Professional Documents
Culture Documents
INTRODUCTION
Developmental psychologists have long looked for ways to predict childhood
behavioral tendencies from early infant measures. “Inhibited” children, at the
extreme of shyness and fearfulness, have been found to be more responsive to
sensory stimuli as neonates than others [Lagasse et al., 1989], suggesting that
their sensory arousal thresholds may be lower. Kagan and Snidman [1992] and
Calkins et al. [1996] linked differences in inhibition at 14 months to differences
in motor activity and negative affect during a standardized test at 4 months.
Infant temperament has also been related to neonatal stress reactivity. Gunnar
et al. [1995] found that greater neonatal reactivity was later associated with less
negative affect (distress to limitations) at 6 months of age.
Several studies have attempted to relate individual differences in infant ac-
tivity state profiles to later developmental outcomes. Anders et al [1985] found
*Correspondence to: Gayle Byrne, NIH Animal Center, P.O. Box 529, Poolesville, MD 20837. E-mail:
byrneg@lce.nichd.nih.gov
†
© 1998 Wiley-Liss, Inc. This article is a US Government work and, as such, is in
the public domain in the United States of America.
44 / Byrne and Suomi
that temporal patterns and continuity of sleep were related to mental perfor-
mance in Bayley tests, with more mature patterns of sustained long sleep epi-
sodes associated with higher scores on the Bayley mental scale. They speculated
that regulation and sustainment of inhibition was largely determined by the
infant’s physical maturity at birth. Wakefulness, in contrast, seemed to depend
more on environmental factors than biological ones. Keener et al. [1988] linked
parental temperament reports at 6 months to objective sleep measures at that
age, especially those measures related to continuity and stability of sleep. Infants
judged as having easy temperaments had longer sleep periods and spent less
time out of their cribs at night. Fisher and Rinehart [1990] found that children
with sleep disturbances exhibited higher stress cortisol levels and were judged
as less regular in temperament. Halpern et al. [1994] also related sleep mea-
sures to later temperament: infants that spent more time awake, had less active
and more quiet sleep, and showed shorter sustained sleep periods at 3 weeks
were found to be more irritable, difficult to soothe, and less sociable and showed
fewer positive responses during a behavioral assessment at 3 months. Infants
that spent more time awake at night at 3 weeks were rated as being more irri-
table and more inhibited during the 3 month assessment. These authors suggested
that “infant sleep-wake characteristics and infant temperament reflect similar
aspects of infant biological organization” [Halpern et al. 1994:262].
Activity level is one of the most commonly used measures in assessment of
temperament; it is easy to observe and quantify, and it shows a fair amount of
reliability and continuity from infancy through childhood [Hubert et al. 1982;
Huttenen & Nyman, 1982; Worobey & Blajda, 1989]. Activity level has been shown
to have a heritable component [Saudino & Eaton, 1991], providing evidence for a
biological basis upon which an infant’s environment may act. Early differences
in activity level and state may reflect differences in response thresholds, in that
active infants may have lower thresholds of response to stimuli, whereas less
active infants respond only to more intense stimulation. The more inhibited infants
in Kagan and Snidman [1992] and Calkins et al. [1996] were theorized to have a
lower threshold of excitability of the amygdala, and they exhibited characteristi-
cally greater frontal EEG activation than other infants [Calkins et al. 1996].
The concept of temperament has been increasingly employed in studies of
individual differences in behavior and physiology in nonhuman primates (for re-
views see Higley & Suomi, 1989; Clarke & Boinski, 1995). Many of the findings
from human research have their parallels in the nonhuman primate world as
well. Early infant state organization and activity profiles would seem to offer
some promise to primatologists as outward manifestations of the biological dis-
positions with which an infant primate enters the world. These dispositions pre-
sumably affect the nature of the environment’s subsequent impact. The
environment exerts its greatest influence on an infant that is awake and atten-
tive; active infants encounter and interact with a wider range of stimuli in their
environment than do less active ones. The present study was developed to inves-
tigate whether early infant sleep patterns and activity states in capuchin mon-
keys were predictive of home cage behavior over the first year of life and whether
these measures could serve as reliable indicators of infant temperament. Although
home cage behavior in and of itself may not be a standardized measure of infant
temperament, individuals in our colony show considerable differences in levels of
activity, social, and exploratory behavior [Byrne & Suomi, 1995], the extremes of
which may serve as indicators of outgoing or inhibited temperaments such as
those seen in humans.
Early Infant States and Behavior / 45
METHODS
Subjects
Twenty-nine tufted capuchin monkey infants (Cebus apella) from eight social
groups provided the data for this study (20 males, nine females). Seventeen (12
males, five females) lived in indoor cages (.9 × 1.8 × 1.7 m) and 12 (eight males,
four females) lived in indoor/outdoor runs during the fall and winter and in an
outdoor corncrib during the spring and summer.
Procedure
An infant activity state scoring system, first developed by Dorothy Fragaszy
[1989] for use in capuchin monkeys, was adapted for our research. Activity states
were recorded on the day of birth, day 4, week 1, and every week up until week
11. One these days, 3 separate hours of 1 min scan samples of infant and mother
activity were collected and summarized to create the measures listed in Table I.
Observations took place at 1100, 1300, and 1500 h. Interobserver reliability (per-
centage of agreements) of infant and mother activity states was assessed at 90%
at several points over the data collection period. Activity state and mother activ-
ity were expressed as a percentage of samples in which animals were scored in
each category, and weekly scores were combined to obtain monthly means (with
only the last 3 weeks included in month 3).
In addition, home cage behavior over the first year of life was videotaped in
three 10 min sessions each week. The videotapes were scored for the behaviors
listed in Table II by a single observer. Intraobserver reliability was assessed at
90% at the onset of the study. Weekly means for all these measures were ob-
tained, and monthly values were calculated from the weekly means for months
1–6 and months 8, 10, and 12.
Data analysis
Spearman rank correlations were calculated between infant state measures
in the first 3 months and monthly means of behaviors scored from videotapes over
the first year. An alpha level of .01 (two-tailed) was adopted for interpreting re-
sults, due to the large number of variables involved in the correlation analyses.
TABLE I. Infant and Mother Activity State Scoring Categories and Representative
Values Collapsed Over the First 3 Months*
Runs Cages Males Females
Mean SEM Mean SEM Mean SEM Mean SEM
Infant activity state
Sleep/drowsy 41.8 2.2 38.3 2.3 39.4 1.8 40.5 3.8
Alert–quiet 20.7 0.9 20.0 1.2 19.9 0.8 21.1 1.8
Alert–active 29.2 2.5 35.1 3.3 33.4 2.5 31.1 4.8
Mother activity
Sit 43.5 4.2 35.6 2.8 39.9 3.4 37.3 2.7
Stand 7.4 0.9 16.3 2.0 11.4 1.8 14.5 2.7
Lying down 13.1 2.5 6.1 1.8 10.5 2.2 6.2 1.2
Locomoting 11.6 1.4 21.7 2.7 15.9 2.3 20.4 3.2
*Behaviors are scored as % of samples observed in each category. Alert–quiet is analogous to Prechtl state 3
[Prechtl, 1974]; alert–active is analogous to Prechtl state 4.
46 / Byrne and Suomi
TABLE II. Behaviors Scored During Video Observations and Representative Values
Collapsed Over the First Year*
Runs Cages Males Females
Mean SEM Mean SEM Mean SEM Mean SEM
Interactions with mother
Approach/leave mother 3.0 0.3 7.3 0.9 5.1 0.7 6.4 1.4
Dorsal carry 50.7 2.4 42.1 3.0 44.5 2.7 48.1 3.3
Ventral carry 7.0 0.6 11.3 1.4 10.1 1.3 8.3 0.9
Proximity 13.3 1.0 22.8 2.0 18.4 1.9 19.7 2.6
Nonspecific contact 8.2 0.6 10.0 0.8 8.9 0.5 9.9 1.3
Grooming 3.2 0.5 4.9 0.7 4.3 0.7 3.9 0.6
Total time with mother 64.8 2.0 70.3 2.3 66.9 1.9 70.6 3.1
With other animals
Proximity 43.4 1.0 43.7 3.2 44.0 2.3 42.8 3.5
Contact 9.4 0.7 9.8 1.2 10.2 0.8 8.3 1.5
Total time with others 50.2 1.5 50.2 3.4 51.3 2.5 47.8 3.8
Alone 53.1 1.1 51.5 2.2 52.6 1.7 51.1 2.3
Explore environment 42.4 1.1 44.5 1.7 43.4 1.1 44.1 2.7
Social play 9.8 0.7 7.0 0.9 8.7 0.8 6.9 0.8
Self play 4.6 0.4 6.0 0.5 5.8 0.4 4.7 0.5
Object manipulation 2.9 0.4 3.1 0.6 2.7 0.4 3.7 0.8
(pounding/rubbing food
or objects)
*Most of these measures are arcsin percentages of observation time and are not mutually exclusive. Ap-
proaches/leaves to mothers and object manipulation were scored as frequency measures.
Byrne and Suomi [1995] described several effects of housing on home cage
behavior in these subjects. Mixed design ANOVAs were used to investigate the
effects of sex and housing on infant state and behavioral measures in the present
study. Percentage values were arcsin-transformed for use in ANOVAs [Lorenzen
& Anderson, 1993]. Sex and housing served as between subject factors and month
as a within subject factor in these analyses.
In addition, when correlations between infant state measures and home cage
behaviors were significant, multiple regressions were performed on the ranked
data to test the contributions of sex and housing to the variance in those home
cage measures. In these cases stepwise regressions were performed, which tested
the incremental variance added by sex and/or housing after the infant state mea-
sure had been entered.
Another possible influence on infant activity is the nature of mothers’ activ-
ity: infants’ ability to actively engage their environment could be influenced by
whether or not their mothers were sitting still or moving around. Thus, infant
activity scores were also correlated with maternal activity scores within each
month of observation.
RESULTS
Figures 1 and 2 present the significant rank correlations (P<.01) between
mean activity state scores for months 2 and 3 and behaviors scored from video
over the first year (none of the activity state variables in month 1 were predic-
tive of later behavior). These and the following results are presented as direc-
Early Infant States and Behavior / 47
Fig. 1. Significant (P<.01) rank correlations of infant activity state measures in month 2 (a) and month 3
(b) with interactions with mothers. Central axis indicates months; positive correlations are shown on top of
the axis, while negative ones are shown below. Numbers on the axis correspond to the month(s) in which the
corresponding behavior was correlated with the activity state measure indicated. Ap/lv, approach/leave; Con-
tact, contact with mother; Proximity, proximity to mother; Dorsal, dorsal contact with mother; Total, total
time spent mother.
Early Infant States and Behavior / 49
Fig. 2. Significant (P<.01) rank correlations of infant activity state measures in month 2 (a) and month 3
(b) with other behaviors. Central axis indicates months; positive correlations are shown on top of the axis,
while negative ones are shown below. Numbers on the axis correspond to the month(s) in which the corre-
sponding behavior was correlated with the activity state measure indicated.
50 / Byrne and Suomi
Fig. 3. a: Scores for alert–active in months 1–3 for each sex/housing category. Vertical bars denote stan-
dard errors. b: Scores for environmental exploration collapsed over the first year for each sex/housing cat-
egory. Vertical bars denote standard errors.
Early Infant States and Behavior / 51
Fig. 4. Housing differences in scores for approach/leave mothers (a), proximity to mothers (b), and ventral
contact with mothers (c) over the first year. Vertical bars denote standard errors.
52 / Byrne and Suomi
As in Byrne and Suomi [1995], several housing effects emerged for behavior
scores over the first year. Animals in cages approached/left mothers and were in
proximity to mothers more than those in runs in later months (Fig. 4). Percent of
time in ventral contact was higher in cages than in runs in the first 2 months.
Multiple regression analyses of those behaviors found to be correlated with
state measures revealed a significant (P<.05) contribution of housing to time alone
in month 2, proximity to mothers and dorsal contact in month 5, approach/leave
in month 6, and exploration in month 8. Sex contributed significantly to the
variance in month 4 values for dorsal contact, proximity and total time with
mothers, time alone, and self-play. Values for these variables in each of the hous-
ing conditions are shown in Table III. In all of the above cases, however, state
variables still contributed significantly to the variance in behavioral measures.
DISCUSSION
Tufted capuchin infants are typically in constant contact with mothers until
sometime in their second or third month, at which time they begin to separate
from mothers and explore their environments independently. By 6 months of
age, infants typically spend only about one-fourth of their time carried by moth-
ers, instead spending the majority of their time alone [Byrne & Suomi, 1995]. In
the present study, infants that spent more time on mothers and less time in
TABLE III. Significant Housing and Sex Differences (P<.05) Found in State–Behavior
Correlations as Revealed by Multiple Regression Analyses*
Runs Cages
Mean SEM Mean SEM
*Ap/lv, approach/leave mother (frequency/10 min); Prox, proximity to mother (this and following scores are
arcsin percentages of observation time, presented as means and standard errors).
Early Infant States and Behavior / 53
TABLE IV. Correlations Between Maternal Activity and Infant State Scores in
Months 2 and 3*
Month 2 Month 3
Mother activity Sl/dr Al–qu Al–ac Sl/dr Al–qu Al–ac
Sit .087 –.096 .636a .427 –.130 .718a
Stand .108 .047 –.124 .448 .812a .144
Lie –.204 –.503 .164 –.463 –.476 .190
Locomotion .335 .516 –.429 .449 .775a .054
*Sl/dr, sleep/drowsy; Al–qu, alert–quiet; Al–ac, alert–active.
a
Spearman rank correlations significant at P<.01.
maternal movement and a negative one between sleep and maternal movement.
Examination of mothers’ activity in this study suggests that, at least in older
infants, a more sedentary mother may actually promote greater activity in an
infant, whereas a more active mother, in constant motion, may discourage inde-
pendent infant activity. In our colony, the more sedentary mothers appear to be
the calmer, more equable ones, less likely to overreact to external stimuli, while
the ones in constant motion appear to be more excitable, nervous, hyperreactive
ones. If this characterization is accurate, calmer mothers would have infants
that appeared more independent, more exploratory, and less fearful, whereas
mothers that tended to be more skittish would have infants that appeared to
stay with them longer and explore less. The correspondence between mother and
infant activity is not absolute, however; in month 3, both sleep and alert–active
were correlated with mothers’ inactivity. In infants under 1 month of age, who
were only active 10–20% of the time [Byrne & Suomi, 1995], there was no signifi-
cant relationship between mother and infant activity.
Whatever infant tendencies may be, however, they must by necessity inter-
act with the existing environment. Sleepy infants end up staying closer to moth-
ers and may lose out on exploration, play experience, and contact with conspecifics.
More active infants engage in more exploration and therefore may be exposed to
more learning experiences in their social and physical environments, with all the
concomitant risks and benefits of those experiences. An interesting, varied envi-
ronment may encourage sustainment of an awake state beyond that seen in a
more monotonous one [Wolff, 1965]. Differences in reactivity affect behavior, which
in turn can affect how the environment exerts its impacts, which may then affect
future responsivity, etc.
Findings from the human literature cited earlier [e.g., Kagan & Snidman,
1992; Halpern et al. 1994] link extremely high levels of wakefulness and activity
to temperaments judged as difficult or inhibited, citing low arousal thresholds as
a possible mechanism underlying these aspects of development. In the present
study, however, increased waking activity appeared to be associated with infants
that were more exploratory and independent of mothers in early months, sug-
gesting a less fearful temperament. In the confines of this study, it is not pos-
sible to separate the effects of genetic vs. environmental contributions (i.e., calmer
mothers may have more independent infants because of a genetic disposition or
because of the differential ease of sustaining activity on a moving vs. sedentary
mother). Ongoing studies in our laboratory assessing infants’ adrenocortical re-
activity and subjective personality ratings are examining the relationship of re-
activity to early behavior and temperament in more detail.
CONCLUSIONS
1. Capuchin infants that appeared sleepier or quieter during early infant
activity state scoring (slept more, spent more time in a quiet and less time in an
active state) spent more time on mothers and less time in exploration during the
period from 2–6 months of age.
2. Conversely, those infants who appeared more awake or active during ac-
tivity state scoring (slept less, spent more time in an active state) spent more
time separated from mothers, alone or with other animals, and in exploration
and play over the first 6 months of life.
3. A relationship between early activity and later behavior was less apparent
in the second half of the first year, whereas effects of differing housing environ-
ments became more pronounced at this time.
Early Infant States and Behavior / 55
ACKNOWLEDGMENTS
We thank Sarah Williams, Georgina Slavoff and Sarah Thrasher for assis-
tance in data collection during parts of this study and Evan Byrne for consulta-
tion on graphical presentation of data. This research was supported by the
Division of Intramural Research, National Institute of Child Health and Hu-
man Development.
REFERENCES
Anders, T.F.; Keener, M.A.; Kraemer, H. reactivity in non-human primates. Pp.
Sleep-wake state organization, neonatal 153–167 in TEMPERAMENT IN CHILD-
assessment and development in premature HOOD. G.A. Kohnstamm; J.E. Bates; M.K.
infants during the first year of life. II. Rothbart, eds. New York, John Wiley and
SLEEP 8:193–206, 1985. Sons, 1989.
Byrne, G.; Suomi, S.J. Development of ac- Hubert, N.C.; Wachs, T.D.; Peters-Martin,
tivity patterns, social interactions and P.; Gandour, M. The study of early tem-
exploratory behavior in infant tufted perament: Measurement and conceptual
capuchins (Cebus apella). AMERICAN issues. CHILD DEVELOPMENT 53:571–
JOURNAL OF PRIMATOLOGY 35:255– 600, 1982.
270, 1995. Huttenen, M.O.; Nyman, G. On the conti-
Calkins, S.D.; Fox, N.A.; Marshall, T.R. Be- nuity, change and clinical value of infant
havioral and physiological antecedents of temperament in a prospective epidemio-
inhibited and uninhibited behavior. CHILD logical study. Pp. 240–247 in TEMPERA-
DEVELOPMENT 67:523–540, 1996. MENTAL DIFFERENCES IN INFANTS
Clarke, A.S.; Boinski, S. Temperament in non- AND YOUNG CHILDREN. R. Porter;
human primates. AMERICAN JOURNAL G.M. Collins, eds. London, Pitman, 1982.
OF PRIMATOLOGY 37:103–125, 1995. Kagan, J.; Snidman, N. Infant predictors
Fisher, B.E.; Rinehart, S. Stress, arousal, psy- in inhibited and uninhibited children.
chopathology and temperament: A multidi- Pp. 71–88 in FUTURE DIRECTIONS IN
mensional approach to sleep disturbance in INFANT DEVELOPMENT RESEARCH.
children. PERSONALITY AND INDI- G.J. Suci; S.S. Robertsin, eds. New York,
VIDUAL DIFFERENCES 11:431–438, 1990. Springer-Verlag, 1992.
Fragaszy, D.M. Activity states and motor Keener, M.A.; Zeanah, C.H.; Anders, T.F.
activity in an infant capuchin monkey Infant temperament, sleep organization,
(Cebus apella) from birth through eleven and nighttime parental interventions. PE-
weeks. DEVELOPMENTAL PSYCHOBI- DIATRICS 81:762–771. 1988.
OLOGY 22:141–157, 1989. Lagasse, L.L.; Gruber, C.P.; Lipsitt, L.P. The
Fragaszy, D.M. State organization and activ- infantile expression of avidity in relation
ity in infant cebid monkeys (Cebus and to later assessments of inhibition and
Saimiri) in two rearing conditions. INTER- attachment. Pp. 159–176 in PERSPEC-
NATIONAL JOURNAL OF COMPARATIVE TIVES ON BEHAVIORAL INHIBITION.
PSYCHOLOGY 8:150–167, 1995. J.S. Reznick, ed. Chicago, University of
Gunnar, M.R.; Porter, F.L.; Wolf, C.M.; Chicago Press, 1989.
Rigatuso, J.; Larson, M.C. Neonatal stress Lorenzen, T.J.; Anderson V.L. DESIGN
reactivity: Predictions to later emotional OF EXPERIMENTS: A NO-NAME AP-
temperament. CHILD DEVELOPMENT PROACH. New York, Marcel Dekker,
66:1–13, 1995. Inc., 1993.
Halpern, L.F.; Anders, T.F.; Coll, C.G.; Prechtl, H.F.R. The behavioural states of the
Hua, J. Infant temperament: Is there a newborn infant (a review). BRAIN RE-
relation to sleep-wake states and mater- SEARCH 76:185–212, 1974.
nal nighttime behavior? INFANT BE- Saudino, K.J.; Eaton, W.O. Infant tempera-
HAVIOR AND DEVELOPMENT 17:255– ment and genetics: An objective twin study
263, 1994. of motor activity level. CHILD DEVELOP-
Higley, J.D.; Suomi, S.J. Temperamental MENT 62:1167–1174, 1991.
56 / Byrne and Suomi
Wolff, P.H. The development of attention in Worobey, J.; Blajda, V.M. Temperament
young infants. ANNALS OF THE NEW ratings at 2 weeks, 2 months, and 1 year:
YORK ACADEMY OF SCIENCES 118:815– Differential stability of activity and emo-
830, 1965. tionality. DEVELOPMENTAL PSYCHOL-
OGY 25:257–263, 1989.