Background <p>Depression has been related to difficulties in updating negative beliefs in response to novel positive information. Research on this issue in the context of social feedback is scarce, though, as previous work mostly used scenario-based experimental tasks.</p> Methods <p>This research investigated biased belief updating in relation to depressive symptoms in the context of real social interactions in two non-clinical samples. Participants rated their beliefs about themselves in terms of a number of personality traits. In groups of 3–5 persons, they subsequently played a parlour game, after which they gave each other anonymised feedback. In Study 1 (<i>N</i> = 117), this feedback was manipulated by the experimenter so that it was a bit more positive than participants’ initial self-view. In Study 2 (<i>N</i> = 119), participants received the real feedback of their co-players. Finally, participants rated their self-beliefs again.</p> Results <p>In Study 1, high depressive symptom severity was associated with less belief updating following positive feedback. In Study 2, the correlations of depression with updating following positive and negative feedback were not significant. However, in terms of the learning rate, participants with elevated levels of depression learned more from negative than from positive feedback, whereas the opposite was found for people without depressive symptoms.</p> Conclusions <p>The results provide insights into the psychopathology of depressive symptomatology, as they point to deficits in updating self-beliefs in response to positive social feedback. At the same time, depressive symptoms are related to an increased learning from negative social feedback, provided that the feedback is honest.</p>