Common and distinct equity preferences in children and adults

Abstract

Fairness plays a crucial role in children's social life and has garnered considerable attention. However, previous research and theories primarily examined the development of children's fairness behaviors in the conflict between self-interest motivation and fairness-complying motivation, neglecting the influence of advantage-seeking motivation. Moreover, despite the well-established role of gain/loss frame in human decision-making, it remains largely unclear whether the framing effect modulates fairness behaviors in children. It was hypothesized that children would exhibit advantage-seeking motivation resulting in more selfish behaviors in the loss context. To examine the hypothesis, we combined an adapted dictator game and computational modeling to investigate various motivations underlying fairness behaviors of children in both loss and gain contexts and to explore the developmental directions by contrasting children and adults. In addition, the current design enabled the dissociation between fairness knowledge and behaviors by asking participants to decide for themselves (the first-party role) or for others (the third-party role). This study recruited a total of 34 children (9-10 years, M_age = 9.82, SD_age = 0.38, 16 females) and 31 college students (M_age = 19.81, SD_age = 1.40, 17 females). The behavioral results indicated that children behaved more selfishly in first-party and more fairly in third-party than adults, without any significant framing effects. The computational results revealed that both children and adults exhibited aversion to advantageous and disadvantageous inequity in third-party. However, they showed distinct preferences for advantageous inequity in first-party, with advantage-seeking preferences among children and aversion to advantageous inequity among adults. These findings contribute to a deeper understanding of children's social preferences and their developmental directions.

Keywords: advantage-seeking; computational model; fairness; framing effect; inequity aversion.

Figure 1

Experimental paradigm. (A) The dictator game. Participants acted as dictators in the role of first-party or third-party. Inequity was defined as the difference between the payoffs of both players. (B) Frames of the game. In the gain frame, allocatable tokens were exchanged for cents, while in the loss frame, tokens were exchanged to a loss in cents. Tokens were multiplied by a ratio (3:1, 2:1, 1:1, 1:2, and 1:3) when exchanged for cents. (C) The timeline of the DG task. Each round started with a jitter of 0.6 s ~ 1.0 s, followed by a decision phase, in which participant made their decisions within 6 s [The figure was inspired by (Sáez et al., 2015)].

Figure 2

Behavioral results. (A) The selfish deviations. For children, selfish deviations were higher in the role of first-party than third-party, while there was no significant difference between roles for adults. Moreover, children showed lower selfish deviations than adults in first-party role, but higher deviations than adults in third-party role. (B) The equal deviations. Equal deviations were higher for children than adults in first-party role, but lower for children than adults in third-party role. ^*p < 0.05, ^***p < 0.001. n.s., not significant.

Figure 3

Results of model comparison. (A) The scores and weights of LOOIC and WAIC across models applied to child participants. Dots and triangles indicate summed LOOIC and WAIC scores for each model from largest (green) to smallest (orange) respectively, while dots and triangles indicate summed LOOIC and WAIC weights for each model from largest (orange) to smallest (green) respectively. (B) The scores and weights of LOOIC and WAIC across models applied to adult participants. These results indicate that the Fehr-Schmidt inequity aversion model (Model 1) best captured choices of children and adult participants compared against the other three models.

Figure 4

Behavioral results of simulated data. (A) The average of selfish deviation. For children, selfish deviations were higher in the role of first-party than third-party, while there was no significant difference between roles for adults. Moreover, children showed lower selfish deviations than adults in first-party role, but higher deviations than adults in third-party. (B) The average of equal deviation. Equal deviations were higher for children than adults in first-party role, but exhibited no significance between children than adults in third-party role. ^**p < 0.005, ^***p < 0.001. n.s., not significant.

Figure 5

Results of the difference between group-level parameters and 0. (A) Children’s advantageous aversion parameters (α) were lower than 0 in the loss frame but not significant different in the gain frame for the role of first-party. Children’s advantageous aversion parameters (α) were higher than 0 in the gain frame and loss frame for third-party role. (B) adults’ advantageous aversion parameters (α) were higher than 0 in first-party and gain frame, first-party and loss frame, third-party and gain frame and third-party and loss frame. (C) Children’s disadvantageous aversion parameters (β) were higher than 0 in first-party and gain frame, first-party and loss frame, third-party and gain frame and third-party and loss frame. (D) Adults’ disadvantageous aversion parameters (β) were higher than 0 in first-party and gain frame, first-party and loss frame, third-party and gain frame and third-party and loss frame. FP: first-party; TP: third-party; CS, children; AS, adults; GF, gain frame; LF, loss frame.

Figure 6

Results of the difference between group-level parameters. (A) The effect of Role. Children’s advantageous aversion parameters (α) were lower in the role of first-party than third-party, while disadvantageous aversion parameters (β) were not significant in the role of first-party than third-party. In adults’ condition, advantageous aversion parameters (α) were lower in the role of first-party than third-party, while disadvantageous aversion parameters (β) were not significant in the role of first-party than third-party. (B) The effect of Group. Advantageous aversion parameters (α) were lower for children than adults in first-party role, while were higher for children than adults in third-party role. Moreover, there were no significance in disadvantageous aversion parameters (β) for children and adults in first-party role and third-party role. The shading and asterisks indicating the percentage of samples from the posterior greater than or less than 0. FP, first-party; TP, third-party; CS, children; AS, adults.