Abstract
Prosocial behavior is a highly heterogeneous construct, and young children use distinct prosocial actions in response to differing emotional needs of another person. This study examined whether toddlers’ prosocial responses differed in response to two understudied emotional contexts—whether or not children caused a victim’s distress and the specific emotion expressed by the victim. Toddlers (N = 86; M age = 35 months) and their parent participated in two separate mishap paradigms in which parents feigned pain and sadness, respectively. Half of the sample was led to believe they had transgressed to cause their parent’s distress, whereas the other half simply witnessed parent distress as bystanders. Results indicated that toddlers were overall equally prosocial when they were transgressors compared to when they were bystanders and significantly more prosocial in response to sadness than pain. Toddlers were significantly more likely to use affection as transgressors than bystanders, information seeking as bystanders than transgressors, and affection in response to pain than sadness. All children used greater helping in response to sadness than pain, and this was especially true when they were bystanders. Findings add to mounting evidence of the complexity of prosocial action in early childhood by identifying that two, distinct emotional contexts influence the amount and type of prosocial behaviors that toddlers use to help others.
Keywords
Prosocial behaviors are voluntary actions intended to benefit others (Hay & Cook, 2007). Subtypes of prosocial behavior—that is, instrumental helping, sharing, and comforting behaviors—have historically been treated as similar “kinds,” resulting in inconsistencies in identifying the emergence and correlates of prosocial responding (Dunfield, 2014). Recently, however, studies have found that these distinct subtypes emerge at different ages and have different correlates, supporting that prosocial behavior is heterogeneous as early as toddlerhood, when prosocial actions rapidly develop (Hay & Cook, 2007). Importantly, different prosocial subtypes reflect distinct responses to another person’s differing emotional needs; for example, sharing is a response to a person’s material need, whereas comforting is a response to another’s emotional distress (Dunfield, 2014). Thus, the emotional context of another’s need influences children’s prosocial responses. This study examined whether toddlers’ prosocial responses also differ depending on two, understudied emotional contexts: whether or not the child caused a victim’s distress and the emotion expressed by the victim.
Witnessing another’s distress as a bystander is a distinct emotional context from causing another’s distress as a transgressor, and the prosocial responses children use as bystanders differ from those they use as transgressors (referred to as reparative behaviors). Indeed, although both prosocial and reparative behaviors emerge during the second year of life (Zahn-Waxler et al., 1992), they have been associated with distinct socialization mechanisms and developmental trajectories (Donohue et al., 2019). Several studies have examined young children’s reparative prosocial responses to transgressions, finding that children use more reparative behaviors when they experience guilt rather than shame (e.g., Drummond et al., 2017). Relatively, few studies have compared children’s prosocial responding in transgressor versus bystander contexts. Most studies have found that young children are less frequently prosocial as transgressors than as bystanders (e.g., Dunn & Brown, 1994; Zahn-Waxler et al., 1992), though others found that children were more prosocial as transgressors (Demetriou & Hay, 2004; Vaish et al., 2016). Notably, most of these studies examined prosocial responses to peers’ distress (for an exception, see Vaish et al., 2016). Toddlers’ prosocial responses in the parent–child context are important to study, as by age 2, children are more empathic toward parents’ than others’ emotions (van der Mark et al., 2002), and toddlers may be particularly motivated to use reparative behaviors to mend the parent–child relationship. Thus, although findings are mixed, whether children cause or witness distress appears to influence their mean levels of prosocial responding.
Although there is a literature supporting toddlers’ ability to understand others’ negative emotions and respond with appropriate empathic behaviors (Chiarella & Poulin-Dubois, 2013), only one known study has compared toddlers’ prosocial responses to specific negative emotions. Bandstra et al. (2011) found that toddlers used more frequent prosocial behaviors in response to an experimenter’s sadness versus pain. Interestingly, this study also examined the specific prosocial strategies children used (e.g., comforting statements, sharing) and found that toddlers used significantly more instrumental helping (i.e., problem-solving) when responding to sadness versus pain. This finding raises the intriguing possibility that the emotional context of another’s distress may influence not only toddlers’ overall rate of prosocial responding but also the specific prosocial strategies that they choose.
The present study sought to add to a small body of literature that documents how emotional contexts influence children’s prosocial responding by comparing children’s prosocial responses to their parent’s distress (a) when they were bystanders versus transgressors and (b) in response to pain versus sadness. Following most previous studies, we hypothesized that toddlers would be overall more prosocial as bystanders than transgressors. Consistent with the one previous study on the topic (Bandstra et al., 2011), we also hypothesized that toddlers would be more prosocial in response to their parent’s sadness than pain. We also explored the possibility that toddlers would use differing specific prosocial strategies depending on the emotional context of the distress. Due to lack of research on the topic, this hypothesis was exploratory.
Method
Participants
Participants were a community sample (N = 86) of typically developing toddlers and their parent. Participants were recruited from a research participant pool, community postings, and word of mouth. Children were either 2 (range = 26.60–34.30 months; M age = 31.41 months; SD = 1.40 months) or 3 years old (range = 33.70–47.80 months; M age = 38.27 months; SD = 3.21 months). Demographic characteristics are presented in Table 1.
Descriptive Statistics and Main Effects of Fault on Prosocial Behavior.
Note. OR = odds ratio; CI = confidence interval. Item anchors for global prosocial variable: 0 = none; 4 = strong. Bold indicates significant results.
a This value reflects the significant interaction between fault and emotion on helping.
Experimental Design
We used mishaps adapted from previous research to create situations where parents feigned distress (e.g., Kochanska et al., 2002). The first half of the children who participated in the study chronologically (n = 43) were assigned to the bystander condition, in which parents pretended they caused their own distress; the second half (n = 43) were assigned to the transgressor condition, in which parents pretended the child caused their distress. All children (i.e., children in both conditions) experienced two trials, one in which the mishap caused the parent’s pain and the other in which it caused parent’s sadness. The order of the trials (i.e., pain, sadness) was fully counterbalanced. Thus, condition was a between-subjects variable and trial was a within-subjects variable.
Procedures
This study took place within a larger study of children’s socio-cognitive development. Informed consent was obtained from parents. The experimenter (“E”) trained parents on acting out the mishaps while a research assistant (RA) engaged in a 5-min warm-up with the child in a separate room. The remainder of the session was videotaped. The first trial (i.e., pain or sadness) occurred as the first paradigm of the lab visit. Children then engaged in paradigms related to the larger study over the next 30 min. The second trial then occurred as the last paradigm of the lab visit. Children in the transgressor condition were debriefed to minimize distress (e.g., Kochanska et al., 2002). Note that about half of the children in this study viewed a video before each trial related to the larger study—and unrelated to any variable examined in this study. For more details regarding procedures of the larger study and current study, see Supplemental Materials.
Materials
A pounding toy composed of blocks and a hammer and a framed photo that was rigged to break when handled were used to cause the mishaps.
Experimental Conditions
Bystander condition
In the bystander pain trial, the parent played with the pounding toy and pretended to hammer her own finger. The parent then said “Ow! I hit my finger. It really hurts.” Parents repeated this script after 30 s. Throughout the entire mishap (60 s), the parent pretended to be in pain by rubbing her finger and using pained facial expressions and vocalizations. E entered the room, cueing the parent to recover and say “I feel better now.”
In the bystander sadness trial, E gave the parent a framed photo as a gift. The parent said “Oh I love this picture frame. It’s my new favorite present. It makes me happy” to demonstrate to the toddler that the frame was valuable to the parent. The parent attempted to stand the picture frame and in the process it fell apart. The parent said “Oh, no! I broke my present. I’m really sad. Oh, no.” Parents repeated this script after 30 s. Throughout the entire incident (60 s), the parent pretended to be sad using sad facial expressions and vocalizations. E entered the room, cueing the parent to recover.
Transgressor condition
In the transgressor pain trial, the parent slipped a finger on a block as her child hammered, pretending her child hammered her finger. She said “Ow! You hit my finger. It really hurts.” All other aspects were identical to the bystander pain trial. In the transgressor sadness trial, the parent asked the child to hold the rigged frame. When the child held the frame, it fell apart. The parent said “Oh, no. You broke my present. I’m really sad. Oh, no.” All other aspects were identical to the bystander sadness trial. Note that all parent scripts were matched for approximate duration, body positions, and vocalizations.
Scoring and rating of observational data
RAs who were blind to study hypotheses rated several measures of prosocial behaviors. Inter-rater reliability was assessed with two-way, mixed, absolute agreement intraclass correlation coefficients (ICCs). ICCs reported subsequently are averages of each measure across conditions and trials.
Global prosocial behaviors
RAs rated how globally prosocial children were by rating the overall frequency and sophistication of children’s prosocial behaviors on a 5-point scale (0 = none; 4 = strong) based on systems used by Williamson et al. (2013) and Zahn-Waxler et al. (1995). RAs made one global prosocial rating per trial; ICC = .92.
Specific prosocial strategies
RAs also rated the frequency and sophistication of children’s displays of five specific prosocial strategies using the same 5-point scale. Apologies included children’s verbal apologies (e.g., “I’m sorry”); comforting statements included statements to reassure or comfort the parent (e.g., “You’re all better”); information seeking included children’s attempts to gain information about the parent’s emotions using questions (e.g., “You’re very sad?”); affection included children’s affectionate behaviors (e.g., hugging, kissing, patting); and helping included children’s problem-solving behaviors (e.g., attempting to retrieve a Band-Aid, fix the frame) and statements about these helping behaviors (e.g., “I’ll get a bandaid”). Thus, RAs made one rating of each specific strategy per trial. The five strategies were chosen based on previous work (e.g., Bandstra et al., 2011); ICC: M = 0.92; range = .86–.99.
Data Analytic Plan
To answer whether children were more prosocial (a) in the bystander versus transgressor context and (b) in response to parent’s pain versus sadness, we conducted one, two-way mixed analysis of variance (ANOVA) with condition as the between-subjects variable and trial as the within-subjects variable predicting global prosocial ratings. Trial order and whether or not children watched a video related to the larger study were included as covariates. Second, we examined whether toddlers used more frequent and sophisticated specific prosocial strategies (a) in the bystander versus transgressor context and (b) in response to pain versus sadness. Assumptions for mixed ANOVA were not met; for example, the data for each specific strategy were non-normal (i.e., positively skewed wherein for any given strategy, many children did not use the strategy). Thus, we converted each specific prosocial strategy variable from the continuous scale to a dichotomous variable (i.e., presence or absence of the strategy). We then used generalized estimating equations with condition as the between-subjects variable and trial as the within-subjects variable to predict the probability of children’s use of given specific strategy. Again, covariates included trial order and whether or not children watched a video related to the larger study, and p-values were false discovery rate (FDR) corrected to account for multiple comparisons. For readability, we refer to condition (i.e., bystander or transgressor) as “fault” and trial (i.e., pain or sadness) as “emotion” when reporting results of hypothesis tests.
Results
Descriptive Statistics
Boys (M = 36.24, SD = 4.88) were older than girls (M = 33.70, SD = 3.03), t(84) = 2.55, p = .005. Independent samples t-tests and χ2 tests did not reveal significant differences in any dependent variable by trial order or any demographic characteristic of the child.
Children’s greater global prosocial behaviors on the pain trial were moderately and significantly associated with greater global prosocial behaviors on the sadness trial (r = .55; p ≤ .001), suggesting that they are related but distinct measures. Across pain and sadness trials, children used an average of 3.0 (SD = 1.27) distinct specific prosocial strategies; 10 children used all 5 strategies, and only 3 children used 0 strategies (i.e., did not use a prosocial strategy at any time during the laboratory visit). Children used an average of 1.78 (SD = 1.18) distinct strategies on the pain trial and 2.12 (SD = 1.03) on the sadness trial. The χ2 tests revealed that the two ratings of apologies and affection were significantly related; for example, children who used affection on the pain trial were also likely to use affection on the sadness trial. The two ratings of comforting statements, information seeking, and helping were not significantly associated. Thus, there was some evidence that individual children tended to employ the same specific prosocial strategies between pain and sadness trials.
Hypothesis Tests
Effects of fault and emotion on global prosocial behaviors
There was not a significant interaction between fault and emotion on toddlers’ global prosocial ratings, F(1, 78) = 0.15, p = .70, partial η2 = .002. There was not a significant main effect of fault, F(1, 78) = 1.15, p = .29, partial η2 = .014; children’s mean global prosocial rating was higher, but not significantly so, in the transgressor (M = 2.11; SE = .16) than bystander context (M = 1.87; SE = .16). There was a significant main effect of emotion, F(1, 78) = 11.31, p = .001, partial η2 = .13; toddlers were significantly more prosocial in response to sadness (M = 2.19; SE = .13) than pain (M = 1.79; SE = .12).
Effects of fault and emotion on specific prosocial strategies
Tables 1 and 2 present significant main effects of fault and emotion on specific prosocial strategies. There was not a significant interaction between fault and emotion on information seeking, Wald χ2(1) = 2.37, p = .12, or main effect of emotion, Wald χ2(1) = 2.74, p = .10. However, there was a main effect of fault, Wald χ2(1) = 6.25, p = .01; children were 3.41 (95% CI: 1.30, 8.92) times more likely to use information seeking as bystanders than transgressors. There was not a significant interaction between fault and emotion on affection, Wald χ2(1) = 3.81, p = .05. However, there was a significant main effect of emotion, Wald χ2(1) = 16.19, p ≤ .001; children were 8.55 (95% CI: 3.01, 24.32) times more likely to use affection in response to pain than sadness. There was also a significant main effect of fault, Wald χ2(1) = 4.77, p = .03; children were 4.08 (95% CI: 1.16, 14.45) times more likely to use affection as transgressors than bystanders.
Main Effects of Emotion on Prosocial Behavior (N = 86).
Note. SD = standard deviation; OR = odds ratio; CI = confidence interval. Item anchors for global prosocial variable: 0 = none; 4 = strong. Bold indicates significant results.
a This value reflects the significant interaction between fault and emotion on helping.
There was not a significant interaction between fault and emotion on apologies, Wald χ2(1) = 0.03, p = .87, or a main effect of emotion, Wald χ2(1) = 1.14, p = .29. However, there was a main effect of fault, Wald χ2(1) = 4.59, p = .03; children were 4.57 (95% CI: 1.14, 18.38) times more likely to apologize as transgressors than bystanders. Regarding comforting statements, there was not a significant interaction between fault and emotion, Wald χ2(1) = 3.55, p = .06, nor was there a main effect of emotion, Wald χ2(1) = 1.42, p = .23, or fault, Wald χ2(1) = 1.91, p = .17.
Interestingly, there was a significant interaction between fault and emotion on helping, Wald χ2(1) = 7.48, p = .006. There was a significant difference in helping depending on emotion for both transgressors, Wald χ2(1) = 20.88, p ≤ .001, and bystanders, Wald χ2(1) = 30.22, p ≤ .001; helping was more likely on the sadness than pain trial. There was also a significant difference in helping depending on fault on the sadness trial, Wald χ2(1) = 8.344, p = .004 but not the pain trial, Wald χ2(1) = 0.382, p = .65. Bystanders were 10.07 (95% CI: 2.10, 48.22) times more likely to use helping than transgressors on the sadness trial. In other words, children were most likely to use helping when responding to their parent’s sadness as bystanders.
Discussion
Building on recent literature demonstrating the complexity and multidimensional nature of prosocial responding in toddlerhood, this study examined whether two emotional contexts influenced toddlers’ mean levels of prosocial responding as well as the specific prosocial strategies they used to help their parent. Overall, toddlers were equivalently prosocial when they caused their parent’s distress as transgressors as when they witnessed the distress as bystanders. In contrast, toddlers engaged in more frequent and sophisticated prosocial behaviors in response to their parent’s sadness than pain. Toddlers also used different specific prosocial strategies depending on the emotional context of their parent’s distress—both when comparing bystander versus transgressor contexts and pain versus sadness contexts—and the two emotional contexts interacted to predict children’s use of instrumental helping.
First, we found that toddlers’ global prosocial behaviors did not differ significantly between transgressor and bystander contexts. Previous findings related to differences in prosocial responses between the two contexts are quite mixed. Most studies have found that toddlers were significantly more prosocial as bystanders (e.g., Dunn & Brown, 1994); however, most of these studies observed toddlers interacting naturalistically with peers and did not report whether children’s transgressions were intentional or unintentional. Thus, it is possible that toddlers who transgressed in these studies engaged in intentional transgressions that reflected aggression, which is consistently associated with less prosociality (e.g., Krahé & Möller, 2011). A smaller number of studies have found that children were more prosocial as transgressors (e.g., Demetriou & Hay, 2004), with one study finding that 3-year-olds, but not 2-year-olds, were more prosocial as transgressors (when the experimenter was the victim; Vaish et al., 2016). None of these studies compared children’s prosocial responses between bystander and transgressor contexts when the victim was a parent. Toddlers may be particularly motivated to use prosocial behaviors to repair the all-important parent–child relationship, which may help explain why 2- and 3-year-old toddlers in our study were somewhat—but not significantly—more prosocial as transgressors than as bystanders. Future studies that disentangle both unintentional from aggressive transgressions and toddlers’ prosocial responses to peers from responses to parents will be important in further comparing toddlers’ prosocial behaviors as bystanders versus transgressors.
Second, toddlers in our study were significantly more prosocial when parents expressed sadness than when they expressed pain. This result is consistent with one existing study that found that children were more prosocial to an experimenter’s sadness than pain and extends this finding to prosocial responses to parent distress (Bandstra et al., 2011). Future studies should attempt to determine why young children appear more prosocial to sadness than pain. An alternative explanation for our finding may be that children might perceive that one scenario reflects greater emotional need (alleviate pain) and one reflects greater instrumental need (fixing an item) and thus, our results might reflect that toddlers were more globally prosocial in response to instrumental than emotional need. Yet, the distress scripts in this study did not communicate that there was one correct way of solving the problem in either scenario, and indeed children responded in a variety of ways that included both instrumental helping and comforting behaviors in both scenarios. Future work should attempt to tease apart whether our effect reflects a difference in children’s prosociality based on the differing emotional content (pain vs. sadness) or possible differences in perceived needs (instrumental vs. emotional), perhaps by examining children’s responses to parents’ pain and sadness without any evident cause and/or accompanying items.
Despite evidence that distinct prosocial subtypes are associated with differing correlates and developmental trajectories (Hay & Cook, 2007), studies typically use only a global measure of prosocial responding that groups all prosocial behaviors together. In this study, individual toddlers who used a specific prosocial strategy in response to one parent emotion tended to use that same strategy in response to the other emotion, suggesting that as early as toddlerhood, children are developing their own go-to strategies for helping others. Most critically, we found that the emotional context of another’s distress influenced the specific prosocial strategies toddlers chose. Children were more likely to use affection as transgressors than bystanders, information seeking as bystanders than transgressors, and affection in response to pain than sadness. Moreover, all toddlers used greater instrumental helping in response to sadness than pain, replicating previous work (Bandstra et al., 2011), and this was particularly true among bystander children. Future studies should examine factors that may explain children’s use of differing prosocial strategies in distinct contexts, including mechanisms through which the contexts interact to predict helping behaviors. Children may socially learn specific strategies to use in particular distress situations through parental modeling; for example, toddlers may frequently witness parents respond to children’s pain with affection (e.g., “a kiss to make it feel better”). It is also possible that even this early in development, toddlers choose strategies that seem most likely to successfully reduce distress—for example, alleviating sadness by fixing the broken item, the cause of the sadness.
This study was limited primarily by the facts that half of the children watched a video before each trial related to the larger study, and that the children were assigned to a condition based on when they participated chronologically rather than being randomly assigned. The exact situation that caused pain and sadness was also different; although it would be ideal to hold this constant between pain and sadness trials, it is difficult to imagine a situation that could be used for both trials that would be equally believable. For instance, a parent expressing pain due to a broken item might confuse children, creating a different confound in that one trial would be more ecologically valid than the other. As previously discussed, the sadness trial included a broken item, which children might have perceived to be more of an instrumental need than that of the pain trial; thus, it is possible that greater rates of the instrumental helping strategy reflect this methodological difference. Yet, it should be noted that children’s helping was greatest when children were bystanders and responding to sadness; thus, there appear to be some context effects on helping that cannot be explained by mere methodological differences.
In sum, this study adds to findings that demonstrate the complexity of toddlers’ prosocial responding by establishing that two distinct emotional contexts influenced both toddlers’ mean levels of prosocial behaviors and the specific prosocial strategies they chose to help a distressed parent. Toddlers were overall more prosocial in response to their parent’s sadness than pain and were specifically likely to use affection both when they witnessed their parent’s pain and when they transgressed to hurt their parent, to seek information about their parent’s distress when they witnessed the distress as bystanders, and to use helping behaviors when they witnessed their parent’s sadness as bystanders.
Supplemental Material
Supplemental Material, Context_Supplement_ - Emotional contexts influence toddlers’ prosocial strategies
Supplemental Material, Context_Supplement_ for Emotional contexts influence toddlers’ prosocial strategies by Meghan Rose Donohue, Rebecca A. Williamson and Erin C. Tully in International Journal of Behavioral Development
Supplemental Material
Supplemental Material, JBD912007_supplement - Emotional contexts influence toddlers’ prosocial strategies
Supplemental Material, JBD912007_supplement for Emotional contexts influence toddlers’ prosocial strategies by Meghan Rose Donohue, Rebecca A. Williamson and Erin C. Tully in International Journal of Behavioral Development
Footnotes
Acknowledgment
The authors would like to thank the families who participated in the study and the research assistants who assisted with coding.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work of Dr. Donohue was supported by NIH Grant T32 MH100019 (PIs: Barch & Luby).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
