Participants completed a series of Dictator Games where they were asked how much money they wanted to donate to people who had previously rated them in the scanner. Participants donated on average 23% of their £5 endowment. Their donation behavior was modulated by the feedback raters gave them (F(2, 78)=34. 91, p<0.001, ηp2 = 0.472) such that they donated more to raters who gave approval than those whose feedback was not displayed (p<0.001) and those who disapproved of them (p<0.001). Donation behavior also interacted with probability of approval (F(6, 234)=3.33, p=0.007, ηp2 = 0.079). Follow-up comparisons showed that raters for whom feedback was not displayed were treated differently based on their group membership, F(3, 117)=5.88, p=0.001, ηp2 = 0.131. Raters from different groups who gave approval feedback were treated similar to each other (F(3, 117)=0.326, p=0.789, ηp2 = 0.008) as were raters from different groups who gave disapproval feedback (F(3, 117)=0.843, p=0.445, ηp2 = 0.021). These results provide evidence that participants treated the raters as real people and that social feedback had a significant impact on subsequent prosocial behavior. Data are represented as mean ± SEM.