1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments

Early individual differences in prosocial behaviors are pivotal for children's peer relationships. To investigate the interplay among verbal ability, emotion understanding, and mother–child mutuality as predictors of prosocial behaviors, we observed 102 children at the ages of two, three, and four. All time points included verbal ability and emotion understanding tests and both video-based and maternal ratings of prosocial behavior. The first two time points also included video-based ratings of mother–child mutuality. The third time point included teacher ratings of prosocial behavior and an experimental task. Regression analysis demonstrated robust associations between emotion understanding at the age of three and prosocial behavior at the age of four. Path analysis showed that emotion understanding at the age of three mediated associations between verbal ability/mother–child mutuality at the age of two and prosocial behavior at the age of four.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments

Prosocial behaviors are actions, such as sharing and helping, that benefit others (Eisenberg, 1982). Numerous studies have shown that prosocial behaviors also reap their own rewards. Prosocial behaviors are pivotal for the quality of friendships (Wojslawowicz Bowker, Rubin, Burgess, Booth-Laforce, & Rose-Krasnor, 2006) and peer relations (Lansford, Putallaz, Grimes, Schiro-Osman, Kupersmidt, & Coie, 2006); equally, the absence of prosocial behaviors predicts peer rejection (Vitaro, Gagnon, & Tremblay, 1990) and deviant peer group affiliation (Lacourse, Nagin, Vitaro, Cote, Arseneault, & Tremblay, 2006). Together, these findings highlight the importance of understanding the origins of individual differences in prosocial behaviors.

For most children, displays of prosocial behaviors are occasional in the first 2 years of life (Hay & Cook, 2007) but more frequent thereafter (Eisenberg & Fabes, 1998). Between the second and fourth years of life, children's prosocial behaviors also show important qualitative changes (Hay & Cook, 2007). Specifically, newly acquired skills and sensitivities enable toddlers to tailor prosocial behaviors (e.g., Levitt, Weber, Clark, & McDonnell, 1985). The likelihood of prosocial behaviors also depends upon whether the recipient is a friend or acquaintance (Farver & Branstetter, 1994), or particularly prosocial themselves (Dunn, Cutting, & Fisher, 2002). Despite these examples of context sensitivity, stable individual differences in prosocial behaviors emerge by the preschool years (Eisenberg, Guthrie, Murphy, Shepard, Cumberland, & Carlo, 1999).

Early theoretical accounts of prosocial behaviors (e.g., Eisenberg & Mussen, 1989; Hoffman, 1982) highlighted the importance of perspective taking, such as understanding visual perspectives, thoughts, and feelings. Children make impressive gains in perspective taking in the toddler and preschool years (Pons, Harris, & de Rosnay, 2004). Specifically, from around 18 months, children can identify simple emotions such as happiness (Borke, 1971), anger, and sadness (Michalson & Lewis, 1985), and use others' emotional expressions to interpret desires (Repacholi & Gopnik, 1997). By three years of age, children start to understand both how situations relate to emotions (Harris, 1989) and the subjectivity of emotions (Harris, Johnson, Hutton, Andrews, & Cooke, 1989). Although their nature and magnitude in later years remain poorly understood, individual differences in perspective-taking skills in the preschool years are striking (de Rosnay & Hughes, 2006).

Variation in perspective-taking skills is associated with individual differences in preschoolers' prosocial behaviors (Denham, 1986; Iannotti, 1985). A simple explanation is that children's linguistic skills are associated with both emotion understanding (e.g., Hughes, Lecce, & Wilson, 2007) and positive social behaviors (e.g., Cassidy, Werner, Rourke, Zubernis, & Balaraman, 2003). In other words, the relationship between emotion understanding and prosocial behavior may simply reflect common associations with verbal ability. Ruling out this ‘third factor’ account is thus an important step toward elucidating the processes that underpin individual differences in prosocial behaviors. Moreover, evidence from two separate studies of school-age children suggests that predictive relationships between emotion understanding and later social skills are independent of effects of verbal ability (Denham et al., 2003; Mostow, Izard, Fine, & Trentacosta, 2002). However, findings from studies of younger children are less consistent. In one study, verbal ability and emotion understanding showed overlapping associations with children's co-operative play with friends (Dunn & Cutting, 1999); in another, covarying effects of verbal ability fully explained relations between emotion understanding and prosocial behaviors (Cassidy et al., 2003). However, in a third study, the association between emotion understanding and prosocial behaviors was statistically independent from effects of verbal ability (Ensor & Hughes, 2005). These mixed findings highlight the need for further research to disentangle relations among young children's verbal ability, emotion understanding, and prosocial behaviors.

By definition, prosocial behaviors are directed toward others: Elucidating the origins of individual differences in such acts therefore requires attention to children's relationships. In particular, children's prosocial behaviors are nurtured through the parent–child relationship (Grusec, 1991). For example, there is extensive evidence that authoritative/democratic disciplinary style (e.g., Dekovic & Janssens, 1992) and victim-centered/other-orientated reasoning (Krevans & Gibbs, 1996) each promote children's prosocial behaviors, perhaps because such strategies catch children's attention without provoking anxiety/distress (Hoffman, 1975). Conversely, a large-scale multilevel study reported reduced prosocial behaviors in children whose mothers engaged in punitive disciplinary strategies (Romano, Tremblay, Boulerice, & Swisher, 2005).

However, disciplinary strategies are not equally effective with all children. In particular, children who are temperamentally fearless (e.g., Kochanska, 1997a) or who show callous unemotional traits (Frick & White, 2008) typically show little discomfort about wrongdoing, even when parents engage in victim-centered reasoning (but see also Valiente, Fabes, Eisenberg, & Spinrad, 2004). For such children, parental responsiveness appears critical (e.g., Kochanska). Parents who are particularly responsive to their children's needs are likely to engender children's wholehearted compliance to parental requests (Kochanska & Aksan, 2004), leading to the development of prosocial behaviors (Maccoby, 1992).

Although research on parent–child dyadic synchrony focuses on infancy, there are striking and readily observable individual differences in the toddler years (Harrist & Waugh, 2002). Specifically, studies highlight that both parents and toddlers modulate their behaviors (Rutter & Durkin, 1987) to work toward common goals (Rocissano, Slade, & Lynch, 1987). Longitudinal studies have demonstrated that children who engage in such mutually responsive interactions with their parents appear more prosocial than their peers (e.g., Kochanska, 1997b). This may be because consistent parental practices are cumulative in their effects (Ruffman, Slade, Devitt, & Crowe, 2006) such that, over extended periods of time, parenting style becomes a strong predictor of behavior (Rothbaum & Weisz, 1994).

How should the effect of the parent–child relationship on children's prosocial behaviors be explained? That is, what processes underpin the association between mutually responsive parent–child interactions and later prosocial behaviors? The importance of these questions is underscored by evidence that individual differences in the parent–child relationship and in children's emotion understanding are closely entwined. For example, good emotion understanding is associated with secure attachment (e.g., Laible & Thompson, 1998) and emotionally contingent (Denham, Zoller, & Couchod, 1994), high-quality affective interactions (Laible & Song, 2006) between parents and children. Thus, the specificity of emotion understanding as a correlate of prosocial behaviors remains unknown: Addressing this issue is an important first step toward elucidating the role of emotion understanding in prosocial behavior. A preliminary aim in this study was to test whether emotion understanding remains associated with children's prosocial behaviors when effects of parent–child relationships and verbal ability are taken into account (H1).

The next step in elucidating the role of emotion understanding in explaining individual differences in prosocial behavior is to test whether good emotion understanding explains why verbally able children or children who enjoy mutually responsive interactions with their parents are more likely than their peers to be prosocial. In other words, we hypothesized that young children's emotion understanding mediated the associations between prosocial behaviors and both verbal ability (H2) and mutually responsive parent–child interactions (H3). With regard to H2, evidence that emotion understanding mediates the relationship between verbal ability and prosocial behaviors (H2) is available for school-age children (Mostow et al., 2002); there is also consistent (albeit cross-sectional) evidence from toddlers (Ensor & Hughes, 2005). For preschoolers, it is worth noting that the marked developmental changes in the quality and quantity of prosocial behaviors, described earlier, appear in tandem with rapid developments in preschoolers' emotion vocabulary (Bretherton & Beeghly, 1982; Hughes et al., 2007). In sum, there is suggestive, but not conclusive, evidence that emotion understanding is the proximal process through which verbal ability predicts prosocial behaviors.

Turning to H3, there is only indirect evidence that emotion understanding mediates the relationship between parent–child mutuality and prosocial behaviors (H3). Specifically, emotional regulation (a construct that overlaps in part with emotion understanding—Thompson & Goodvin, 2007) has been reported to mediate associations between (1) parental emotional expressivity and children's social competence in the early school years (e.g., Eisenberg et al., 2001), and (2) attachment and peer relationships in middle childhood (Contreras, Kerns, Weimer, Gentzler, & Tomich, 2000). Two further findings also deserve mention. Firstly, young children's effortful control (a construct closely allied to emotion regulation—Eisenberg, Zhou, Spinrad, Valiente, Fabes, & Liew, 2005) mediates links between supportive parenting and social competence (Spinrad et al., 2007). Second, in middle childhood, children's use of display rules (a construct that reflects emotion regulation—Campos, Campos, & Barrett, 1989) mediates links between parental style and children's social competence (McDowell & Parke, 2005). The current study builds on these findings to assess directly whether emotion understanding mediates relations between prosocial behaviors and both verbal ability and parent–child mutuality.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments


This study included longitudinal data from a socially diverse sample of children seen at the ages of two, three, and four, and capitalized on the availability of several indexes of prosocial behaviors. At the ages of two and three, measures included (1) maternal questionnaire ratings, and (2) video-based coding of the children sharing toys or food with unfamiliar peers. At the age of four, measures included (1) maternal and teacher questionnaire ratings, (2) video-based coding of the children sharing toys and/or helping a friend, and (3) an experimental assessment of the children's intentions to share task rewards with a friend (Moore, Barresi, & Thompson, 1998). Individual differences in aggregates were examined in relation to the children's performance on well-validated tests of verbal ability and emotion understanding (at the ages of two, three, and four), and video-based coding of parent–child mutuality (at the ages of two and three).

Recruitment and Participants

A socially diverse sample was recruited via face-to-face contact in mother–toddler groups or support groups for mothers in Cambridge, UK. Core inclusion criteria included: toddler age between 24 months and 36 months at the first wave of data collection and English as a home language. Of the 192 eligible families, 140 gave informed consent and took part in the first wave of the study (hereafter ‘time 1’ or ‘t1’); of these, 125 participated in the second and third waves of data collection (hereafter ‘time 2/3’ or ‘t2/3’). Information about family background and structure was collected from the mothers during home visits at time 1 and time 2 and phone calls prior to school visits at time 3. Note that, compared with the 102 families with complete experimental and observational data, the 23 families with missing data had marginally lower levels of maternal education: t= 1.70, p= .09; but did not significantly differ on head of household occupational status: t= 1.45, p= .15; number of children: t= 1.30, p= .20; and lone parents: χ2= 1.30, p= .25. At time 3, teachers nominated a friend for each of the target children: 11 target children were friends with another target child (one child was a nominated friend of two children), and we recruited 91 further children (hereafter ‘friends’). Reflecting the local population, all but four target children and two friends were White. Table 1 shows the target children's and friends' characteristics, family background, and structure; note that, at time 3, 15 target children were seen between 1 month to 4 months prior to their fourth birthday because of the constraints of the school year.

Table 1. Sample Characteristics of Study Participants
  Time point‘Target’‘Friend’
  1. GCSE = General Certificate of Secondary Education; GED = general educational development; SD, standard deviation.

Child characteristicsAge (years)–2.98–4
Boys : Girls 60:4243:47
Head of household occupation (percent)Professional/managerial 4857
Non-manual 2426
Manual/unemployed 2817
Maternal education (percent)Degree 3233
A-levels (high school diploma) 3121
GCSEs (GED) 2440
No qualification 135
Family structure (percent)Lone mothers 1717
Singletons 713
One sibling 4853
Two siblings 3120
Three siblings 1010
Four or five siblings 44


At times 1 and 2 (interval: M= 12.18 months, SD= 1.2 months), two researchers visited each family, typically on a weekday evening; as part of each visit, target children completed an emotion understanding task. Within approximately 6 weeks of the home visit (intervals: t1 M= 1.10 months, SD= .82 months; t2 M= 1.25 months, SD= .99 months), pairs of unfamiliar children matched for sex visited a lab, accompanied by their mothers (who received free transport or travel cost refunds). In a playroom, one child was filmed with his/her mother in 10 minutes free play (with a train track and doll set), approximately 5 minutes tidy up, and 10 minutes structured play (with Duplo, jigsaws and Mrs Potato Head). In an adjacent room, the other child completed a verbal ability assessment. The mother–child dyads then swapped rooms and completed the other activity before being filmed together for 30 minutes in the playroom. Here, the children were introduced and invited to play together with pretend play toys (e.g., umbrellas, puppets, tools) with their mothers (5 minutes); the mothers were then asked to (1) sit on a sofa (5 minutes), and (2) leave the room (5 minutes). The children then had a 5-minute snack break before playing with a rocking horse (5 minutes) and a double trampoline (5 minutes). At time 3 (t2–t3 interval: M= 8.81 months, SD= 1.44 months), researchers visited each child and a friend (N = 97 pairs; 40 boy–boy, 32 girl–girl, 25 boy–girl) at their preschool and filmed the children for 15–20 minutes playing with dressing-up materials and role-play toys (e.g., nurse and doctor outfits/kits; light sabers; fairy wand). The children also completed a vocabulary assessment and an emotion understanding task. Finally, all teachers (N = 82) completed questionnaires. Questionnaires were mailed with a prepaid envelope to the families; 95 (93 percent) mothers completed and returned the questionnaires. At each time point, each family received a DVD of their child, as well as £20 at times 1 and 2, and a £10 gift voucher at time 3; each school also received a £10 gift voucher.

Verbal Ability

At time 1 and time 2, the children completed the vocabulary and comprehension subtests from the British abilities scales (Elliott, Murray, & Pearson, 1983). Standard scoring procedures were applied and possible scores range from 0 to 20 points for vocabulary and 0 to 27 points for comprehension. For simplicity, we summed scores from the two subtests. At time 3, the British picture vocabulary scale (Dunn, 1997) was used to assess the children's vocabulary.

Emotion Understanding

We used Denham's (1986) widely used puppet task, a measure that taps individual differences in young children's recognition and identification of emotional expressions and understanding of ties between situations and emotions. The task materials included a puppet with a blank face (male for boys, female for girls) and four felt faces portraying happy, sad, angry, and scared expressions. In the emotion-labeling subtask, the children were asked to name how the puppet felt when each face was placed on the puppet; the children were then told how the puppet felt and asked to choose the appropriate face. In the emotion perspective-taking subtask, vignettes associated with feelings of happiness, sadness, anger, or fear were enacted (four of each type) with matching affective tone and the children were asked to choose the face that matched how the puppet was feeling in each vignette. Eight unambiguous vignettes reflected what most children typically feel in a given situation and eight ambiguous vignettes presented scenarios in which the puppet's emotion differed from the emotion that the mother had previously stated his/her child would typically express in that situation. Fully correct responses received 2 points, whereas responses correct in valence but not specific emotion (e.g., calling the angry face sad) received 1 point. Possible scores on this task range from 0 to 32 points at time 1 (note here that, despite providing clear measures of perspective taking, the ambiguous vignettes were not administered at time 1 because of their complexity) and 0 to 48 points at time 2 and time 3.

Mother–Child Mutuality

The parent–child interaction system (Deater-Deckard, Pylas, & Petrill, 1997) was used to code each 1-minute segment from videos of mother–child interaction at time 1 and time 2. We focused on four 7-point Likert scale ratings:

  • 1
    mother responsiveness to child's comments, questions, and behaviors (0 = ignored child's comments, questions, and behaviors; 3 = responded to about half of child's comments, questions, or behaviors, although some responses may be delayed; 6 = always responds immediately to child; expands on some comments made by child);
  • 2
    child responsiveness to mother's comments, questions, and behaviors;
  • 3
    dyadic reciprocity: shared positive affect, eye contact, a ‘turn taking’ (conversation like) quality of interaction; and
  • 4
    dyadic co-operation: explicit agreement and discussion about how to proceed with and complete task.

Within each time point, we averaged, across intervals, ratings for each behavior separately. Following the example of other studies (Deater-Deckard & O'Connor, 2000; Deater-Deckard & Petrill, 2004), we then averaged the mother/child responsiveness and dyadic reciprocity/co-operation scores to create mother–child mutuality scores (Cronbach's α at t1 = .78 and t2 = .73).

Prosocial Behaviors

Observations.  We coded videos of the children's play with unfamiliar peers at time 1 and time 2, and with friends at time 3 for two types of prosocial behaviors:

  • 1
    share: child offers object previously in their possession or gives object into other's hands or laps, or adds object to array within which other is situated; and
  • 2
    help: child provides physical or verbal assistance.

Note that sharing and helping could be spontaneous or in response to peer requests (e.g., peer points to or reaches for an object) or distress. We converted the raw frequencies of the two behaviors into hourly rates, for two reasons. Firstly, the length of observations at times 1 and 2 (M= 30.25 and 27.17 minutes, respectively) were longer than those at time 3 (M= 18.74 minutes). Secondly, at time 1 only, three children became distressed when their mothers left the room; as a result, this part of the observation was omitted for these children.

Experimental Assessment.  At time 3, the children received reward trials in which they could choose between an immediate reward (one sticker now), a delayed reward for self (two stickers later), and a delayed shared reward (two stickers to share with a friend later) (Moore et al., 1998). Each child was offered the choice of reward after he/she had completed a cognitive task (such as the British picture vocabulary scale); there were 14 trials in total. We focused on the frequency with which the children elected to share the reward with a friend. Possible scores ranged from 0 to 14 points.

Questionnaire.  The strengths and difficulties questionnaire (SDQ) (Goodman, 1997) was completed by the mothers at all three time points and by the teachers at time 3. For each of the 25 items, the mothers/teachers rated ‘not true’ (0), ‘sometimes true’ (1), or ‘certainly true’ (2). Our focus is on the five questions that make up the prosocial subscale (e.g., considerate of other people's feelings; shares readily with other children; helpful if someone is upset, hurt, or feeling ill; kind to younger children; often volunteers to help others) (Cronbach's α at t1 = .73, t2 = .62, t3 mother ratings = .69, and t3 teacher ratings = .82); possible scores range from 0 to 10 points.

Reliability of Coding

Twenty percent of videos were double coded, with good to excellent agreement between coders. Intra-class correlations for mother–child mutuality at time 1 and time 2 were .99 and .82, respectively. Kappa coefficients were 1.00, 1.00, and .79 for sharing at time 1, time 2, and time 3, respectively, and 1.00, 1.00, and .72 for helping at time 1, time 2, and time 3, respectively.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments

Descriptive Statistics

Sharing behaviors were observed in over 70 percent of the children at each of the three time points; 43 percent of the children also helped their friends at time 3 (see Table 2). In contrast, helping was very infrequent at time 1 (<7 percent of the children) and time 2 (<13 percent of the children), and so these measures were excluded from our analyses.

Table 2. Descriptive Statistics for Study Measures
MeasureTime pointXSDRange
  1. BAS = British abilities scales; BPVS = British picture vocabulary scale; SDQ = strengths and difficulties questionnaire.

BAS: vocabulary/comprehension120.3010.020–38
BPVS: vocabulary335.0211.3911–72
Emotion understanding19.007.270–26
Mother–child play: mutuality13.21.322.17–3.92
Unfamiliar peer play: share15.195.410–22.76
Friend play: share312.9511.070–60
Friend play: help32.854.330–22.86
Sticker task: share35.694.810–14
Mother SDQ: prosocial16.012.260–10
Teacher SDQ: prosocial35.922.970–10

Gender Differences.  Girls showed higher frequencies of sharing/helping than did boys. These gender differences were marginal for sharing at time 1: t= 1.64, p= .10; and time 2: t= 1.82, p= .07; non-significant for sharing at time 3: t= 1.28, p= .20; and significant for helping at time 3: t= 3.63, p < .001. In addition, compared with boys, girls elected to share marginally more often on the sticker task: t= 1.74, p= .09; and received marginally higher SDQ prosocial behavior ratings from the mothers at time 2: t= 1.77, p= .08; and teachers at time 3: t= 1.74, p= .09. In contrast, the mother SDQ prosocial behavior ratings showed no gender differences at times 1 and 3: t≤ 1.54, p≥ .13. With the exception of time 2 emotion understanding scores, which were significantly higher in girls than in boys: t= 2.43, p < .05; there were no gender differences on any other measure: t≤ 1.61, p≥ .11.

Developmental Change.  Here, we focus on measures available at all three time points. Repeated measure analyses of variance showed age-related increases in hourly rates of sharing: F= 53.11, p < .001 (reflecting significantly higher frequencies of the children's sharing with friends at time 3 vs. unfamiliar peers at times 1: t= 7.29, p < .001; and 2: t= 6.87, p < .001); the mothers' SDQ prosocial behavior ratings: F= 22.48, p < .001; and the children's performance on emotion understanding task (with scores at time 1 prorated to a possible maximum score of 48): F= 367.17, p < .001.

Data Reduction

At times 1 and 2, hourly rates of sharing were significantly correlated with the mothers' SDQ prosocial behavior ratings: t1 r= .20, p < .05; t2 r= .27, p < .01. We therefore converted these measures into z-scores and averaged them to create times 1 and 2 prosocial behavior composites. At time 3, a principal components analysis with varimax rotation showed that the five measures (hourly rates of sharing with and helping friends; planned sharing within sticker task; the mothers' and teachers' SDQ prosocial behavior ratings) loaded onto a single factor that accounted for 42 percent of the variance (with loadings that ranged from .54 to .73). The analysis was rerun with standardized residuals to control for the effects of age: A single factor was again evident, with moderate to substantial loadings (.56–.73) that explained 43 percent of the variance. To maximize reliability, the observational, experimental, and questionnaire measures of prosocial behaviors at time 3 were therefore converted into z-scores and averaged to create a multi-informant and multi-measure aggregate with modest but acceptable internal consistency (Cronbach's α= .65).


Table 3 shows the full correlations between all study measures; partial correlations with effects of concurrent/later age controlled are shown in parentheses. Note that although individual differences in prosocial behaviors were correlated across all time points, this correlation was especially strong between times 1 and 2. Individual differences in prosocial behaviors were significantly correlated (both concurrently and longitudinally) with individual differences in: (1) emotion understanding at time 2 and time 3; (2) verbal ability (with the exception of a marginally significant correlation between prosocial behaviors at time 1 and verbal ability at time 3); and (3) mutuality at time 1 (mutuality at time 2 was only significantly correlated with time 3 prosocial behaviors). In contrast, individual differences in prosocial behaviors were unrelated to maternal education, head of household occupational status, and family structure (mean r= .04); for reasons of space and simplicity, these family background measures were excluded from our analyses.

Table 3. Correlations between Study Measures (Correlations with Concurrent/Later Age Partialled)
 Verbal ability (V)Emotion understanding (E)Mutuality (M)Prosocial behavior (P)
  • p < .10;

  • ** 

    p < .05;

  • *** 

    p < .01.

V t2.77***          
V t3.65***.69***         
E t1.61***.53***.34***        
E t2.70***.69***.59***.53***       
E t3.61***.64***.58***.34***.71***      
M t1.39***.48***.27***.28***.43***.47***     
M t2.23**.17*.22**.14.32***.27**.27***    
P t1.30***.30***.19*.19*.24***.26***.35***.08   
P t2.30***.32***.22**.21**.25**.39**.31**.20*.52**  
P t3.28**.30**.37**.17*.37***.40***.31***.29***.21**.21** 

Regression and Mediation Analyses

Rather than examining all possible cross-lagged associations, our next analyses focused on the significant correlations reported above. We selected independent variables (verbal ability and mother–child mutuality at time 1), a posited mediator (emotion understanding at time 2), and a dependent variable (prosocial behavior at time 3) assessed at different study time points. Before conducting mediation analyses, we addressed our first study hypothesis by using a hierarchical regression analysis to establish whether emotion understanding at time 2 remained significantly associated with prosocial behaviors at time 3 when effects of verbal ability and mutuality at time 1 were controlled. Specifically, we entered age, gender, and prosocial behaviors at time 1 at step 1, and verbal ability and mutuality at time 1 at step 2. At step 3, emotion understanding at time 2 explained an additional 8 percent of variance in prosocial behaviors at time 3. Table 4 shows the coefficients. In sum, emotion understanding at time 2 predicted unique variance in prosocial behaviors at time 3 beyond effects of age, prosocial behaviors, verbal ability, and mutuality at time 1.

Table 4. Regression Analysis: Emotion Understanding at Time 1 Predicts Prosocial Behavior at Time 3, Beyond Effects of Age, Gender, Prosocial Behavior, Verbal Ability, and Mother–Child Mutuality at Time 2
DV: Prosocial behavior t3BSE Bβp
  • p < .10;

  • ** 

    p < .05;

  • *** 

    p < .01.

Step 1    
 Prosocial behavior t1.01.01.24**
R2= .08, F= 2.77, p < .05
Step 2    
 Prosocial behavior t1.
 Verbal ability t1.02.01.20*
 Mutuality t1.27.12.24**
ΔR2= .11, ΔF= 6.33, p < .01
Step 3    
 Prosocial behavior t1.
 Verbal ability t1.
 Mutuality t1.
 Emotion understanding t2.05.02.36***
ΔR2= .08, ΔF= 9.93, p < .01

To test our mediation hypotheses, we used a path analysis and applied the multivariate delta method (Sobel, 1982) to estimate indirect effects and derive their standard errors. We selected MPlus 5 (Muthén & Muthén, Los Angeles, CA) to analyze the sample variance–covariance matrix because this program directly estimates specific indirect effects. We tested the model that emotion understanding at time 2 mediated the impact of verbal ability and mother–child mutuality at time 1 on prosocial behaviors at time 3. In addition, on the basis of the results of the regression analysis, we covaried prosocial behaviors at time 1. Each coefficient shown in Figure 1 is the ratio of the parameter estimate to its standard error. The model was saturated; thus, there were no degrees of freedom and the χ2 test of fit equaled zero. In order to obtain model fit statistics, we omitted the (non-significant) path between verbal ability at time 1 and prosocial behavior at time 3. Goodness of model fit was evaluated using the standardized root mean square residual ≤.08, root mean square error of approximation ≤.06, comparative fit index ≥.95, and the Tucker–Lewis index ≥.95 (Brown, 2006). The model fitted the data well: χ2 (1) = .05, p= .82, standardized root mean square residual = .00, root mean square error of approximation = .00, comparative fit index = 1.00, Tucker–Lewis index = 1.03.


Figure 1. Mediation Analysis: Emotion Understanding at Time 2 Mediates Associations between Verbal Ability and Mother–Child Mutuality at Time 1 and Prosocial Behavior at Time 3. **p < .01.

Download figure to PowerPoint

In relation to H2, the indirect effect of verbal ability at time 1 on prosocial behaviors at time 3 through emotion understanding at time 2 was .02 (SE= .01) with 95 percent confidence intervals of .01 to .03. The confidence intervals did not include zero, indicating that the difference between the indirect effect of verbal ability at time 1 on prosocial behaviors at time 3 through emotion understanding at time 2 and the direct effect of verbal ability at time 1 on prosocial behaviors at time 3 was significantly different from zero. This analysis supports our hypothesis that emotion understanding at time 2 mediated the effect of verbal ability at time 1 on prosocial behaviors at time 3.

Turning now to H3, the indirect effect of mutuality at time 1 on prosocial behaviors at time 3 through emotion understanding at time 2 was −.86 (SE= .32) with confidence intervals of −1.48 to −.23. Thus, the difference between the indirect effect of mutuality at time 1 on prosocial behaviors at time 3 through emotion understanding at time 2 and the direct effect of mutuality at time 1 on prosocial behaviors at time 3 was again significantly different from zero. This analysis supported our hypothesis that emotion understanding at time 2 mediated the effect of mutuality at time 1 on prosocial behaviors at time 3.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments

This study examined three hypotheses. Our first, preliminary hypothesis was that variation in children's emotion understanding would be robustly associated with later individual differences in prosocial behaviors. In support of this hypothesis, emotion understanding at the age of three was significantly associated with prosocial behaviors at the age of four, even when effects of both verbal ability and parent–child mutuality at the age of two were controlled. Note, however, that interpretations of this association must be tentative, given the considerable concurrent and longitudinal correlations between verbal ability and emotion understanding. Next, we hypothesized that emotion understanding would mediate the associations between early verbal ability and parent–child mutuality, and later prosocial behaviors. Our results supported each of these hypothesized mediation effects. Specifically, including emotion understanding at the age of three in our regression analyses attenuated relations between verbal ability/parent–child mutuality at the age of two and prosocial behaviors at the age of four. A path analysis confirmed these mediation effects, which are discussed in turn below.

In relation to the first mediation effect, it is worth noting that the use of emotion terms is a key milestone in early language development (Bretherton & Beeghly, 1982; Hughes et al., 2007). In the present study, verbal skills appeared necessary but not sufficient for the children's prosocial behaviors, because intervening individual differences in emotion understanding carried the association between verbal ability and prosocial behaviors. Indeed, in our path analysis, the direct association between verbal ability at the age of two and prosocial behaviors at the age of four was not significant. Such a link between the predictor and outcome is not necessarily required to test a mediation effect (Kenny, Kashy, & Bolger, 1998); however, studies often lack power to detect the direct association between distal predictors and outcomes (Shrout & Bolger, 2002).

One simple interpretation of the above finding is that verbal ability enables children to initiate discussions about emotions and to assimilate others' talk about feelings. A second, more complex interpretation hinges on the regulatory role of children's verbal ability (Winsler, Diaz, McCarthy, Atencio, & Chabay, 1999). Specifically, some prosocial acts require children to simultaneously observe others' emotional expressions, monitor their own affective response, and plan an adaptive action. From a Vygotskian perspective, these higher order processes depend heavily upon children's early language skills. This perspective is consistent with the view that emotional regulation plays a key role in prosocial activity (e.g., Blair, Denham, Kochanoff, & Whipple, 2004).

Turning to the second mediation effect, it is worth noting that responsive, reciprocal, and co-operative parent–child interactions (i.e., parent–child mutuality) are traditionally thought to lead directly to children's prosocial behaviors. The findings from this study suggest that children's abilities to detect and reflect on feelings are implicated in the association between parent–toddler mutuality and preschool prosocial behaviors. In other words, early close relationships may influence later prosocial behaviors by motivating children to apply their emotion understanding to prosocial ends. That said, direct associations with parent–child mutuality may have been weakened in our study because our measures focused on prosocial behaviors directed toward peers rather than parents.

The findings described above also provide a fresh perspective on the results of earlier studies of mother–child conversations. Secure attachment is associated with types of communication that promote young children's abilities to identify feelings and understand their causes (e.g., Harris, 1999). In particular, mutually responsive mother–child dyads have been shown to refer frequently to feelings (Farrar, Fasig, & Welch-Ross, 1997) and to engage in fluent and connected conversations (Askan, Kochanska, & Ortmann, 2006); moreover, such connectedness predicts later individual differences in emotion understanding (Ensor & Hughes, 2008). These prior findings suggest a conversationally mediated pathway from mother–child mutuality to emotion understanding. Our current findings extend this model by demonstrating that, via effects on emotion understanding, parent–child mutuality may also encourage children's prosocial behaviors.

We turn now to a few methodological issues. There is a relative lack of research addressing the processes that underpin individual differences in prosocial behaviors (Eisenberg, Fabes, & Spinrad, 2006). The current study followed recommendations for tests of mediation (Maxwell & Cole, 2007) as far as sample size and power permitted but did not attempt a full autoregressive mediation model. Finally, a basic prerequisite for testing mediation is the availability of reliable measures of individual differences. However, establishing convergence between different informants and methods is a particular challenge for studies of young children (Ladd & Kochenderfer-Ladd, 2002); we were therefore encouraged by the (albeit modest) agreement, not only between maternal and teacher questionnaire ratings, but also with disparate measures of preschoolers' observed sharing and helping friends, and experimental assessments of planned sharing.

Future studies should address the shortfalls of our own study. Chief among these is the lack of similarity between our aggregate measures of prosocial behaviors at the ages of two and three vs. the age of four. Constructing reliable and equivalent measures of prosocial behaviors at several ages would enable future studies to assess whether emotion understanding matters not only for individual differences but also for developmental change in prosocial functioning. Nevertheless, we believe that the current study provides insights into the mechanisms that underlie individual differences in prosocial behaviors. This transactional perspective allows us to go beyond demonstrating the importance of parent–child interactions to identify the specific mechanisms through which these everyday family exchanges may shape children's propensities to be prosocial.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments
  • Askan, N., Kochanska, G., & Ortmann, M. (2006). Mutually responsive orientation between parents and their young children: Toward methodological advances in the science of relationships. Developmental Psychology, 42, 833848.
  • Blair, K., Denham, S., Kochanoff, A., & Whipple, B. (2004). Playing it cool: Temperament, emotion regulation, and social behavior in preschoolers. Journal of School Psychology, 42, 419443.
  • Borke, H. (1971). Interpersonal perception of young children: Egocentrism or empathy? Developmental Psychology, 5, 263269.
  • Bretherton, I., & Beeghly, M. (1982). Talking about internal states: The acquistion of a theory of mind. Developmental Psychology, 18, 906921.
  • Brown, T. (2006). Confirmatory factor analysis for applied research (1st ed.). New York: Guilford Press.
  • Campos, J., Campos, R., & Barrett, K. (1989). Emergent themes in the study of emotional development and emotion regulation. Developmental Psychology, 25, 394402.
  • Cassidy, K., Werner, R., Rourke, M., Zubernis, L., & Balaraman, G. (2003). The relationship between psychological understanding and positive social behaviors. Social Development, 12, 198221.
  • Contreras, J., Kerns, K., Weimer, B., Gentzler, A., & Tomich, P. (2000). Emotion regulation as a mediator of associations between mother-child attachment and peer relationships in middle childhood. Journal of Family Psychology, 14, 111124.
  • de Rosnay, M., & Hughes, C. (2006). Conversation and theory of mind: Do children talk their way to socio-cognitive understanding? British Journal of Developmental Psychology, 24, 737.
  • Deater-Deckard, K., & O'Connor, T. (2000). Parent-child mutuality in early childhood: Two behavioral genetic studies. Developmental Psychology, 36, 561570.
  • Deater-Deckard, K., & Petrill, S. (2004). Parent-child dyadic mutuality and child behavior problems: An investigation of gene-environment processes. Journal of Child Psychology and Psychiatry, 45, 11711179. Behavioral genetic studies. Developmental Psychology, 36, 561–570.
  • Deater-Deckard, K., Pylas, M., & Petrill, S. (1997). Parent-child interactive system (PARCHISY). Unpublished.
  • Dekovic, M., & Janssens, J. (1992). Parents' child-rearing style and child's sociometric status. Developmental Psychology, 28, 925932.
  • Denham, S. (1986). Social cognition, prosocial behavior, and emotion in preschoolers: Contextual validation. Child Development, 57, 194201.
  • Denham, S., Blair, K., DeMulder, E., Levitas, J., Sawyer, K., Auerbach-Major, S., et al. (2003). Preschool emotional competence: Pathway to social competence. Child Development, 74, 238256.
  • Denham, S., Zoller, D., & Couchod, E. (1994). Socialisation of preschoolers' emotion understanding. Developmental Psychology, 30, 928936.
  • Dunn, J., & Cutting, A. (1999). Understanding others, and individual differences in friendship interactions in young children. Social Development, 8, 201219.
  • Dunn, J., Cutting, A., & Fisher, N. (2002). Old friends, new friends: Predictors of children's perspective on their friends at school. Child Development, 73, 621635.
  • Dunn, L. (1997). British picture vocabulary scale—revised. Windsor: NFER-Nelson.
  • Eisenberg, N. (1982). Introduction. In N.Eisenberg (Ed.), The development of prosocial behavior (pp. 121). New York: Academic Press.
  • Eisenberg, N., & Fabes, R. (1998). Prosocial development. In W.Damon (Ed.), Handbook of child psychology (Vol. 3, pp. 701778). New York: Wiley.
  • Eisenberg, N., Fabes, R., & Spinrad, T. (2006). Prosocial development. In N.Eisenberg, W.Damon, & R.Lerner (Eds.), Handbook of child psychology: Social, emotional, and personality development (6th ed., Vol. 3, pp. 646718). Hoboken, NJ: John Wiley & Sons Inc.
  • Eisenberg, N., Gershoff, E., Fabes, R., Shepard, S., Cumberland, A., Losoya, S., et al. (2001). Mother's emotional expressivity and children's behavior problems and social competence: Mediation through children's regulation. Developmental Psychology, 37, 475490.
  • Eisenberg, N., Guthrie, I., Murphy, B., Shepard, S., Cumberland, A., & Carlo, G. (1999). Consistency and development of prosocial dispositions: A longitudinal study. Child Development, 70, 13601372.
  • Eisenberg, N., & Mussen, P. (1989). The roots of prosocial behavior in children. New York: Cambridge University Press.
  • Eisenberg, N., Zhou, Q., Spinrad, T., Valiente, C., Fabes, R., & Liew, J. (2005). Relations among positive parenting, children's effortful control, and externalizing problems: A three-wave longitudinal study. Child Development, 76, 10551071.
  • Elliott, C., Murray, D., & Pearson, L. (1983). British abilities scales. Windsor: NFER-Nelson.
  • Ensor, R., & Hughes, C. (2005). More than talk: Relations between emotion understanding and positive behaviour in toddlers. British Journal of Developmental Psychology, 23, 343363.
  • Ensor, R., & Hughes, C. (2008). Content or connectedness? Early family talk and theory of mind in the toddler and preschool Years. Child Development, 79, 201216.
  • Farrar, M., Fasig, L., & Welch-Ross, M. (1997). Attachment and emotion in autobiographical memory development. Journal of Experimental Child Psychology, 67, 389408.
  • Farver, J., & Branstetter, W. (1994). Preschoolers' prosocial responses to their peers' distress. Developmental Psychology, 30, 334341.
  • Frick, P., & White, S. (2008). Research review: The importance of callous-unemotional traits for developmental models of aggressive and antisocial behavior. Journal of Child Psychology and Psychiatry, 49, 359375.
  • Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581586.
  • Grusec, J. (1991). Socializing concern for others in the home. Developmental Psychology, 27, 338342.
  • Harris, P. (1989). Children and emotion. Oxford: Basil Blackwell.
  • Harris, P. (1999). Individual differences in understanding emotion: The role of attachment status and psychological discourse. Attachment & Human Development, 1, 307324.
  • Harris, P., Johnson, C., Hutton, D., Andrews, G., & Cooke, T. (1989). Young children's theory of mind and emotion. Cognition and Emotion, 3, 379400.
  • Harrist, A., & Waugh, R. (2002). Dyadic synchrony: Its structure and function in children's development. Developmental Review, 22, 555592.
  • Hay, D., & Cook, K. (2007). The transformation of prosocial behavior from infancy to childhood. In C.Brownell, & C.Kopp (Eds.), Socioemotional development in the toddler years: Transitions and transformations (pp. 100131). New York: Guilford Press.
  • Hoffman, M. (1975). Altruistic behavior and the parent-child relationship. Journal of Personality and Social Psychology, 31, 937943.
  • Hoffman, M. (1982). Development of prosocial motivation: Empathy and guilt. In N.Eisenberg (Ed.), The development of prosocial behavior (pp. 281313). New York: Academic Press.
  • Hughes, C., Lecce, S., & Wilson, C. (2007). ‘Do you know what I want?’ Preschoolers' talk about desires, thoughts and feelings in their conversations with sibs and friends. Cognition and Emotion, 21, 330350.
  • Iannotti, R. (1985). Naturalistic and structured assessments of prosocial behavior in preschool children: The influence of empathy and perspective taking. Developmental Psychology, 21, 4655.
  • Kenny, D., Kashy, D., & Bolger, N. (1998). Data analysis in social psychology. In D.Gilbert, S.Fiske, & G.Lindzey (Eds.), The handbook of social psychology (pp. 233265). New York: McGraw-Hill.
  • Kochanska, G. (1997a). Multiple pathways to conscience for children with different temperaments: From toddlerhood to age 5. Developmental Psychology, 33, 228240.
  • Kochanska, G. (1997b). Mutually responsive orientation between mothers and their young children: Implications for early socialisation. Child Development, 68, 94112.
  • Kochanska, G., & Aksan, N. (2004). Development of mutual responsiveness between parents and their young children. Child Development, 75, 16571676.
  • Krevans, J., & Gibbs, J. (1996). Parents' use of inductive discipline: Relations to children's empathy and prosocial behavior. Child Development, 67, 32633277.
  • Lacourse, E., Nagin, D., Vitaro, F., Cote, S., Arseneault, L., & Tremblay, R. (2006). Prediction of early-onset deviant peer group affiliation: A 12-year longitudinal study. Archives of General Psychiatry, 63, 562568.
  • Ladd, G., & Kochenderfer-Ladd, B. (2002). Identifying victims of peer aggression from early to middle childhood: Analysis of cross-informant data for concordance, estimation of relational adjustment, prevalence of victimization, and characteristics of identified victims. Psychological Assessment, 14, 7496.
  • Laible, D., & Song, J. (2006). Constructing emotional and relational understanding: The role of affect and mother-child discourse. Merrill-Palmer Quarterly, 52, 4469.
  • Laible, D., & Thompson, R. (1998). Attachment and emotional understanding in preschool children. Developmental Psychology, 34, 10381045.
  • Lansford, J., Putallaz, M., Grimes, C., Schiro-Osman, K., Kupersmidt, J., & Coie, J. (2006). Perceptions of friendship quality and observed behaviors with friends: How do sociometrically rejected, average, and popular girls differ? Merrill-Palmer Quarterly, 52, 694720.
  • Levitt, M., Weber, R., Clark, M., & McDonnell, P. (1985). Reciprocity of exchange in toddler sharing behavior. Developmental Psychology, 21, 122123.
  • Maccoby, E. (1992). The role of parents in the socialization of children: An historical overview. Developmental Psychology, 28, 10061017.
  • Maxwell, S., & Cole, D. (2007). Bias in cross-sectional analyses of longitudinal mediation. Psychological Methods, 12, 2344.
  • McDowell, D., & Parke, R. (2005). Parental control and affect as predictors of children's display rule use and social competence with peers. Social Development, 14, 440457.
  • Michalson, M., & Lewis, M. (1985). What do children know about emotions and when do they know it? In M.Lewis, & C.Saarni (Eds.), The socialization of emotions (pp. 117139). New York: Plenum.
  • Moore, C., Barresi, J., & Thompson, C. (1998). The cognitive basis of future-oriented prosocial behaviour. Social Development, 7, 198218.
  • Mostow, A., Izard, C., Fine, S., & Trentacosta, C. (2002). Modeling emotional, cognitive, and behavioral predictors of peer acceptance. Child Development, 73, 17751787.
  • Pons, F., Harris, P., & de Rosnay, M. (2004). Emotion comprehension between 3 and 11 years: Developmental periods and hierarchical organization. European Journal of Developmental Psychology, 1, 127152.
  • Repacholi, B., & Gopnik, A. (1997). Early reasoning about desires: Evidence from 14- and 18-month-olds. Developmental Psychology, 33, 1221.
  • Rocissano, L., Slade, A., & Lynch, V. (1987). Dyadic synchrony and toddler compliance. Developmental Psychology, 23, 698704.
  • Romano, E., Tremblay, R., Boulerice, B., & Swisher, R. (2005). Multilevel correlates of childhood physical aggression and prosocial behavior. Journal of Abnormal Child Psychology, 33, 565578.
  • Rothbaum, F., & Weisz, J. (1994). Parental caregiving and child externalising behaviour in nonclinical samples: A meta-analysis. Psychological Bulletin, 116, 5574.
  • Ruffman, T., Slade, L., Devitt, K., & Crowe, E. (2006). What mothers say and what they do: The relation between parenting, theory of mind, language and conflict/cooperation. British Journal of Developmental Psychology, 24, 105124.
  • Rutter, D., & Durkin, K. (1987). Turn-taking in mother-infant interaction: An examination of vocalizations and gaze. Developmental Psychology, 23, 5461.
  • Shrout, P., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422445.
  • Sobel, M. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. In S.Leinhardt (Ed.), Sociological methodology (pp. 290312). San Francisco, CA: Jossey-Bass.
  • Spinrad, T., Eisenberg, N., Gaertner, B., Popp, T., Smith, C., Kupfer, A., et al. (2007). Relations of maternal socialization and toddlers' effortful control to children's adjustment and social competence. Developmental Psychology, 43, 11701186.
  • Thompson, R., & Goodvin, R. (2007). Taming the tempest in the teapot: Emotion regulation in toddlers. In C.Brownell, & C.Kopp (Eds.), Socioemotional development in the toddler years: Transitions and transformations (pp. 320341). New York: Guilford Press.
  • Valiente, C., Fabes, R. A., Eisenberg, N., & Spinrad, T. L. (2004). The relations of parental expressivity and support to children's coping with daily stress. Journal of Family Psychology, 18, 97106.
  • Vitaro, F., Gagnon, C., & Tremblay, R. (1990). Predicting stable peer rejection from kindergarten to grade one. Journal of Clinical Child Psychology, 19, 257264.
  • Winsler, A., Diaz, R., McCarthy, E., Atencio, D., & Chabay, L. (1999). Mother-child interaction, private speech, and task performance in preschool children with behaviour problems. Journal of Child Psychology and Psychiatry, 40, 891904.
  • Wojslawowicz Bowker, J., Rubin, K., Burgess, K., Booth-Laforce, C., & Rose-Krasnor, L. (2006). Behavioral characteristics associated with stable and fluid best friendship patterns in middle childhood. Merrill-Palmer Quarterly, 52, 671693.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. References
  8. Acknowledgments

This research was supported by grants from the Health Foundation and the Economic and Social Research Council.