Michel Boivin, Ph.D.
GRIP, École de psychologie
Université Laval, Québec, Canada
Social wariness and preference for solitude, two dimensions of social withdrawal, show unique associations with various socioemotional difficulties in childhood, including internalizing and peer problems. However, their early childhood predictors remain vastly undocumented. The present study aimed to examine whether early indicators of reactivity in situations of unfamiliarity such as behavioral inhibition, affect and cortisol independently, or in interaction with emotion regulation as indexed by vagal tone, predict later social wariness and preference for solitude. Participants were 1209 children from the Quebec Newborn Twin Study. Vagal tone was assessed at 5 months, and behavioral inhibition, affect and cortisol were assessed at 19 months in situations of unfamiliarity. Mothers, teachers, and peers evaluated social wariness and preference for solitude repeatedly from 4 to 10 years old. Findings show that three temperamental dimensions, social inhibition, non-social inhibition, and affect accounted for the variability in reactions to unfamiliarity. Behavioral inhibition to social unfamiliarity at 19 months predicted social wariness during the preschool years. Poor vagal regulation at 5 months exacerbated the risk associated with negative affect at 19 months to predict preference for solitude during the preschool years. Overall, results show that social wariness and preference for solitude may follow different developmental pathways.
Keywords: social wariness, preference for solitude, behavioral inhibition, cortisol, vagal tone, longitudinal study, social withdrawal
Social interactions with peers contribute to shape socioemotional skills that are central to children’s development (Bukowski et al., 2015). Yet, some children consistently refrain from initiating interactions with other children and often find themselves isolated from the peer group (Rubin et al., 2009). Social withdrawal is an overarching term for all behaviors leading to self-isolation, including social wariness and preference for solitude (Rubin et al., 2009). Social wariness and preference for solitude are distinct dimensions of social withdrawal, as they are manifested in different social contexts (i.e., unfamiliar versus familiar) and differ with respect to their associations with socioemotional difficulties (Coplan et al., 2013; Degnan et al., 2008; Morneau-Vaillancourt et al., 2021). Social wariness is presumed to reflect an ambivalence between high approach and high avoidance motivations, where children typically want to approach their peers but are too anxious to do so (Asendorpf, 1990). Social wariness is thus characterized by heightened levels of anxiety when meeting new people for the first time. By comparison, preference for solitude reflects an inclination for solitary activities in the presence of peers. According to Asendorpf (1990), children may prefer solitude for two reasons: either because they enjoy spending time on their own without necessarily avoiding peer interactions (i.e., unsociability, characterized by low approach and low avoidance motivations) or because they feel emotionally overwhelmed in social situations and therefore avoid interacting with peers (i.e., social avoidance, characterized by low approach and high avoidance motivations; Asendorpf, 1990). Studies have showed that social wariness and preference for solitude are uniquely associated with indices of socioemotional adjustment such as internalizing and peer problems (Coplan et al., 2004, 2018; Coplan & Weeks, 2010; Morneau-Vaillancourt et al., 2021). For instance, social wariness or related constructs such as shyness are consistently associated with internalizing difficulties in early and middle childhood (Coplan et al., 2004, 2018; Coplan & Weeks, 2010). Although this may also be true for children who prefer solitude because they actively avoid others (i.e., social avoidance), it is not always the case for those who are unsociable (Coplan et al., 2015). Nonetheless, preferring solitude in the presence of peers may violate social norms later in childhood and in early adolescence as social affiliations become increasingly valued among peers, thereby leading to peer difficulties (Morneau-Vaillancourt et al., 2021). Therefore, social wariness and preference for solitude may have distinct implications for children’s socioemotional adjustment at different developmental periods. These distinctions further suggest that social wariness and preference for solitude could result from different developmental pathways. Identifying their early developmental factors will contribute to identify children at risk for social wariness and preference for solitude. However, these factors remain vastly undocumented as there is a lack of research examining these questions.
Difficulties in adapting to daily social demands, for instance getting acquainted with new peers at school or asking a question to adults, may be rooted in temperamental dispositions such as behavioral inhibition (Buss, 2011; Fox et al., 2005). Some toddlers react strongly and manifest signs of elevated discomfort (e.g., by crying) in contexts of novelty, such as meeting new people or being exposed to new objects (Kagan et al., 1984). Children’s reactions to unfamiliarity have often been considered under the same unidimensional umbrella as a temperamental trait (Kagan et al., 1984). However, evidence shows that some inhibited children do not always manifest negative emotions in situations of unfamiliarity (Kochanska & Radke-Yarrow, 1992; Putnam & Stifter, 2005). Thus, although behavioral and emotional reactivity to novelty are related, they may reflect distinct components of temperament. In turn, these distinct components of temperament may be uniquely related to specific dimensions of social withdrawal. For instance, social wariness is presumably characterized by an ambivalence between approach and avoidance (Asendorpf, 1990); socially wary children may be interested in interacting with new people but may be too anxious to do so. Young children who find meeting new people unsettling and stressful, but not necessarily unpleasant, may therefore be inhibited but calm in situations of social unfamiliarity (Kochanska & Radke-Yarrow, 1992). On the other hand, children who actively avoid social interactions and turn to solitary activities instead (i.e., preference for solitude) are likely to suffer from emotional difficulties, including depressive symptoms and loneliness (Coplan et al., 2013, 2018). Accordingly, a temperamental disposition for low positive or negative affect could forecast an emerging preference for solitude in childhood. However, this question remains undocumented due to a lack of longitudinal information on the predictors of preference for solitude. Therefore, behavioral inhibition may be more specifically related to social wariness, whereas negative affect may forecast later preference for solitude. Clearly, there is a need for more longitudinal studies documenting behavioral inhibition and affect independently to more precisely depict their role in the development of social wariness and preference for solitude.
In addition, there is also evidence that inhibition to unfamiliar people (i.e., social inhibition) and inhibition to unfamiliar settings or objects (i.e., non-social inhibition) are only weakly correlated. Accordingly, they may represent distinct behavioral tendencies and differentially predict specific dimensions of social withdrawal (Dyson et al., 2011; Kochanska & Radke-Yarrow, 1992; Rubin et al., 1997). For instance, Dyson and colleagues (2011) showed that young children who were inhibited when meeting an adult stranger were more likely to be socially wary, shy and socially anxious than children who were inhibited when exposed to new objects. Further, Kochanska and Radke-Yarrow (1992) found that social inhibition in toddlerhood, defined as fear of strangers, was predictively associated with social wariness, whereas non-social inhibition, defined as reticence to explore new environments or objects, rather predicted a lack of involvement in group play. These studies indicate that early social inhibition (i.e., behavior inhibition regarding a stranger) and non-social inhibition (i.e., behavior inhibition when exposed to new objects) may play different roles in the development of specific dimensions of social withdrawal.
In addition to those behavioral markers of temperament, the cortisol response to stress may be one key etiological factor in the onset and persistence of social withdrawal. Cortisol is a glucocorticoid stress hormone secreted by hypothalamic-pituitary-adrenal (HPA) axis activity that, among other functions, help individuals gather energy to cope with environmental demands (Gunnar & Quevedo, 2007; Ulrich-Lai & Herman, 2009). Although cortisol secretion in challenging situations is considered adaptative, some individuals show dysregulated cortisol responses in mildly stressful situations, which may indicate a higher sensitivity to stress (van den Bos et al., 2017). Children who show a higher cortisol response in social situations may thus remove themselves from these situations to alleviate unpleasant physiological symptoms due to stress. To the extent that socially wary children experience greater stress and anxiety when facing typical social demands or are more sensitive to mildly stressful social situations, they should show a higher cortisol response in that context than children who prefer solitude (Coplan et al., 2013; Davis & Buss, 2012; Schmidt et al., 1997). High reactivity to unfamiliarity in toddlerhood, as captured by elevations in cortisol levels detected in the saliva (Ouellet-Morin et al., 2008), may therefore be key in understanding how early inhibition specifically forecast later social wariness (Kagan et al., 1987; Schmidt et al., 1997). In other words, an early cortisol response to stress could differentiate the risk of developing social wariness versus preference for solitude. However, this question remains unanswered as most studies focused on anxious forms of withdrawal (e.g., shyness) or on social withdrawal broadly. Whether there is a unique association between cortisol and preference for solitude therefore remains an open question.
Prior research has shown mixed evidence regarding the association between cortisol reactivity and social withdrawal. While some studies revealed that children prone to anxiety showed increased cortisol in social-evaluative situations (van den Bos et al., 2017; van West et al., 2008), other studies indicated that shyness and fearfulness were not related to higher levels of cortisol (Blair et al., 2004; Schmidt et al., 1997). These mixed findings could be due to developmental differences across studies, as well as to the small sample sizes and thus limited power in some studies. There is also extensive evidence pointing to childhood maltreatment and quality of care as putative factors affecting cortisol regulation (Gunnar & Donzella, 2002; Lupien et al., 2009; MacMillan et al., 2009; Ouellet-Morin et al., 2008; Poole et al., 2018). Therefore, a more comprehensive overview of the link between cortisol and social behavior calls for well-powered studies that focus on specific developmental periods and considers multiple familial factors such as parenting behaviors and socioeconomic background.
Apart from these potential contextual influences activating the response to stress, the child’s capacity to regulate emotion may play a key role in attenuating early behavioral (i.e., behavioral inhibition and affect) and physiological (i.e., cortisol) responses (Smith et al., 2019). Emotion regulation through vagal tone, a marker of parasympathetic regulation activated by the vagus nerve, has been linked to behavior, social problems and cognitive outcomes in children (Graziano & Derefinko, 2013). According to Porges’ polyvagal theory (Porges, 2007), a release of the vagal brake (i.e., vagal withdrawal) reflects an adaptative allocation of resources to cope efficiently with environmental demands (i.e., by increasing cardiac activity). For children who already experience elevated stress in social situations, having difficulties in regulating heart responses, reflected by an absence of vagal regulation, is likely to exacerbate their emotional (e.g., distress) responses (Buss et al., 2018; Viana et al., 2021). As these children experience more discomfort, they may prefer to avoid social interactions and, over time, be more prone to exhibit social withdrawal (Buss et al., 2018). Although maturational changes in cardiac vagal tone occur during the first years of life, individual differences in vagal tone are moderately stable over this period (Bornstein & Suess, 2000; Fracasso et al., 1994; Friedman, 2007; Patriquin et al., 2014; Porges et al., 1994; Wagner et al., 2021). As such, vagal tone, an early marker of individual differences in emotion regulation, may contribute to explain why some inhibited and reactive infants are more likely to become socially withdrawn (Wagner et al., 2021).
Findings regarding vagal tone as a marker of emotion regulation are however inconsistent and again, mostly relate to social behaviors broadly or to anxious forms of withdrawal (e.g., inhibition), thereby limiting our understanding of how vagal tone may shape risk for preference for solitude. On the one hand, a larger vagal modulation, that is, a larger decrease in heart rate variability from resting to stressful situations, has been related to children’s capacity to regulate emotions in stressful situations and more generally, to low internalizing difficulties (Graziano & Derefinko, 2013; Perry et al., 2016; Porges et al., 1996). Displaying little change in vagal tone when challenged may reflect greater difficulty to physiologically adapt to social stressors. Accordingly, withdrawal behaviors could be an attempt to preempt or reduce these stress responses. On the other hand, larger vagal modulation has also been linked to higher childhood internalizing problems and comorbid externalizing problems, suggesting that excessive vagal modulation may reflect emotional liability (Beauchaine, 2001; Calkins et al., 2007; Kossowsky et al., 2012; Shanahan et al., 2014). Finally, some studies report no association between vagal tone and internalizing problems or inhibition (Hastings et al., 2008; Marshall & Stevenson-Hinde, 1998; Schmidt et al., 1999). As for cortisol response, multiple child-specific and familial factors may affect vagal regulation, including age and sex of the child, parenting behaviors and parental mental health (Graziano & Derefinko, 2013; Hastings et al., 2008). Again, considering parenting behaviors and socioeconomic background may help better comprehend the role of vagal tone in regulating responses to social demands.
In addition to these mixed findings, some issues have not been properly addressed in previous studies. Research suggests that there are various indicators of reactivity to novelty in early childhood (e.g., behavioral inhibition, emotional reactivity, cortisol response), and that their manifestation may depend on the type of unfamiliar setting to which the child is exposed (i.e., social versus non-social). In addition, emotional regulation, as indexed by vagal tone, may play a role in attenuating or exacerbating the risk associated with these early temperamental and biological dispositions. Yet, little research has considered these different facets of reactivity altogether in both social and non-social contexts, as well as the extent to which emotion regulation modulates their impact. In addition, despite the well-documented distinctions between social wariness and preference for solitude (e.g., Morneau-Vaillancourt et al., 2021), no study has yet documented the unique patterns of associations between these early temperamental and biological indicators and specific dimensions of social withdrawal. As prior studies mostly focused on social withdrawal broadly or exclusively on anxious dimensions of withdrawal (e.g., social wariness), it is unclear whether social wariness and preference for solitude have unique developmental pathways. Examining various indicators of reactivity in both social and non-social contexts and the extent to which emotional regulation modulates their impact will help better understand the early risk factors associated with dimensions of social withdrawal.
Finally, these questions have rarely been examined over an extended period in childhood, for instance before and after the transition to formal schooling. This transition period is critical, as it brings additional social demands, including adults’ expectations about children’s capacity to interact with peers and regulate their emotions in social contexts (Buss, 2011). If this is true for all children, the transition to kindergarten may be particularly stressful for socially withdrawn children (Rubin et al., 2009). As previous studies were mostly cross-sectional or included follow-ups conducted over a short period of time, the way in which these associations unfold from the preschool to the grade school years remains undocumented.
The main goal of this study was to examine the extent to which early temperamental and biological dispositions in early childhood later predict the development of social wariness and preference for solitude. To address current knowledge gaps, the present study used follow-up data from a prospective longitudinal cohort, collected over a 10-year period using observations of behavior in unfamiliar situations, biological samples, and behavioral evaluations from mothers, teachers, and peers. The specific objectives were to examine 1) whether behavioral and emotional reactions across situations of unfamiliarity at 19 months reflect different dispositions rather than a unidimensional trait, 2) the extent to which these dispositions and the cortisol response in the context of unfamiliarity at 19 months predict social wariness and preference for solitude in early (4 and 5 years old) and middle childhood (6 to 10 years old), that is, before and after transition to formal schooling, and 3) whether these predictions vary as a function of the physiological capacity to regulate emotions, indexed by vagal tone assessed at 5 months. We expected that inhibition to social unfamiliarity specifically would be associated with later social wariness (Dyson et al., 2011), whereas negative affect would be uniquely related to later preference for solitude, as social avoidance (one form of preference for solitude) may indicate a high risk for emotional difficulties (Coplan et al., 2015). Here, preference for solitude is seen as a general inclination for solitary activities, which encompasses both social avoidance and unsociability. Finally, considering the lack of consistency in findings on cortisol and vagal tone, we did not posit any hypothesis as to their roles in the development of social wariness and preference for solitude.
Participants were from the Quebec Newborn Twin Study (QNTS), an ongoing prospective longitudinal study. The QNTS recruited 662 families after getting access to birth records of newborn twins in the greater Montreal area, Canada, between April 1995 and December 1998. At recruitment, families in the sample were representative of the greater Montreal area population in the mid-1990s in terms of sociodemographic characteristics (Boivin et al., 2019). Participants were assessed on familial, behavioral, social and educational characteristics from birth to early adulthood. Among the participants included in the present study (51% girl), 86% were White, 5% were Black, 3% were Asian and less than 1% were from First Nations communities in Canada.
The present study used data collected over multiple waves, starting with laboratory assessments of vagal modulation at 5 months and of behavioral inhibition, affect and cortisol response at 19 months. Due to limited financial resources available, a subsample of participants were assessed in the laboratory at 5 and 19 months (N = 437 to 462). In addition, mothers reported their children’s social wariness and preference for solitude at 4 and 5 years old, whereas teachers and peers completed these reports when children were 6, 7 and 10 years old. The number of participants for whom social wariness and preference for solitude was evaluated at 4, 5, 6, 7 and 10 years old varied from 721 to 880 (descriptive statistics are presented in Table 1). In the final model, 1209 participants were included using full information maximum likelihood estimation, as these participants had available data for at least one variable. The pattern of missing data was missing at random (MAR) for the most part, satisfying conditions for our models. Most variables of interest were unrelated to missing data, and inconsistent patterns of associations with sociodemographic variables were found. For instance, at 19 months, children for whom data was missing came from higher socioeconomic backgrounds (e.g., with fathers who were more educated). At other waves, children for whom data were missing were more often from minority and lower socioeconomic backgrounds (e.g., from families with lower income). Missing data analyses are described in the Supporting information.
For all follow-up assessments, parental informed consent and child assent were obtained, and ethics review boards at Université Laval, Québec City, and at Université de Montréal and Ste-Justine Hospital, Montréal approved the procedure.
At approximately 5 months (mean [M] = 5.27, standard deviation [SD] = .64), mothers and twins visited the laboratory in Montreal, Canada. An assigned team of research assistants assessed each twin individually in separate rooms. Visits took place between 10:00 a.m. and 3:00 p.m., at least one hour after eating. We assessed vagal tone by recording heart rate variability through electrocardiograms during two resting and two unfamiliar and cognitively challenging situations. High heart frequency within 0.15-1.0 Hz were obtained by the mean of the autoregressive forward-backward linear least-squares method with a 20th-order using Heart Rate Variability Analysis Software 1.1 (Poliakova et al., 2014). Descriptive statistics for individual heart rate variability measures are presented in Table S1 of the Supporting Information.
Resting Heart Rate Variability. To assess resting heart rate variability, electrocardiograms were recorded twice over 4-minute periods when young children were sleeping in a reclining car seat in a horizontal posture and then in a vertical posture (after a period of 60 seconds following vertical tilt to allow heart rate variability to stabilize first). This procedure was adapted from a task previously used with children and adolescents (Mezzacappa et al., 1997, 1998). We assessed electrocardiograms during quiet sleep to limit physiological changes due to irregular movement and breathing in awake infants. Throughout the procedure, a research assistant assessed whether the infant was in a state of quiet sleep based on limb, eye and facial muscular movements. Distribution of heart rate variability in horizontal and vertical resting positions were positively skewed and were thus log-transformed. Since they were highly correlated (r = .78, p < .01), we averaged both resting positions and used this mean as the resting heart rate variability score.
Heart Rate Variability in Challenging Tasks. We then recorded heart rate variability in two challenging unfamiliar situations, which aimed to engage the child cognitively. During a first habituation task, young children were held on a research assistant’s lap and faced a projection screen, on which a rotating multicoloured circular cross moved vertically. The number of seconds the child looked directly at the stimulus was recorded for the first two sustained looks. Habituation was achieved once subsequent visual fixation was half that of the average of the first two looks. Then, a recovery phase followed, where a black and white checkerboard pattern travelled horizontally on the screen and alternated with the now familiar multicoloured cross. During a second mobile task, adapted from the Kagan battery (Kagan & Snidman, 1991), three toy mobiles were presented to the infant at approximately 20cm from his or her face by a research assistant standing behind. The mobiles respectively included one, three and seven moving toys and were presented for 20 seconds, separated by 10-second intervals. This sequence was repeated a second time. Since heart rate variability was strongly correlated across the two challenging tasks (r = .60, p < .01), we used the average across both tasks as the challenge heart rate variability score.
Computing Vagal Modulation Scores. As in previous studies, we computed vagal modulation scores by calculating the change in heart rate variability from resting to challenging tasks (i.e., vagal modulation = challenge - resting) and used these scores in the following analyses (Graziano et al., 2007). Negative scores indicated larger decrease in heart rate variability from resting to challenging tasks, an adaptative process for coping with environmental demands (Porges, 2007). Vagal modulation was not associated with birth weight, weight at 5 months, gestational age, and Apgar score assessed 5 minutes after birth (r = -.00 to .05, p > .27), and did not vary as a function of sex (t(426) = .61, p = .54).
When participants were approximately 19 months old (M = 18.50, SD = .82), mothers and children visited the laboratory again. Visits took place in the morning, between 8:05 a.m. and 12:15 p.m. After arriving at the laboratory, the mother and the child entered a room in which the child was exposed to two unfamiliar situations. The first social situation involved a female adult dressed as a clown inviting the child to approach by offering a set of familiar toys. The second non-social situation involved an odd-looking and noisy mechanical robot placed on a platform in the opposite corner of the room. Both situations, each lasting 4 minutes and 40 seconds, were separated from a 5-minute free play period. In the first 2 minutes and 20 seconds of each situation, the mother was instructed to limit interactions with her child and to answer only briefly when the child asked for support (e.g., by saying “It’s okay, it’s a clown”). For the remaining time, the mother was allowed to soothe her child or do whatever she thought would alleviate the child’s distress, if necessary. The procedure was recorded with two unobtrusive cameras to allow for later double-blind coding of behaviors for the first 2 minutes and 20 seconds of the clown and robot situations, respectively.
The systematic coding of behaviors and emotional reactivity was analogous to previous studies on behavioral inhibition (Garcia-Coll et al., 1984; Kochanska & Radke-Yarrow, 1992; Rubin et al., 1997). Six behavioral indices were coded for each situation: 1) time spent in physical proximity to the mother (number of seconds spent within two feet of the mother), 2) latency leaving the mother (number of seconds before leaving the mother, i.e., the two-feet proximal zone), 3) latency approaching the stimulus (number of seconds before approaching within two feet of the unfamiliar stimulus), 4) interaction with the stimulus (sum of occurrences where the child interacted directly or indirectly with the stimulus), 5) negative affect and 6) positive affect. Research assistants coded negative and positive affect over 20-second intervals using an intensity scale ranging from 0 to 3 (0 = neutral, 1 = mild, 2 = moderate, 3 = intense). We then computed the average negative and positive affect intensity, respectively, over the 7 intervals. Descriptive statistics for the coded behavioral reactions are presented in Table S1 of the Supporting Information.
Observations were coded by two research assistants who went through a 3-week training prior to the codification. To ensure independent and unbiased coding, each twin of a pair was coded by a different coder, and all coders were kept blind regarding the zygosity status of the twin. Research assistants first double-coded 6% of the sample. Then, throughout coding, inter-rater agreement was monitored once a week using the following criteria: correlations for observations of behavior above .90 and kappa coefficients for the ratings of emotional reactions above .80.
Sub-Dimensions of Behavioral Inhibition. Given prior work suggesting the multidimensional nature of the behavioral reactions to unfamiliarity (Kochanska & Radke-Yarrow, 1992; Putnam & Stifter, 2005), the twelve behavioral scores (six for each situation) were submitted to a principal component analysis using SPSS Statistics (Version 27). As presented in Table S2, three factor components accounting for 63% of the variance were identified. The first factor represented avoidance of the robot, the second, approach of the clown, and the third, positive and negative affect across both clown and robot situations. Factor scores were saved and used in the following analyses. Scores for approach of the clown were reversed (i.e., multiplied by -1) to represent avoidance of the clown (i.e., social inhibition). Positive affect scores represented positive emotional reactions, and vice versa for negative scores. Avoidance of the robot (i.e., non-social inhibition) was weakly correlated with social inhibition (r = .22, p < .01), and negatively correlated with affect (r = -.27, p < .01). Social inhibition was weakly and negatively correlated with affect (r = -.09, p = .052).
To assess cortisol response, saliva samples were collected before and 20 minutes after the clown and robot situations. All samples were collected between 8:05 a.m. and 12:15 p.m. and the exact time of collection was recorded. Samples were later stored at -80 °C until analysis. This procedure is described in more details elsewhere (Ouellet-Morin et al., 2008). We computed cortisol response scores by calculating the difference between pretest and posttest cortisol and then dividing this change score by pretest cortisol, in accordance with the law of initial values (Wilder, 1957). As presented in Ouellet-Morin and colleagues (2008) either pretest nor posttest cortisol levels varied as a function of zygosity, sex, ethnicity, or exact time of day when samples were collected. Descriptive statistics for pretest and posttest cortisol levels are presented in Table S1 of the Supporting Information.
Social wariness and preference for solitude were assessed through mother ratings when the participants were aged 4 and 5 years. Mothers were asked to answer questions on a 0-to-2 scale (0 = never, 1 = sometimes, or 2 = often), starting with Over the past 12 months, at which frequency would you say that your child… The social wariness scale included 6 questions (e.g., was too fearful or anxious) and was adapted from Asendorpf’s scale of situational shyness among unfamiliar peers and Achenbach’s Teacher Report Form (Achenbach, 1991; Asendorpf, 1987). The preference for solitude scale included 3 questions (e.g., preferred to play alone rather than with other children) and was adapted from the Teacher Report Form and from the Child Behavior Scale (Achenbach, 1991; Ladd & Profilet, 1996). Questions for both scales are presented in Table S3 of the Supporting information. Internal consistency, evaluated using ordinal alpha, was satisfactory for both social wariness and preference for solitude at 4 (α = .79, .66, respectively) and 5 years old (α = .78, .68, respectively). We averaged items to compute final scores at each assessment, requiring at least 4 out of 6 items to be non-missing for social wariness, and 2 out of 3 for preference for solitude.
In school, social wariness and preference for solitude were assessed through a combination of teacher ratings and peer nominations when participants were 6 (kindergarten), 7 (first grade) and 10 years old (fourth grade).
Teacher Ratings. For social wariness, teachers were asked to rate the participants’ behavior over the past six months according to the same items used by the mothers at 4 and 5 years old, with an additional item, avoided the company of other children (7 items in total). For preference for solitude, teachers were asked similar questions as those used by mothers, plus an additional question: demonstrated little interest in activities with other children (4 items in total). Ordinal alpha for teacher measures of social wariness and preference for solitude were satisfactory at 6 (α = .83 and .82, respectively), 7 (α = .82 and .80) and 10 (α = .81 and .81) years old. At each timepoint, we averaged items for social wariness and preference for solitude, requiring at least 5 out of 7 items to be non-missing for social wariness, and 3 out of 4 for preference for solitude. Teacher-rated social wariness and preference for solitude were moderately correlated at each assessment wave (r = .51 to .54, p < .01), suggesting that teachers ratings indicated partially independent constructs (see Table S4 in the Supporting information for correlations between teacher and peer ratings). Correlations across assessment waves were mostly moderate for social wariness (r = .13 to .25, p < .01), and the same was found for preference for solitude (r = .21 to .28, p < .01), indicating moderate stability in both variables over time.
Peer Ratings. At 6 and 7 years old, children in the classroom were asked to nominate from a roster of photos two classmates who were the shyest (Boivin et al., 2013). At 10 years old, peers had to choose two classmates who 1) were the shyest, 2) were the most fearful, and 3) worried all the time. The number of nominations received at 10 years old for the three items were moderately to strongly correlated (r = .27 to .54, p < .01). We averaged these items to create a social wariness peer nomination score at 10 years old, requiring at least 2 out of 3 items to be non-missing. As for preference for solitude at 6, 7 and 10 years old, children had to nominate two classmates who preferred to play by themselves. Peer-rated social wariness was slightly more stable across time (r = .19 to .28, p < .01) than preference for solitude (r = .08 to .13, p < .07; see Table S4). The two behaviors were weakly to moderately correlated at each assessment wave (r = .14 to .42, p < .05), suggesting that peers in the classroom perceived social wariness and preference for solitude as two partially different behaviors.
Composite Teacher and Peer Ratings. Cross-informant (teachers and peers) correlations for social wariness and preference for solitude were moderate at 7 and 10 years old (r = .21 and .38, p < .05; see Table S4), except for a weak correlation for preference for solitude at 6 years old (r = .07, p = .053). To create composite scores based on multiple informants, we z-standardized and then averaged teacher ratings and peer nominations at each assessment wave. These composite scores were used in the following analyses.
Social wariness and preference for solitude scores were then used to estimate latent variables using structural equation modeling. Latent variables typically reduce measurement error, especially if based on repeated assessments over time. Here, mother ratings of social wariness were moderately correlated at 4 and 5 years old, and the same was found for preference for solitude (see Table 2). Accordingly, we estimated one latent variable for social wariness in preschool and the same was done for preference for solitude. Similarly, composite scores of social wariness at 6, 7 and 10 years old (based on both teacher ratings and peer nominations) were weakly to moderately correlated, and the same was found for preference for solitude (see Table 2). Therefore, social wariness scores at 6, 7 and 10 years old were used to estimate a latent variable in grade school, and the same was done for preference for solitude. Latent variables for preschool and grade school were included in the models to examine potential changes in predictive associations and to account for temporal stability across the two developmental periods.
Descriptive analyses were conducted in SPSS Statistics (Version 27). We first inspected variables for outliers and performed transformations to improve distributions (see the Supporting information for more details). Table 1 presents means, standard deviations, skewness, kurtosis, and number of participants for all variables. We performed independent samples t-tests to examine sex differences (presented in the Supporting information). Girls had significantly lower preference for solitude scores than boys at age 6, 7 and 10 years, and significantly higher social wariness scores than boys at age 10 years. Therefore, these variables were regressed on sex in the structural equation models.
We then ran two sequential structural equation models to examine pathways through which early behavioral inhibition, affect and cortisol response predicted social wariness and preference for solitude, and the extent to which vagal modulation moderated these associations. Models were sequential in that they considered predicting social wariness and preference for solitude in preschool, and then the same outcomes in grade school. Models were first estimated with full information maximum likelihood and the cluster option in Mplus 8.2 to include participants with at least one available data (N = 1209) and to handle missing data, non-normal distributions of variables and dependency of twin data (Muthen & Muthen, 2017). We then estimated the models by bootstrapping 1,000 times to obtain confidence intervals.
We tested sequential models in two steps. First, we examined how social and non-social inhibition, affect, cortisol response and vagal modulation predicted dimensions of social withdrawal in preschool and grade school. All predictors were considered as a block to examine their unique contributions. Second, we examined whether vagal modulation at 5 months moderated these associations. Interaction terms were computed using z-standardized variables. Again, all predictors and their interactions with vagal modulation were entered as a block to examine their unique contributions.
In addition, the following parameters were included in both sequential models: 1) correlations between predictors to account for their interrelatedness; 2) concurrent correlations between social wariness and preference for solitude at 4, 5, 6, 7 and 10 years old and correlations between the latent constructs during the preschool and grade school years to account for the phenotypic overlap between both dimensions of social withdrawal; 3) regressive paths from the preschool to the grade school years to account for the stability in both social wariness and preference for solitude over time; 4) regressive paths from sex to social wariness at age 10 years and preference for solitude at ages 6, 7 and 10 years to account for observed sex differences.
As early temperamental and biological dispositions are often related to familial adversity factors (e.g., Graziano & Derefinko, 2013), both sequential models accounted for other confounding sociodemographic and parenting factors by regressing all observed variables on both parents’ education level, first language spoken, ethnicity, age at child’s birth, overprotective parenting behaviors when children were 30 months old, family income and family status (i.e., single-parent versus two-parent family). These covariates are presented in more details in the Supporting information.
To examine whether results were affected by the lower number of participants assessed at 5 and 19 months, we examined models (without covariates, to simplify models) using only the participants evaluated in the laboratory at 5 or 19 months (N = 584). Path estimates and model fit indices were analogous to models including participants who had at least one available data (N = 1209). We thus present the findings drawn from all available data to maximize statistical power and representativeness.
Since vagal modulation scores were computed using the average of heart rate variability across two challenging tasks (i.e., habituation and mobile), we ran two separate models using vagal modulation for each task separately. As results were highly similar in both models, we present findings using the average vagal modulation across both tasks.
At each step, we also ran models including only grade school latent variables to examine whether early predictors were directly associated with social wariness and preference for solitude during the grade school years. Findings were analogous to those of the models including preschool variables, suggesting that accounting for the stability across the two developmental periods did not significantly impact the results.
We first tested a model examining the predictive associations from social and non-social inhibition, affect, cortisol and vagal modulation to social wariness and preference for solitude (Figure S2). Then, we tested a model including all predictors and their respective interactions with vagal modulation (Figure 1). The models were sequential in that they accounted for the temporal stability for both social wariness and preference for solitude from the preschool (i.e., 4-5 years old) to the grade school years (i.e., 6 to 10 years old). The final model including the interactions was conducted using data from 1209 participants and had a good fit: Root Mean Square Error of Approximation (RMSEA) = .02 (90% confidence intervals [CI] = .01-.02); Comparative Fit Index (CFI) = .98; Tucker Lewis Index (TLI) = .91. To help the model overcome convergence issues, we fixed the residual variance of preference for solitude at 5 years old (i.e., mother ratings) to 0.
Figure 1 shows the final model including the putative interactions. Results show that observed social inhibition significantly predicted social wariness at 4-5 years old (β = .16, p = .02). Young children who were reluctant to approaching the clown at 19 months were more likely to be seen as manifesting higher social wariness with unfamiliar peers later in preschool years. No other significant association between predictors and preschool social wariness and preference for solitude was found. Vagal modulation interacted with affect in predicting preschool preference for solitude (β = -.16, p = .02). Specifically, when this interaction was decomposed, negative affect (as reflected by negative scores on the affect component) predicted higher preschool preference for solitude at lower levels of vagal modulation (see Figure 2). No other interaction with vagal modulation significantly predicted social wariness or preference for solitude in preschool.
As shown in Figure 1, social wariness and preference for solitude were moderately stable between preschool and grade school (β = .36; β = .24, respectively). After accounting for this stability, no early predictor was significantly associated with social wariness and preference for solitude in grade school. We also found that none of these predictors interacted with vagal modulation to predict grade school social wariness and preference for solitude. However, the mediation tests performed with bootstrapping showed a significant and a marginally significant indirect prediction via preschool variables. First, social inhibition at 19 months predicted social wariness at 6-10 years old via social wariness at 4-5 years old (β = .06; 95% CI = .01, .13). Second, the interaction between affect and vagal modulation marginally predicted preference for solitude at 6-10 years old via preference for solitude at 4-5 years old (β = -.02; 95% CI = -.06, .01).
The goal of the present study was to examine 1) the ways in which different early childhood temperamental and biological dispositions, that is, behavioral inhibition, emotional reactivity, and cortisol response in situations of unfamiliarity at 19 months, predicted social wariness and preference for solitude in preschool and in grade school, and 2) whether these predictive associations varied depending on vagal tone at 5 months. First, we found that three behavioral facets, that is inhibition in a social novel context (social inhibition), inhibition in a non-social novel context (non-social inhibition), and affect in both contexts, accounted for most of the variability in behavioral and emotional reactions across situations of unfamiliarity at 19 months. Second, social inhibition specifically predicted social wariness in the preschool years, as well as indirectly in grade school. Children who were reluctant in approaching an unfamiliar clown (i.e., who were highly inhibited) were more likely to manifest social wariness later. Third, children’s vagal modulation at 5 months moderated the association between affect and preference for solitude. Children who manifested more distress (negative affect) in unfamiliar situations at 19 months were more likely to prefer solitude in preschool years if they had low vagal modulation. However, non-social inhibition (i.e., inhibition during the robot situation) and the cortisol response across situations of unfamiliarity did not predict social wariness and preference for solitude.
Our results extend previous studies in several ways. First, we found that early temperamental and biological dispositions differentially predicted dimensions of social withdrawal. Inhibition during the clown situation (i.e., social inhibition) was associated with higher social wariness, but not preference for solitude, at 4-5 years old. This association was robust to the different methods used (i.e., observation of social inhibition in experimental situations predicting mother ratings of social wariness) and to the considerable time gap between both assessments (i.e., from 19 months to 4-5 years old). The prediction of preschool social wariness supports and extends prior evidence suggesting that inhibition to social unfamiliarity forecast later social withdrawal and that a temperamental profile characterized by complete avoidance unfamiliar people may be an early sign of later behavioral difficulties (Dyson et al., 2011; Garcia-Coll et al., 1984; Kochanska & Radke-Yarrow, 1992). One component linking early social inhibition to later shyness or social wariness may be sensitivity to social stress (Lobue & Pérez-Edgar, 2014). For instance, Lobue and Pérez-Edgar (2014) showed that extremely shy children may detect social threats (e.g., images of angry faces) more quickly than non-shy children, but that both shy and non-shy children did not differ regarding sensitivity to non-social threats (e.g., images of snakes), suggesting that shy children may be more sensitive to social stress specifically. Therefore, the consistent manifestation of inhibition in the presence of unfamiliar people at 19 months (i.e., never leaving the proximity of the mother) may indicate heightened sensitivity to social stress, which in turn, may be an early temperamental indicator differentiating social wariness from preference for solitude (as social inhibition was not related to preference for solitude).
Second, as most studies on early temperament and social behaviors focused on anxious types of social withdrawal (e.g., social wariness, shyness or anxiety), the predictors of preference for solitude remained undocumented. Here, we found that affect across situations of unfamiliarity at 19 months was uniquely associated with preference for solitude during the preschool years, at 4 and 5 years old, but only at certain levels of emotional regulation, as indexed by vagal modulation in challenging situations at 5 months. Children who manifested more negative affect in reaction to both social and non-social unfamiliarity showed higher preference for solitude if they previously showed low vagal modulation at 5 months. In contrast, high vagal modulation, which reflected a strong parasympathetic response in challenge situations, attenuated the risk associated with negative affect. This finding is in line with the polyvagal theory, which suggests that release of the vagal “brake” favors the allocation of resources, such as increase of cardiac activity, necessary to regulate emotions and cope with environmental demands (Porges, 2007). Our study shows that the physiological capacity to regulate emotions, indexed by vagal modulation, may play a role in differentiating which young children later develop preference for solitude. Indeed, a temperamental profile characterized by vulnerabilities in emotional reactivity and regulation may forecast the development of preference for solitude. One pathway by which this temperamental profile predicts preference for solitude could be through behavioral problems, which could lead to peer problems and eventually to preference for solitude. A temperamental profile characterized by negative affect (or low positive affect) and low self-regulation may be related to multiple behavioral problems such as internalizing and externalizing difficulties (Nigg, 2006). These behavioral problems often lead to negative peer experiences (Arseneault et al., 2010; Nigg, 2006), which in turn can increase children’s tendency to prefer solitude (Ladd et al., 2019). However, a similar pathway could lead to social wariness, where social inhibition in early childhood could predict internalizing and peer difficulties, which could in turn enhance risk for social wariness (Coplan et al., 2018). One alternative explanation for why the combination of low vagal modulation and negative affect may uniquely predict preference for solitude is through a higher risk for psychopathology. This risk could manifest through the expression of one specific form of preference for solitude that is, social avoidance. Researchers have suggested that social avoidance, which presumably reflects a combination of low approach and high avoidance motivations, may be the most problematic expression of social withdrawal in terms of its potential consequences on children's mental health (Asendorpf, 1990; Coplan et al., 2015, 2018). Findings show that, compared to other dimensions of social withdrawal, social avoidance is the most strongly related to depressive symptoms (Coplan et al., 2013, 2018; Ding et al., 2019). Researchers have thus suggested that social avoidance may be an early marker of social anhedonia, a key symptom of depression. This proposition that social avoidance may carry a high psychopathological risk is further supported by recent genetic findings showing that, compared to socially wary children, those who prefer solitude are characterized by a higher polygenic risk for multiple mental health problems (e.g., depression, autism spectrum disorder, attention deficit hyperactivity disorder; Morneau‐Vaillancourt et al., 2021). The present study therefore contributes to this emerging literature by suggesting that the combination of temperamental and biological vulnerabilities observed during the first years of life may be an early marker of psychopathological risk, which could in turn manifest behaviourally under the form of preference for solitude. However, our assessment of preference for solitude did not distinguish between social avoidance and unsociability, and our findings will need to be replicated while considering social avoidance specifically.
In addition, we found that informants only showed moderate agreement in their reports. Correlations between teacher and peer reports of social wariness and preference for solitude were weak to moderate (r = .25 to .30 for social wariness; r = .07 to .38 for preference for solitude; see Table S4). We also found that correlations within each construct were only slightly higher than correlations across constructs (see Table 2), which suggests that raters may have had difficulty distinguishing social wariness from preference for solitude, as they have limited access to children's internal motivations for isolating themselves (Spangler & Gazelle, 2009). This is consistent with prior studies indicating that it may be difficult for mothers, teachers, and peers to accurately assess specific dimensions of social withdrawal, and that self-reported assessments may provide better discrimination among dimensions of withdrawal (Eggum-Wilkens et al., 2020). Yet, as different social contexts may trigger different behavioral responses, social wariness and preference for solitude may overlap and coexist within individuals (Eggum-Wilkens et al., 2020). Here, as social wariness and preference for solitude were evaluated in different situations, some children may have been socially wary when meeting new peers while also choosing to play alone in the presence of familiar peers. However, since we focused on the developmental factors differentiating social wariness from preference for solitude, we conducted our analyses while accounting for this overlap (as is most often done in other studies; e.g., Coplan et al., 2013). It would be relevant for future research to examine the factors underlying the co-occurrence of both dimensions of social withdrawal.
Finally, despite evidence showing that young children’s temperamental expressions may vary across contexts (Dyson et al., 2011; Kochanska & Radke-Yarrow, 1992; Majdandžić & Van Den Boom, 2007; Rubin et al., 1997), prior studies often conceptualized behavioral inhibition as a unidimensional trait characterized by heightened inhibition and emotional distress in unfamiliar situations (e.g., Kagan et al., 1984). Yet, we found that early reactions to social and non-social novelty may reflect different facets of temperament, which in turn, may uniquely predict specific dimensions of social withdrawal (Kochanska & Radke-Yarrow, 1992; Putnam & Stifter, 2005). Our findings are in line with evidence showing that early inhibition in situations of social unfamiliarity (e.g., meeting an unfamiliar person) is not always correlated with inhibition in situations of non-social unfamiliarity (e.g., being exposed to new objects; Dyson et al., 2011; Kochanska & Radke-Yarrow, 1992). We showed that avoidance of an unfamiliar clown (i.e., social inhibition) was only weakly correlated with avoidance of a mechanical robot (i.e., non-social inhibition), indicating that social and non-social inhibition may reflect different dimensions of temperament.
In addition to social and non-social inhibition, we found that both positive and negative affect manifested across situations of unfamiliarity loaded on a single affect dimension. This finding is inconsistent with evidence suggesting that positive and negative affect are orthogonal constructs reflecting different dimensions of temperament (Kessel et al., 2017; Putnam & Stifter, 2005). One reason for why we found that a single dimension represented both positive and negative affect may be that most participants manifested low levels of emotional reactivity (as shown in Figure S3). This lack of variability in emotional reactions may be explained by the fact that we relied on only two high intensity tasks (i.e., clown and robot), thereby limiting the potential for individual differences to emerge across various contexts and thus the power to differentiate positive from negative affect.
The present study extends previous research in several ways. Prior longitudinal studies were either limited in sample size, conducted over short periods of time, or did not compare specific dimensions of social withdrawal. Our study addressed this gap by using data from a large population-based longitudinal study and by covering a 10-year period, thereby offering a broader window into the etiological pathways underlying dimensions of social withdrawal. It is the first study to test the predictive value of temperamental and biological dispositions on specific dimensions of social withdrawal from early childhood to grade school. Further, we collected measures through different approaches (biological samples, observation, reports) and from different informants (research assistants, mothers, teachers, peers), which increased the validity and reliability of assessments and decreased measurement error and bias from shared method variance.
Still, these findings must be interpreted while considering some limitations. First, vagal withdrawal was assessed prior (i.e., at 5 months) to the behavioral inhibition and cortisol response (i.e., at 19 months). Despite studies suggesting that individual differences in vagal tone are relatively stable during infancy (Bornstein & Suess, 2000; Fracasso et al., 1994; Friedman, 2007; Patriquin et al., 2014; Porges et al., 1994; Wagner et al., 2021), we were not able to examine the stability of vagal tone from 5 to 19 months. Therefore, the significant interaction between vagal modulation and affect needs to be interpreted while considering this limitation. Second, although children included in the sample were representative of the population in the greater Montreal area in the mid 1990s, most of them were of European descent (Boivin et al., 2019). Thus, results can hardly be generalized to children from diverse ethnic backgrounds. Third, as we assessed social inhibition at 19 months by exposing participants to an unfamiliar adult dressed as a clown, we cannot conclude that participants would have reacted similarly had they met unfamiliar peers. Similarly, for participants who had never seen a clown before, elements of the clown costume (e.g., colorful clothes and wig) may represent non-social novelty. Fourth, we cannot exclude the possibility that the significant association between social inhibition and mother-rated social wariness and preference for solitude at 4 and 5 years old may be partly attributed to characteristics of the methodology. Mothers were also present during the exposition to unfamiliarity at 19 months. Thus, mothers’ perception of their child’s behavior at 4 and 5 years old could have been affected by their previous experience at the laboratory.
The present study contributes to extend current knowledge on the early temperamental and biological dispositions linked to social wariness and preference for solitude, two dimensions of social withdrawal. Findings show that some early dispositions may help differentiate young children at risk of developing social wariness versus preference for solitude and thereby contribute to early intervention strategies. Overall, results underscore the necessity of considering early behavioral, emotional and biological dispositions as well as indicators of physiological regulation to provide a broader and more complete window into the developmental pathways to specific dimensions of social withdrawal.
Achenbach, T. M. (1991). Manual for the child behavior checklist/4-18 and 1991 profile. University of Vermont, Department of Psychiatry.
Arseneault, L., Bowes, L., & Shakoor, S. (2010). Bullying victimization in youths and mental health problems: ‘Much ado about nothing’? Psychological Medicine, 40, 717-729. https://doi.org/10.1017/S0033291709991383
Asendorpf, J. B. (1987). Social competence. In F. E. Weinert & W. Schneider (Eds.), LOGIC - Report No. 2: Assessment procedures used in waves ones to three. Max- Planck Institut fur psychologische Forschung.
Asendorpf, J. B. (1990). Beyond social withdrawal: Shyness, unsociability, and peer avoidance. Human Development, 33(4–5), 250–259. https://doi.org/10.1159/000276522
Beauchaine, T. P. (2001). Vagal tone, development, and Gray’s motivational theory: Toward an integrated model of autonomic nervous system functioning in psychopathology. Development and Psychopathology, 13(2), 183–214. https://doi.org/10.1017/S0954579401002012
Blair, C., Peters, R., & Granger, D. (2004). Physiological and neuropsychological correlates of approach/withdrawal tendencies in preschool: Further examination of the behavioral inhibition system/behavioral activation system scales for young children. Developmental Psychobiology, 45(3), 113–124. https://doi.org/10.1002/dev.20022
Boivin, M., Brendgen, M., Dionne, G., Ouellet-Morin, I., Dubois, L., Pérusse, D., Robaey, P., Tremblay, R. E., & Vitaro, F. (2019). The Quebec Newborn Twin Study at 21. Twin Research and Human Genetics, 22(6), 475–481. https://doi.org/10.1017/thg.2019.74
Boivin, M., Brendgen, M., Vitaro, F., Dionne, G., Girard, A., Pérusse, D., & Tremblay, R. E. (2013). Strong genetic contribution to peer relationship difficulties at school entry: Findings from a longitudinal twin study. Child Development, 84(3), 1098– 1114. https://doi.org/10.1111/cdev.12019
Bornstein, M. H., & Suess, P. E. (2000). Child and mother cardiac vagal tone: continuity, stability, and concordance across the first 5 years. Developmental Psychology, 36(1), 54–65. https://doi.org/10.1037/0012-1622.214.171.124
Bukowski, W. M., Castellanos, M., Vitaro, F., & Brendgen, M. (2015). Socialization and experiences with peers. In J. E. Grusec & P. D. Hastings (Eds.), Handbook of socialization: Theory and research (pp. 228–250). The Guilford Press.
Buss, K. A. (2011). Which fearful toddlers should we worry about? Context, fear regulation, and anxiety risk. Developmental Psychology, 47(3), 804–819. https://doi.org/10.1037/a0023227
Buss, K. A., Davis, E. L., Ram, N., & Coccia, M. (2018). Dysregulated fear, social inhibition, and respiratory sinus arrhythmia: A replication and extension. Child Development, 89(3), e214–e228. https://doi.org/10.1111/cdev.12774
Calkins, S. D., Graziano, P. A., & Keane, S. P. (2007). Cardiac vagal regulation differentiates among children at risk for behavior problems. Biological Psychology, 74(2), 144–153. https://doi.org/10.1016/j.biopsycho.2006.09.005
Coplan, R. J., Ooi, L. L., & Nocita, G. (2015). When one is company and two is a crowd: Why some children prefer solitude. Child Development Perspectives, 9(3), 133– 137. https://doi.org/10.1111/cdep.12131
Coplan, R. J., Ooi, L. L., Xiao, B., & Rose-Krasnor, L. (2018). Assessment and implications of social withdrawal in early childhood: A first look at social avoidance. Social Development, 27(1), 125–139. https://doi.org/10.1111/sode.12258
Coplan, R. J., Prakash, K., O’Neil, K., & Armer, M. (2004). Do you “want” to play? Distinguishing between conflicted shyness and social disinterest in early childhood. Developmental Psychology, 40(2), 244–258. https://doi.org/10.1037/0012-16126.96.36.199
Coplan, R. J., Rose-Krasnor, L., Weeks, M., Kingsbury, A., Kingsbury, M., & Bullock, A. (2013). Alone is a crowd: Social motivations, social withdrawal, and socioemotional functioning in later childhood. Developmental Psychology, 49(5), 861–875. https://doi.org/10.1037/a0028861
Coplan, R. J., & Weeks, M. (2010). Unsociability in middle childhood: Conceptualization, assessment, and associations with socioemotional functioning. Merrill-Palmer Quarterly, 56(2), 105–130. https://doi.org/10.2307/23098037
Davis, E. L., & Buss, K. A. (2012). Moderators of the relation between shyness and behavior with peers: Cortisol dysregulation and maternal emotion socialization. Social Development, 21(4), 801–820. https://doi.org/10.1111/j.1467- 9507.2011.00654.x
Degnan, K. A., Henderson, H. A., Fox, N. A., & Rubin, K. H. (2008). Predicting social wariness in middle childhood: The moderating roles of childcare history, maternal personality and maternal behavior. Social Development, 17(3), 471–487. https://doi.org/10.1111/j.1467-9507.2007.00437.x
Ding, X., Coplan, R. J., Deng, X., Ooi, L. L., Li, D., & Sang, B. (2019). Sad, scared, or rejected? A short-term longitudinal study of the predictors of social avoidance in Chinese children. Journal of Abnormal Child Psychology, 47(7), 1265–1276. https://doi.org/10.1007/s10802-018-0476-9
Dyson, M. W., Klein, D. N., Olino, T. M., Dougherty, L. R., & Durbin, C. E. (2011). Social and non-social behavioral inhibition in preschool-age children: Differential associations with parent-reports of temperament and anxiety. Child Psychiatry and Human Development, 42(4), 390–405. https://doi.org/10.1007/s10578-011- 0225-6
Eggum-Wilkens, N. D., An, D., Zhang, L., & Costa, M. (2020). Co-occurrence of and cross-informant agreement on shyness, unsociability, and social avoidance during early adolescence. Social Development, 29(1), 73–88. https://doi.org/10.1111/sode.12398
Fox, N. A., Henderson, H. A., Marshall, P. J., Nichols, K. E., & Ghera, M. M. (2005). Behavioral inhibition: Linking biology and behavior within a developmental framework. Annual Review of Psychology, 56(1), 235–262. https://doi.org/10.1146/annurev.psych.55.090902.141532
Fracasso, M. P., Porges, S. W., Lamb, M. E., & Rosenberg, A. A. (1994). Cardiac activity in infancy: Reliability and stability of individual differences. Infant Behavior and Development, 17(3), 277–284. https://doi.org/10.1016/0163-6383(94)90006-X
Friedman, B. H. (2007). An autonomic flexibility–neurovisceral integration model of anxiety and cardiac vagal tone. Biological Psychology, 74(2), 185–199. https://doi.org/10.1016/j.biopsycho.2005.08.009
Garcia-Coll, C., Kagan, J., & Reznick, J. S. (1984). Behavioral inhibition in young children. Child Development, 55(3), 1005–1019.
Graziano, P. A., & Derefinko, K. (2013). Cardiac vagal control and children’s adaptive functioning: A meta-analysis. Biological Psychology, 94(1), 22–37. https://doi.org/10.1016/j.biopsycho.2013.04.011
Graziano, P. A., Keane, S. P., & Calkins, S. D. (2007). Cardiac vagal regulation and early peer status. Child Development, 78(1), 264–278. https://doi.org/10.1111/j.1467- 8624.2007.00996.x
Gunnar, M. R., & Donzella, B. (2002). Social regulation of the cortisol levels in early human development. Psychoneuroendocrinology, 27(1–2), 199–220. https://doi.org/10.1016/S0306-4530(01)00045-2
Gunnar, M. R., & Quevedo, K. (2007). The neurobiology of stress and development. Annual Review of Psychology, 58, 145–173. https://doi.org/10.1146/annurev.psych.58.110405.085605
Hastings, P. D., Sullivan, C., McShane, K. E., Coplan, R. J., Utendale, W. T., & Vyncke, J. D. (2008). Parental socialization, vagal regulation, and preschoolers’ anxious difficulties: Direct mothers and moderated fathers. Child Development, 79(1), 45– 64. https://doi.org/10.1111/j.1467-8624.2007.01110.x
Kagan, J., Reznick, J. S., Clarke, C., Snidman, N., & Garcia-Coll, C. (1984). Behavioral inhibition to the unfamiliar. Child Development, 55(6), 2212–2225.
Kagan, J., Reznick, J. S., & Snidman, N. (1987). The physiology and psychology of behavioral inhibition in children. Child Development, 58(6), 1459–1473.
Kagan, J., & Snidman, N. (1991). Infant predictors of inhibited and uninhibited profiles. Psychological Science, 2(1), 40–44. https://doi.org/10.1111/j.1467- 9280.1991.tb00094.x
Kessel, E. M., Kujawa, A., Goldstein, B., Hajcak, G., Bufferd, S. J., Dyson, M., & Klein, D. N. (2017). Behavioral observations of positive and negative valence systems in early childhood predict physiological measures of emotional processing three years later. Journal of Affective Disorders, 216(2017), 70–77. https://doi.org/10.1016/j.jad.2016.10.044
Kochanska, G., & Radke-Yarrow, M. (1992). Inhibition in toddlerhood and the dynamics of the child’s interaction with an unfamiliar peer at age five. Child Development, 63(2), 325–335. https://doi.org/10.2307/1131482
Kossowsky, J., Wilhelm, F. H., Roth, W. T., & Schneider, S. (2012). Separation anxiety disorder in children: Disorder-specific responses to experimental separation from the mother. Journal of Child Psychology and Psychiatry, 53(2), 178–187. https://doi.org/10.1111/j.1469-7610.2011.02465.x
Ladd, G. W., Ettekal, I., & Kochenderfer-Ladd, B. (2019). Longitudinal changes in victimized youth’s social anxiety and solitary behavior. Journal of Abnormal Child Psychology, 47(7), 1211–1223. https://doi.org/10.1007/s10802-018-0467-x
Ladd, G. W., & Profilet, S. M. (1996). The Child Behavior Scale: A teacher-report measure of young children’s aggressive, withdrawn, and prosocial behaviors. Developmental Psychology, 32(6), 1008–1024. https://doi.org/10.1037/0012- 16188.8.131.528
Lobue, V., & Pérez-Edgar, K. (2014). Sensitivity to social and non-social threats in temperamentally shy children at-risk for anxiety. Developmental Science, 17(2), 239–247. https://doi.org/10.1111/desc.12110
Lupien, S. J., McEwen, B. S., Gunnar, M. R., & Heim, C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature Reviews Neuroscience, 10(6), 434–445. https://doi.org/10.1038/nrn2639
MacMillan, H. L., Georgiades, K., Duku, E. K., Shea, A., Steiner, M., Niec, A., Tanaka, M., Gensey, S., Spree, S., Vella, E., Walsh, C. A., De Bellis, M. D., Van der Meulen, J., Boyle, M. H., & Schmidt, L. A. (2009). Cortisol response to stress in female youths exposed to childhood maltreatment: Results of the Youth Mood Project. Biological Psychiatry, 66(1), 62–68. https://doi.org/10.1016/j.biopsych.2008.12.014
Majdandžić, M., & van den Boom, D. C. (2007). Multimethod longitudinal assessment of temperament in early childhood. Journal of Personality, 75(1), 121–168. https://doi.org/10.1111/j.1467-6494.2006.00435.x
Marshall, P. J., & Stevenson-Hinde, J. (1998). Behavioral inhibition, heart period, and respiratory sinus arrhythmia in young children. Developmental Psychobiology, 33(3), 283–292. https://doi.org/10.1002/(SICI)1098- 2302(199811)33:3<283::AID-DEV8>3.0.CO;2-N
Mezzacappa, E., Steingard, R., Kindlon, D., Philip Saul, J., & Earls, F. (1998). Tricyclic antidepressants and cardiac autonomic control in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 37(1), 52– 59. https://doi.org/10.1097/00004583-199801000-00017
Mezzacappa, E., Tremblay, R. E., Saul, J. P., Seguin, J., Pihl, R. O., Kindlon, D., Arseneault, L., & Earls, F. (1997). Anxiety, antisocial behavior, and heart rate regulation in adolescent males. Journal of Child Psychology and Psychiatry and Allied Disciplines, 38(4), 457–469. https://doi.org/10.1111/j.1469- 7610.1997.tb01531.x
Morneau‐Vaillancourt, G., Andlauer, T. F. M., Ouellet‐Morin, I., Paquin, S., Brendgen, M. R., Vitaro, F., Gouin, J., Séguin, J. R., Gagnon, É., Cheesman, R., Forget‐Dubois, N., Rouleau, G. A., Turecki, G., Tremblay, R. E., Côté, S. M., Dionne, G., & Boivin, M. (2021). Polygenic scores differentially predict developmental trajectories of subtypes of social withdrawal in childhood. Journal of Child Psychology and Psychiatry, 62(11), 1320–1329. https://doi.org/10.1111/jcpp.13459
Morneau-Vaillancourt, G., Matte-Gagné, C., Cheesman, R., Brendgen, M., Vitaro, F., Tremblay, R., Dionne, G., & Boivin, M. (2021). Social wariness, preference for solitude, and peer difficulties in middle childhood: A longitudinal family- informed study. Developmental Psychology, 57(3), 410–420. https://doi.org/10.1037/dev0000961
Muthen, L. K., & Muthen, B. O. (2017). Mplus user’s guide. Muthen & Muthen.
Nigg, J. T. (2006). Temperament and developmental psychopathology. Journal of Child Psychology and Psychiatry and Allied Disciplines, 47(3–4), 395–422. https://doi.org/10.1111/j.1469-7610.2006.01612.x
Ottenbreit, N. D., & Dobson, K. S. (2004). Avoidance and depression: the construction of the Cognitive–Behavioral Avoidance Scale. Behaviour Research and Therapy, 42(3), 293–313. https://doi.org/10.1016/S0005-7967(03)00140-2
Ouellet-Morin, I., Boivin, M., Dionne, G., Lupien, S. J., Arsenault, L., Barr, R. G., Perusse, D., & Tremblay, R. E. (2008). Variations in heritability of cortisol reactivity to stress as a function of early familial adversity among 19-month-old twins. Archives of General Psychiatry, 65(2), 211. https://doi.org/10.1001/archgenpsychiatry.2007.27
Patriquin, M. A., Lorenzi, J., Scarpa, A., & Bell, M. A. (2014). Developmental trajectories of respiratory sinus arrhythmia: Associations with social responsiveness. Developmental Psychobiology, 56(3), 317–326. https://doi.org/10.1002/dev.21100
Perry, N. B., Calkins, S. D., & Bell, M. A. (2016). Indirect effects of maternal sensitivity on infant emotion regulation behaviors: The role of vagal withdrawal. Infancy, 21(2), 128–153. https://doi.org/10.1111/infa.12101
Poliakova, N., Dionne, G., Dubreuil, E., Ditto, B., Pihl, R. O., Pérusse, D., Tremblay, R. E., & Boivin, M. (2014). A methodological comparison of the Porges algorithm, fast Fourier transform, and autoregressive spectral analysis for the estimation of heart rate variability in 5-month-old infants. Psychophysiology, 51(6), 579–583. https://doi.org/10.1111/psyp.12194
Poole, K. L., Van Lieshout, R. J., McHolm, A. E., Cunningham, C. E., & Schmidt, L. A. (2018). Trajectories of social anxiety in children: Influence of child cortisol reactivity and parental social anxiety. Journal of Abnormal Child Psychology, 46(6), 1309–1319. https://doi.org/10.1007/s10802-017-0385-3
Porges, S. W. (2007). The polyvagal perspective. Biological Psychology, 74(2), 116–143. https://doi.org/10.1016/j.biopsycho.2006.06.009
Porges, S. W., Doussard‐Roosevelt, J. A., Lourdes Portales, A., & Suess, P. E. (1994). Cardiac vagal tone: Stability and relation to difficultness in infants and 3‐year‐ olds. Developmental Psychobiology, 27(5), 289–300. https://doi.org/10.1002/dev.420270504
Porges, S. W., Doussard-Roosevelt, J. A., Portales, A. L., & Greenspan, S. I. (1996). Infant regulation of the vagal “brake” predicts child behavior problems: A psychobiological model of social behavior. Developmental Psychobiology, 29, 697–712. https://doi.org/10.1002/(SICI)1098-2302(199612)29:8<697::AID- DEV5>3.0.CO;2-O
Putnam, S. P., & Stifter, C. A. (2005). Behavioral approach-inhibition in toddlers: Prediction from infancy, positive and negative affective components, and relations with behavior problems. Child Development, 76(1), 212–226. https://doi.org/10.1111/j.1467-8624.2005.00840.x
Rubin, K. H., Coplan, R. J., & Bowker, J. C. (2009). Social withdrawal in childhood. Annual Review of Psychology, 60(1), 141–171. https://doi.org/10.1146/annurev.psych.60.110707.163642
Rubin, K. H., Hastings, P. D., Stewart, S. L., Henderson, H. A., & Chen, X. (1997). The consistency and concomitants of inhibition: Some of the children, all of the time. Child Development, 68(3), 467-483. https://doi.org/10.2307/1131672
Schmidt, L. A., Fox, N. A., Rubin, K. H., Sternberg, E. M., Gold, P. W., Smith, C. C., & Schulkin, J. (1997). Behavioral and neuroendocrine responses in shy children. Developmental Psychobiology, 30(2), 127–140. https://doi.org/10.1002/(SICI)1098-2302(199703)30:2<127::AID- DEV4>3.0.CO;2-S
Schmidt, L. A., Fox, N. A., Schulkin, J., & Gold, P. W. (1999). Behavioral and psychophysiological correlates of self-presentation in temperamentally shy children. Developmental Psychobiology, 35(2), 119–135. https://doi.org/10.1002/(SICI)1098-2302(199909)35:2<119::AID- DEV5>3.0.CO;2-G
Shanahan, L., Calkins, S. D., Keane, S. P., Kelleher, R., & Suffness, R. (2014). Trajectories of internalizing symptoms across childhood: The roles of biological self- regulation and maternal psychopathology. Development and Psychopathology, 26(2014), 1353–1368. https://doi.org/10.1017/S0954579414001072
Smith, K. A., Hastings, P. D., Henderson, H. A., & Rubin, K. H. (2019). Multidimensional emotion regulation moderates the relation between behavioral inhibition at age 2 and social reticence with unfamiliar peers at age 4. Journal of Abnormal Child Psychology, 47(7), 1239–1251. https://doi.org/10.1007/s10802-018-00509-y
Spangler, T., & Gazelle, H. (2009). Anxious solitude, unsociability, and peer exclusion in middle childhood: A multitrait-multimethod matrix. Social Development, 18(4), 833–856. https://doi.org/10.1111/j.1467-9507.2008.00517.x
Ulrich-Lai, Y. M., & Herman, J. P. (2009). Neural regulation of endocrine and autonomic stress responses. Nature Reviews Neuroscience, 10(6), 397–409. https://doi.org/10.1038/nrn2647
van den Bos, E., Tops, M., & Westenberg, P. M. (2017). Social anxiety and the cortisol response to social evaluation in children and adolescents. Psychoneuroendocrinology, 78(2017), 159–167. https://doi.org/10.1016/j.psyneuen.2017.02.003
van West, D., Claes, S., Sulon, J., & Deboutte, D. (2008). Hypothalamic-pituitary-adrenal reactivity in prepubertal children with social phobia. Journal of Affective Disorders, 111(2008), 281–290. https://doi.org/10.1016/j.jad.2008.03.006
Viana, A. G., Trent, E. S., Raines, E. M., Woodward, E. C., Storch, E. A., & Zvolensky, M. J. (2021). Childhood anxiety sensitivity, fear downregulation, and anxious behaviors: Vagal suppression as a moderator of risk. Emotion, 21(2), 430–441. https://doi.org/10.1037/emo0000713
Wagner, N. J., Holochwost, S. J., Lynch, S. F., Mills-Koonce, R., & Propper, C. (2021). Characterizing change in vagal tone during the first three years of life: A systematic review and empirical examination across two longitudinal samples. Neuroscience and Biobehavioral Reviews, 129(2021), 282–295. https://doi.org/10.1016/j.neubiorev.2021.07.025
Wilder, J. (1957). The Law of Initial Value in neurology and psychiatry. The Journal of Nervous and Mental Disease, 125(1), 73–86. https://doi.org/10.1097/00005053- 195701000-00009
All authors declare no conflict of interest.
Data cannot be shared publicly because of privacy and ethical restrictions but may be obtained by filling a request to access from the Research Unit on Children’s Psychosocial Maladjustment Website: https://www.gripinfo.ca/grip/public/www/etudes/en/dadprocedures.asp
This research was supported by many grants received over the years, including grants were from the Fonds de Recherche du Québec - Société et Culture (FQRSC), the Fonds de Recherche du Québec en Santé (FRQS), the Social Science and Humanities Research Council of Canada (SSHRC), the National Health Research Development Program, the Canadian Institutes for Health Research (CIHR), Ste. Justine Hospital’s Research Center, Université Laval, and Université de Montréal. Geneviève Morneau-Vaillancourt was supported by a Doctoral Research Scholarship from the FRQSC, and Lysandre Provost by a Master’s Scholarship from the SSHRC. Isabelle Ouellet-Morin, Célia Matte-Gagné, Amélie Petitclerc and Michel Boivin are supported by the Canada Research Chair Program. We are grateful to the children and parents of the QNTS and the participating teachers and schools. We thank Jocelyn Malo for coordinating the data collection and Hélène Paradis for data management.
Parental informed consent and child assent was obtained from ethics review boards at Université Laval, Québec City, and at Université de Montréal and Ste-Justine Hospital, Montréal for all data collections.
Table 1: Descriptive Statistics
Vagal modulation 5 months
Non-social inhibition 19 months
Social inhibition 19 months
Affect 19 months
Cortisol response 19 months
Social wariness age 4
Social wariness age 5
Social wariness age 6
Social wariness age 7
Social wariness age 10
Preference for solitude age 4
Preference for solitude age 5
Preference for solitude age 6
Preference for solitude age 7
Preference for solitude age 10
Note. Social and non-social inhibition and affect at 19 months were extracted from a principal components analysis. Positive affect scores represent positive affect, vice versa for negative scores. Social wariness and preference for solitude at age 4 and 5 were evaluated by mothers; at age 6, 7 and 10, by both teachers and peers (average of both z-standardized scores).
Table 2: Correlations between Variables
1. VM 5 mo
2. SI 19 mo
4. AFF 19 mo
5. CR 19 mo
6. SW 4 yo
7. SW 5 yo
8. SW 6 yo
9. SW 7 yo
10. SW 10 yo
11. PFS 4 yo
12. PFS 5 yo
13. PFS 6 yo
14. PFS 7 yo
15. PFS 10 yo
Note. *** p < .001; ** p < .01; * p < .05. Abbreviations: VM = vagal modulation; SI = social inhibition; NSI = non-social inhibition; AFF = affect; CR = cortisol response; SW = social wariness; PFS = preference for solitude; mo = months; yo = years old. Positive affect scores represent positive affect and vice versa for negative scores.
Table 3: Estimates from the Sequential Model Predicting Social Wariness and Preference for Solitude (Figure 1)
Estimate (95% confidence interval)
SW 4-5 years old
SW 6-10 years old
PFS 4-5 years old
PFS 6-10 years old
Vagal modulation 5 months
.11 (-.01, .23)
-.04 (-.21, .12)
-.08 (-.26, .10)
Non-social inhibition 19 months
.06 (-.07, .19)
-.09 (-.24, .07)
-.06 (-.18, .06)
-.08 (-.28, .11)
Social inhibition 19 months
.16 (.04, .29)
-.07 (-.23, .09)
.01 (-.11, .12)
.06 (-.14, .26)
Affect 19 months
-.08 (-.20, .04)
-.03 (-.18, .12)
-.10 (-.20, .01)
.06 (-.12, .23)
Cortisol response 19 months
.02 (-.12, .16)
.10 (-.06, .26)
-.04 (-.16, .08)
-.05 (-.28, .17)
Vagal modulation*non-social inhibition
.06 (-.07, .18)
.03 (-.13, .19)
-.02 (-.15, .12)
-.02 (-.26, .22)
Vagal modulation*social inhibition
-.12 (-.25, .01)
-.15 (-.33, .03)
-.03 (-.18, .11)
-.19 (-.44, .06)
.05 (-.07, .17)
.01 (-.18, .20)
-.16 (-.29, -.03)
-.02 (-.26, .22)
Vagal modulation*cortisol response
-.06 (-.21, .09)
-.16 (-.32, .01)
.00 (-.17, .18)
-.11 (-.32, .11)
Note. Estimates with 95% confidence intervals that do not include 0 are in bold. Abbreviations: SW = social wariness; PFS = preference for solitude.
Figure 1: Sequential Model Predicting Social Wariness and Preference for Solitude
Note. *p < .05. To simplify the figure, we do not show 1) regression paths from covariates to observed variables, from sex to social wariness at 10 years old and to preference for solitude at 6, 7, and 10 years old (as we found sex differences), and 2) correlations between observed social wariness and preference for solitude variables within assessment waves, between early childhood predictors and their interaction terms, and between covariates. Covariates were both parents' education, first language, ethnicity, age at birth of the twins, overprotective behaviors when children were 30 months old, family income when children were 5 months old, and family structure at birth (single- or two-parent). Non-significant path estimates are presented in Table 3.
Figure 2: Vagal Modulation Interacts with Affect in Unfamiliar Situations to Predict Preschool Preference for Solitude
Note. Affect included both negative and positive affect. Lower affect scores represent higher intensity negative affect (or low positive affect), whereas higher affect scores represent higher intensity positive affect (or low negative affect).