分配意图与上行间接互惠：来自行为与ERP的证据

Upstream indirect reciprocity, a widespread phenomenon observed both in real-world settings and controlled experimental environments, extends beyond conventional reciprocity systems and plays a crucial role in fostering large-scale human cooperation and maintaining social order. Although this social phenomenon has garnered significant scholarly attention, existing research remains insufficient in uncovering its underlying mechanisms.Previous studies typically use a two-stage dictator game to investigate upstream indirect reciprocity. According to meta-analyses of experimental literature on dictator games, dictators typically allocated 28% of the total to recipients, with allocations exceeding 50% being extremely rare (Engel, 2011). Surprisingly, most existing research considers equal distribution (where A allocates 50% to B) as the threshold for determining whether A has good intentions (e.g., Gray et al., 2014; Hu et al., 2018), despite lacking sufficient justification for this criterion. Different from prior work, we propose that individuals assess others intentions relative to the social mean as a reference point. Based on this premise, we hypothesize that when individuals receive an allocation above the social mean, they are more likely to pass on a value above the mean to third parties, whereas allocations below the mean will result in values passed below the mean. If individuals indeed pass on a higher or lower value based on whether they received above- or below-mean allocations, this result might also be explained by an alternative hypothesis: the income effect, where people give more when they have more resources. Therefore, this study investigates whether intention-based indirect reciprocity persists event after controlling for the income effect.We recruited 42 undergraduate participants for the experiment, which consists of two parts: a standard dictator game followed by an indirect reciprocity experiment. The second experiment employed a 2 (distribution below the social mean vs. above the social mean) 2 (human allocation vs. computer allocation) within-subject design. The main experiment featured both a human allocation task and a computer allocation task, with task order counterbalanced among participants. Each task included 156 trials, for a total of 312 trials.The results show that both distribution outcomes and perceived intentions significantly influence upstream indirect reciprocity. Specifically, participants allocated more to third parties after receiving above-mean distributions from a human compared to a computer, while below-mean distributions from a human led to lower allocations than those from a computer. EEG data revealed that N1 components were modulated by perceived intention, with human allocations eliciting greater N1 responses. The feedback-related negativity (FRN) component was influenced by distribution outcomes, with below-mean distributions evoking larger FRN responses than above-mean distributions. Finally, the P3 component was regulated by the interaction between distribution outcome and intention.