Abstract: An individual’s social class is a context rooted in both the material substance of social life (wealth, education, work) and the individual’s construal of his or her class rank, and is a core aspect of how he or she thinks of the self and relates to the social world. In psychological science, research has shown that social class influences a lot of domains that include language (Bernstein, 1971), aesthetic preferences (Snibbe &Markus, 2005), subjective well-being (Diener & Suh, 1997; Howell & Howell, 2008), physical health (Adler et al., 1994; Gallo & Matthews, 2003), and cognitive performance (Nisbett, 2008). It suggests that social class can shape the individual’s basic psychological processes and behaviour. In recent years, there have been quite a few researches on social class and prosocial behavior, but the results are controversial (Dovidio, Piliavin, Schroeder, & Penner, 2006; Sober & Wilson, 1998; Piff, Kraus, C?té, Cheng, & Keltner, 2010; Piff, Stancato, C?té, Mendoza-Denton, & Keltner, 2012 ; Stellar, Manzo, Kraus, & Keltner, 2012). Based on these findings, we argued that social class differences in prosocial behavior were associated with return prediction. To test the hypothesis mentioned above, the present study recruited 40 subjects who were all adults. Researchers made the experiment online, which was a 2 (social class: upper, lower) × 2 (return prediction: higher, lower) mixed-design, in which social class was a between-subjects variable, and explored the participants of different social classes in two return prediction inductions (high, low), the prosocial behavior variation, and the interaction. Firstly we filtered the subjects of 2 types based on their occupations. The simulation experiment is adapted from the famous “dictator game,” a single-trial economic game that yields a behavioral measure of generosity (Forsythe, Horowitz, Savin, & Sefton, 1994; Fowler & Kam,2007). In this game, participants were informed that they had been paired with an anonymous partner seated in a different room (Forsythe et al., 1994; Fowler & Kam, 2007). Participants were given 10 points and told that their task was to decide how many of these points they wanted to keep for themselves and how many (if any) they wanted to transfer to their partner. Participants were further told that their partner would have no strategic input into the game’s outcome, that their responses in the game would remain anonymous. They played this game with two different experiment assistants respectively. Every time before the game, the participants were informed that how many points does the experiment assistant give on average in the past experiments. Their gifts of the study would depend on how many points they had remaining. In the dictator game, higher allocations reflect higher levels of altruism in that they represent participants’ willingness to sacrifice their own material interests in favor of the well-being of their partner. The results indicated that: (1) different social classes have main effect on the prosocial behavior; (2)the main effect of return prediction was significant; (3)the interaction of social class and prosocial behavior is significant. There were no significant differences between social classes in higher return prediction, we found significant differences between social classes in lower return prediction. These findings not only high light the importance of the return prediction in social class differences in prosocial behavior, but also have important contributions to developing strategies in dealing with people．

Key words: social class, prosocial behavior, return prediction

摘要： 近年来关于社会阶层和亲社会行为关系的研究不少,但结论不一致。基于此，本研究假设亲社会行为的社会阶层差异可能与对回报水平的预期有关。实验改编自“独裁者”游戏，通过网络模拟情境进行,采用2（社会阶层：高、低）×2（回报预期：高、低）的混合设计（其中社会阶层为被试间变量），考察了不同社会阶层被试在两种不同回报预期条件下，亲社会行为的差异。实验结果表明：回报预期和社会阶层之间存在交互作用，进一步的简单效应分析表明，当预期目标有较高的回报水平时，高低社会阶层的亲社会行为差异不显著。当预期目标有较低的回报水平时，与高社会阶层相比，低社会阶层的亲社会行为明显减少。说明回报预期对不同社会阶层的亲社会行为起显著的调节作用。

关键词: 社会阶层, 亲社会行为, 回报预期