objects brought from home. Subordinates
were very active in their attempts to retain possession of the item, and
dominants ceded. Additional support for the claim that such a principle can
operate between infant peers comes from four sources: 1) the daycare workers
advocated rights of possession by reversing agonistically determined
outcomes which went against this principle (e.g., making the intruder return
an object to its “rightful” owner, saying “give that back, it’s hers”; see
Russon, Waite, and Rochester, 1990); 2) language acquisition studies have
found that notions of possession appear early in the second year (Brown
1973); 3) the subordinate of an infant chimpanzee dyad was similarly determined
and successful in winning object struggles against the dominant for
objects which he temporarily possessed (Russon 1985); and 4) Bakeman and
Brownlee (1982) found a principle of prior possession to operate in a toddler
peer group, in the sense that attempts to take objects were more likely to
succeed if the taker had used the contentions object in the recent past.
Whether or not one includes encounters involving rights of possession
in dominance assessments (as we did) or excludes them (as other researchers
have done), the fact remains that both agonistic and possession principles
can determine asymmetries in outcomes, even in infant peer groups. We
concluded that at least two processess, agonism and primitive rights of possession,
constituted the basis for dominance in our group.
Because of the relatively low levels of linearity and rigidity found when
individual ranks were assigned and because of the probable influence of
several principles, we considered whether subgroup ranks might be more
appropriate for this group. Principles of possession rights develop later than
agonism, and, at the ages we studied, they constitute a relatively primitive
and weaker means of resource acquisition (Bakeman and Brownlee 1982).
Their effect would most likely be to weaken the asymmetries created agonistically.
Simply empirically, the large number of ties we found in our group
essentially created subgroups of like-ranked individuals. Like a number of
other researchers, we therefore decided that the dominance patterns in our
group were better represented by subgroup rather than individual ranks
(Abramovitch 1976; Missakian 1980; Silk 1987; Walters and Seyfarth 1987).
As individuals in the extreme ranks tend to show distinct behavioral patterns,
we grouped our subjects into high, mid, and low rank subgroups (Table 2).
These other researchers adopted similar subgroupings.
Dominance and Imitation in Infant Peers 67
Table 4. Distribution of Types of Peer Imitation
Type Frequency Percentage Examples
Vocal
Verbal
Gestural
Actions on objects
Multiple
(coordinated)
Totals
22 13.7
7 4.3
46 28.6
76 47.2
10 6.2
161 100.0
Shrieks, cries, squeals, yells,
“mmmmuh”, “uh-uh”,
“ahhhh”
“Bubboo” or “bubah” (for
bubble), “da-da,” “daow” (for
down)
Clap, point, bounce, wave,
reach, stamp, butt, slap, hit,
“dance”
Bang table, draw, put toy in box,
take toy out of box, offer
object, shovel sand, “play”
xylophone, shake toy, throw
ball
Raise arms and shriek, offer toy
and vocalize “beee”
Dominance and Imitation
We identified a total of 161 peer-modeled imitative events. Their distribution
is shown in Table 4. Only one event constituted delayed imitation. The
distribution of peer model choice in relation to dominance is shown in Table
5; individual cells show the number of imitation events in which the row
partner chose the column partner as a model. Average rates of model choice
summarized by imitator and model rank are shown in Table 6.
We first assessed preference for high rank. We used Friedman’s 2-way
ANOVA by ranks to test for differences in model preference relating to
dominance rank. For each imitator, we ranked preferences for each dominance
subgroup on the basis of total frequency of imitations to models of
Table 5. Peer Imitation and Dominance: Distribution of Model Choice by Dominance Rank
Rank Models
Class Individual Imitator AX EN CR PL KT CT LE WJ SM ER KA ST Total
High 2 AX 6 IO 5 2 1 I 25
High 2 EN 2 4 6 I I3
High 2 CR 8 3 8 4 I 2 I I 28
High 4 PL 5 I 9 6 3 1 I 26
Mid 5 KT 1 2 2 5
Mid 6 CT 2 4 2 4 I 1 14
Mid 7 LE I 2 4 I 2 I 11
Mid 8.5 WJ 2 5 1 I 2 I 12
Mid 8.5 SM I I 1 1 I 5
L.OW IO ER I I 2 I I 2 I I I II
Low II KA 1 I 2 1 1 I I 8
Low I2 ST I 2 3
Totals 22 I4 38 25 23 9 4 5 3 8 6 4 I61
Blanksin cells represent frequency of 0 (zero) incidents.
68 A. E. Russon and B. E. Waite
Table 6. Average Frequency of Model Choice by Dominance Rank, Per Imitator in Each
Dominance Rank Subclass
Model’s Dominance Rank
Imitator’s High
Dominance Rank (n = 4)
Mid
(n = 5)
Low
(n = 3)
High 5.60 0.95 0.50
Mid 1.05 0.85 0.60
Low 0.92 0.53 0.50
that subgroup. This analysis showed reliable differences in model preference
related to dominance rank (Friedman’s xR2 = 7.625, n = 12, k = 3, p <
0.05). Page’s L trend test, an extension of Friedman’s ANOVA, showed a
reliable linear trend in preferences; infant imitators preferred, in order, high,
then mid, then low rank models (Page’s L = 41, n = 3, p < 0.05; see Cohen
and Holliday 1982). Wilcoxon’s sign test, comparing observed versus expected
proportions of imitations directed to high rank models, indicated a
reliable preference for high rank models (T- = 12.5, n = 12, p < 0.05).
Expected proportions were calculated on the assumption of uniform distribution
of model choice with respect to rank. These results support the claim
that imitators prefer high rank peers over lower rank peers as models.
To test whether infants prefer “like-self’ models, we tested whether
they tended to choose models “close” to themselves in dominance rankfirst,
same rank models, then, next higher rank models. Wilcoxon’s sign
test, comparing observed versus expected proportions of imitations to same
rank models, indicated a preference for same rank models (T- = 15, n =
12, p < 0.05). The same test, based on imitations to models in the next higher
rank for mid and low rank imitators only, was nonsignificant (T- = 15, n =
8, p > 0.05).
However, for both high rank and same rank analyses, inspection of
individual contributions to the T statistic showed strong rank-related model
preferences in high rank infants but much more diffuse patterns in mid and
low rank infants. Since this could reflect a ceiling effect in high rank imitators,
who chose themselves as models because there were no higher rank
models available, we recalculated Wilcoxon tests for mid and low rank imitators
only. Results showed a tendency to prefer high rank models, which
was not statistically reliable; there was no preference for same rank models.
Other than a ceiling effect, these differences between imitators could reflect
individual differences related to developmental levels; lower rank imitators,
who in our group tended to be younger, may simply choose models on different
bases than high rank ones. However, small sample sizes at the individual
level makes comments about individual differences highly speculative.
Our conclusion was that infants as a group did show a tendency to
prefer high rank peers as models, but that because of the probable ceiling
artifact and/or developmental differences, the tendency was probably
weaker than the group-level statistical tests would suggest.
Dominance and Imitation in Infant Peers 69
Finally, we looked briefly at the occasions when infants did choose
lower rank models (n = 34), and found one unexpected pattern. Imitation
served as a means of “taking”‘used by the dominant partner is some possession
struggles, especially those involving large objects or social partners.
By performing the same action on the contentious item as the current owner,
the “taker” usurped the owner’s use of the object, or “showed up” the
owner (Observation 2). Imitation can then function as a direct means of
resource acquisition for dominant individuals as well as for subordinates.
Observation 2. AX and PL each picked up small plastic cars, then carried
them to a radiator. AX, then PL placed their cars on the radiator grate.
Soon PL rubbed hers across the grate, making a scratching noise. Immediately
AX rubbed his own car back and forth 8-10 times, watching PL.
PL continued her own rubbing. AX watched her, grabbed her car, rubbed
it back and forth, then banged it on the grate, paused to look at PL who
turned away briefly, then resumed banging. When PL tried to regain access
to the radiator beside AX, he glanced at her and resumed rubbing his car
on the grate. PL detoured around him to a part of the radiator farther away
and started to rub her car there. AX paused his own rubbing, glanced at
PL, moved nearer her and banged his car on the radiator. PL moved back
to her old place at the radiator; AX followed and she finally left altogether.
AX resumed banging and rubbing his car briefly but soon stopped to watch
other activities.
Dominance has some predictive value for infant peer imitation, but it
may not influence imitation directly. Although this study was not designed
to explore underlying mechanisms, it is possible to make limited comments
on the role that competence, age, and other social factors played in dominance-related
model choice.
Dominance could simply reflect either competence and/or age: competent
individuals are more likely to win encounters; and competence in
infancy correlates strongly with developmental level, which itself regularly
correlates with age (Yando, Seitz, and Zigler 1978). However, our finding
that imitators did not show preferences for closely-ranked models suggests
that competence was not a central factor influencing their model choices.
Concerning age, we used Friedman’s 2-way ANOVA by ranks to test for
age-based differences in rate of model choice, with ages grouped into
subgroups like those used for dominance. Results were nonsignificant (Friedman’s
XR* = 6 0 n = 12, k = 3, p > 0.05), although patterns lay in the
predicted dire&&. Moreover, age was not reliably correlated with rate of
being chosen as a model (Pearson’s r = 0.49, n = 12, p > 0.05) although
dominance was (Spearman’s p = -0.71, n = 12, p < 0.05). The findings
for age also confirm the conclusion concerning competence. Neither competence
nor age appeared to be major factors in peer model choice in this
infant peer group.
In efforts to tease apart other possible mechanisms, we looked at the
qualities of imitations directed to models of equal or higher rank (n = 127).
We used additional data on these events that judges had coded for another
70 A. E. Russon and B. E. Waite
part of the project (reliabilities were of the same order as those reported
earlier). Actions imitated were, overwhelmingly, already known to imitators
(94%) and very easy for them to perform (73%); only 27% were moderately
or very difficult. This suggests that the actions imitated were simply elicited
via processes such as social facilitation, rather than explicitly constructed
in response to knowledge acquisition motives. Infant imitators’ motives for
copying peers were judged as fused in 69% of events, social in 21%, and
knowledge-related in 9%. This indicates stronger interpersonal than knowledge
acquisition processes. However, task-relevant features were reproduced
in 84% of events, which suggests some focus on the nature of the
activity itself. In fact, infants often chose to copy activities which were foci
for social exchange in this group (e.g., “reading” books, singing songs,
taking turns jumping off a springboard), and by imitating, joined both the
activity and a social encounter. Perhaps most importantly, the fact that observers
could not isolate a single motive in the majority of cases suggests
diffuse or fused rather than clearly differentiated processes. Peer dominance
appeared to serve in leading or coordinating activities. Although highly speculative,
one scenario on the underlying processes is as follows. Because
dominance attracts attention, infants tend to notice dominant individuals,
their activities, and their resources; although dominant, these individuals
are peers so that some of their actions are already know to observers; in
one-year old infants, observation of known actions very readily elicits their
imitation via social facilitation; then, because imitation is experienced as
ingratiating, it functions to allow the imitator to share in the dominant’s
activities and resources.
CONCLUSIONS
Dominance structures could be identified in this infant peer group, although
the patterns we found did not fit well with traditional models. Our data
suggested that the processes underlying infant peer dominance could be more
complex than agonistic success. Although agonistic success may be in some
sense a more “basic” process in dominance, it was not the only factor
operating in this human infant peer group nor did it always take priority.
Daycare workers’ efforts were towards discouraging agonism as a way of
settling disputes and at the same time encouraging conventionalized principles-in
particular, rights of possession. The infants themselves did behave
in accordance with this principle on occasion. We join Charlesworth
(1988) in emphasizing the multiplicity of means which can be used in competing
for resources and the complexity of their interactions. Like Bakeman
and Brownlee (1982) our findings point to the importance of rights of possession
in later infancy. Our findings also highlight the value of using an
outcome basis to assess dominance rather than the traditional agonistic process
basis. The outcome basis allows the identification of alternative pro-
DominanceandImitationinInfantPeers 71
cesses, whereas the process basis narrowly focused on agonism precludes
that possibility.
The occurrence of peer imitation was patterned in relation to the model’s
dominance risk. At the group level, infants showed a preference for high
rank peers as models over others. At the individual level, however, there
was considerable variability in model choice; it is not clear to what extent
this was artifactual. This study was not designed to identify the processes
underlying the dominance-imitation pattern, particularly whether infants are
relating to social dominance per se or to other associated features, but a few
speculations are possible. Either finer or more global processes could equally
well produce the model choice patterns we identified. Of the finer mechanisms
which have been suggested, we considered competence and age. Our
data provided no evidence for either as a central factor in infant peer model
choice; if anything, social factors were more important. The other possibility
is that global rather than finer processes underlie model choice patterns.
Given the ages of the infants studied, global in the sense of undifferentiated
processes are in fact more likely. The orthogenetic principle of developmental
movement, from initial fusion or lack of differentiation to increasing
specialization and differentiation, probably applies to the processes driving
imitation as it does elsewhere. This echoes the claim of researchers taking
a social cognition perspective, that especially in infancy social and practical
knowledge cannot readily be separated. In this light, dominance constitutes
only one component of a constellation of the characteristics of models and
of the actions they perform. It is probable that all but the oldest infants in
our group were responding to this larger constellation as a gestalt rather
than to dominance per se or to any finer dimension (Aronfreed 1969; Piaget
1962). This possibility is certainly supported by the large number of events
where observers could not identify a single motive.
In keeping with this view, our findings do not imply that other social
factors, such as friendships, do not influence the occurrence and direction
of peer imitation. If models represent one upit of choice, it must be considered
fhat each model may represent several roles vis a vis a given imitator.
Although these various roles can be separated, they do not necessarily represent
mutually exclusive factors, but may all influence any one incident;
what probably varies is the relative contribution of each and the nature of
their interaction (Aronfreed 1969; Uzgiris 1981). Thus, in choosing one
model, an imitator may be responding to one or to any combination of the
roles that model represents. Our finding that the relationship between dominance
and imitation was weaker rather than stronger is directly in line with
this view.
Finally, none of these cautions lessons the value of dominance rank in
predicting patterns in infant peer imitation, nor of the demonstration that
model choice is one process by which peer imitation could influence development.
These results demonstrate that the infant peer system should be
considered as a partially integrated or perhaps poorly differentiated whole
72 A. E. Russon and B. E. Waite
rather than as a collection of independent and isolated themes. Other interdependences,
such as between friendships and imitation, may then very
well exist. Although little systematic research has been conducted on such
interdependencies to date, such studies are clearly important to understanding
how the totality of peer experience affects social development in infancy.
This research was funded by SSHRC operating grants No. 410-85-l 130 and No. 410-87-0688.
Thanks to Lorraine Chaisson, Gilles Simon, and Brendan McCabe who participated in data
collection and coding and to Edna, Greg, Ann, Eileen, Char, and the infants themselves for
tolerating our intrusion into their daycare class.
REFERENCES
Abramovitch, R. The relation of attention and proximity to rank in preschool children. In The
Social Structure ofAttention M.R.A. Chance and R.R. Larson (Eds.). London: Wiley,
1976.
Abramovitch, R. and Grusec, J.E. Peer imitation in a natural setting. Child Deve/opment 49:
60-65, 1978
Apolloni, T. and Cooke T.P. Peer behavior conceptualized as a variable influencing infant and
toddler development. American Journal of Orthopsychiatry 45: 4-17, 1975.
Aronfreed, J. The problem of imitation. In Advances in child developmenf and behavior, Vol.
4, L.P. Lipsitt and H.W. Reese (Eds.). New York: Academic Press, 1969.
Bakeman, R. and Brownlee, J.R. Social rules governing object conflicts in toddlers and preschoolers.
In Peer relationships and social skills in childhood, K.H. Rubin and H.S.
Ross (Eds.). New York: Springer-Verlag, 1982, pp. 99-l 11.
Belsky, J. and Steinberg, L.D. The effects of day care: A critical review. Child Development
49: 929-949, 1978.
Brenner, H. and Mueller, E. Shared meaning in boy toddlers’ peer relations. ChildDevelopment
53(2): 380-391, 1982.
Bridges, K.B.M. A study of social development in early infancy. Child Developmenf 4: 36-
49, 1933.
Brown, R. AJirst language: The early stages. Massachusetts: Harvard University Press, 1973.
Charlesworth, W.R. Resources and resource acquisition during ontogeny. In Sociobiological
perspectives on human development, K.B. MacDonald (Ed.). New York: SpringerVerlag,
1988, pp. 24-77.
Cohen, L. and Holliday, M. Statisticsfor social scientists. London: Harper & Row, 1982.
Daudelin, N. Establissement d’une relation de dominance entre deux chimpanzes (Pan troglodytes)
durant leur premiere an&e de vie. Unpublished doctoral dissertation, University
of Montreal, 1980.
Deag, J.M. Aggression and submission in monkey societies. Animal Behavior 25: 465-474,
1977.
Hartrup, W.W. and Coates, B. The role of imitation in childhood socialization. In Ear/y experiences
and the process ofsocialization, R.A. Hoppe, G.A. Milton, and E.C. Simmel
(Eds.). New York: Academic Press, 1970.
Hay, D.F. Learning to form relationships in infancy: Parallel attainments with parents and
peers. Developmental Review 5: 122-161, 1985.
Hinde, R.A. Towards understanding relationships. London: Academic, 1979.
Howes, C. Patterns of friendship. Child Development 54: 1041-1053, 1983.
LaFreniere, P. and Charlesworth, W.R. Dominance, attention, and affiliation in a peer group:
a nine-month longitudinal study. Ethology and Sociobiology 4: 55-67, 1983.
Maccoby, E.E. Social-emotional development and response to stressors. In Stress, Coping and
Dominance and Imitation in Infant Peers 73
Development in Children, N. Garmezy and M. Rutter (Eds.). New York: McGrawHill,
1983.
McCabe, M.A. and Uzgiris, I.C. Effects of model and action on imitation in infancy. MerrillPalmer
Quarter/y 29(l): 69-82, 1983.
Missakian, E. Gender differences in agonistic behavior and dominance relations of Synanon
communally reared children. In Dominance relations: An ethological view ofhumnn
confhct and social interaction, D.R. Omark, F.F. Strayer, and D.G. Freedman (Eds.).
New York: Garland, 1980, pp. 397-413.
Mueller, E. and DeStefano, C. Sources of toddlers’ peer interaction in a playgroup setting.
Unpublished manuscript, Boston University, 1973.
Mueller, E. and Lucas, T. A developmental analysis of peer interaction among toddlers. In
Friendship andpeerrelations, M. Lewis and L.A. Rosenblum (Eds.). New York: Wiley
Interscience, 1915, pp. 223-258.
Nadel, J. Imitation et communication entre jeunes enfants. Paris: PUF, 1986.
Noe, R., de Waal, F.B.M., and van Hooff, J.A.R.A.M. Types of dominance in a chimpanzee
colony. Folio Primatologica 34: 90-l 10, 1980.
Piaget, J. Play, dreams, and imitation. New York: Norton, 1962.
Rubenstein, J.L. and Howes, C. Caregiving and infant behavior in daycare and in homes.
Developmental Psychology 15(l): l-24, 1979.
Russon, A. The ontogeny of peer social interaction between infant chimpanzees: A description
and comparative analysis. Unpublished doctoral dissertation, University of Montreal,
1985.
-. The development of peer social interaction in infant chimpanzees: Comparative social,
Piagetian and brain perspectives. In “Language” and intelligence in monkeys and
apes: Comparative developmentalperspectives, S. Parker and K. Gibson (Eds.). Cambridge:
Cambridge University Press, 1990.
Russon, A.E., Waite, B.W., and Rochester, M.J. Direct caregiver intervention in infant peer
social encounters. American Journal of Orthopsychiatry,60(3): 428-439, 1990.
Silk, J.B. Social behavior in evolutionary perspective. In Primate Societies, B.B. Smuts, D.L.
Cheney, R.M. Seyfarth, R.W. Wrangham, and T.T. Struhsaker (Eds.). Chicago: University
of Chicago Press, 1987, pp. 318-329.
Smollar, J. and Youniss, J. Social development through friendship. In Peer relationships and
social skills in childhood, K.H. Rubin and H. S. Ross (Eds.). New York: SpringerVerlag,
1982, pp. 279-298.
Strayer, F.F. The organization and coordination of asymmetrical relations among young children:
A biological view of social power. New Directions for Methodology of Social
and Behavioral Science 7: 33-49, 1981.
Strayer, F.F. and Trudel, M. Developmental changes in the nature and function of social
dominance among preschool children. Ethology and Sociobiology 5: 279-295, 1984.
Thelen, M.H. and Kirkland, J.D. On status of being imitated: effects on reciprocal imitation.
Journal of Personality and Social Psychology 33: 691-697, 1976.
Uzgiris, I. Two functions of imitation during infancy. International Journal of Behavioral Development
4: l-12, 1981.
de Waal, F.B.M. and Luttrell, L.M. Mechanisms of social reciprocity in three primate species:
symmetrical relationship characteristics or cognition? Ethology and Sociobiology 9:
101-118, 1988.
Walters, J.R. and Seyfarth, R.M. Conflict and cooperation. In Primate Societies, B.B. Smuts,
D.L. Cheney, R.M. Seyfarth, R.W. Wrangham, and T.T. Struhsaker (Eds.). Chicago:
University of Chicago Press, 1987, pp. 306-317.
Wishart, J.B. Siblings as models in early infant learning. Child Development 57: 1232-1240,
1986.
Yando, R., Seitz, V., and Zigler, E. Imitation: A developmental approach. New Jersey: Erlbaum.
1978.