Satisfying personal preference or being considerate? The protagonist's preference matters in children's evaluation of social mindfulness

Abstract

Children around ages 5–6 begin valuing social mindfulness, where one leaves a diverse choice for others when selecting an item for themselves. This study investigated how Han Chinese 4‐ to 9‐year‐olds and adults may consider protagonists' preference (unknown, preferring the unique, or identical items) in their evaluations of socially mindful actions. Results show that, with age, children increasingly rate leaving a choice as nicer, both when the protagonist's preference is unknown and for the unique item. However, when the protagonist prefers identical items, neither children nor adults differentially evaluate between leaving a choice or not. Additionally, children become increasingly sensitive to the costs associated with preferences when evaluating mindful actions. They increasingly evaluated high‐cost actions (sacrificing one's preferred item) most favourably, followed by unknown‐cost actions, and low‐cost actions (when preference aligns with leaving a choice) least favourably. These findings highlighted children's growing ability to integrate complex understandings of preference and cost into social evaluations.

INTRODUCTION

Imagine you are waiting in line at a buffet. When it is your turn, there are three cakes left on the table: one strawberry‐flavoured and two lemon‐flavoured, and there is one other person waiting behind you. In this case, choosing the unique strawberry cake would limit options for the next person, while picking the lemon one respects the other person's freedom to choose. Such small, considerate actions reflect social mindfulness, defined as thinking about others' needs and respecting their choices (Sierksma et al., 2022; Van Doesum et al., 2013; Zhao et al., 2021). Social mindfulness is considered an indirect form of prosocial behaviour that conveys a prosocial intention − respecting others' autonomy and maximizing their opportunities to choose what they prefer (Zhao et al., 2021; Zhao et al., 2022). Adults' socially mindful behaviours have been linked to empathic concern (Van Doesum et al., 2013). Compared with more direct forms of prosocial behaviour (e.g., helping, sharing), social mindfulness is subtle and centres on preserving others' freedom of choice as a welfare concern.

Recent evidence shows that adults and 5‐ to 6‐year‐old children recognize socially mindful action as nice and understand the prosocial intention behind it (Sierksma et al., 2022; Van Doesum et al., 2013; Zhao et al., 2021). In most of these studies, it is assumed that the protagonist had no salient preference for the items. However, in real life, people usually have preferences and prosocial actions often involve balancing one's own desires with the needs of others. So, a question naturally emerges: how does knowledge of the protagonist's preference influence children's and adults' evaluation of the act of leaving a choice (or not)?

For example, in the buffet scenarios above, what if the first person prefers the unique strawberry cake over the non‐unique lemon cake? Then, leaving two different cakes for the next person would require sacrificing her own preference, making her act of leaving a choice more costly and arguably more prosocial. What if the first person prefers the identical lemon cake? Then, do we still consider her act of leaving a choice as prosocial? Perhaps she is just picking the cake she prefers, without thinking much about the next person's choice.

In the present study, we ask these questions as they pertain to children. How do children consider the protagonist's preference in their social evaluations of the act of leaving a choice for others? And how might such evaluations vary by age? Investigating these questions among children is particularly important because it can shed light on whether children's evaluations of socially mindful actions focus solely on the outcome (e.g., whether a choice is left) or whether they also flexibly incorporate mental state information, such as the protagonist's preference. Investigating these questions can reveal how children use their social‐cognitive capacities to integrate multiple pieces of information when evaluating subtle social actions. We propose that children's social evaluations and reasoning in this context may depend on several interrelated social‐cognitive capacities: their understanding of the value of diverse choice (Echelbarger & Gelman, 2017), their understanding of others' individual preferences (Zang et al., 2023), and their sensitivity to behavioural costs associated with actions (Liu & Spelke, 2017; Radovanovic et al., 2023; Zhao & Kushnir, 2023). Specifically, to understand and evaluate these scenarios, children must first recognize that leaving a choice for others is socially desirable. They must then integrate the protagonist's own preference into this evaluation to determine whether the act of leaving a choice is genuinely prosocial or merely self‐serving. Moreover, children's understanding of the costs involved is also important. They need to consider that leaving a choice for others may entail a higher personal cost for the protagonist if the protagonist prefers the unique item, compared to scenarios where the protagonist prefers the identical item.

The development of these aspects suggests that, by the late preschool years, children may already possess the prerequisite social‐cognitive capacities to understand the value of diverse choice and reason about preferences and cost (Echelbarger & Gelman, 2017; Liu & Spelke, 2017; Sierksma et al., 2022; Starmans & Bloom, 2016; Zhao et al., 2021; Zhao & Kushnir, 2023). However, it remains an open question how children may incorporate these understandings into their evaluations of the act of leaving a choice. Below, we review prior literature examining children's development of these social‐cognitive abilities and motivate the current hypotheses.

First, an increasing body of research shows that children across cultures appreciate autonomy and recognize the value of having the freedom to choose during early and middle childhood (Echelbarger & Gelman, 2017; Sierksma et al., 2022; Zhao et al., 2021). U.S. children around age 4 begin to prefer diverse options (e.g., two different toys) over identical ones, and they also provide varied options when selecting items for others, reflecting their understanding that having choices matters (Echelbarger & Gelman, 2017). Children aged 5–6 in Germany and Israel also understand the importance of scarce or unique items (Ferera et al., 2020; John et al., 2018). Consistent with this emerging appreciation for choice, research on social mindfulness further proves that, by ages 5–6, children across various cultures increasingly evaluate leaving a choice for others as nicer than limiting others' choice (the Netherlands: Sierksma et al., 2022; the U.S. and China: Zhao et al., 2021). Even when the story items are damaged (undesirable), recent evidence shows that 6‐year‐olds in China still view leaving a choice as preferable to restricting it (Li & Zhao, 2024). Together, these findings indicate that early in development, children value others' autonomy and understand that preserving options for others is nice.

Second, during early childhood, children also develop sophisticated abilities to infer others' preferences through social cues (Lucas et al., 2014). This is a key part of theory of mind (ToM) − the capacity to reason about others' mental states, including preferences, desires, beliefs and emotions (Wellman, 2014). Abundant research on ToM has shown that children from a very young age recognize that others' preferences may differ from their own (Repacholi & Gopnik, 1997), and use what others choose (Ma & Xu, 2011; Pesowski et al., 2016) and statistical sampling to infer preferences (Kushnir et al., 2010). Children also take preference into account in social evaluations. Even 10‐month‐old infants evaluate helping behaviour based on whether it is intentionally aligned with the beneficiary's preferences (Hamlin et al., 2013). Particularly relevant to the current study, recent research on social mindfulness indicates that children increasingly integrate the beneficiary's preferences into their social evaluations (Zang et al., 2023). When the beneficiary prefers the unique item, children as young as age 4 understand that it is nicer to leave the preferred item for the beneficiary. However, when the beneficiary prefers the identical items, younger children tend to judge that leaving a choice and leaving two identical items are equally nice, whereas children older than 6 years old tend to view leaving the two identical items the beneficiary preferred as the nicer act. These findings reveal children's developing ability to incorporate others' preferences into their evaluations of socially mindful behaviours.

However, this line of research has focused primarily on the beneficiary's preferences rather than the protagonist's. Recently, researchers have started to examine the role of the protagonist's preference. Chai et al. (2025) showed 4‐ and 6‐year‐old Chinese children videos of peers choosing a peanut from a plate with four peanuts and a lollipop, and manipulated whether the characters did so because they wanted to leave the lollipop for others or because they themselves preferred the peanut. They found that children who observed peers leaving the lollipop out of prosocial intent later shared more with others than those who observed peers acting based on their own preference. This research suggests that children do consider the protagonist's preference when observing socially mindful actions. However, in that study, the protagonist's intentions were made explicit (e.g., the peer character said ‘Because I like peanuts’), so children did not need to infer them. Thus, it remains unclear whether children can spontaneously integrate the protagonist's preference and their acts of leaving (or not leaving) a choice to infer underlying intentions and make social evaluations.

Considering the protagonist's preference in such inferences and evaluations may require reasoning about the personal costs involved for the protagonist. Recent research reveals that children are sensitive to costs incurred in actions in their social inferences and evaluations (Liu & Spelke, 2017; Starmans & Bloom, 2016; Zhao & Kushnir, 2023). Children use cost to make inferences about others' actions from quite young. Infants as young as 6 months expect agents to choose efficient actions and avoid physical cost (Liu & Spelke, 2017). During infancy and toddlerhood, children also flexibly consider the cost required when making their own prosocial decisions. For example, 18‐month infants' willingness to help carry a block is influenced by the physical cost (e.g., how heavy the block is) involved in the action (Sommerville et al., 2018). Toddlers are less willing to help when the task involves a higher cognitive cost, such as matching an egg with a hidden colour dot to the correct location (Radovanovic et al., 2024). When evaluating others' prosocial actions, between ages 4 and 9, children also increasingly recognize the value of incurring psychological costs to do the right thing, viewing sacrificing something one likes to help others as more praiseworthy than helping without having to forgo something they like (Zhao & Kushnir, 2023).

In summary, previous research has shown that during early childhood, children increasingly recognize that leaving a choice for others can be a prosocial act. They also begin to incorporate information about others' preferences into their social reasoning and evaluations. Moreover, children gradually understand that prosocial behaviours may involve personal costs, which in turn influences how they evaluate others' actions. However, no study has integrated these elements to examine how children use knowledge of the protagonists' preferences to infer and evaluate socially mindful behaviour.

Socially mindful behaviours offer an interesting case for studying whether and how children incorporate the protagonist's preferences into their evaluations. By examining cases in which the protagonist's preference is unknown, for the unique item, or for the identical items, and manipulating whether the protagonist leaves a choice or not, we can systematically investigate how children reason about scenarios where a person's behaviour aligns with their own preference or conflicts with it, whether the behaviour is considerate or inconsiderate of others' choices, and whether leaving a choice incurs a high or low cost.

Therefore, the present study investigates whether and how 4–9‐year‐old children and adults consider the protagonist's preferences when inferring and evaluating socially mindful and unmindful actions. We focused on this age range because children's understanding of social mindfulness develops between ages 4 and 6 (Sierksma et al., 2022; Zhao et al., 2021) and their considerations of preference and cost in social evaluations continue to develop through age 9 (Chai et al., 2025; Zang et al., 2023; Zhao & Kushnir, 2023).

We ask children to evaluate the acts of leaving a choice for others or leaving no choice, while manipulating the protagonist's preference information (unknown, prefers the unique item, or prefers the identical item). We hypothesize that with age, children increasingly take the protagonists' preference into account when evaluating socially mindful acts. Specifically, we examine two aspects of this hypothesis. First, we investigate how children infer and evaluate the act of leaving a choice versus not leaving a choice under each preference condition. When the protagonist's preference is unknown, based on previous research (e.g., Zhao et al., 2021; Zhao et al., 2022), we expect that with age, children will increasingly evaluate the act of leaving a choice as nicer than not leaving a choice. When the protagonist prefers the unique item—where taking an identical item and leaving a choice differs from the protagonist's preference and entails a higher cost—we also expect children to increasingly evaluate the act of leaving a choice as nicer than not leaving a choice. In contrast, when the protagonist prefers the identical items, taking an identical item and leaving a choice aligns with the protagonist's preference and could therefore reflect a self‐serving rather than prosocial intent. Accordingly, we do not have a clear prediction as to whether children and adults will evaluate leaving a choice differently from not leaving a choice.

Second, we examine how children's evaluations of the same act (e.g., leaving a choice) vary across different preference conditions. We hypothesize that, as children's understanding of cost develops, they may increasingly view socially mindful acts in which someone forgoes a preferred item to leave a choice for others (when the cost is highest) as nicer than similar acts in which the protagonist's preference is unknown (when the cost is unclear) and even nicer than acts in which the protagonist prefers the identical items (when the cost is lowest).

We also examine how adults and children evaluate socially unmindful acts (i.e., leaving no choice), and whether this evaluation varies depending on the protagonist's preference. Previous research suggests that when the protagonist's preference is unknown, young children (around age 4) tend to evaluate leaving no choice as ‘nice’, but with age, they gradually realize that this act limits others' choices, and their niceness evaluations decrease towards the ‘neutral’ level. Notably, neither children nor adults typically view the act of not leaving a choice as ‘antisocial’ or ‘mean’, perhaps because the beneficiary still has two items to choose from (Zang et al., 2023). We will examine whether evaluations of leaving no choice differ when the protagonist's preference is unknown versus when the protagonist prefers the unique or the identical item. Given the lack of clear prior evidence, we do not have strong predictions for this act, and will explore participants' intention inferences to help explain their evaluations.

METHODS

Participants

Informed by previous research on similar topics (Zang et al., 2023; Zhao et al., 2021), we set our adult sample size as 30. Our final adult sample included 30 participants (age range: 22–28 years, M_age = 24.2, SD = 1.61; 27 female; all Han Chinese) recruited via social media in China.

For the children's sample size, we conducted an a priori power analysis for a repeated‐measures ANOVA. Prior research on similar topics (Zang et al., 2023) reported medium‐to‐large effects (η ² _p = 0.08–0.25, corresponding to f = 0.30–0.58). Because these effects are larger than typical ones in developmental psychology, we decided to use a commonly used estimate of a small‐to‐medium effect size (f = 0.15) for the power analysis. The analysis indicated that a minimum of 75 participants would be required to detect effects of f = 0.15 in a 2 × 3 within‐subjects design, with power = 0.95 and α = .05. To ensure model stability and an even distribution of participants across age groups, we ultimately recruited 95 typically developing Han children aged 4–9 years (4.15–9.85 years, M_age = 6.90, SD = 1.75, 46 boys) from an online lab database (N = 29) and preschools and primary schools in Shanghai (N = 66). The study was approved by the University Committee on Human Research Protection at East China Normal University. All the parents provided written consent and the children provided oral consent. The study was not preregistered. However, the study design and sample size were determined prior to data collection and were not altered thereafter. The study materials and the data are made publicly available at the Open Science Framework (https://osf.io/9a4rg).

Materials and procedures

All child participants were tested individually in preschools or online by VooV meeting (a video conferencing software similar to Zoom) by trained experimenters. Video conferencing and screen sharing were used. Parents were typically present when using online meeting, but were politely asked not to interfere. Adult participants participated via Qualtrics, where they read scenarios and answered questions that were identical to those presented to the children (see the story setup in Table 1 and Figure 1, and the detailed script in the Data S1).

TABLE 1.

Design of the study.

	Story 1	Story 2	Story 3	Story 4	Story 5	Story 6
Knowledge of the protagonist's preference	Unknown	Unknown	Preferring unique item	Preferring unique item	Preferring identical item	Preferring identical item
Choice left	Leaving a choice	leaving no choice	Leaving a choice	leaving no choice	Leaving a choice	Leaving no choice

FIGURE 1.

Story setup.

Before the main task, children were first trained to use a 1–20 scale to answer questions on how ‘nice’ someone is to a friend (adapted from Bass et al., 2022, see Figure 2). We presented children with a colourful 1–20 scale, explaining that 1 meant ‘very bad’ and 20 meant ‘very nice,’ and that larger numbers with greener colours represented being ‘nicer’ to a friend, whereas smaller numbers with redder colours represented being ‘less nice’ to a friend. Then, we asked each child to name a friend and say (or point to) a number on the scale to show how nice they were to that friend.

FIGURE 2.

The 1–20 scale measuring children's evaluations of the niceness of the protagonist.

The main task employs a 2 (Choice left: leaving a choice, leaving no choice) × 3 (Knowledge of the protagonist's preference: Unknown, preferring the unique item, preferring the identical item) within‐subjects design. Children were shown six stories on PowerPoint, each depicting two child characters selecting fruits during snack time at school. Note that the characters were different in each story. To distinguish them, each character was referred to by a different name (e.g., Little A, Little B, Little C, etc). In each story, children were told that during snack time, two characters were waiting in line to pick a snack. The character standing in front (i.e., the protagonist) was always asked to choose from three fruits: one was unique, and the other two were identical (e.g., one apple and two bananas). We manipulated the protagonist's preference information (preference unknown, preferring the unique item, preferring the identical item) and whether the protagonist left a choice (leaving a choice, leaving no choice). The two stories with unknown preferences were presented before the four stories with known preferences to avoid potential influence of preference information on children's responses in the stories with unknown preferences. The orders of the first two stories and the remaining four stories were counterbalanced across participants, which resulted in a total of eight different presentation orders.

After each story, the experimenter first asked memory‐check questions to ensure the children were paying attention to and understood the story: ‘What fruit does [the protagonist's name] like?’ (omitted for stories with unknown preferences), ‘What did she/he choose?’, ‘What did she/he leave for [the next character's name] to choose?’ If the children answered any of the memory check questions incorrectly, the experimenter would repeat the story and ask the same question again. All children answered the memory check questions correctly within three attempts.

After that, the experimenter asked the main measure question (niceness question): ‘How nice do you think [the protagonist's name] is to [the next character's name]?’ We then asked children to infer the intention of the first character (intention question): ‘Why did [the protagonist's name] choose an apple?’

Coding

We analysed qualitative responses to the intention question (‘Why did [the protagonist's name] choose an apple?’). Given our primary focus on whether children inferred prosocial or self‐serving intention from each character's action, we coded responses into three categories: prosocial intention, self‐serving intention and other. The ‘prosocial intention’ category included inferences that the protagonist wanted to leave a choice for the beneficiary (e.g., ‘She left an apple and a banana for her to choose’) or inferred and considered the beneficiary's preference (‘She [the protagonist] thinks [the beneficiary] prefers bananas’). The ‘self‐serving intention’ category mainly included inferences that the protagonist acted to satisfy their own preference (e.g., ‘She [the protagonist] prefers apples’). All other responses were coded as ‘other’, which included negative intentions (e.g., ‘She did not want to leave a choice for her,’ ‘She is not nice to her’) and uninformative responses (e.g., ‘I don't know,’ ‘She just wants to’). Note that a single response could be coded into multiple categories if applicable. See Table 2 for details of coding categories and examples. The first and second authors independently coded all responses. The inter‐coder reliability, measured using Cohen's kappa, was 0.98. Any disagreements between the coders were resolved through discussion.

TABLE 2.

Details of coding categories and examples of how children infer the intention of the protagonist (intention question).

Coding category	Content	Examples
Prosocial intention	Leaving a choice for the beneficiary; Inferring and considering the beneficiary's preferences	‘She left an apple and a banana for her to choose’ or ‘She (the protagonist) thinks (the beneficiary) prefers bananas’
Self‐serving intention	Satisfying the protagonist's own preference	‘She (the protagonist) prefers apples’
Other	Negative intention/No response/I don't know/Unrelated responses	‘She didn't want to leave a choice for her’ ‘I have no idea’

RESULTS

Preliminary analysis revealed no effect of test modality (i.e., online or in person) or story orders (8 orders, ps > .073) on the dependent measure, thus we did not consider these in the subsequent analyses.

Adults' niceness evaluation

We performed a linear mixed‐effects model predicting adults' niceness evaluation of the characters (ranging from 1 to 20) as a function of choice left (leaving a choice vs. leaving no choice), protagonist's preference (unknown, preferring the unique item, preferring the identical item), and their interaction with a random intercept for participant (see Figure 3). The model revealed a significant main effect of choice left (B = −3.67, SE = 0.88, 95% CI [−5.40, −1.94], p < .001) that adults' niceness evaluations for the act of leaving a choice were in general higher than for the act of leaving no choice. There was also a significant main effect of protagonist's preference, that adults' niceness evaluations were higher in the condition where the protagonist preferred the unique item than when the protagonist's preference was unknown (B = 3.73, SE = 0.88, 95% CI [2.00, 5.46], p < .001), and when the protagonist preferred the identical item (B = 6.00, SE = 0.88, 95% CI [4.27, 7.73], p < .001). Adults' niceness evaluations were also higher in the condition where the protagonist's preference was unknown than when the protagonist preferred the identical item (B = 2.27, SE = 0.88, 95% CI [0.54, 4.00], p = .011). Importantly, these main effects were qualified by a significant interaction between choice left and protagonist's preference (χ ² (2, N = 30) = 31.81, p < .001).

FIGURE 3.

Adults' niceness evaluation split by choice left (leaving a choice, leaving no choice) and protagonist's preference (unknown, preferring the unique item, preferring the identical item). Each point in the figure represents the actual response of each participant. Error bars represent 95% confidence intervals. ***p < .001, *p < .05.

Pairwise comparison showed that when the protagonist's preference was unknown, adults rated the protagonist who left a choice (M = 13.63, SD = 3.62) significantly higher than the protagonist who did not leave a choice (M = 9.97, SD = 2.13), B = 3.67, SE = 0.88, 95% CI [1.92, 5.41], p < .001. When knowing that the protagonist preferred the unique item, adults also rated the protagonist leaving a choice (M = 17.37, SD = 3.17) significantly higher than the protagonist leaving no choice (M = 9.37, SD = 2.77), B = 8.00, SE = 0.88, 95% CI [6.25, 9.75], p < .001. However, when knowing that the protagonist preferred the identical item, there was no significant difference in adults' evaluations of the protagonist who left a choice (M = 11.37, SD = 3.09) and the protagonist who did not leave a choice (M = 10.63, SD = 6.32), p = .408.

We further analysed the interaction from another perspective by comparing adults' evaluations of the protagonists leaving a choice across different preference conditions. Adults rated the protagonist leaving a choice significantly higher when knowing that the protagonist preferred the unique item (M = 17.37, SD = 3.17) than when the protagonist's preference was unknown (M = 13.63, SD = 3.62; B = 3.73, SE = 0.88, 95% CI [1.64, 5.83], p < .001) and when the protagonist preferred the identical item (M = 11.37, SD = 3.09; B = 6.00, SE = 0.88, 95% CI [3.91, 8.09], p < .001). They also rated the protagonist leaving a choice significantly higher when the preference was unknown than when knowing that the protagonist preferred the identical item, B = 2.27, SE = 0.88, 95% CI [0.18, 4.36], p = .030. However, for adults' evaluations of the protagonists who did not leave a choice, there was no significant difference among the preference conditions (all ps > .326).

We then ran one‐sample t‐tests comparing to the neutral point (10.5) for each story. Adults rated the protagonist leaving a choice as significantly nicer than the neutral point both when the preference was unknown (M = 13.63, t(29) = 4.75, 95% CI [1.78, 4.48], p < .001) and when the protagonists preferred the unique item (M = 17.37, t(29) = 11.87, 95% CI [5.68, 8.05], p < .001). When the protagonist preferred the identical item, adults rated leaving a choice as not significantly different from the neutral point (M = 11.37, t(29) = 1.54, p = .135). When the protagonist did not leave a choice, adults rated the act as less nice than the neutral point when the protagonist preferred the unique item (M = 9.37, t(29) = −2.24, 95% CI [−2.17, −0.10], p = .033). Ratings did not significantly differ from the neutral point when the protagonist's preference was unknown (p = .180) or when the protagonist preferred the identical item (p = .909).

Children's niceness evaluation

We then analysed children's responses to the niceness question (see Figure 4). We performed a linear mixed‐effects model predicting children's niceness evaluations of the characters (ranging from 1 to 20) as a function of choice left (leaving a choice vs. leaving no choice), protagonist's preference (unknown, preferring the unique item, preferring the identical item), age (continuous standardized predictor) and their interactions, with a random intercept for participant. The three‐way interaction among choice left, protagonist's preference, and age was trending but not significant, χ ² (2, N = 95) = 4.83, p = .089, η ² _p  = 0.01. Therefore, this three‐way interaction term was dropped from the model. The final model revealed a significant main effect of choice left (B = −8.71, SE = 1.11, 95% CI [−10.89, −6.53], p < .001) and a significant main effect of protagonist's preference, such that children's niceness evaluations were higher in the condition where the protagonist preferred the unique item than in the two other preference conditions (ps < .007). The main effect of age was not significant (p = .154). Importantly, there was a significant interaction between protagonist's preference and choice left (χ ² (2, N = 95) = 61.51, p < .001), a significant interaction between choice left and age (χ ² (1, N = 95) = 12.68, p < .001) and a significant interaction between protagonist's preference and age (χ ² (2, N = 95) = 7.70, p = .021).

FIGURE 4.

Children's age (continuous) predicting average niceness evaluation varying by protagonist's preference (unknown vs. preferring the unique item versus preferring the identical item) and choice left (leaving a choice vs. leaving no choice). Each point in the figure represents the actual response of each participant. The lines represent linear fits, and the shaded areas indicate the 95% confidence intervals. The black square indicates the age at which children's niceness evaluations of leaving a choice versus leaving no choice begin to differ significantly.

We then further examine the interactive effects. As we were interested in how children's evaluations of the protagonists' acts of leaving versus not leaving a choice varied across preference conditions, we fit separate linear mixed‐effects models in each preference condition predicting children's niceness evaluation of the characters (1–20) as a function of choice left, age (continuous standardized predictor), and their interactions with a random intercept for participant.

When the protagonist's preference was unknown, the results revealed a significant main effect of choice left (B = −9.18, SE = 1.89, 95% CI [−12.89, −5.47], p < .001) and no significant main effect of age (p = .313). Importantly, there was a significant interaction between choice left and age (B = −0.88, SE = 0.42, 95% CI [−1.70, −0.06], p = .038). Simple slope analysis showed that children's rating of the protagonist leaving no choice significantly decreased with age (B = −1.21, SE = 0.32, p < .001), while their rating of the protagonist leaving a choice did not significantly vary by age (p = .313). A Johnson‐Neyman analysis further revealed that when the protagonist's preference was unknown, children started to rate leaving a choice as significantly nicer than leaving no choice across all ages tested (beginning at 4.2 years).

When knowing that the protagonist preferred the unique item, the model revealed a significant main effect of choice left (B = −11.86, SE = 1.84, 95% CI [−15.46, −8.25], p < .001) and no significant main effect of age (p = .085). Importantly, there was a significant interaction between choice left and age (B = −1.28, SE = 0.41, 95% CI [−2.08, −0.48], p = .002). Simple slope analysis showed that when knowing the protagonist preferred the unique item, children's rating of the protagonist leaving no choice significantly decreased with age (B = −0.74, SE = 0.31, p = .016), while their evaluations of the act of leaving a choice tended to increase with age but were not significant (B = 0.53, SE = 0.31, p = .085). A Johnson‐Neyman analysis indicated that when the protagonist preferred the unique item, children rated leaving a choice as significantly nicer than leaving no choice across all ages tested (beginning at 4.2 years).

When knowing that the protagonist preferred the identical item, there was no significant interaction between choice left and age (p = .722), so the interaction term was dropped. The final model revealed a significant main effect of age (B = −0.70, SE = 0.23, 95% CI [−1.14, −0.26], p = .003), suggesting that with age, children's niceness evaluations decreased. We did not find a significant main effect of choice left (p = .550).

As we were also interested in whether children's evaluations of the same act (e.g., leaving a choice) varied by preference conditions, we then conducted two separate linear mixed‐effects models on children's niceness evaluation (one for evaluations of protagonists leaving a choice, and one for evaluations of protagonists leaving no choice) with protagonist's preference (preference unknown, preferring the unique item, preferring the identical item) and age (continuous standardized predictor) as predictors. For evaluations of the protagonists leaving a choice, there was a significant main effect of the protagonist's preference, such that children's evaluations were significantly higher when the protagonist preferred the unique item than when the protagonist's preference was unknown (B = 4.28, SE = 1.20, 95% CI [1.92, 6.64], p < .001), and evaluations when protagonist's preference was unknown were higher than when the protagonist preferred the identical item (B = 3.81, SE = 1.20, 95% CI [1.45, 6.16], p = .002). There was no significant main effect of age (p = .177). Importantly, there was a significant interaction between protagonist's preference and age (χ ² (2, N = 95) = 19.43, p < .001). We then used the Johnson–Neyman procedure to identify the ages at which children's niceness evaluations for protagonists who left a choice became significantly different across preference conditions (see Figure 5). We found that, starting at 4.8 years old, children rated the protagonist whose preference was unknown as significantly nicer than the protagonist who preferred the identical item. Starting at 5.4 years old, children rated the protagonist who preferred the unique item as significantly nicer than the protagonist who preferred the identical item. Starting at 7.5 years old, children rated the protagonist who preferred the unique item as significantly nicer than the protagonist whose preference was unknown.

FIGURE 5.

Children's niceness evaluations (1–20) of leaving a choice (left) and leaving no choice (right) split by protagonist's preference as a function of age. Each point in the figure represents the actual response of each participant. The lines represent linear fits, and the shaded areas indicate the 95% confidence intervals. The black squares mark the specific ages at which significant differences emerged in pairwise comparisons between different conditions of the protagonist's preference.

We conducted the same analysis for the evaluations of the protagonists leaving no choice. There was no significant interaction between the protagonist's preference and age (p = .364), so the interaction term was dropped. The final model revealed a significant main effect of age (B = −0.91, SE = 0.30, 95% CI [−1.50, −0.33], p = .003), indicating that children's evaluations of the acts of leaving no choice significantly decreased with age. The main effect of the protagonist's preference was also significant: that children's evaluations of leaving no choice were significantly higher when the protagonists preferred the identical item than when the preference was unknown (B = 3.22, SE = 0.65, 95% CI [1.94, 4.51], p < .001) and when the protagonists preferred the unique item (B = 3.56, SE = 0.65, 95% CI [2.28, 4.84], p < .001). There was no significant difference between when the preference was unknown and when the protagonists preferred the unique item (p = .603).

Intention question

We explored how participants inferred the protagonist's intention (i.e., the intention question: ‘Why did [the protagonist's name] choose an apple?’). The percentages of children's explanations categorized under each condition can be seen in Table 3. Overall, the findings indicate that children's inferences about the protagonist's intention varied depending on whether a choice was left and the knowledge of the protagonist's preference, aligning with the patterns observed in their responses to the niceness question.

TABLE 3.

Percentage of children's explanations that fell into each category (see the coding category in Table 2) in each condition, with correlation with age shown in brackets.

Percentage (correlation with age in month)	The protagonist leaving a choice			The protagonist leaving no choice
	Prosocial intention	Self‐serving intention	Other	Prosocial intention	Self‐serving intention	Other
Preference unknown	50.5% (.57***)	24.2% (−.20)	14.2%	4.2% (.09)	53.7% (.24*)	21.2%
Protagonist prefers unique item	77.9% (.53***)	0%	32.6%	3.2% (.11)	80% (.07)	16.8%
Protagonist prefers identical item	26.3% (.19)	75.8% (.09)	11.6%	52.6% (.06)	7.4% (−.01)	45.3%

Note: Shading indicates the most prevalent category in each condition.

^* p < .05.

^*** p < .001.

When the preference was unknown, the most prevalent explanation (50.5%) children provided for the protagonist leaving a choice referred to prosocial intentions (e.g., ‘She wanted to let her [the beneficiary] choose between an apple and a banana’). This type of explanation increased with age (r = .57, p < .001). Conversely, the most prevalent explanation (53.7%) for the protagonist leaving no choice referred to the self‐serving intention (e.g., ‘She [the protagonist] likes apples’). This type of explanation also increased with age (r = .24, p = .018).

When it was known that the protagonist preferred the unique item, the most prevalent explanation (77.9%) children provided for the protagonist leaving a choice referred to prosocial intentions (e.g., ‘She wanted to give her [the beneficiary] more choices’). This type of explanation increased with age (r = .53, p < .001). In contrast, the majority of explanations (80%) for the protagonist's leaving no choice referred to the self‐serving intention, such as satisfying the protagonist's own preference (e.g., ‘She [the protagonist] likes apples’).

When it was known that the protagonist preferred the identical item, the most prevalent explanation (75.8%) children provided for the protagonist leaving a choice referred to self‐serving intention to satisfy their own preference (e.g., ‘She [the protagonist] likes apples’). Additionally, 26.3% of the explanations referred to prosocial intentions (e.g., ‘She wanted to leave two fruits for her [the beneficiary] to choose from’). This suggests that under this context, most children inferred that the protagonist chose the identical (leaving a choice) to satisfy their own preference, while a smaller proportion of children believed the act was motivated by a prosocial intention.

For the protagonist preferring the identical items but leaving no choice, the most prevalent explanation (52.6%) referred to prosocial intentions (e.g., ‘Maybe she [the beneficiary] also likes bananas, so she [the protagonist] left two bananas for her [the beneficiary]’). In other words, in this context, children inferred that the beneficiary shared the same preference as the protagonist, attempting to rationalize the behaviour and infer prosocial intentions. Given that a significant proportion of children's intention inferences in this context were categorized as ‘Other’, we further refined the coding of this category by adding a new subcategory (negative intention). We found that 16.1% of children's responses indicated that the protagonist intentionally harmed the beneficiary's interests (e.g., ‘She [the protagonist] didn't want to leave two fruits for her [the beneficiary]’; ‘She did it on purpose’; ‘She [the protagonist] was not nice to her [the beneficiary]’). This type of explanation also increased with age (r = .31, p = .003), reflecting that, as children grew older, more of them inferred negative intentions from this behaviour.

DISCUSSION

Recent studies on social mindfulness have found that adults and children by 5–6 understand that leaving a diverse choice for others is nicer than leaving no choice (Sierksma et al., 2022; Van Doesum et al., 2013; Zang et al., 2023; Zhao et al., 2021). However, previous research has only examined cases where the preference of the protagonist was unknown. The current study examines whether the protagonist's preferences influence adults' and children's evaluations of socially mindful and unmindful actions. Our findings show that with age, children increasingly take the protagonist's preference into account when evaluating the act of leaving a choice.

We found that adults evaluated the act of leaving a choice as nicer than the act of leaving no choice, both when the protagonist's preference was unknown and when the protagonist preferred the unique item. However, when the protagonist preferred the identical item, adults evaluated the act of leaving a choice and the act of leaving no choice similarly. Moreover, adults' evaluations of the act of leaving a choice varied by the protagonist's preference. They evaluated the act of leaving a choice as the nicest when the protagonist preferred the unique item, followed by when the protagonist's preference was unknown, and the least nice when the protagonist preferred the identical item. In contrast, adults' evaluations of the act of leaving no choice did not differ based on the protagonist's preference.

Children also gradually develop into a pattern similar to that of adults. When the protagonist's preference was unknown, with age, children increasingly evaluated the act of leaving a choice as nicer than the act of leaving no choice. When the protagonist preferred the unique item, with age, children also increasingly evaluated the act of leaving a choice as nicer than leaving no choice. However, when the protagonist preferred the identical item, children across ages evaluated the act of leaving a choice as similar to the act of leaving no choice.

We further found that, with age, children increasingly evaluated the act of leaving a choice differently depending on the protagonist's preference. With age, they gradually evaluated leaving a choice as the nicest when the protagonist preferred the unique item (i.e., incurring a high cost), followed by when the preference was unknown (i.e., unknown cost), and least nice (though still above chance) when the protagonist preferred the identical item (i.e., incurring a low cost). By around 7.5 years of age, children reliably distinguished among these three conditions. These findings together reveal that, during early and middle childhood, children gradually incorporate their understandings of preference and cost to flexibly make evaluations on socially mindful actions.

Children also integrated knowledge of the protagonist's preference into their inferences about intentions underlying the act of leaving a choice. When inferring the intentions of the protagonists who left a choice, children attributed the highest rate of prosocial intention (77.9%) when the protagonist preferred the unique item, higher than when the preference was unknown (50.5%). This may be because the preference information made the prosocial intent more apparent—she must be truly prosocial to sacrifice her preferred item to leave a choice for others. In contrast, when the protagonist's preference was unknown, children had less evidence to infer her intentions. Children attributed the lowest rate of prosocial intention (26.3%) when the protagonist preferred the identical items, perhaps reasoning that she simply took what she liked without much consideration for others. These inference responses are consistent with their niceness ratings.

Children's inferences about the intentions underlying socially unmindful acts (i.e., the act of leaving no choice) appear to be more complex. When the protagonist's preference was unknown, most children (53.7%) inferred self‐serving intentions. When the protagonist preferred the unique item and left no choice, an even higher percentage of children (80%) inferred self‐serving intentions, suggesting that this preference information made the self‐serving motive more salient. Interestingly, neither children's nor adults' niceness ratings for these two characters seemed to differ. We believe this is because people tend to be lenient towards unmindful acts (Van Doesum et al., 2013; Van Lange & Van Doesum, 2015), often considering them to be neutral rather than ‘mean’ (Zang et al., 2023). Even when the act was based on a self‐serving intention, children and adults still tended to consider it as neutral.

When the protagonist preferred the identical items but instead took the unique item and left no choice, children's inferences became more complex. First, a substantial proportion of children (52.6%) inferred prosocial intention (e.g., assuming the beneficiary shared the protagonist's preference: ‘Maybe she [the beneficiary] also likes bananas, so she [the protagonist] left two bananas for her [the beneficiary]’). These responses seem to reflect children's tendency to try to rationalize seemingly counterintuitive behaviour, which may have contributed to their more favourable niceness evaluations. This tendency aligns with previous work showing that young children tend to show a ‘positive bias’ in inferring others' intentions (Sato & Wakebe, 2014). Second, some children (16.1%) inferred negative intentions (e.g., that the protagonist did not want to leave a choice for the beneficiary), and this response increased with age. Such inferences likely contributed to more negative evaluations and may explain why evaluations of this character declined with age. Taken together, these opposing inferences likely contributed to the pattern observed in the average niceness ratings, such that younger children rated the behaviour above chance, and with age, these ratings declined towards the neutral point.

Our findings also reveal an interesting asymmetry in evaluations of socially mindful versus unmindful actions. When the protagonist preferred the unique item and still left a choice, both children and adults clearly inferred prosocial intentions and rated the action as particularly nice. In contrast, evaluations of the act of leaving no choice varied less across preference conditions. When the protagonist preferred the identical item but did not leave a choice, neither children nor adults showed clearly negative evaluations; instead, as noted above, participants' intention inferences were mixed, including both prosocial and negative interpretations. This pattern is broadly consistent with previous findings showing that socially unmindful actions are often judged as neutral rather than mean when mental‐state cues (e.g., preference, intentions) are unclear (e.g., Van Doesum et al., 2013; Zang et al., 2023). Both adults and children often interpret leaving no choice as a result of inattention or unawareness of the consequences for the next person, rather than as an intentional effort to restrict others' choice. Future research is needed to further clarify why children's and adults' evaluations of socially unmindful actions did not vary across preference conditions.

The current findings add to a growing body of work showing that children's evaluation of social mindfulness is flexible. Recent evidence suggests that children do not simply rely on the outcome (leaving a choice or not for the beneficiary) when evaluating social mindfulness. Many factors, such as the desirability of the story item (Li & Zhao, 2024), the knowledge of the beneficiary's preference (Zang et al., 2023), reputation concern (Wang & Zhao, 2024), friendship (Koçyiğit‐Sarı et al., 2025), and the group context (Tao et al., 2024) between the protagonist and the beneficiary could influence children's and adolescents' evaluation and behaviour of social mindfulness. Our study advanced this literature by revealing how the social knowledge of the protagonist's preference influences children's evaluation.

Our findings reveal that, with age, children develop an increasingly mature understanding of preferences when evaluating social mindfulness. These findings are consistent with and extend recent research on the consideration of preference on social mindfulness. For example, Zang et al. (2023) found that during the ages 4–10, children increasingly consider the beneficiary's preferences in evaluating the act of leaving a choice. Another recent study (Chai et al., 2025) found that observing socially mindful actions driven by different intentions influenced 6‐year‐olds' own sharing behaviours. Children shared more tokens after observing a socially mindful action with explicit prosocial intention (e.g., I want to leave a lollipop [unique] for the last child), compared to a similar action that was driven by personal preferences (e.g., ‘I like peanuts [identical]’). The current study further found that when intentions were not directly provided, children were able to infer the protagonist's intentions based on the protagonist's preference and actions (leaving a choice vs. leaving no choice) and made corresponding social evaluations.

The current study also reveals children's outcome‐to‐intent developmental trajectory in evaluations of social mindfulness. This developmental shift has been documented in children's evaluations of prosocial and antisocial actions: younger children focus more on outcome but gradually come to consider intentions, evaluating the same act differently depending on the underlying intentions (Cushman et al., 2013; Margoni & Surian, 2017; Nobes et al., 2009; Wainryb et al., 2005). The present study extends this research by demonstrating a similar shift in evaluations of social mindfulness. Younger children tend to evaluate acts of leaving a choice similarly across different preference conditions, whereas with age, they gradually differentiate the same acts based on the protagonist's preference information.

Relevantly, this work also adds to previous work on the role of cost in social inference and evaluation (Finiasz et al., 2025; Jara‐Ettinger et al., 2015; Liu et al., 2017; Zhao & Kushnir, 2023). Our findings reveal that, between 5–7 years old, children can increasingly differentiate the subtle differences in personal cost incurred in the act of leaving a choice depending on different information about the protagonist's preference. This is largely consistent with recent work showing that children gradually differentiate high‐cost prosocial actions versus low‐cost prosocial actions (Radovanovic et al., 2023; Zhao & Kushnir, 2023). By focusing on a subtler form of prosocial action (i.e., socially mindful action) and a more indirect form of cost (caused by the protagonist's preference), our study reveals that children's understanding of cost and their ability to evaluate prosocial behaviours are quite flexible and complex. In the current study, we use social evaluation as the first step; future work could also examine whether children's own socially mindful behaviours may vary by cost.

Previous studies indicate that factors such as need and preference influence how children evaluate others' direct prosocial behaviours (Hamlin et al., 2013; Heyman et al., 2014; Sierksma et al., 2014). Extending this literature, our study demonstrates that even with indirect prosocial behaviours like social mindfulness, children also rely on information about preference to guide their social inferences and evaluations. Future work could examine both forms of prosociality within the same paradigm to determine whether there are age differences in the consideration of preference across these contexts.

We have conceptualized social mindfulness as an indirect form of prosocial behaviour that conveys prosocial intentions toward others. However, it is possible that adults' and children's evaluations of such behaviour as ‘nice’ may stem from consideration of autonomy‐support norms rather than purely prosocial concern. Research with adults found that social mindfulness is associated with prosocial value orientation and empathy concern, and is also grounded in self‐determination theory (Van Doesum et al., 2013). Specifically, a protagonist may ‘see’ the beneficiary's needs and wishes, and leave a diverse choice to convey prosocial concern; at the same time, they may also want to respect the beneficiary's decision‐making autonomy. These two motivations−prosocial concern to promote others' welfare and sensitivity to others' autonomy−are conceptually distinct, yet they can be intertwined in socially considerate actions. Future work could further explore how children weigh prosocial concern versus respect for others' autonomy within the context of social mindfulness.

The current study also has several limitations. First, we did not counterbalance the assignment of unique and identical items. Participants' preferences for certain fruits may have introduced item‐specific biases that are independent of the experimental manipulation. Future studies should counterbalance item identity to minimize the potential effects of individual preferences. Second, our findings are based on a sample of Han Chinese children and should be interpreted within this specific cultural context. Future work could examine this question in other cultures to assess the generalizability of the findings.

CONCLUSION

The current study examines how knowledge of the protagonist's preference influences adults' and 4‐ to 9‐year‐old Chinese children's social evaluation of leaving a choice or not. Results show that children increasingly consider the protagonist's preference in their evaluations. With age, children increasingly evaluate the act of leaving a choice as nicer than leaving no choice when the preference is unknown or when the protagonist prefers the unique item (that is, leaving a choice requires forgoing the preferred item). However, children evaluate leaving a choice and leaving no choice similarly when the protagonist prefers the identical item. Furthermore, our findings suggest that, with age, children increasingly consider the subtle cost associated with different preference conditions when evaluating socially mindful actions. They increasingly evaluate the act of giving up the preferred item to leave a choice as the nicest, followed by when preference is unknown, and rate it least nice when the protagonist prefers the identical item, for which leaving a choice is relatively low‐cost. These findings indicate that with age, children gradually and flexibly incorporate understandings of preference and cost into their social evaluations.

AUTHOR CONTRIBUTIONS

Dandan Li: Writing – original draft; investigation; formal analysis; data curation; conceptualization. Lu Zang: Writing – review and editing; formal analysis. Xin Zhao: Writing – review and editing; supervision; project administration; methodology; funding acquisition; conceptualization.

CONFLICT OF INTEREST STATEMENT

None.

Supporting information

Figure S1.

DATA AVAILABILITY STATEMENT

All study materials and data information have been made publicly available via the Open Science Framework (OSF) and can be accessed at https://osf.io/9a4rg/?view_only=f5496893a92949569b65ac2b38585d3a.