Physical Activity Play: The Nature and Function of a Neglected Aspect of Play
Author(s): A. D. Pellegrini and Peter K. Smith
Source: Child Development, Vol. 69, No. 3 (Jun., 1998), pp. 577-598
Published by: Blackwell Publishing on behalf of the Society for Research in Child
Stable URL: http://www.jstor.org/stable/1132187
Accessed: 13/05/2009 19:07
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact firstname.lastname@example.org.
Blackwell Publishing and Society for Research in Child Development are collaborating with JSTOR to digitize,
preserve and extend access to Child Development.
Child Development,June 1998,Volume 69, Number3, Pages 577-598
Physical Activity Play: The Nature and Function of a Neglected
Aspect of Play
A. D. Pellegrini and Peter K. Smith
In this review, we considerthe natureand possible developmentalfunctionsof physical activityplay, defined
as a playful contextcombinedwith a dimension of physicalvigor. We distinguish 3 kinds of physical activity
play, with consecutiveage peaks:rhythmicstereotypiespeaking in infancy,exerciseplay peaking during the
preschoolyears, and rough-and-tumble play peaking in middle childhood.Genderdifferences(greaterpreva-
lence in males) characterizethe latter 2 forms. Functionis considered in terms of beneficialimmediate and
deferred consequences in physical, cognitive, and social domains. Whereasmost theories assume that chil-
dren's play has deferredbenefits, we suggest that forms of physical activity play serve primarilyimmediate
developmental functions. Rhythmicstereotypiesin infancy are hypothesized to improve control of specific
motor patterns.Exerciseplay is hypothesized to functionprimarilyfor strengthand endurancetraining;less
clear evidence exists for possible benefits for fat reductionand thermoregulation. addition, there may be
benefitsof exerciseplay that we hypothesizeto be largelyincidentalto its playful or physicalnature.
play has a distinctivesocial component;we hypothesizethat it servesprimarilydominance
functions;evidence for benefits to fighting skills or to emotionalcoding are more equivocal.Furtherresearch
is indicated, given the potentially importantimplicationsfor children'seducation,health, and development.
INTRODUCTION physical activity, and even for aspects of social orga-
nization and social skills. Physical activity play may,
Over the past 30 years, the study of children's play
in some senses, matter psychologically.
has been a popular topic of scientific inquiry (see the
In this review, we discuss the definition of physi-
chapter on play in the fourth edition of the Manual cal activity play, and age and gender trends; we then
of ChildPsychology;Rubin, Fein, & Vandenberg, 1983). consider the evidence regarding its functional bene-
Pretend play is the aspect of children's play most
thoroughly studied (see Fein, 1981). Indeed, the para-
digmatic study of young children's play has probably
been the study of children's symbolic use of play ob- WHAT IS PHYSICAL ACTIVITY PLAY?
jects in a developmental progression (e.g., McCune, GENERAL DEFINITIONAL ISSUES
1995). Both child developmentalists and animal ethologists
Yet these analyses ignore some of the most com-
agree that play behavior is enjoyable, and that play-
mon forms of play, as well as some basic theoretical
ers, typically children or juveniles, are concerned
assumptions regarding the functions of play. Chil- with means over ends, and that the activity appears
dren's play often has a vigorous physical component,
to be quot;purposeless,quot; or to occur for its own sake
and thus it may variously be called physical activity
(Martin & Caro, 1985; Rubin et al, 1983; Smith & Voll-
play, locomotor play, or exercise play. Much of chil- stedt, 1985). Physical activity play, specifically, may
dren's physical activity can be seen as playful in the
involve symbolic activity or games with rules; the ac-
sense that it is minimally constrained by adult de-
tivity may be social or solitary, but the distinguishing
mands. Adults, however, often show some ambiva-
behavioral features are a playful context, combined
lence toward children's high levels of physical activ-
with what Simons-Morton et al. (1990) describe as
ity. This ambivalence may also be reflected in the moderate to vigorous physical activity, such that
relative paucity of research on children's physical ac-
metabolic activity is well above resting metabolic
tivity generally (Pellegrini & Smith, 1993; Welsh & rate. Paradigm examples of physical activity play in-
Labbe, 1994), and on physical activity play and
clude running, climbing, chasing, and play fighting,
rough-and-tumble play in particular (Humphreys &
Smith, 1984). Yet physical activity levels may be im-
portant, not only for physical development, but also @ 1998 by the Society for Research in Child Development, Inc.
perhaps for cognitive performance subsequent to All rights reserved. 0009-3920/98/6903-0002$01.00
these behaviors is probably controlled by general
the latter being a component of rough-and-tumble
play (R & T).
Stereotypic behaviors tend to peak during the
The criterion of purposelessness in the definition
midpoint of the first year of life; at 6 months, some
of play seems to present a logical problem when play
infants spend as much as 40% of a 1 hr observational
is also considered to be functional (Martin & Caro,
period in stereotypic behavior (Thelen, 1980). After
1985). How can play simultaneously serve no pur-
this point, the behaviors gradually disappear from
pose and serve a developmental function? One way
to resolve this problem is to consider play as serving normal children's behavioral repertoires (Thelen,
1979). Across the first year of life, infants spend 5.2%
minimal immediate functions during childhood,
of their time in stereotypic behaviors (Thelen, 1980).
with benefits deferred until maturity. Alternatively,
play may serve immediate functions about which Some early parent-infant interactions probably pro-
players and others are unaware. We will hypothesize vide other physical play opportunities. For example,
Roopnarine, Hooper, Ahmeduzzaman, and Pollack's
about such functions. We first consider age and gen-
(1993) examination of play between parents and 1-
der trends, because these may indicate different types
year-old infants in India suggests that play, such as
of physical activity play that may have different func-
tossing the infant in the air and bouncing on the knee
tional significance (Byers & Walker, 1995).
(but also including R & T), accounted for 13% of all
play, whereas object play accounted for 80%. Simi-
AGE AND GENDER TRENDS IN PHYSICAL larly low rates of American parent-infant physical
ACTIVITY PLAY play (or vestibular stimulation) were reported by
Studies relevant to age and gender trends for physi-
cal activity play are presented in Table Al of the Ap-
By exercise play we mean gross locomotor move-
ments in the context of play. The distinguishing fea-
ture of this play is its physical vigor; it may or may
Play generally follows an inverted-U develop- not be social, but the distinctively specialized social
mental course: It begins in early infancy, peaks dur- form of R & T is discussed later.
ing childhood, then declines during adolescence, and Exercise play in this sense can start at the end of
all but disappears by adulthood (Byers & Walker, the first year. It can be solitary or with parents or
1995; Fagen, 1981; Rubin et al., 1983). However, the
peers. In fact, much of the research on parent-infant
trends in physical activity play in humans appear to
play does not distinguish between exercise play and
show three successive peaks, reflecting three types of R & T, but appears to be describing quot;rough physical
play, probably with different functions. We designate playquot; (Carson, Burks, & Parke, 1993; Roopnarine et
these as (1) rhythmic stereotypies, (2) exercise play, al., 1993), which we consider under R & T later. Low
and (3) R & T. We discuss each in turn. rates of American parent-infant physical play have
been reported by MacDonald and Parke (1986), with
rates peaking at around 4 years of age. A few cases
Rhythmic of infant exercise play without parents have been re-
ported. Konner (1972), for example, observed that a
Although most studies in the infancy literature re-
Botswana foraging group encouraged infants to
late to the ontogeny of symbolic play with parents
chase after and catch large insects.
(e.g., Bornstein & Tamis-LeMonda, 1995) and sensori-
motor exploration / play (e.g., Ruff & Saltarelli, 1993), As we move to the preschool period, greater inci-
dences of exercise play are reported. As was the case
there is limited evidence documenting infants' physi-
in infancy, most of the peer play literature for this
cal activity play. Thelen's (1979, 1980) longitudinal
period focuses on pretend play, not physical activity
study of infants' quot;rhythmical stereotypiesquot; during
play. Where the latter is reported, it is often in the
the first year of life provides basic and important de-
form of R & T, which tends to co-occur with pretend
scriptive information. Rhythmical stereotypic behav-
during this period (Pellegrini & Perlmutter, 1987;
iors are similar to our definition of physical activity
Smith, 1973; Smith & Connolly, 1980). In a few stud-
play to the extent that they are gross motor move-
ies, quot;gross motorquot; play other than R & T is reported,
ments, quot;and it is difficult to ascribe goal or purpose
sometimes occurring alone, sometimes with peers
to those movementsquot; (Thelen, 1979, p. 699); examples
(rather than with parents).
include body rocking and foot kicking. The onset of
Pellegrini and Smith 579
most children engaged in one or two long (>10 min)
Exercise play seems to increase from the toddler
to preschool period and then declines during the pri- MVPAs/day.
In general, then, exercise play is quite common in
mary school years, with a likely peak at around 4 to
early/middle childhood, and appears to peak in the
5 years (Eaton & Yu, 1989; Routh, Schroeder, & O'Tu-
preschool and early primary grades, although more
ama, 1974). Specifically, for 2-year-olds, Rosenthal
evidence is certainly needed to clarify exactly when
(1994) reports that it accounts for about 7% of behav-
ior observed in day-care settings. For children 2 to 4 the age peak occurs.
years of age, Field (1994) reports physical activity
play accounting for 10% of all day-care behavior.
Similarly, Bloch's (1989) observational study in Sene-
gal found that gross motor activities accounted for Rough-and-tumble play refers to vigorous behav-
iors such as wrestling, grappling, kicking, and tum-
11% and 13% of children's play in the home at 2 to
bling that would appear to be aggressive except for
4 years of age and 5 to 6 years of age, respectively.
In two ethological studies in British nursery the playful context; chasing is sometimes included
within this definition, but here we restrict it to the
schools, using a variety of samples, McGrew (1972)
former contact behaviors. Whereas exercise play may
and Smith and Connolly (1980) observed children's
or may not be social, R & T is necessarily so.
behavior at a micro-analytical level. In McGrew's
The earliest cases of children's R & T play are sup-
sample, with a mean age of 49.2 months, approxi-
ported by a parent, often the father in the case of
mately 20% of children's activity was physically vig-
rough physical play (Carson et al., 1993; Roopnarine
orous, such as run, flee, wrestle, chase, jump, push
and pull, lift, and climb. Similarly, in Smith and Con- et al., 1993). As reviewed earlier, rates of such play
are low. At around 4 years of age, R & T accounts for
nolly's (1980) sample, with a mean age of 43.3
months, vigorous activities, such as run, chase, and about 8% of observed parent-child behavior (Jacklin,
climb (but also including R & T), accounted for 21% DiPietro, & Maccoby, 1984).
Rough-and-tumble play with peers is more thor-
of their behavior.
oughly documented; its frequency follows an in-
As children move into primary school, a decline
in physical activity is witnessed. For children aged 6 verted-U developmental curve. For preschool chil-
to 10 years, quot;exercise playquot; accounted for only 13%of dren, it accounts for approximately 3%-5% of play
behavior (Pellegrini, 1984); at 6 to 10 years, it ac-
all outdoor behavior observed during school recess
periods (Pellegrini, 1990). This relative decrease in counts for 7%-8% of recess behavior (Boulton, 1992;
Pellegrini, 1988); and during the period from 7 to 11
play might be underestimated, however, because the
years, for about 10% (Humphreys & Smith, 1987). At
primary school observations occurred on school
11 to 13 years it falls to 5% (Boulton, 1992; Pellegrini,
playgrounds, rather than in classrooms, unlike most
1995a), and there is a further decrease to only 3% at
studies of preschoolers; the relative spatial density of
14 years (Pellegrini, 1995b).
classrooms, compared to playgrounds, inhibits gross
These studies used similar definitions and meth-
motor activity (Smith & Connolly, 1980).
odologies, so that this age curve can be advanced
It will be important for our later argument to dem-
with reasonable confidence. Rough-and-tumble play
onstrate that during the preschool/primary school
appears to increase through the preschool and pri-
years children engage in substantial amounts of exer-
mary years and peak in the later primary years,
cise play. Blatchford (1996) described the general lev-
around 8 to 10 years, just prior to early adolescence;
els of English primary school children's activity on
this appears to be a distinctively later peak than that
the school playground and found that most (60% of
for exercise play, provisionally described as being at
the children) are engaged in some form of physically
active play or games during their daily break times,
which lasted between 65 and 75 min.
What about physical activity outside the school
Summary of Age Trends
context? Simons-Morton and colleagues (1990) stud-
In summary, forms of physical activity play are
ied children aged 9 and 10 years, using children's
quite common in childhood. In primary schools, most
self-reported frequency of moderate to vigorous
children engage in active play during their daily
physical activity (MVPA) over 3 days. Major sources
break times; further, levels of physical activity play
of MVPA were running, walking fast, games and
are moderate for most children when they are out
sports, and cycling. MVPAs generally were slightly
of school. An analysis of age trends suggests three
more common before or after school (2.3/day) than
successive inverted-U curves describing three differ-
during school (1.6/day). Throughout the whole day,
580 Child Development
ent forms of physical activity play. Rhythmic stereo- many other mammalian species (Meaney, Stewart, &
typies peak in infancy, at around 6 months of age; Beatty, 1985; Smith, 1982). This is true for parent-
exercise play peaks during the preschool years, ac- child rough play (Carson et al., 1993; Roopnarine et
counting for up to 20% of observed school recess be- al., 1993) and for peer play (Humphreys & Smith,
havior, and declines during the primary school years, 1984; Pellegrini, 1989a); this gender difference is
accounting for about 13% of observed behavior. more marked for contact R & T than for play chasing
Rough-and-tumble increases during the late pre- (Smith & Connolly, 1980).
school and early primary school years, accounting for Differences in boys' and girls' initiation of and re-
about 5% of observed recess behavior, peaks in later sponse to R & T may be implicated in these gender
primary years at around 10%, and then declines dur- differences. Both the animal and child literatures of-
ing early adolescence, accounting for less than 5% of ten show males' higher rates of initiation of R & T
play. We postulate that these successive age peaks bouts and females' higher withdrawal from bouts
reflect different forms of play with different func- (Fabes, 1994; Meaney et al., 1985; Pellis, Field,
tions. Smith, & Pellis, 1996). Females may withdraw from
R & T initiations as they react differently from males
to tactile stimulation of the sort that characterizes
Gender Differences R & T (Meaney et al., 1985; Pellis et al., 1996). The
physical vigor and roughness typical of boys' play
groups seem to be important factors for girls segre-
No sex differences are reported in the incidence of
gating themselves from boys' play groups (Fabes,
rhythmic stereotypies. Thelen (1980) compared rates 1994; Maccoby, 1986).
for 10 male and 10 female infants; these averaged 35.1
and 34.4 bouts per hour, respectively, a very small
Causal Factors in GenderDifferences
and nonsignificant difference.
Gender differences appear to be absent in rhyth-
mical stereotypies, but appreciable for both exercise
play and R & T. Hypotheses about the functions of
Males tend to engage in exercise play at higher play must take account of such gender differences
rates than females. Eaton and Enns's (1986) meta- and the causation of such differences.
analysis of 90 studies of gender differences in motor Hormonal influences on play have been impli-
activity level reported a significant difference in favor cated in gender differences in R & T. Hormonal in-
of males, with the effect size tending to increase from fluences typically center around the effects of en-
infancy to mid-adolescence. Part of this gender differ- dogenous and exogenous androgens on neural
ence may be due to differential maturation rates. Ea- organization and behavior (Meaney et al., 1985). Nor-
ton and Yu (1989) found that relative maturity (per- mal exposure to androgens during fetal development
cent of estimated adult height attained) interacted predisposes boys, compared to girls, toward physical
with gender, being negatively related to activity activity and R & T. Excessive amounts of these male
level, with girls being both less active and more phys- hormones are hypothesized to quot;masculinizequot; fe-
ically mature than boys. males' play (Collaer & Hines, 1995). The experimen-
In these studies, R & T was probably included tal literature involving mice, rats, hamsters, and
with exercise play; although R & T takes up a smaller monkeys supports the androgenization hypothesis
percentage of time through the early primary years (Collaer & Hines, 1995; Quadagno, Briscoe, & Qua-
(see discussion of age trends), some of the gender dif- dagno, 1977). For obvious ethical reasons, the effects
ference in exercise play may be attributable to the of androgens on human behavior can only be studied
well-documented gender difference in R & T (next through natural experiments, where fetuses receive
section); it will be important for future research to abnormally high levels of these male hormones be-
delineate gender differences with an unconfounded cause of genetic defects (e.g., Congenital Adrenal Hy-
measure of exercise play. perplasia [CAH]) or difficulties during pregnancy
(e.g., where mothers take synthetic progestins).
Most human studies of CAH support the andro-
genization hypothesis. These studies typically have
Males exceed females in frequency of R & T in vir- used questionnaire methodology to ask parents or
tually all cultures that have been examined (DiPietro, children about their preferences for various activities,
including physically active sports and R & T. The fre-
1981; Humphreys & Smith, 1984) and also among
Pellegrini and Smith 581
quently cited research of Money and colleagues (e.g., on the long-held emphasis among child develop-
Money & Ehrhardt, 1972) has shown that andro- mentalists on developmental continuity (Bateson,
genized girls are more quot;tomboyishquot;: They prefer 1981; Gomendio, 1988; Kagan, 1971). Bateson's (1976)
male activities more than do nonandrogenized girls. metaphor for the deferred-benefit view of play is
Using observations of toy preferences in CAH chil- quot;scaffoldingquot;: Play functions in skill assembly, and
dren compared to controls, Berenbaum and Snyder then is disassembled when the skill is mastered.
(1995) found that CAH girls showed greater prefer- Alternatively, play may be viewed not as an in-
ence for boys' toys and activities. complete or imperfect version of adult behavior, but
Although Hines and Kaufman (1994) found no in- as having immediate benefits during childhood. This
crease of R & T in CAH girls, the children were ob- quot;metamorphicquot; (Bateson, 1976) view posits that play
served for only one 12 min session in an experimental and its consequences are unique to the niche of child-
room of undisclosed size-conditions that may have hood, and that later benefits are not necessary for its
inhibited children's exhibition of normal behavior, explanation (Bjorklund & Green, 1992; Gomendio,
including R & T. 1988; Pellegrini, Horvat, & Huberty, 1998). This view
Socialization interacts with hormonal events to af- is consistent with recent discussions suggesting that
fect gender differences (Ehrhardt, 1984; Fabes, 1994; play occurs at specific periods during which develop-
Maccoby, 1986; Meaney et al., 1985; Quadagno et al., ment may be modified (Byers & Walker, 1995;
1977). Beginning with interactions with their parents, Thelen, 1979). Accordingly, the previously discussed
boys and girls are socialized into different, and often age distribution of physical activity play may be use-
segregated, worlds that tend to reinforce these gen- ful in evaluating functional hypotheses.
der differences (Maccoby, 1986; Meaney et al., 1985). Different forms and dimensions of physical activ-
For example, fathers spend more time with their sons ity play may serve specific developmental functions
than with their daughters (Parke & Suomi, 1981), and (Gomendio, 1988; Smith, 1982). We discuss the func-
when with their sons, they engage in physically vig- tion(s) of physical activity play, considering the age
orous play (Carson et al., 1993; MacDonald, 1993; trends and gender differences summarized above.
MacDonald & Parke, 1986). That girls are more We also consider both the dimension of physical ac-
closely supervised by parents and teachers (Fagot, tivity itself and the dimension of social participation
1974, 1994) may further inhibit their physically vigor- that distinguishes exercise play from R & T. Al-
ous behavior (Maccoby, 1986). though some authors list up to 30 possible functions
for play (Baldwin & Baldwin, 1977), certain functions
(physical training, cognitive, and social) are most
FUNCTIONS OF PHYSICAL ACTIVITY PLAY
commonly advanced. Appendix Table A2 summa-
In his discussion of ways in which behavior can be rizes the extant studies, by function.
explained, Tinbergen (1963) described the quot;Four As a first step in establishing the functional impor-
Whysquot; of behavior as being immediate causation, de- tance of physical activity play during childhood,
velopmental history, immediate function, and ulti- we present evidence from a small number of play-
mate function in terms of evolutionary history. For deprivation studies, either natural or experimental.
the purposes of this article, we do not consider func- These suggest that a lack of opportunity to engage
tion in its quot;ultimatequot; sense (Hinde, 1980, p. 102), that in physical activity play leads to compensation later,
is, in terms of survival or reproductive success over concluding that play is of functional benefit.
succeeding generations. To directly address ultimate In Thelen's (1980) study of rhythmic stereotypies
function, a more wide-ranging analysis would be in the first year of life, infants who engaged in a lot
necessary, to an extent that would necessarily be very of these spontaneous physical activities when given
the opportunity were those observed to receive less
speculative given the limited evidence available. In-
stead, we consider function in terms of the quot;benefi- vestibular stimulation from caregivers and those who
were more often restricted in natural movements
cial consequencesquot; of the behavior to the individual
(Hinde, 1980). These consequences can be either im- (e.g., placed in infant seats). Thelen (1980, p. 148) con-
mediate or deferred. cluded that quot;deprivation of active as well as passive
For almost a century, the dominant view in child movement may ... promote stereotypy.quot;
development (Groos, 1898, 1901) has been that play Three sets of field experiments have looked at dep-
has deferred benefits. That is, during the period of rivation of physical activity play during childhood;
extended childhood, children engage in play to learn in these, although exercise play was the primary out-
come variable, R & T would also have been included.
and practice those skills necessary to be functioning
adult members of society. This assumption is based Smith and Hagan (1980) studied English preschool
582 Child Development
are based on the systematic onset of specific behav-
children (3-4 years) who were deprived of vigorous
iors and their co-occurrence with milestones of motor
exercise by varying the amount of time they re-
mained in their classrooms engaged in sedentary development. Thelen found that the individual be-
haviors appeared during a restricted period; onset
seatwork. After deprivation periods, they played out-
was not randomly distributed. This pattern, she ar-
doors. On the long, compared to short, deprivation
days, children's play was more vigorous in the imme- gued, is indicative of neuromuscular maturation. To
support this claim further, she presented significant
diate postdeprivation period.
correlations between the age of onset of stereotypic
Utilizing a similar deprivation paradigm with
groups (e.g., legs, arms, hands and knees) and the
American primary school children (5-9 years), Pelle-
age of passing items from the Bayley Scales of Infant
grini and colleagues (Pellegrini & Davis, 1993; Pelle-
grini, Huberty, & Jones, 1995) replicated Smith and Development reflecting neuromuscular, not cogni-
tive, development. Rhythmic movements of given
Hagan's results: Long, compared to short, depriva-
tion periods resulted in higher levels of physical ac- body systems appear to increase just before the infant
tivity. Deprivation, however, interacted with gender achieves voluntary control of that system.
We postulate that infants' rhythmic stereotypies
of the child; boys, compared to girls, were especially
are primarily functional for the immediate benefits
active after long deprivation.
of improving control of specific motor patterns. The
These results support the following generaliza-
tion: If children are deprived of opportunities for correspondence between the ages at which these
movements occur and cerebral development sug-
physical activity play, they will, when given the op-
portunity to play, engage in more intense and sus- gests that, initially, rhythmic stereotypies may be
manifestations of immature sensorimotor integra-
tained bouts of physical activity play than they
would have done if not so deprived. This generaliza- tion. Play may modify or eliminate irrelevant syn-
apse formations; with maturation, these patterns are
tion, in turn, suggests that physical activity play is
used in more goal-directed ways (Byers & Walker,
serving some developmental function(s) such that a
1995; Thelen, 1979). Such a hypothesis is consistent
lack of it leads to compensation.
with the lack of gender differences in these behaviors,
In the remainder of this section, we first consider
because there is no reason to suppose that control of
physical training (and related) functions of rhythmi-
motor patterns at this very basic level of generality
cal stereotypies and of exercise play. We then exam-
is more important for boys than for girls.
ine possible cognitive functions of exercise play. We
assume that any physical training functions and cog-
nitive functions of exercise play would also apply to
R & T, which has some of the same physical compo-
With the onset of locomotion, another develop-
nents. We then examine different hypotheses for so-
mental course may begin, as evidenced by the corre-
cial functions for R & T only, because this is the form
spondence between exercise play and muscle differ-
of physical activity play that is characteristically and
entiation, strength, and endurance. Brownlee (1954)
was the first to propose that animal play was related
to juvenile muscle development. Fagen (1976) ex-
Physical Training tended this argument by proposing deferred benefits
of exercise play for motor training, specifically, mus-
Both rhythmic stereotypies and exercise play seem
cle strength, general cardiopulmonary functioning,
to have physical training benefits.
and metabolic capacity. He suggested that the forms
of exercise play, often involving varied, interrupted,
Rhythmic Stereotypies and repeated use of muscle groups, as well as whole-
body activities, would be well suited to these de-
Activities such as waving the arms and kicking
ferred benefits. Byers and Walker (1995), in a thor-
peak at around 6 months of age. The onset of these
ough review of the animal play and motor training
actions is probably controlled by general matura-
literatures, evaluated the issue of immediate or de-
tional processes, which correspond to neuromuscular
ferred benefits of exercise play for three aspects of
maturation (Field, Ting, & Shuman, 1979). Thelen
motor training: endurance, strength, and skill and
(1979), in her naturalistic longitudinal study of in-
economy of movement. They suggest that exercise
fants' rhythmic stereotypies, suggests that this is a
play may improve skill and economy of movement
sensitive period in neuromuscular development,
due to the effects of exercise on muscle fiber differen-
similar to the argument of Byers and Walker (1995).
tiation and cerebellar synaptogenesis. They present
Her functional inferences about physical activities
Pellegrini and Smith 583
developmental data from house mice, rats, cats, and functional inferences, they do provide some evidence
in a very restricted literature. This evidence supports
giraffes and conclude that physical activity in the ju-
the hypothesis that children who habitually engage
venile period, beginning in the early postnatal period
in vigorous games and sports show immediate bene-
and declining at mid-lactation, is a sensitive period
fits in terms of being fitter than children who do not.
in the development of these functions. Exercise play
Smoll and Schutz (1985) studied 3,000 students in
during this period has a lasting effect on subsequent
British Columbia aged 9, 13, and 17 years. Athletes
economy and skill of movement.
emerged as significantly fitter than nonathletes on all
In human children, exercise play may help shape
physical fitness tests. This was true for both boys and
the muscle fibers used in later physically vigorous
girls. Differences between athletes and nonathletes
activities. This could improve the economy and skill
were small at 9 years, but increased substantially by
of movement along the lines suggested by Byers and
13 and 17 years. This would be consistent with physi-
Walker in other species, although we know of no di-
cal training effects of such participation, although
rect evidence for this. However, the evidence sug-
as selective participation
gests that endurance and strength may be developed
and dropout-are also possible.
through sustained exercise bouts. The age course of
Relatedly, Lussier and Buskirk (1977) examined
exercise play also corresponds to the growth of arm
the effects of a 12 week endurance training program
and leg muscles and bones during the preschool pe-
(distance running) on 8- to 12-year-old boys and girls.
riod (Tanner, 1970). Exercise play during the school
Training decreased heart rate during submaximal
years and beyond might continue to benefit muscle
workloads and increased maximum oxygen uptake.
and bone remodeling and strength and endurance
Other studies of endurance training have had posi-
training; physiological effects have been observed
tive immediate results, and reviews (e.g., Rowland,
into adulthood in numerous species (Byers & Walker,
1985; Simons-Morton, O'Hara, Simons-Morton, &
Parcel, 1987) concur in concluding that regular, high-
Byers and Walker (1995, p. 29) were skeptical that
intensity training can improve cardiorespiratory
exercise play functions to support strength or endur-
ance. They concluded that quot;in many species, it is un- functioning.
The evidence presented thus far suggests that chil-
likely that play is a form of endurance or strength
dren are given opportunities for exercise play that are
training because play bouts are too brief to prompt
probably adequate for endurance training; but before
such benefits of exercise.quot; They suggested that chil-
firm conclusions are made, even for the preschool
dren would need to engage in daily bouts of exercise
and primary school periods, more research is needed
play, lasting 1 hr, 4 to 5 days per week, to increase
to document the intensity and duration of exercise
endurance significantly. However, based on previous
play. Further, we do not know, beyond the preschool
data from research on preschool and primary school
period, the extent to which vigorous physical activi-
children's playtime and activities out of school, we
ties are playful, per se, or not. This remains an impor-
postulate that children at these ages may well engage
tant task for future research.
in exercise play at levels meeting these criteria. Etho-
In summary, exercise play, in the preschool years
logical preschool studies discussed above show that
especially, seems sufficiently frequent that it can
20% of children's behavior during free-play periods,
serve an immediate function for endurance and
usually lasting the whole morning, was classified as
strength training. It may also improve skill and econ-
vigorous (McGrew, 1972; Smith & Connolly, 1980).
omy of movement, although specific evidence for this
Although some primary school children's play op-
is lacking. This hypothesized set of functions is con-
portunities may be more limited, due to school regi-
sistent with the age curve for exercise play (Tanner,
mens, their recess periods, typically accounting for
1970). More intriguing is its relation to the gender dif-
20-60 min / day in American schools and 65-75 min /
ference observed. Strength training at least would be
day in English primary schools (Blatchford, 1996), are
more important for males in the quot;environment of
also characterized by exercise play (Blatchford, 1996;
evolutionary adaptednessquot; for fighting and hunting
Pellegrini, 1990). These activities are supplemented
skills (e.g., Boulton & Smith, 1992); however, it could
by activities in other contexts, as children usually en-
also be argued that endurance training is equally or
gage in more than two physically vigorous activities
more important for females (for gathering activities).
daily outside of school (Simons-Morton et al., 1990).
Although very speculative at present, such hypothe-
Additional evidence for the role of physical activ-
ses could lead to more differentiated predictions con-
ity play and endurance training comes from compari-
cerning gender differences in types of exercise play,
sons of athletes and nonathletes. Although we recog-
analogous to Silverman and Eals's (1992) differentia-
nize the limitations of these comparisons for making
hypothesis does not explain the specific age course
tion of types of spatial ability in relation to gender
of exercise play, nor does it explain the gender differ-
Two other functional hypotheses regarding the ence found. We hypothesize that thermoregulation is
an incidental benefit of exercise play, and that ther-
physical component of exercise play can be taken
moregulation, when important, can also be achieved
from the animal literature; these are the fat-reduction
by other means so far as humans are concerned.
hypothesis and the thermoregulation hypothesis.
In summary, we postulate that exercise play func-
Initial studies of physical activity play in children
were theoretically framed in variants of Spencer's tions primarily to develop physical strength, endur-
(1898) Surplus Energy theory, in which such play was ance, and economy of movement; there is some evi-
dence to support the fat-reduction theory; and
seen as a way of dissipating energy surplus to bodily
thermoregulation is postulated to be an incidental
requirements. Although little logical or empirical
benefit of exercise play. We next examine possible
support currently exists for this theory (Burghardt,
1984, 1988; Smith, 1982; Smith & Hagan, 1980), Barber cognitive functions for exercise play.
(1991) develops a variant of the argument. He sug-
gests that energy is not usually in short supply for
Cognitive Performance and Exercise Play
young mammals, and that play prevents obesity by
Less obvious than the likely benefits of exercise
ensuring that quot;surplusquot; energy is not stored as un-
necessary fat. In particular, some young animals may play for physical development are possible effects on
psychological, and especially cognitive, factors. Here
need to consume large amounts of food to get enough
we review whether engaging in exercise play has
protein, and play can quot;burn offquot; the excess require-
proximal consequences for cognitive performance.
ments. In addition, play, by generating heat, may
Effects on cognitive tasks might be expected from
provide defense against cold exposure.
These postulated functions could be applied to hu- several theoretical viewpoints; however, these theo-
ries link any cognitive benefits to outcomes one step
man children as well as to other young mammals.
The childhood period, which involves high rates and removed from exercise play per se, namely, arousal,
breaks from cognitive tasks, and sense of mastery or
levels of physical activity play, also corresponds neg-
atively to gains in fat (Tanner, 1970, p. 86). That is, well-being.
First, exercise play can lead to heightened arousal,
after 9 months of age, when play involving locomo-
which might influence performance following the in-
tion is increasing, fat gain has a negative velocity un-
verted-U hypothesis (Tomporowski & Ellis, 1986).
til 6 to 8 years of age. Although these age trends are
That is, moderate levels of arousal lead to better per-
consistent with Barber's hypothesis applied to hu-
formance than do higher or lower levels. Alterna-
man children, gender differences are not; there seems
tively, increased arousal may lead to a narrowing of
no reason to suppose that fat reduction should be
attention to core task components (Easterbrook,
more important for boys in the preschool years.
The vigorous dimension of exercise play may also 1959).
serve an immediate function in relation to thermo- Second, exercise play might, by breaking up cogni-
tive tasks, provide spaced or distributed practice
ability of individuals to regulate
rather than massed practice. In another version of
their body temperatures (Burghardt, 1988). Ac-
this approach, stemming from the cognitive immatu-
cording to this hypothesis, children would engage in
rity hypothesis, the specifically playful nature of the
exercise play, which expends stored caloric energy,
break could be considered important. Relatedly, ef-
to raise their body temperature when the ambient
fects on performance, possibly mediated by breaks
temperature is low (Barber, 1991). The empirical rec-
and enhanced attention, might be dependent upon
ord provides some support for this claim. Studies of
enhanced feelings of mastery, or of well-being, after
outdoor play have found that exercise play is in-
creased by cool ambient temperatures for preschool
Unfortunately, much of the available research on
(Smith & Hagan, 1980) and primary school children
exercise and cognitive performance is on adults.
(Pellegrini et al., 1995), and low levels of exercise play
Tomporowski and Ellis (1986) provided a compre-
are observed in tropical climates (Cullumbine, 1950).
hensive review of 27 such studies directly linking ex-
Whereas exercise play might be used by children to
ercise intervention-only some of which can be de-
raise body temperature, nonplayful physical activity
fined as exercise play-to cognitive performance.
is another way in which human beings of all ages
They considered effects of different types of exercise
(and other mammals; Barber, 1991) can raise their
(short or long duration; anaerobic or aerobic) on cog-
body temperatures in cold climates. Therefore, these
benefits are not limited to childhood. In addition, this nitive tasks given during and / or after the exercise.
and Smith 585
The pattern of findings is conflicting. The only con- with primary school children, Pellegrini and col-
sistent trends emerged from studies of brief, high- leagues (1995) examined the relation between the
level of physical activity in exercise play on the
intensity anaerobic exercise (hand dynamometer, or
school playground during recess and subsequent at-
weight-pulling); here, moderate levels of muscular
tention to standardized classroom tasks. They found
exertion usually improved cognitive task perfor-
no relation between post-recess attention and levels
mance assessed during the exercise (e.g., participants
of physical activity in recess exercise play, as mea-
would grip a hand dynamometer in each hand and
sured behaviorally by an indicator of caloric expendi-
recall nonsense syllables). This generally inconclu-
ture ranging from a resting state to a highly vigorous
sive pattern of findings may be due both to inconsis-
state. Attention to standardized tasks after recess,
tencies across studies and to confounding factors,
however, was higher than before recess. They con-
such as different levels of participants' motivation to
participate or different initial levels of physical fit- cluded that the break between tasks, not physical ac-
tivity in exercise play per se, was responsible for the
Although information on children is limited, a increased attention. From this viewpoint, any break
that provided an activity different from the class-
large-scale study of the benefits of guided physical
room activities could serve this function.
activity (physical education classes) on children's
school performance was conducted by Shephard and
colleagues (1983; Volle et al., 1982). As part of the The Cognitive Immaturity Hypothesis
Trois Rivieres project in Canada, entire primary
An adaptation of the cognitive immaturity hy-
school classrooms (grades 2-6) received an addi-
pothesis (Bjorklund & Green, 1992; Pellegrini &
tional 5 hr of physical education per week; control
Bjorklund, 1997) would suggest that not only is
classrooms received no physical education. Teachers
apparently were aware of children's condition as- spaced or distributed practice important for attention
signments, but the influence of teacher bias was and cognitive performance, but that the nature of the
minimized by the administration of independent, intervening break period is crucial; and, specifically,
that activity of a playful nature might be important.
province-wide examinations. The academic perfor-
This hypothesis holds that immaturity is not just
mance of children in the experimental group was su-
something to overcome, but that children's immature
perior to that of the control children. The authors sug-
gest that the benefits shown in this project may have nervous systems may be adapted to deal effectively
with the cognitive demands they face in their daily
been due either to arousal caused by the enhanced
exercise or by shortening of class work time (and lives at those particular developmental periods. Spe-
hence quot;spaced practicequot;) by inserting physical edu- cifically, young children have difficulty keeping ex-
traneous information from entering short-term mem-
cation classes (Shephard, 1983).
ory store. As a result, their working memories are
The latter interpretation is consistent with Steven-
son and Lee's interpretation (1990) of achievement in often cluttered with irrelevant information, leaving
less mental capacity for task-relevant information or
Japanese, Taiwanese, and American schools. They
for the execution of cognitive strategies. Younger
suggested that the frequent breaks between periods
children may require a greater change in activity or
of intense work in Japanese schools (usually 10 min
stimulus materials before they experience a release
every hour) maximize children's cognitive perfor-
from interference (Bjorklund, 1978); hence, a shorter
mance. That children's task vigilance increases when
attention span in childhood, and an inclination to
the time spent on the task is distributed, rather than
play, may have adaptive value.
massed, is a consistent finding in the animal and hu-
From this perspective, there would be a general
man learning literatures (Dempster, 1988). The
increase in interference when children perform a se-
massed versus distributed practice literature consis-
ries of cognitive tasks, regardless of the nature of
tently shows few differences across age from pre-
those tasks. Although one would predict, from a
school to old age (Dempster, 1988). Individual differ-
massed versus distributed practice perspective, that
ences in children, such as quot;distractibilityquot; or
changing from one type of cognitive activity to an-
quot;activityquot; levels, may mediate the effectiveness of
other would yield some attentional benefit, children,
distributed practice regimens, however, and should
be addressed in future research. especially young children, may experience a contin-
ued buildup of interference with repeated perfor-
It thus remains questionable whether it is exercise
mance of even very different cognitive/academic
play per se, rather than just a break from sustained
tasks, all of which require concentration; thus, chil-
classroom work, that is responsible for any increased
dren may experience greater benefit from a drastic
cognitive performance. In a series of experiments
586 Child Development
change in activity, such as is afforded by exercise childhood. Rough-and-tumble play peaks later than
play at recess. exercise play, during the middle childhood period,
Exercise play could be beneficial in this way, both when it accounts for about 10%of free-play behavior;
because it is nonacademic and also because it is self- it then declines in adolescence, accounting for less
motivated, playful, and associated with enjoyment, than 5%. This peak period co-occurs with children's
self-efficacy, and mastery. The cognitive immaturity establishing social status in peer groups (Waters &
hypothesis (Bjorklund, in press; Bjorklund & Green, Sroufe, 1983).
1992), as well as other theories of play (Pellis & Pellis, The second is an argument by design. Rough-and-
in press; Piaget, 1962; Vygotsky, 1978), postulate that tumble play is a distinctive form of behavior. It is su-
children's play is associated with a sense of mastery perficially similar to real fighting; however, it is dif-
and well-being, and that this has cognitive implica- ferent in many respects and should be regarded as a
tions. separate construct (Pellis & Pellis, in press). A num-
On this viewpoint, the playful nature of exercise ber of distinguishing criteria have been identified by
play would be important for its attentional benefits in observational and ethological research (Smith, 1989);
providing a break in academic, teacher-led cognitive in addition, young children from a variety of nations
tasks; especially for preschool and young elementary also consider them to be distinct (Boulton, 1993; Cos-
school children, such a function is consistent with the tabile et al., 1991; Pellegrini, 1989b). That R & T and
age distribution of exercise play. Once again, how- aggression are independent systems during child-
ever, a gender difference would not be predicted. hood also has support from behavioral observations
In summary, evidence for the direct role of exer- of children from numerous cultures showing that
cise play in cognitive performance is incomplete. It R & T, at least before adolescence, is not correlated
does not seem likely that any such benefits are a pri- with, and does not escalate to, aggression for most
mary function of exercise play. They do not explain children (Blurton Jones, 1972; Fry, 1987; Pellegrini,
the gender difference observed. In addition, the orga- 1988). Finally, the endocrinological and neural con-
nized massed cognitive practice of schooling is a re- trols for fighting and for R & T appear to be different
cent cultural invention, and therefore the comple- in many mammalian species (Meaney et al., 1985).
mentary benefits of spaced practice provided by We review here evidence relating R & T to fighting
exercise play-even if specifically related to its play- skills, to dominance functions, and to skills of emo-
ful nature, as the cognitive immaturity hypothesis tional encoding and decoding.
predicts-would also be recent in human history. We
hypothesize that the cognitive benefits associated R & T and FightingSkills
with exercise play are in fact incidental or serendipi-
tous, due to the fact that it provides a playful break The most traditional view in the animal and hu-
from demanding tasks. man literature (Smith, 1982; Symons, 1978) is that
Thus far we have considered the benefits of the R & T functions to provide safe practice for fighting
physical component of exercise play, irrespective of (and possibly hunting) skills that will be useful in
its social nature. However, children often engage in later life. This hypothesis would be consistent with
physical activity play with partners; and rough-and- the strong gender difference observed, if one as-
tumble play (play fighting, wrestling) is a specialized sumes that fighting and hunting skills were and are
form of physical activity play with a distinctive age more characteristically male activities (Boulton &
curve, suggesting that it has its own functions. We Smith, 1992). It does not, however, predict the age
propose that such functions will be primarily social, curve for R & T, because quot;safequot; practice of such skills
because purely physical or cognitive functions of the might be especially important in adolescence, when
kind reviewed earlier could be provided by exercise R & T declines. Also, there is little or no direct evi-
play generally and may not require the specific forms dence linking R & T to fighting or hunting skills, in
of R & T observed. We also recognize, as noted either the animal or human literature. Finally, this
above, the possibility that R & T may have motor hypothesis does not predict the age changes in
training benefits. quot;cheatingquot; observed in human R & T. Players
quot;cheatquot; when they violate the rules of play by manip-
ulating the nonserious tenor of the activity for their
Social Functions and Rough-and-Tumble Play
own, exploitive ends. We do not dismiss this argu-
The distinct functional significance of R & T is sug- ment-indeed, we suspect that it may be a phyloge-
gested by two arguments. The first is the relative and netically prior function with some remaining rele-
peak frequencies with which it is observed during vance for younger children. However, we turn to the
and Smith 587
argument for dominance functions of R & T, which which are related to physical prowess, to regulate ac-
cess to resources (e.g., struggling over access to a
have not received the same attention but which may
toy). Physical prowess, or toughness, when used in
yield new insights in the case of human R & T.
conjunction with more affiliative skills is an impor-
tant dimension of boys' peer group status (Strayer,
R & T and Dominance Relationships
1980), popularity (Pellegrini, 1995b; Vaughn & Wa-
ters, 1981), and social leadership (Hartup, 1983).
Dominance is defined as a dyadic, affiliative rela-
Age trends in R & T are also consistent with a
tionship between individuals, not in terms of one's
dominance function if we consider that the immedi-
aggressiveness (Hinde, 1974). Dominance hierarchies
ate preadolescent period is one in which it is impor-
are generally unique to specific groups and ecologies,
tant to establish peer group dominance. At this age,
the implication being that individuals might have
youngsters experience rapid change in body size,
different dominance status in different groups and
along with changes in environment, as they move
ecologies (deWaal, 1985; Hartup, 1983; Strayer, 1980).
from primary to secondary school. The argument that
Although explicable in terms of advantage to domi-
males use R & T to establish dominance as they move
nant individuals, dominance hierarchies also medi-
into adolescence is also consistent with theory and
ate group members' access to valued resources and
research in the animal literature (Dunbar, 1988; Fa-
reduce intragroup aggression in many situations
Data from observations and interviews suggest
We postulate that R & T may serve a social func-
that R & T may be involved in dominance in two
tion in peer groups, for boys especially, by assisting
ways, each of which is age-related. The first is indi-
in the establishing and maintaining of dominance re-
rect; R & T may provide a way of assessing the
lationships. The idea that R & T is related to estab-
strength of others so as to decide one's strategy vis-
lishing and maintaining dominance status is consis-
a-vis dominance competition-a form of quot;ritualized
tent with arguments from design. Males often use
aggression,quot; as described in other mammalian spe-
quasi-agonistic displays (e.g., soft or no-contact kicks
cies, which leads to real fighting in only certain cir-
and punches, light pushes) in the service of domi-
nance. Very similar behaviors are also displayed in cumstances. Observations by Paquette (1994) support
the idea that young chimpanzees can use R & T to
R & T (Blurton Jones, 1972), but these behaviors are
learn the strengths and weaknesses of others in plan-
embedded in a nonserious context: Kicks and
ning challenges for dominance, suggesting that
punches do not make contact, and if they do they are
R & T is more useful for this than real fights because
soft; players are smiling; and they often handicap
of the lesser chance of injury or intervention by a
themselves (e.g., let the player on the bottom of a pile
get on top).
Similarly, children's R & T occurs in symmetrical
Because self-handicapping seems to occur in the
groups, or with children of similar dominance status,
R & T of both nonhuman primates and children, Sy-
and many children say they can determine their own
mons (1978) was critical of the hypothesis that R & T
as well as peers' strength from these encounters
is related to dominance. However, subsequent find-
(Smith, Hunter, Carvalho, & Costabile, 1992). R & T
ings counter this argument in two ways. First, chil-
tends to occur between friends (Humphreys & Smith,
dren can often evaluate the strength of others from
1987; Smith & Lewis, 1985) and in groups of three to
R & T bouts, despite self-handicapping and restraint.
four children (Pellegrini, 1993); these findings indi-
Second, in some youngsters, especially by adoles-
cate that it is a safe and relatively visible venue to
cence, it now appears that subtle or not so subtle
test and exhibit physical strength. Thus, through
forms of quot;cheatingquot; may occur, demonstrating
R & T, children can, in an indirect way, assess their
clearly to opponents and to onlookers that one partic-
own strength and that of others; in this way they pre-
ipant is in fact stronger (Neill, 1976; Pellegrini, 1995b;
pare for dominance encounters through the primary
Smith & Boulton, 1990). In fewer cases, youngsters
initiating R & T turn it into aggression (Pellegrini,
The second way in which R & T may provide the
A dominance function is also consistent with the context for establishing or maintaining dominance is
more direct. Participants may use an R & T bout to
gender differences in R & T. Children establish and
get their partners in a position where they can actu-
maintain dominance in different ways. Girls use pri-
ally display their superior strength or assert domi-
marily verbal rather than physical means to gain and
nance, for example, by pinning or intimidating a
keep resources (Charlesworth & Dzur, 1987). Boys,
playmate. Participants doing this may have lulled
on the other hand, use a variety of skills, some of
(1995b) shed further light on this age trend. In a lon-
their partners into a false sense of security by using
the predominantly playful nature of R & T, or may gitudinal study of adolescent boys, he found that
asymmetrical choices for R & T were observed dur-
have used the self-handicapping and reversal criteria
ing the first year of middle school (12 years), but not
of R & T to get themselves into a quot;winningquot; position.
This could be called a quot;cheatingquot; use of R & T for the second (13 years). He also found that during the
first year of middle school, boys' R & T was corre-
For preadolescents, in contrast to adolescents, lated with peer-nominated dominance; only with so-
R & T is not correlated with peer-nominated domi- ciometrically rejected boys, not popular or average
boys, did R & T lead to and relate to aggression, how-
nance, and it occurs with partners of similar domi-
nance status (Humphreys & Smith, 1987; Pellegrini, ever. During the second year of middle school,
R & T continued to relate to dominance status, but
1993); also, in most cases, R & T is not exploited for
it did not lead or relate to aggression. These results
immediate aggressive ends (Pellegrini, 1988). This
suggest that R & T is used to establish dominance in
suggests that R & T may not be used to establish
dominance in this second direct way before adoles- early adolescence; once established, the hierarchy re-
duces aggression, and R & T declines.
Rough-and-tumble play and actual fighting re-
main separate for most children during the primary
R & T and EmotionalCoding Skills
school years; there are cases, however, involving so-
An important dimension of social skill is the abil-
ciometrically rejected children especially (Pellegrini,
1988), where R & T and fighting are linked. The eth- ity to encode and decode social signals. Successful
nographic record also provides illustrations. Sluck- encoding and decoding of messages, such as quot;This is
in's (1981) in-depth study of British 5- to 9-year-old play,quot; is necessary if play is to be initiated and main-
children's behavior and perceptions of their lives in tained (Bateson, 1972; Bekoff, 1995). Behaviors that
the school playground provides examples of R & T send the message quot;This is playquot; are typically exag-
being used to deceive and manipulate peers. Simi- gerated, compared to more functional counterparts
(Biben & Suomi, 1993); for example, play fighting,
larly, the work of Oswald and colleagues (Oswald,
Krappmann, Chowduri, & Salisch, 1987) in Germany compared to real fighting, might be characterized by
an open mouth, hunched shoulders, and rhythmic
with children aged 6 to 10 years found instances of
hurtfulness in the play of the older children in this movement of the hands.
Research by Parke and colleagues indicates that
However, a different picture emerges in early ado- the ability to encode and decode play signals can
originate in vigorous play between parents (primar-
lescence. Neill (1976, p. 219) was the first to suggest
that adolescent boys' R & T might be used to estab- ily fathers) and their children (primarily sons) begin-
ning in infancy and continuing throughout child-
lish dominance. His factor-analytic study of 12- to 13-
year-old boys' playground behavior found that hood (Carson et al., 1993; Parke, Cassidy, Burks,
R & T and aggression often co-occurred. Neill stated Carson, & Boyum, 1992). They found the amount of
time spent in vigorous play bouts to be positively re-
that R & T might be a quot;means of asserting or main-
taining dominance; once a weaker boy has registered lated to preschool children's ability to decode emo-
distress the bond can be maintained by the fight tak- tional expressions, such as happy, sad, angry, scared,
ing a more playful form, but if he does not do so at and neutral (Parke et al., 1992). Further, children's ex-
pression of emotional states was also related posi-
the start of the fight, the stronger boy may increase
the intensity of the fight until he does.quot; tively to bout length (Parke et al., 1992). Involvement
in R & T with peers, expressed in terms of proportion
This age change in the function of R & T received
some support from Humphreys and Smith (1987). of total behavioral output, has also been found to re-
They found that at 11 years, but not at 7 and 9 years, late to primary school children's ability to decode
dominance was a factor in partner choice in R & T. play signals (Pellegrini, 1988). It may thus be the case
that parent-child play provides the groundwork for
When the younger children engaged in R & T, they
children's ability to encode and decode emotions,
did so in symmetrical groups, or with peers of similar
dominance status; for the older children, dominant with this ability later being used in physical activity
play with peers.
youngsters initiated R & T with less dominant
However, there are difficulties with this hypothe-
youngsters, or in asymmetrical groups.
sis. First, these correlational studies do not establish
This finding would be consistent with stronger
cause and effect; it is equally plausible that the causal
children using R & T to exhibit dominance with
relation is such that those children less able to
weaker children. Results from a study by Pellegrini
and Smith 589
tions of strength training, endurance training, and
encode/decode emotions are less willing to engage
skill and economy of movement.
in R & T. Second, and more conclusively, these
Related to this, we need more data on gender dif-
hypotheses are inconsistent with observed gender
ferences in exercise play, unconfounded by R & T.
differences; encoding and decoding of emotions
We can be confident of the gender differences in
should be just as important for girls as for boys, and
R & T, and they fit readily with the predominant
they certainly are no worse at it than boys. Yet the
functional hypothesis for R & T-social dominance.
gender difference in R & T is a well established
Gender differences in exercise play are more intri-
guing with respect to functional hypotheses. Con-
In summary, we hypothesize that the primary
ceivably, these differences may be more complex and
function of R & T through the primary years is to pro-
differentiated than previously thought, as Silverman
vide a way in which boys assess the strength of others
and Eals (1992) found in the domain of spatial ability;
for dominance purposes; it may also provide practice
for example, gender differences might be different for
in fighting skills, but little direct evidence exists to
aspects of exercise play relevant to strength as op-
support this. There is good evidence that in early ad-
posed to those relevant to endurance.
olescence (and perhaps earlier for rejected children)
R & T can function to actually establish dominance
status in boys' peer groups. The contemporaneous
Functions of Exercise Play for Strength, Endurance,
correlations between R & T and dominance and
and Economy and Skill of Movement
R & T and popularity for adolescent boys suggest
The benefits of exercise play for two dimensions
that R & T is only one behavioral strategy used by
of motor training-strength and endurance-should
boys to gain and maintain status. Finally, we hypoth-
be immediate and occur across the life span. By im-
esize that any benefits for emotional encoding or de-
mediate we mean that strength and endurance will
coding are incidental benefits of R & T, achievable in
result from repeated activity bouts, usually across the
other ways, rather than functions.
span of a number of weeks (Byers & Walker, 1995).
However, we need more information on duration,
frequency, and intensity of physical exercise of both
SUGGESTIONS FOR FUTURE RESEARCH
the playful and nonplayful variety from infancy
In light of the relative paucity of research in the area
through adulthood, and the correspondence between
of physical activity play, we suggest directions for
these data and measures of immediate and sustained
future research on the nature of such play and its hy-
fitness. To test the hypothesis that exercise play re-
pothesized benefits in physical, cognitive, and social lates to bone remodeling and physical endurance and
strength, as measured by decreased heart rate during
exercise or VO2Max (the maximum rate of oxygen
uptake during exercise), we need to measure the sep-
The Need for Naturalistic Data on Forms and
arate contributions of nonplayful and playful vigor-
Amounts of Physical Activity Play
ous activity to physiological measures of endurance
There is a need for more descriptive data on the and strength.
forms of physical activity play and their age trends
through childhood and adolescence. The review by Function of Exercise Play for Cognitive
Byers and Walker (1995) and our own analyses here Performance
show the importance of age trends in examining
The beneficial consequences of exercise play for
functional hypotheses. Yet, for exercise play espe-
cially, data are scanty. Although we can be reason- children's cognitive functioning are not firmly estab-
ably confident of the inverted-U curve with age, we lished. Although exercise play breaks are related to
cannot be confident that the peak is at 4-5 years as children's attention to classroom tasks (Pellegrini &
we have assumed. Also, different types of exercise Davis, 1993; Pellegrini et al., 1995), alternative expla-
nations abound. Future research should examine the
play may peak at different ages, and the develop-
mental course may be different for boys and girls. extent to which cognitive performance-for example,
task attention measured by gaze and heart-rate vari-
The difference in age peaks between exercise play
and R & T, together with other differences, has led us ability and possibly vagal tone (Obrist, Howard, Sut-
terer, Hennis, & Murrell, 1973; Porges, 1992) for chil-
to separate them for functional consideration. Simi-
larly, different aspects of exercise play and R & T dren of different ages-is increased as a function of
may relate to somewhat different, if related, func- quot;breaksquot; during cognitive tasks.
590 Child Development
Determining the character of the breaks-how have been concerned primarily with pretend play,
playful, how physically vigorous, or how aca- with some attention given to functional and construc-
demic-is crucial for explaining extant and future tive play (Rubin et al., 1983). Given the co-occurrence
findings. Distributed practice theory predicts that of R & T and pretend and the theoretical bias toward
any break in the duration of cognitive tasks, whether studying pretend, it may have been the case that the
it be playful or nonplayful, should increase perfor- occurrence of R & T during the preschool period has
mance on the criterion task. The cognitive immaturity been underreported. For example, children's play fol-
hypothesis predicts, on the other hand, that playful lowing a quot;Star Warsquot; theme is usually coded as so-
breaks between demanding tasks where close atten- cial pretend, not R & T (as reported in Rubin et al.,
tion is expected, such as computing mathematics 1983, p. 723). Future research should reevaluate the
problems, are necessary to reduce cognitive interfer- place of R & T during the preschool period by consid-
ence and facilitate preschool children's performance. ering its pretend and nonpretend dimensions, as well
However, the theory does not make a prediction re- as the play fighting and chase dimensions.
lated to the physically vigorous dimension of play. For a number of reasons, the distinction between
Age of the participant is an important factor in play fighting and play chasing is an important one
testing these theories. Distributed practice affects for future work. First, most young children enjoy
cognitive performance across the life span (e.g., chasing, but fewer, mostly boys, enjoy play fighting
Dempster, 1988), whereas the effects of play, follow- (Smith et al., 1992). Second, chasing and play fighting
ing the cognitive immaturity hypothesis, should be are statistically independent of each other and have
limited to the niche of childhood. According to the different consequences (Pellegrini, 1993). Further, for
cognitive immaturity hypothesis only, the playful na- older and rejected youngsters, play fighting relates
ture of the break should become less crucial with in- to dominance status, whereas chasing does not (Pel-
The cognitive immaturity hypothesis also posits Related to this, we need to know more about the
cognitive benefits associated with increased feelings ways in which R & T is used by boys to establish and
of self-efficacy. Young children should overestimate maintain social leadership and dominance in their
their competence in cognitive and social domains; peer groups. We hypothesize that R & T is used pri-
this orientation should then lead to increased self- marily to assess the strength of others in the peer
efficacy, which should, in turn, encourage children to group. This hypothesis is based on several untested
explore new arenas and to persist at difficult tasks assumptions. First, those boys who engage in more
(Bjorklund, in press). That preschool children overes- R & T should have more accurate knowledge of rela-
timate their own social and cognitive performance on tive strength or quot;toughnessquot; in the peer group (see
a number of social and cognitive tasks, and that these Sluckin & Smith, 1977). Second, particular choices of
estimates become more realistic by the late elemen- R & T partners should relate to particular knowledge
tary school years, is well established (Stipek & of relative strength vis-a-vis those partners. It would
MacIver, 1989). Thus, the increase in children's self- be particularly relevant to engage in R & T with a boy
efficacy attributed to play may result from this more joining the peer group, whose strength was not
general orientation to overestimate their own perfor- known. Relatedly, R & T would be an important pre-
mance. At the simplest level, this hypothesis could dictor of peer leadership, including dominance, espe-
be tested by examining the correlations between the cially as boys enter new social institutions, such as a
frequency with which children engage in R & T and new school.
the degree to which they overestimate their own We posit that an alternative pathway for boys to
toughness, after more general levels of overestima- use R & T for dominance purposes is directly via
tion are controlled. quot;cheating.quot; However, quot;cheatingquot; is a double-edged
tool; public display of dominance in this way both
enhances the success of that particular bout for domi-
Functions of R & T
nance assertion, but also starts to get the boy who
More naturalistic data are needed for R & T in the cheated a particular reputation in this respect. Thus,
preschool and primary school periods, and especially children should not cheat repeatedly at R & T. Where
at the transition into adolescence. Even during the cheating does occur, it may be in the presence of a
preschool and primary school period, R & T is not crowd who can witness their exhibition. It may also
sufficiently well understood. Most studies of pre- be the case that boys initially cheat at R & T by in-
schoolers' play, following Piagetian (1962) theory, flicting pain and thereby gaining public notice of
and Smith 591
mance, and social dominance skills. A complemen-
their quot;toughness,quot; then apologize (under the guise
of an quot;honest mistakequot;) and resume another form of tary approach to the study of function, cost-benefit
analysis, has been advanced by behavioral ecologists
play or social interaction.
An uninvestigated topic in this context is chil- (e.g., Krebs & Davies, 1993) and applied to animal
and children's play with generally supportive results
dren's ability to detect cheating in R & T. Ability to
(e.g., Caro, 1995; Pellegrini et al., 1998). From an evo-
detect cheating could be measured by children's re-
lutionary perspective, costs associated with play
sponses to filmed play and real fighting bouts. Alter-
should have corresponding benefits for the individu-
natively, naturalistic data could be gathered on chil-
als of the species in which the play behavior is typi-
dren's actual responses (either as participants or
onlookers) to cheating in R & T bouts. Such direct ob- cally observed. If this were not the case, play would
not have been naturally selected for and maintained
servations would be difficult to collect given the rela-
tive infrequency of cheating in R & T, so use of hypo- across generations.
Costs associated with physical activity play can be
thetical situations, although less ecologically valid,
expressed in terms of time spent playing, calories, or
may be more practical.
energy expended during play, and in terms of survi-
There may also be cognitive implications of chil-
dren's ability to detect cheating in social interactions vorship where death or injury result from play (Mar-
(i.e., accepting the benefits of a social contract with- tin & Caro, 1985). High costs should be associated
with high benefits, and low costs should more likely
out paying the accepted cost). Ethological theory
be associated with low benefits. Benefits for play
(e.g., Humphrey, 1976) and recent evidence (Cos-
need not be absolutely high, but merely greater than
mides, 1989) suggest that detection of cheating may
be an important component in the evolution of quot;so- associated costs (Caro, 1995).
Application of a cost-benefit analysis to children's
cial intelligence.quot; There may be a relation between
play would be useful on a number of fronts. First, it
engaging in R & T and the ability to later detect
cheating or violations, not only in the specific context would empirically test the widely held assumption
that play during childhood consumes a substantial
of R & T, as just described, but also on socially
portion of children's time and energy budgets. Sec-
framed conditional logic problems as used by Cos-
ond, a description of the time and energy expendi-
mides (i.e., do people test hypotheses by searching
ture on physical activity play across childhood would
for evidence that could falsify them).
Gender differences in R & T also warrant future complement the information provided in this review
and that provided by Byers and Walker (1995). Func-
research. Specifically, direct observations should dis-
tinguish initiation of R & T bouts, and response to tional hypotheses could be evaluated by relating dif-
such initiations. For example, do boys and girls re- ferent measures of cost to measures of motor training,
spond differently to R & T initiations? Gender differ- cognitive performance, and social dominance during
ences in the preference for R & T may be related to childhood and into adulthood.
differential responses to physical stimulation gener-
ally (Meaney et al., 1985). CONCLUSION
But it also may be the case that there are individual
differences, associated with factors such as CAH and We have undertaken a functional analysis of a ne-
temperament, within each gender. Longitudinal ob- glected aspect of play, physical activity play. It has
servations could be made of CAH and non-CAH not been extensively studied in the child develop-
girls' and boys' sensitivity to tactile stimulation, as ment literature, yet the ethological literature suggests
well as their R & T with parents and then peers. Early that of all forms of play, it is among the best candi-
observations of tactile sensitivity and subsequent dates for serving developmental function(s) (Fagen,
play with parents should provide information on the 1981).
In the literature reviewed here, certain functions
specific and interactive contributions of each factor
to children's engagement in R & T with peers. emerge as being better supported than others. We re-
late these functions to three main forms of physical
activity play, distinguished by successive age peaks
Testing Functional Hypotheses with Cost-Benefit in incidence. We postulate that rhythmic stereotypies
Analyses function primarily for establishing voluntary motor
We have made functional inferences based on the
Exercise play peaks in the preschool years. The
co-occurrence of physical activity play and beneficial
vigorous physical component of exercise play has im-
consequences in physical training, cognitive perfor-
stress on deferred rather than immediate benefits of
mediate beneficial consequences for children in mo-
play deserves reevaluation.
tor training. This is consistent with ethological/evo-
Even if benefits of physical activity play are more
lutionary reasoning as well as with the evidence, and
immediate than deferred, they may still be impor-
might be hypothesized to be the earliest ultimate
tant. There are public health implications for the role
function for exercise play in mammals. There may be
of physical activity play for the physical fitness of
additional benefits of fat reduction and thermoregu-
children growing up in a modern industrial society.
lation, which we hypothesize to be incidental. The
Children have limited opportunities for physical ac-
physical component of exercise play also may have
tivity, due to shortage of play spaces, dangerous
benefits for cognitive performance. We view the evi-
neighborhoods, and the increased demands of formal
dence here as more tenuous; cognitive benefits at-
schooling. That children seem to quot;needquot; physical ac-
tributed to physically vigorous dimensions of play
tivity is supported by the rebound effects observed
may be epiphenomenal to breaks from work, which
in deprivation studies. The evidence suggests that if
need not necessarily be physically vigorous, or even
children are deprived of physical activity play for
long periods of time, their health, in terms of cardio-
Rough-and-tumble is a form of play with a strong
vascular and physical fitness, may suffer. There may
gender difference, which peaks later, in middle child-
also be social consequences of R & T, a common form
hood. The predominant hypothesis has been that
of physical activity play that may be necessary in the
R & T may provide training for fighting/hunting
normal developmental sequence of boys' peer
skills, especially in males (Smith, 1982; Symons,
groups. We hope that future research efforts can be
1978). We do not dismiss this hypothesis; it may, in
directed both to the theoretical issues of function and
fact, have been an originating or ultimate function for
to the practical implications affecting children's
R & T as a social form of physical play found in cer-
health and development.
tain species, but direct support for this hyposthesis
is minimal. Instead, we develop the hypothesis that,
in human children especially, R & T can function to
develop and maintain leadership and dominance in
We acknowledge the comments of G. M. Burghardt
the peer group, especially for adolescents. These tra-
and L. Galda, as well as beneficial discussions with
jectories interact with sociometric status in complex
D. J. Bjorklund, P. Blatchford, T. M. Caro, R. Dis-
ways still to be fully explored. Less well supported
hman, R. Fagen, C. Hamilton, and B. Sutton-Smith.
is the hypothesis that parent-infant rough play and
This work was partially supported by grants to the
R & T may function to improve skills of encoding and
first author from the W. T. Grant Foundation and the
decoding emotional signals.
School of Education, University of Wales, Cardiff.
In summary, we believe that physical activity play
deserves greater attention from psychologists and
educators. In general, our conclusions have been
ADDRESSES AND AFFILIATIONS
strongly tempered by the insufficiency of available
Corresponding author: A. D. Pellegrini, Department
evidence. What evidence there is has not infrequently
of Educational Psychology, Burton Hall, University
come from areas such as sports science, rather than
of Minnesota, Minneapolis, MN 55455. P. K. Smith is
psychology, with a consequent neglect of psychologi-
at Goldsmiths College, the University of London.
cal variables. There is scope for considerable concep-
tual rethinking in the area; in particular, the usual
Empirical Studies of Age and Gender Trends
Physical activity, Other play
0-6 years Observations
Playground behavior, Peer group
11 and 12 years Observations
4 years Observations, Actometer
Eaton & Keats Physical activity
1989 83 6.6 years
Eaton & Yu Physical activity
Physical activity, Other play
1994 25-60 months Observations
Hovell et al. 1978 Physical activity
R & T, Friendship, Dominance
7-11 years Observations, Peer
Humphreys & Smith 1987
Movement, Subsistence play
1972 newborn-5 years
R & T, Vigorous play, Social behavior
36-58 months Observation
Play, R & T
1984 20 3-4 years Observations
R & T, Games, Aggression
1988 5-12 years
32 Observations, Sociometry
R & T, Games, Aggression
94 5-12 years Observations
Vigorous play, R & T
1990 5-12 years
R & T, Aggression
82 12-13 years Observations, Sociometry,
R & T, Physical play
Roopnarine et al. Observations
Object play, Motor play, Peer play
1994 82 Observations
Routh et al. 3-9 years Observations, Physical activity
Simons-Morton et al. 870 9-11 years
1990 Self-report Physical activity
R & T, Vigorous play, Social behavior
1980 28-56 months Observations
Smith & Connolly
1979 28-52 weeks Observations
Thelen Rhythmic stereotypies
1980 28-52 weeks
Thelen Observations, Rhythmic stereotypies
Motor tests, IQ
Studies of Function
Authors N Activity
Cullumbine Douglas bag expenditure, Energy
Pellegrini et al. 5-9 years Observations activity, Attention
45 7-9 years Observations activity, Attention
44 Observations activity, Attention
36 2-4 years Observations
Smith & Hagan play
Strength, Endurance, and Economy of Movement Function
1969 89 11-14 years Aerobic capacity
Cummings et al.
1977 26 8-12 years Aerobic capacity
Lussier & Buskirk
Pellegrini & Davis 1993 Physical activity
Pellegrini et al.a 5-9 years
1995 62 Observations Physical activity, Attention
45 7-9 years Observations Physical activity, Attention
44 9 years Observations Physical activity, Attention
1980 36 2-4 years Observations Exercise play
Smith & Hagana
17 years Strength, Endurance
Smoll & Schutz 1985 Athlete / Nonathlete
1979 20 28-52 weeks Observations
Thelena Rhythmic stereotypies
Table A2 (Continued)
Authors N Method
Date Age Activity
Dominance and Emotional Encoding
2-4 years Social behavior, R & T, Aggression
Blurton Jones 1972 13 Observations
9-11 years Observations, Interviews
Boulton 1991 89
R & T, Aggression
8-11 years Observations, Interviews
Boulton 1993 89
7 months-10 years
Carson et al. 1993 Interviews, Observations Physical play
R & T, Aggression
Costabile et al. 1991 256 8 and 11 years
4 years R & T, Physical activity, Playmates
DiPietro 1981 52
R & T, Aggression
1987 24 3-8 years Observations
R & T, Friendship, Dominance
Humphreys & Smitha 94
R & T, Parent roles
MacDonald 1987 36 3-5 years Observations, Sociometry
R & T, Parent roles, Affect displays
MacDonald & Parke 27 3-4 years
1984 Observations, Sociometry,
MacDonald & Parke 746 0-10 years
1986 Survey Physical play
R & T, Aggression
Neill 1976 34 12-13 years Observations
Social behavior R & T, Fighting
Oswald et al. 52 6 and 10 years
1987 Ethnography: Observations,
Parke et al. 1992 3-4 years Physical play, Parental role
158 Observations, Sociometry
R & T, Games, Aggression
1988 32 Observations, Sociometry
R & T, Games, Aggression
94 5-12 years Observations
R & T, Dominance
1993 Observations, Play partners,
R & T, Dominance, Aggression
1994 54 13 years Observations, Play partners,
R & T, Aggression, Dominance
82 12-13 years Observations, Sociometry
Sluckin 1981 100 Ethnography: Observations, Playground behavior, Aggression
Interviews, Play partners
Social behavior, R & T, Aggression
1973 38 months
R & T, Aggression
Smith & Lewis 1985 Observations, Sociometry,
Interview, Play partners
Cognitive Performance Function
9 years Observations
Pellegrini & Davisa Physical activity
5-9 years Observations Physical activity, Attention
Pellegrini et al.a
7-9 years Observations
45 Physical activity, Attention
44 9 years Observations Physical activity, Attention
6-11 years Break periods
Stevenson & Lee 480
PE training, Achievement, PE programs
Volle et al. 1982 546
Teacher attitude, Observa-
tions, Student diaries
quot;Indicates that the entry has appeared before.
(Eds.), Growing points in ethology (pp. 401-421). Cam-
bridge: Cambridge University Press.
Baldwin, J. D., & Baldwin, J. I. (1977). The role of learning
Bateson, P. P. G. (1981). Discontinuities in development
phenomena in the ontogeny of exploration and play.
and changes in the organization of play in cats. In
In S. Chevalier-Skolinikoff & F. E. Poirer (Eds.), Primate
K. Immelmann, G. Barlow, L. Petrinovich, & M. Main
biosocial development (pp. 343-406). New York: Gar-
(Eds.), Behavioraldevelopment(pp. 281-295). New York:
Cambridge University Press.
Barber, N. (1991). Play and energy regulation in mammals.
Bekoff, M. (1995). Play signals as punctuation: The struc-
Quarterly Review of Biology, 66, 129-147.
ture of social play in canids. Behaviour,132, 419-429.
Bateson, G. (1972). Steps to an ecologyof mind. San Francisco:
Berenbaum, S. A., & Snyder, E. (1995). Early hormonal
influences on childhood sex-typed activity and
Bateson, P. P. G. (1976). Rules and reciprocity in behav-
playmate preferences: Implications for the development
ioural development. In P. P. G. Bateson & R. Hinde
Pellegrini and Smith 595
Caro, T. M. (1995). Short-term costs and correlates of play
of sexual orientation. DevelopmentalPsychology, 31, 31-
in cheetahs. Animal Behaviour,49, 333-345.
Carson, J., Burks, V., & Parke, R. (1993). Parent-child physi-
Biben, M., & Suomi, S. J. (1993). Lessons from primate play.
cal play: Determinants and consequences. In K. Mac-
In K. MacDonald (Ed.), Parent-childplay (pp. 185-196).
Donald (Ed.), Parent-child play (pp. 197-220). Albany:
Albany: State University of New York Press.
State University of New York Press.
Bjorklund, D. J. (1978). Negative transfer in children's recall
Charlesworth, W. R., & Dzur, C. (1987). Gender compari-
of categorized material. Journalof Experimental
sons of preschoolers' behavior and resource utilization
chology, 26, 299-307.
in group problem solving. Child Development, 58, 191-
Bjorklund, D. J. (in press). The role of immaturity in human
Collaer, M. L., & Hines, M. (1995). Human behavioral sex
Bjorklund, D., & Green, B. (1992). The adaptive nature of
differences: A role for gonadal hormones during early
cognitive immaturity. American Psychologist, 47, 46-54.
development. PsychologicalBulletin, 118, 55-107.
Blatchford, P. (1996, October). A national survey of breaktime
Cosmides, L. (1989). The logic of social exchange: Has natu-
in English schools. Paper presented at the annual meet-
ral selection shaped how humans reason? Cognition, 31,
ings of the British Educational Research Association,
Costabile, A., Smith, P. K., Matheson, L., Aston, J., Hunter,
Bloch, M. N. (1989). Young boys' and girls' play in the
T., & Boulton, M. J. (1991). A cross-national comparison
home and in the community: A cultural ecological
of how children distinguish serious and playful fight-
framework. In M. N. Bloch & A. D. Pellegrini (Eds.), The
ing. DevelopmentalPsychology, 27, 881-887.
ecological context of children's play (pp. 120-154). Nor-
Cullumbine, H. (1950). Heat production and energy re-
wood, NJ: Ablex.
quirements of tropical people. Journalof Applied Physiol-
Blurton Jones, N. G. (1972). Categories of child interaction.
ogy, 2, 201-210.
In N. G. Blurton Jones (Ed.), Ethologicalstudies of child
Cummings, C. R., Goulding, D., & Bagley, G. (1969). Cana-
behaviour(pp. 97-129). London: Cambridge University
dian Medical Association Journal, 101, 69-73.
Dempster, F. (1988). The spacing effect. AmericanPsycholo-
Bornstein, M. H., & Tamis-LeMonda, C. S. (1995). Parent-
gist, 43, 627-634.
child symbolic play: Three theories in search of an effect.
deWaal, F. B. M. (1985). The integration of dominance and
DevelopmentalReview, 15, 382-400.
social bonding in primates. Quarterly Review of Biology,
Boulton, M. J. (1991). Partner preference in middle school
children's playful fighting and chasing. Ethologyand So-
DiPietro, J. A. (1981). Rough-and-tumble play: A function
ciobiology, 12, 177-193.
of gender. DevelopmentalPsychology, 17, 50-58.
Boulton, M. J. (1992). Participation in playground activities
Dunbar, R. I. M. (1988). Primate social systems. Ithaca, NY:
at middle school. EducationalResearch,34, 167-182.
Cornell University Press.
Boulton, M. J. (1993). Children's ability to distinguish be-
Easterbrook, J. A. (1959). The effect of emotion on cue utili-
tween playful and aggressive fighting: A developmental
zation and the organization of behavior. Psychological
perspective. British Journal of DevelopmentalPsychology,
Review, 66, 183-201.
Eaton, W. C., & Enns, L. R. (1986). Sex differences in human
Boulton, M. J., & Smith, P. K. (1990). Affective bias in chil-
motor activity level. PsychologicalBulletin, 100, 19-28.
dren's perceptions of dominance relationships. Child De-
Eaton, W. C., & Keats, J. G. (1982). Peer presence, stress,
velopment, 61, 221-229.
and sex differences in motor activity of preschoolers. De-
Boulton, M. J., & Smith, P. K. (1992). The social nature of
velopmentalPsychology, 18, 534-540.
play-fighting and play chasing: Mechanisms and strate-
Eaton, W. C., & Yu, A. P. (1989). Are sex differences in child
gies underlying cooperation and compromise. In J. H.
motor activity level a function of sex differences in mat-
Barkow, L. Cosmides, & J. Tooby (Eds.), Theadaptedmind
urational status? Child Development,60, 1005-1011.
(pp. 429-444). New York and Oxford: Oxford Univer-
Ehrhardt, A. A. (1984). Gender differences: A biosocial per-
spective. In A. Sondergegger (Ed.), Nebraskasymposium
Brownlee, A. (1954). Play in domestic cattle: An investiga-
on motivation (pp. 33-50). Lincoln: University of Ne-
tion into its nature. British VeterinaryJournal, 110, 48-
Fabes, R. A. (1994). Physiological, emotional, and behav-
Burghardt, G. (1984). On the origins of play. In P. K. Smith
ioral correlates of gender segregation. In C. Leaper (Ed.),
(Ed.), Play in animals and humans (pp. 5-42). Oxford:
Blackwell. gender segregation:Causesand consequences(pp.
19-34). San Francisco: Jossey-Bass.
Burghardt, G. (1988). Precocity, play, and the ectotherm-
Fagen, R. (1976). Exercise, play, and physical training in
endotherm transition. In E. M. Blass (Ed.), Handbookof
mammals. In P. P. G. Bateson & P. H. Klopfer (Eds.),
behavioralneurobiology(Vol. 9, pp. 102-148). New York:
Plenum. Perspectives in ethology (Vol. 2, pp. 189-219). London:
Byers, J. A., & Walker, C. (1995). Refining the motor train-
Fagen, R. (1981). Animal play behavior.New York: Oxford.
ing hypothesis for the evolution of play. AmericanNatu-
Fagot, B. I. (1974). Sex differences in toddlers' behavior and
ralist, 146, 25-40.
596 Child Development
ogy of a foraging people. In N. Blurton Jones (Ed.), Etho-
parental reaction. Developmental Psychology, 10, 554-
558. logical studies of child behaviour(pp. 285-304). London:
Fagot, B. I. (1994). Peer relations and the development of Cambridge University Press.
Krebs, J. R., & Davies, N. B. (1993). An introductionto behav-
competence in boys and girls. In C. Leaper (Ed.), Child-
hoodgender segregation:Causes and consequences(pp. 53- ioural ecology. Oxford: Blackwell.
Lussier, L., & Buskirk, E. R. (1977). Effects of an endurance
66). San Francisco: Jossey-Bass.
Fein, G. (1981). Pretend play: An integrative review. Child training regimen on assessment of work capacity in pre-
Development,52, 1095-118. pubertal children. Annals of the New YorkAcademyof Sci-
Field, T. M. (1994). Infant day care facilitates later social ences, 301, 734-747.
behavior and school performance. In E. V. Jacobs & H. Maccoby, E. E. (1986). Social groupings in childhood: Their
Goelman (Eds.), Children's play in child care settings relationship to prosocial and antisocial behavior in boys
(pp. 69-84). Albany: State University of New York and girls. In D. Olweus, J. Block, & M. Radke-Yarrow
(Eds.), Developmentof antisocial and prosocialbehavior: Re-
Field, T. M., Ting, G., & Shuman, H. H. (1979). The onset search, theory,and issues (pp. 263-280). New York: Aca-
of rhythmic activities in normal and high-risk infants. demic Press.
DevelopmentalPsychobiology,12, 97-100. MacDonald, K. (1987). Parent-child physical play with re-
Fry, D. P. (1987). Difference between play fighting and seri- jected, neglected, and popular boys. DevelopmentalPsy-
ous fighting among Zapotec children. Ethologyand Socio- chology, 23, 705-711.
MacDonald, K. (1993). Parent-child play: An evolutionary
biology, 8, 285-306.
Gomendio, M. (1988). The development of different types perspective. In K. MacDonald (Ed.), Parent-child play
of play in gazelles: Implications for the nature and func- (pp. 113-143). Albany: State University of New York
tions of play. Animal Behaviour,36, 825-836. Press.
Groos, K. (1898). The play of animals. New York: Appleton. MacDonald, K., & Parke, R. D. (1984). Bridging the gap:
Groos, K. (1901). The play of man. New York: Appleton. Parent-child play interactions and peer interactive com-
Hartup, W. W. (1983). Peer relations. In E. M. Hetherington petence. Child Development,55, 1265-1277.
(Ed.), P. H. Mussen (Series Ed.), Handbookof child psy- MacDonald, K., & Parke, R. D. (1986). Parent-child physical
chology:Vol. 4. Socialization,personality,and social develop- play: The effects of sex and age of children. Sex Roles,
ment (pp. 103-196). New York: Wiley. 15, 367-378.
Hinde, R. A. (1974). Biological basis of human social behavior. Martin, P., & Caro, T. (1985). On the function of play and its
New York: Academic Press. role in behavioral development. In J. Rosenblatt, C. Beer,
M. Bushnel, & P. Slater (Eds.), Advances in the study of be-
Hinde, R. A. (1980). Ethology. London: Fontana.
Hines, M., & Kaufman, F. R. (1994). Androgen and the de- havior (Vol 15, pp. 59-103). New York: Academic Press.
velopment of human sex-typical behavior: Rough-and- McCune, L. (1995). A normative study of representational
tumble play and sex of preferred playmates in children play and the transition to language. DevelopmentalPsy-
with Congenital Adrenal Hyperplasia (CAH). Child De- chology, 31, 198-206.
velopment, 65, 1042-1053. McGrew, W. C. (1972). An ecologicalstudy of children'sbehav-
Hovell, M. F., Bursick, J. H., Shockley, R., & McClure, J. iour. London: Academic Press.
(1976). An evaluation of elementary students' voluntary Meaney, M. J., Stewart, J., & Beatty, W. W. (1985). Sex dif-
physical activity during recess. ResearchQuarterlyin Ex- ferences in social play. In J. Rosenblatt, C. Beer, M. C.
ercise Sports, 49, 460-474. Bushnel, & P. Slater (Eds.), Advances in the study of behav-
Humphreys, A. P., & Smith, P. K. (1984). Rough-and- ior (Vol. 15, pp. 2-58). New York: Academic Press.
tumble play in preschool and playground. In P. K. Smith Money, J., & Ehrhardt, A. A. (1972). Man and woman: Boy
(Ed.), Play in animals and humans (pp. 241-270). Oxford: and girl. Baltimore, MD: Johns Hopkins University
Humphreys, A. P., & Smith, P. K. (1987). Rough-and-tum- Neill, S. R. StJ. (1976). Aggressive and non-aggressive
ble play, friendship, and dominance in school children: fighting in twelve-to-thirteen year old pre-adolescent
Evidence for continuity and change with age. Child De- boys. Journalof Child Psychologyand Psychiatry, 17, 213-
velopment, 58, 201-212. 220.
Obrist, P., Howard, J., Sutterer, J., Hennis, R., & Murrell,
Humphrey, N. (1976). The social function of intellect. In
P. P. G. Bateson & R. A. Hinde (Eds.), Growing points in D. (1973). Cardiac-somatic changes during a simple re-
ethology (pp. 303-317). Cambridge: Cambridge Univer- action time study. Journalof ExperimentalChild Psychol-
sity Press. ogy, 16, 346-362.
Jacklin, C. N., DiPietro, J. A., & Maccoby, E. E. (1984). Sex- Oswald, H., Krappmann, L., Chowduri, F., & Salisch, M.
typing behavior and sex-typing pressure in child/par- (1987). Gaps and bridges: Interactions between girls and
ent interaction. Archives of Sexual Behavior, 13, 413- boys in elementary schools. SociologicalStudies of Child
425. Development,2, 205-223.
Kagan, J. (1971). Continuity and changein infancy.New York: Paquette, D. (1994). Fighting and play fighting in captive
adolescent chimpanzees. Aggressive Behaviour,20, 49-65.
Konner, M. J. (1972). Aspects of the developmental ethol- Parke, R. D., Cassidy, J., Burks, Carson, J., & Boyum, L.
Pellegrini and Smith 597
(1992). Familial contributions to peer competence male rats. Neuroscience and BiobehavioralReviews, 21,
among young children: The role of interactive and af- 105-120.
fective processes. In R. D. Parke & G. Ladd (Eds.), Fam- Pellis, S. M., & Pellis, V. C. (in press). The structure-
function interface in the analysis of play fighting. In M.
ily-peer relationships (pp. 107-134). Hillsdale, NJ: Erl-
Bekoff & J. A. Byers (Eds.), Play behavior:Comparative,
Parke, R. D., & Suomi, S. J. (1981). Adult male infant rela- evolutionary, and ethological aspects. New York: Cam-
tionships: Human and nonhuman primate evidence. In bridge University Press.
K. Immelman, G. W. Barlow, L. Petronovitch, & M. Main Piaget, J. (1962). Play, dreams,and imitationin childhood.New
(Eds.), Behavioraldevelopment(pp. 700-725). New York: York: Norton.
Porges, S. W. (1992). Autonomic regulation and attention.
Cambridge University Press.
Pellegrini, A. D. (1984). The social cognitive ecology of pre- In B. A. Campbell, H. Hayne, R. Richardson (Eds.), At-
school classrooms. InternationalJournalof BehavioralDe- processingin infants and adults (pp.
201-223). Hillsdale, NJ: Erlbaum.
velopment, 7, 321-332.
Quadagno, D. M., Briscoe, R., & Quadagno, J. S. (1977). Ef-
Pellegrini, A. D. (1988). Elementary school children's
rough-and-tumble play and social competence. Develop- fects of perinatal gonadal hormones on selected nonsex-
mental Psychology, 24, 802-806. ual behavior patterns: A critical assessment of the non-
Pellegrini, A. D. (1989a). Elementary school children's human and human literature. Psychological Bulletin, 84,
rough-and-tumble play. Early ChildhoodResearchQuar- 62-82.
terly, 4, 245-260. Roopnarine, J. L., Hooper, F., Ahmeduzzaman, A., & Pol-
Pellegrini, A. D. (1989b). What is a category? The case of lack, B. (1993). Gentle play partners: Mother-child and
rough-and-tumble play. Ethology and Sociobiology, 10, father-child play in New Delhi, India. In K. MacDonald
331-341. (Ed.), Parent-childplay (pp. 287-304). Albany: State Uni-
Pellegrini, A. D. (1990). Elementary school children's play- versity of New York Press.
ground behavior: Implications for social cognitive Rosenthal, M. K. (1994). Social and non-social play of in-
development. Children's Environment Quarterly, 7, 8- fants and toddlers in family day care. In E. V. Jacobs &
16. H. Goelman (Eds.), Children's play in child care settings
Pellegrini, A. D. (1993). Boys' rough-and tumble play, so- (pp. 163-192). Albany: State University of New York
cial competence, and group composition. British Journal Press.
Routh, D. K., Schroeder, C. S., & O'Tuama, L. A. (1974).
of DevelopmentalPsychology, 11, 237-248.
Pellegrini, A. D. (1994). The rough play of adolescent boys Development of activity level in children. Developmental
of differing sociometric status. InternationalJournalof Be- Psychology, 10, 163-168.
havioral Development, 17, 525-540. Rowland, T. W. (1985). Aerobic response to endurance
Pellegrini, A. D. (1995a). School recess and playgroundbehav- training in prepubescent children: A critical analysis.
ior. Albany: State University of New York Press. Medicine and Science in Sports and Exercise, 17, 493-497.
Pellegrini, A. D. (1995b). A longitudinal study of boys' Rubin, K. H., Fein, G., & Vandenberg, B. (1983). Play. In
rough-and-tumble play and dominance during early ad- E. M. Hetherington (Ed.), P. H. Mussen (Series Ed.),
olescence. Journalof Applied Developmental Handbook child psychology:Vol. 4. Socialization,personal-
77-93. ity, and social development (pp. 693-774). New York:
Pellegrini, A. D., & Bjorklund, D. F. (1997). The role of re- Wiley.
cess in children's cognitive performance. Educational Ruff, H. A., & Saltarelli, L. M. (1993). Exploratory play with
Psycholgist, 37, 35-40. objects: Basic cognitive processes and individual differ-
Pellegrini, A. D., & Davis, P. (1993). Relations between chil- ences. In M. Bornstein & A. W. O'Reilly (Eds.), The role
dren's playground and classroom behaviour. British of play in the developmentof thought (pp. 5-15). San Fran-
Journalof EducationalPsychology, 63, 86-95. cisco: Jossey-Bass.
Pellegrini, A. D., Horvat, M., & Huberty, P. D. (1998). The Shephard, R. J. (1983). Physical activity and the healthy
relative cost of children's physical activity play. Animal mind. CanadianMedicalAssociationJournal,128, 525-530.
Behaviour,55. Silverman, I., & Eals, M. (1992). Sex differences in spatial
ability: Evolutionary theory and data. In J. Barkow, L.
Pellegrini, A. D., Huberty, P. D., & Jones, I. (1995). The ef-
fects of recess timing on children's classroom and play- Cosmides, & J. Tooby (Eds.), The adaptedmind (pp. 533-
ground behavior. American EducationalResearchJournal, 549). New York: Oxford University Press.
32, 845-864. Simons-Morton, B. G., O'Hara, N. M., Parcel, G. S., Huang,
Pellegrini, A. D., & Perlmutter, J. C. (1987). A re-examina- I. W., Baranowski, T., & Wilson, B. (1990). Children's fre-
tion of the Smilansky-Parten matrix of play behavior. quency of participation in moderate to vigorous physi-
Journal of Researchin ChildhoodEducation,2, 89-96. cal activities. ResearchQuarterly Exerciseand Sport, 61,
Pellegrini, A. D., & Smith, P. K. (1993). School recess: Impli-
Simons-Morton, B. G., O'Hara, N. M., Simons-Morton,
cations for education and development. Review of Educa-
tional Research,63, 51-67. D. G., & Parcel, G. S. (1987). Children and fitness: A pub-
Pellis, S., M., Field, E. F., Smith, L. K., & Pellis, V. C. (1996). lic health perspective. ResearchQuarterlyfor Exerciseand
Multiple differences in the play fighting of male and fe- Sport, 58, 293-302.
Stevenson, H. W., & Lee, S. Y. (1990). Contexts of achieve-
Sluckin, A. M. (1981). Growing up in the playground:The so-
ment. Monographs the Societyfor Researchin Child De-
cial developmentof children. London: Routledge & Kegan for
velopment,55 (1-2, Serial No. 221).
Stipek, D., & MacIver, D. (1989). Developmental changes in
Sluckin, A. M., & Smith, P. K. (1977). Two approaches to
children's assessment of intellectual competence. Child
the concept of dominance in preschool children. Child
Strayer, F. F. (1980). Social ecology of the preschool peer
Smith, P. K. (1973). Temporal clusters and individual differ-
group. In W. A. Collins (Ed.), TheMinnesota symposiaon
ences in the behaviour of preschool children. In R. Mi-
child development:Vol. 13. Developmentof cognition, affect,
chael & J. Crook (Eds.), Comparative
and social relations(pp. 165-196). Hillsdale, NJ: Erlbaum.
752-798). London: Academic Press.
of primates (pp.
Symons, D. (1978). Play and aggression: A study of rhesus
Smith, P. K. (1982). Does play matter? Functional and evo-
lutionary aspects of animal and human play. Behavioral monkeys.New York: Columbia University Press.
Tanner, J. M. (1970). Physical growth. In P. H. Mussen (Ed.),
and Brain Sciences, 5, 139-184.
Smith, P. K. (1989). The role of rough-and-tumble play in Manual of child psychology (Vol. 1, pp. 77-156). New
the development of social competence: Theoretical per-
Thelen, E. (1979). Rhythmical stereotypies in normal hu-
spectives and empirical evidence. In B. H. Schneider, G.
man infants. Animal Behaviour,27, 699-715.
Attili, J. Nadel, & R. P. Weissman (Eds.), Social compe-
Thelen, E. (1980). Determinants of amounts of stereotyped
tence in developmental perspective (pp. 239-255). Dor-
behavior in normal human infants. Ethologyand Sociobi-
ology, 1, 141-150.
Smith, P. K., & Boulton, M. J. (1990). Rough-and-tumble
Tinbergen, N. (1963). On the aims and methods of ethology.
play, aggression and dominance: Perception and behav-
Zeitschriftfur Tierpsychologie, 410-433.
ior in children's encounters. Human Development, 33, 20,
Tomporowski, P. D., & Ellis, N. R. (1986). Effects of exercise
on cognitive processes: A review. PsychologicalBulletin,
Smith, P. K., & Connolly, K. (1980). The ecology of preschool
behaviour.Cambridge: Cambridge University Press. 99, 338-346.
Vaughn, B. E., & Waters, E. (1981). Attention structure, so-
Smith, P. K., & Hagan, T. (1980). Effects of deprivation on
exercise play in nursery school children. Animal Behav- ciometric status, and dominance: Interrelations, behav-
ioral correlates, and relationships to social competence.
iour, 28, 922-928.
Smith, P. K., Hunter, T., Carvalho, A., & Costabile, A. DevelopmentalPsychology, 17, 275-288.
Volle, M., Shephard, R. J., Lavalle, H., LaBarre,R., Jequier,
(1992). Children's perceptions of play-fighting, play-
J. C., & Rajic, M. (1982). Influence of a program of re-
chasing and real fighting: A cross-national interview
study. Social Development, 1, 211-229. quired physical activity upon academic performance. In
H. Lavalle & R. J. Shephard (Eds.), Croissanceet devel-
Smith, P. K., & Lewis, K. (1985). Rough-and-tumble play,
oppement de l'enfant (pp. 91-109). Trois Rivieres: Uni-
fighting, and chasing in nursery school children. Ethol-
versite de Quebec.
ogy and Sociobiology,6, 175-181.
Vygotsky, L. (1978). Mind in society. Cambridge, MA: Har-
Smith, P. K., & Vollstedt, R. (1985). On defining play. Child
vard University Press.
Waters, E., & Sroufe, L. A. (1983). Social competence as a
Smoll, F. L., & Schutz, R. W. (1985). Physical fitness differ-
developmental construct. DevelopmentalReview, 3, 79-
ences between athletes and nonathletes: Do changes oc-
cur as a function of age and sex? Human Movement Sci-
Welsh, M. C., & Labbe, E. F. (1994). Children and aerobic
ence, 4, 189-202.
exercise: A review of cognitive and behavioral effects.
Spencer, H. (1898). The principles of psychology:Vol. 2. New
Journalof ExperimentalChild Psychology, 58, 405-417.