INFANCY, 3(2), 133–152
Copyright © 2002, Lawrence Erlbaum Associates, Inc.
The Effect of Excessive Crying on the
Development of Emotion Regulation
Cynthia A. Stifter
Department of Human Development and Family Studies
Pennsylvania State University
Tracy L. Spinrad
Department of Family and Human Development
Arizona State University
The goal of this study was to examine the effect of excessive crying in early infancy
on the development of emotion self-regulation. Cry diaries were used to categorize
excessive criers and typical criers at 6 weeks of age. At 5 and 10 months of age, in-
fants and mothers participated in procedures to elicit infant reactivity and regulation
during a frustration task and maternal sensitivity and intrusiveness during a free-play
session. Last, maternal ratings of temperament were obtained. Results revealed ex-
cessive criers to show higher levels of negative reactivity than typical criers. Exces-
sive criers also demonstrated lower regulation, but this ﬁnding was only signiﬁcant
for male infants. Boys in the excessive criers group exhibited the lowest level of emo-
tion self-regulation. Maternal behavior and ratings of temperament at 5 and 10
months failed to distinguish the 2 cry groups. The ﬁndings suggest that excessive cry-
ing may inﬂuence the developmental trajectory of the ability of boys to self-regulate
emotion. The hypothesized processes involved in this outcome are discussed.
Since Thomas and Chess (Thomas, Chess, & Birch, 1968) focused the develop-
mental spotlight on individual differences in behavioral style and its moderating
effect on environments, the study of infant and child temperament has ﬂourished.
Subsequently, several theoretical approaches to the study of child temperament
have emerged, as have a number of assessment tools (Rothbart & Bates, 1998).
Requests for reprints should be sent to Cynthia A. Stifter, 105 White Bldg., Department of Human
Development and Family Studies, Pennsylvania State University, University Park, PA 16802.
STIFTER AND SPINRAD
Despite the diversity of perspectives and measures, the general consensus among
temperament researchers is that temperament is a psychobiological construct that
is relatively stable over time (Goldsmith et al., 1987). Whereas it is accepted that
developmental changes may alter the expression of temperament, stability is ex-
pected to be maintained under most circumstances. Thus, even though we might
expect the frequency with which an infant cries to an aversive stimulus to decrease,
temperamental stability would presume that the child will remain high in crying
relative to other infants. Stability also may be found for processes that underlie
temperament behaviors, whereas observed behaviors across time might be differ-
ent. Behavioral inhibition in a young infant is expressed with high negative affect
and motor activity (Calkins, Fox, & Marshall, 1996; Kagan, 1994) but manifests it-
self later in withdrawal, low vocalizations, and high compliance. The underlying
process is hypothesized to be heightened excitability in the brain region known as
the amygdala (Kagan, 1994).
As theorists have noted, stability of temperament is relative. Indeed, in studies
of infant and child temperament using parent ratings, the stability coefﬁcients are
quite modest, around .30 (Slabach, Morrow, & Wachs, 1991). The stability of ob-
served temperament is examined less often, and when reported is lower than
parent-reported temperament (Rothbart, 1986). This level of instability is expected,
as temperament also is proposed to be modiﬁable. There is some disagreement,
however, as to whether it is the expression of temperament that changes while the
underlying nature remains the same or that both change (Goldsmith et al., 1987).
Nevertheless, there is growing empirical evidence that several factors have an
important impact on the longitudinal course of temperament.
One proposed source of change in temperament over time is the emergence of
regulative systems that function to modulate reactivity (Rothbart, 1989). Accord-
ing to Rothbart, temperament is expressed as individual differences in reactivity
and regulation. The ability to shift attention or disengage from orienting, for ex-
ample, is considered a dimension of temperament (Rothbart & Bates, 1998). The
development of this ability in the later part of the ﬁrst year of life may explain
changes in crying behavior that are observed during the same period. Mangelsdorf
and colleagues (Mangelsdorf, Shapiro, & Marzoff, 1995) showed that the infants’
use of distraction during a mildly stressful event increased from 6 to 12 months.
Moreover, when examining the effectiveness of regulatory strategies in the ﬁrst
year of life, Stifter and Braungart (1995) found reorienting to be effective in reg-
ulating anger reactivity at 10 months but not at 5 months of age. Similarly, Diener
and colleagues (Diener, Mangelsdorf, McHale, & Frosch, this issue) found social
referencing to reduce distress during a competing demands task at 12 months of
age. Thus, the emergence and consolidation of one temperament factor can inﬂu-
ence the expression of another.
In the emotion domain, the modulation, maintenance, and enhancement of emo-
tion is deﬁned as emotion regulation (Thompson, 1994). Whereas temperament
EXCESSIVE CRYING AND REGULATION
research would suggest that infants may vary on the degree to which they develop
and utilize certain behaviors to modulate temperamental reactivity, the develop-
ment of the ability to regulate emotions depends largely on the input of the care-
giver (Kopp, 1989). Initially, parents and caregivers reduce infant negative reactiv-
ity through soothing techniques. By relieving distress, caregivers create a context
in which the infants associate their behaviors and the caregiver’s actions with their
accompanying changes in feeling state (Gianino & Tronick, 1988; Kopp, 1989). To
illustrate, a crying infant generally elicits caregiving responses aimed at reducing
the distress. Once levels of reactivity are reduced, the caregiver may then introduce
new behaviors that encourage self-soothing, such as non-nutritive sucking or dis-
traction. If successful, the infant then learns that these behaviors can reduce nega-
tive arousal. As the infant matures, the role of the parent in regulating infant emo-
tions may change to one of support and scaffolding of the infant’s existing skills.
Maternal positive guidance, for example, was contemporaneously related to self-
distraction in toddlers (Calkins & Johnson, 1998). However, when the infant is
highly distressed, the parent may continue to be directly involved in the reduction
of negative affect.
The proposal that parenting is central to the development of regulatory skills is
supported by studies that have demonstrated that parenting behavior either medi-
ates or moderates changes in temperamental reactivity (Eisenberg, Cumberland, &
Spinrad, 1998). For example, warm, responsive, and sensitive mothering has been
shown to be related to decreases in irritability over time (Bell & Ainsworth, 1972;
Fish, Belsky, & Stifter, 1991; van den Boom, 1994). Interestingly, insensitive par-
enting resulted in lack of predictability from negative reactivity to behavioral inhi-
bition (Park et al., 1997), suggesting that the relationship between parenting and
temperament outcomes is conditional on the type of temperament. Expected rela-
tions between temperament and outcome also may be mediated or moderated by
parent behavior (see Kochanska, 1993). In one study, negative temperament was
hypothesized to predict noncompliance (Braungart-Rieker, Garwood, & Stifter,
1997). The ﬁndings revealed that this relationship was mediated by maternal be-
havior, speciﬁcally overcontrolling behavior. Similarly, fearful temperament was
found to interact with gentle discipline to facilitate the development of conscience
Recently it has been proposed that the development of emotion regulatory
strategies is dependent on the degree of reactivity (Kopp, 1989; Stifter, Spinrad, &
Braungart-Rieker, 1999); that is, infants must experience and communicate nega-
tive reactivity to learn that certain behaviors can be used to self-regulate. But at
the extreme level, distress may actually inhibit the infant’s recruitment and effec-
tive utilization of regulatory behaviors (Kopp, 1989; Stifter & Braungart, 1995).
Stifter and Braungart (1995), for example, showed that reorienting was most ef-
fective at low levels of reactivity. Furthermore, high levels of reactivity may limt
the opportunities for parents to introduce new strategies or support existing
STIFTER AND SPINRAD
strategies. This may explain why highly reactive infants exhibited more stable be-
haviors than less reactive infants (Worobey & Lewis, 1989).
One opportunistic method for testing the effect of reactivity on the emergence
of emotion regulation would be to examine the developmental outcomes of in-
fants who exhibit extreme levels of reactivity or excessive crying in early in-
fancy. Developmentally, young infants are expected to increase in their crying,
demonstrating a cry peak (averaging about 2 hr per day) at about 6 to 8 weeks
(Barr, 1990). With the advent of increased alertness, social smiling, and non-
negative vocalizations, crying decreases to less than 1 hr per day by 3 months of
age. However, approximately 10% of infants will cry more than the expected 2 hr
per day and will be difﬁcult to console. These infants are said to have infant colic
(Barr, 1990). Recent attempts to understand this behavioral syndrome suggest
that colic is a transitory condition that may be related to the infants’ ability to reg-
ulate or be regulated (Barr & Gunnar, 2000; Stifter & Braungart, 1992). In addi-
tion, although organic causes for excessive crying during this period have been
identiﬁed, such as cow’s milk allergies, the number is projected to represent only
5% of colic cases (Gormally & Barr, 1997). Regardless of the cause of excessive
crying, intense, inconsolable distress has the potential to affect emotion regula-
tion development. To date, empirical studies that have examined the develop-
mental consequences of infant colic are mixed and, more important, have not
directly addressed this hypothesis. In one study, infants with colic were found to
look “typical” in their responses to aversive stimuli after the colic was resolved
(Stifter & Braungart, 1992). On the other hand, investigations on the outcomes of
colic suggest that excessive crying in early infancy represents the extreme on a
temperamental continuum (Carey, 1972; Jacobson & Melvin, 1995; Lehtonen,
Korhonen, & Korvenranta, 1994) or is the result of continued insensitive, over-
stimulating parenting (Papouˇsek & von Hofacker, 1995). Interestingly, in one study
infants with “persistent crying” were found to show stability in negative reactivity
immediately after the colic resolved, but no differences were observed several
months later (St. James Roberts, Conroy, & Wilsher, 1998b).
In this study, we examined the temperament outcomes of infants with exces-
sive crying measured at 6 weeks of age. Drawing from the literature on emotion
regulation, temperament, and infant colic, we hypothesized that infants who cried
excessively, frequently, and inconsolably in early infancy will have limited expe-
rience with effective soothing, potentially interfering with the development of
their ability to self-regulate their emotional responses. Infants were identiﬁed as
excessive criers at 6 weeks of age using a 4-day, 24-hr cry diary. The standard def-
inition of colic provided by Wessel (Wessel, Cobb, Jackson, Harris, &
Detwiler, 1954) and used predominantly in studies of colic uses the criteria of ex-
cessive crying and fussing behavior. Excessive criers were identiﬁed using an
adaptation of this criteria from the diary data. Outcome data were drawn from lab-
oratory procedures conducted at 5 and 10 months of age. To investigate whether
EXCESSIVE CRYING AND REGULATION
the development of regulatory strategies in excessive criers may have been due to
differences in parent behavior, maternal sensitivity and intrusiveness was as-
sessed during a free-play procedure. Finally, some studies have shown boys to be
more negatively reactive than girls (Moss, 1967; Phillips, King, & DuBois, 1978;
Weinberg et al., 1999), whereas others have indicated the reverse (Mayes &
Carter, 1990). Therefore, gender differences were examined in this study.
Participants were 116 families (65 male infants). Families were recruited through
a local community hospital and university medical center. Leaﬂets asking for vol-
unteers to participate in a study of infant development were distributed to all new
parents. Once the study was explained and consent given for participation, a
2-week home visit was scheduled. Infants recruited into the study were of normal
gestational age and birthweight and did not suffer from any prenatal or perinatal
complications.1 Characteristics of the study participants (reported by group) can
be found in Table 1.
Procedures and Measures
Infants and parents were seen three times throughout the infant’s ﬁrst year of life. An
initial home visit was scheduled for 2 weeks. A 4-day, 24-hr cry diary was com-
pleted by parents when infants were 6 weeks of age. Follow-up laboratory visits
were scheduled for when the infant was 5 and 10 months of age. During these visits
several procedures for eliciting positive and negative reactivity were administered.
Parents completed a 4-day, 24-hr diary when infants were
6 weeks of age. The ruler-like diary has been used extensively in research on in-
fant colic and gives a reliable portrait of the infants’ cry behavior (Barr, Kramer,
Boisjoly, McVey-White, & Pless, 1988; St. James Roberts, Hurry, & Bowyer,
1993). Parents were given crayons, matched to ﬁve different infant states (awake
1Initially, 157 families were recruited into the study. Of the 41 who are not included in this study, 17
failed to complete the diary, 8 were unable to participate in the 5-month visit, and 16 dropped from the
study due to relocation outside the area. Comparisons of these 41 infants on the demographic, group-
ing, and dependent variables revealed that of the families who did not complete the study mothers and
fathers were slightly younger and their infants of slightly older gestational age than the families in this
study. However, these differences were not viewed as having a signiﬁcant impact on the outcome. Im-
portantly, no differences in the dependent measures were found.
STIFTER AND SPINRAD
Characteristics of the Study Sample
(n = 102)
(n = 14)
Gestational age (weeks)
Gender (% male)
Ethnicity (% White)
Parity (% ﬁrstborn)
Only father education was signiﬁcantly different for the cry groups.
and content, sleeping, feeding, fussing, and crying) and were instructed to record
them in 5-min intervals. To insure compliance, a research assistant called parents
2 days into the diary to answer any parental questions or concerns. On receipt, the
cry diaries were entered into a computer program. Four variables were derived
from these data, which were averaged across the 4-day recording period: number
of fuss bouts per day, total fussing per day (in minutes), number of cry bouts per
day, and total crying per day (in minutes).
5- and 10-month laboratory visits.
To elicit frustration, an arm restraint
procedure was administered at both the 5 and 10 month visits. During this proce-
dure, infants were placed in an infant seat or high chair while mothers were es-
corted out of the room. A research assistant, standing behind the infants, gently
held down the infants’ arms from behind for a period of 2 min, after which the in-
fants’ arms were released. If the infants became extremely distressed for a mini-
mum of 10 sec, the research assistant released the infants’ arms. One min after
arm release, mothers were instructed to soothe their infants in a step-wise manner.
First, mothers were instructed to soothe their infants from a distance (of about
10 ft) for 30 sec. Next, mothers were told they could approach their infants and
touch them for 30 sec. Finally, mothers could then pick up their infants and soothe
them in any way necessary.
Infant emotional reactivity was coded from the videotapes on a second-
by-second basis. Behavioral categories were exclusive and exhaustive. Nega-
tive reactivity was coded on a 3-point scale (0 = neutral, 1 = mild to moderate
EXCESSIVE CRYING AND REGULATION
reactivity, 2 = high negative reactivity). Facial and vocal expressions were used to
deﬁne when and to what degree the infant expressed negative reactivity. For ex-
ample, variations in eye closure, furrowing of the brows, mouth openings, and cry
sounds were used to determine the intensity of negative reactivity. Negative reac-
tivity codes were averaged within each of the ﬁve episodes: arm hold, arm release,
distal soothe, proximal soothe, and pickup. Coders were trained to at least .75 re-
liability. Cohen’s kappa computed on 10% of the sample (after training) yielded
reliabilities of .80 for the 5-month data and .83 for the 10-month data.
Behaviors hypothesized and demonstrated to reduce negative arousal (Buss &
Goldsmith, 1998; Stifter & Braungart, 1995) were coded from the videotapes.
Speciﬁcally, reorienting, self-comforting, avoidance, looks to the experimenter,
and non-negative vocalizations were coded during the most intense portion of the
arm restraint procedure—restraint and release. Behaviors were coded continu-
ously on a computer. Two passes of the videotaped arm restraint procedure were
necessary so that behaviors that co-occurred could be noted. Because the time
during restraint and release varied by subject, the proportion of time spent using a
particular strategy was calculated by dividing the total amount a regulatory be-
havior was exhibited by the total time of the restraint or release episode. The mean
proportions (standard deviations) of each regulatory behavior and the percentage
of infants who exhibited the behaviors during arm restraint and arm release can be
seen in Table 2. Coders were trained to .75 reliability. Interrater reliabilities for the
regulatory behaviors coded after training resulted in average Cohen’s kappas of
.75 for the 5-month coding and .83 for the 10-month coding.
Mean Proportions for Each of the Coded Regulatory Behaviors and the Percentage of Infants
Exhibiting That Behavior at 5 and 10 Months of Age During Arm Restraint and Arm Release
Looks to examiner
Looks to examiner
N = 116.
STIFTER AND SPINRAD
A regulation composite score was created based on previous literature (Buss &
Goldsmith, 1998; Stifter & Braungart, 1995), demonstrating their effectiveness
and the statistical relationship between the variables and negative reactivity.
Whereas self-comforting, reorientation, looks to the examiner, and vocalizations
were negatively related to reactivity (all rs > –.19) avoidance behaviors were
found to be positively related to negative reactivity, r = .24 (5 m), r = .26 (10 m).
Due to the low occurrence of three of the four variables (see Table 2), we created
a score that represented the proportion of time any one of the regulatory behaviors
(other than avoidance) was present. Thus, if two regulatory behaviors were used
concurrently, regulation was only counted once. Avoidance was examined sepa-
rately in the analyses.
Maternal behavior during free play.
Mothers and infants participated in a
free-play session at 5 months (10 min) and at 10 months (15 min). Mothers and in-
fants were presented with a basket of toys and instructed to play as they normally
would at home. The mother–infant interaction during the free play was videotaped
and coded for maternal sensitivity and intrusiveness. Maternal sensitivity scores
were based on contingent, infant-centered interactions (Fish, Stifter, & Belsky,
1991). A high score reﬂected baby-centered, synchronous interaction that had an
appropriate level of response and stimulation. Maternal intrusiveness reﬂected
overcontrolling behaviors and ignoring signals. High scores on intrusiveness were
given for overly stimulating behavior, ill-timed responses, and inappropriate de-
mands on the infant. Both sensitivity and instrusiveness were coded every 15 sec
using a 4-point scale (0 = no evidence to 3 = high level). Coders were trained to at
least .75 reliability. Interrater reliability was assessed using Cohen’s kappa on
10% of the sample (after training) and was .69 and .86 for 5-month sensitivity and
intrusiveness, and .68 and .76 for 10-month sensitivity and intrusiveness. All dis-
agreements were resolved through conference.
Infant temperament questionnaires.
An infant temperament question-
naire was completed by mothers when infants were 5 and 10 months of age. The
Infant Behavior Questionnaire (IBQ; Rothbart, 1981), is a 99-item instrument that
assesses the frequency with which certain behaviors occurred during the last
2 weeks. Dimensions of activity level, duration of orientation, distress to limita-
tions, distress to novelty, smiling/laughter, and soothability are derived from the
questionnaire. The IBQ has demonstrated satisfactory reliability and validity
(Goldsmith & Rothbart, 1991).
Creation of cry groups.
To examine whether excessive criers differ in the
development of their regulatory skills, we created two groups of infants based on
their cry behavior taken from the 6-week diaries. Selection of excessive criers was
based on a study by St. James Roberts, Conroy, and Wilsher (1995). Using the
EXCESSIVE CRYING AND REGULATION
Means, Standard Deviations, F Tests, and Effect Sizes for Typical Criers and Excessive
Criers on Cry Behavior at 6 Weeks
(n = 102)
(n = 14)
F test (1,112)
aValues are for average frequency of bouts per day. bValues are for average bout duration (min) per
*p < .001.
data reported by St. James Roberts et al. for the average crying/fussing for per-
sistent (excessive) criers, we categorized infants into two groups. Excessive criers
were those infants who were reported to cry and fuss (distress) in excess of
221 min per day. Twelve infants met criteria for excessive crying and fussing. In
addition, we included 2 infants who narrowly missed the total duration criteria for
distress but scored above the average for the number of minutes of crying per day
(86 min). Our rationale was that excesses in crying, which is qualitatively differ-
ent than fussing, would likely interfere with successful soothing. These 14 infants
(9 boys) represent the 94th percentile in crying and the 92nd percentile in distress.
In addition, their crying and distress was 1.5 SDs above the sample mean. The re-
maining infants comprised the typical criers group (N = 102, 56 boys). Compar-
isons of the means for excessive criers and typical criers show that excessive criers
cried and fussed signiﬁcantly longer and more frequently than typical criers. In-
deed, as can be seen in seen in Table 3, the duration of their daily fuss bouts was
above the mean for fussing (135 min) reported by St. James Roberts and col-
All longitudinal data analyses comparing the two cry groups were conducted us-
ing repeated measures analysis of variance (ANOVA) with cry group and sex as
between-subject factor and age as the within-group factor.
Prior to conducting the following analyses, cry group differences on the de-
mographic data were tested using one-way ANOVAs. Only one difference
emerged between excessive criers and typical criers. Fathers of excessive criers
were more educated than fathers of typical criers, F(1, 115) = 4.47, p < .05 (see
STIFTER AND SPINRAD
Reactivity to Arm Restraint
Our ﬁrst aim was to examine whether infants who cried excessively at 6 weeks
continued to exhibit high levels of negative reactivity at 5 and 10 months of age.
Before testing for differences in negative reactivity during the ﬁve episodes of
the arm restraint task (arm hold, arm release, distal soothe, proximal soothe, and
pickup) the intercorrelation of these variables was examined. As expected, in-
fants were very stable in their level of crying across the ﬁve episodes at both
5 months (rs ranging from .50 to .89) and 10 months (rs ranging from .29 to .73).
To simplify the following analyses, infant reactivity to the arm hold and arm re-
lease episodes were averaged to form the variable reactivity to restraint at each
age. In addition, the three episodes during which the mother soothed the infant
were combined to create a reactivity to mother soothe composite at each age.
Cross-age correlations revealed modest stability between 5 and 10 months for
reactivity to restraint, r = .23, p < .01, but less so for mother soothe, r = .15, p <
.10. Interestingly, when the stability of the reactivity variables was investigated
by group, only the typical criers showed signiﬁcant cross-age stability for reac-
tivity to arm restraint, r = .27, p < .004, whereas the correlations between 5- and
10-month reactivity for the excessive criers was negative, r = –.18, p = .53.
Repeated measures ANOVA with cry group and sex as the between-subject
factors and age as the repeated factor was used to test for differences in reactivity
to arm restraint and mother soothe. A signiﬁcant group effect showed that overall
excessive criers were more negatively reactive than typical criers during restraint,
F(1, 112) = 4.72, p < .05, η2 = .04, and mother soothe, F(1, 111) = 3.96, p < .05,
η2 = .03.2 In addition, a main effect for age was found for negative reactivity dur-
ing arm restraint, F(1, 116) = 46.97, p < .001, η2 = .29. Infants increased in their
reactivity to arm restraint from 5 to 10 months of age. No signiﬁcant interactions
with sex or age emerged. Table 4 reports the means and standard deviations. As re-
activity during restraint and mother soothe were signiﬁcantly related to each other
at 5 months, r = .59, p < .001, and 10 months, r = .57, p < .001, they were aver-
aged to form one reactivity score.
Regulatory Behaviors During Arm Restraint
As with negative reactivity, regulatory behavior exhibited during the arm hold and
regulatory behavior exhibited during the arm release episodes were positively
related at 5 months, r = .44, p < .0001, and 10 months, r = .35, p < .0001. Therefore,
we summed the two regulatory composites before conducting further analyses (recall
that regulation was not coded during mother soothe segments). Cross-age stability
2Data for 1 infant were not available for this analysis because the infant fell asleep after restraint.