Fig. 3: Group-identity and bot-players’ behavior effects on participants’ zapping behavior experimental outcomes.

Zapping rates throughout the experimental block (left panels) and progression of zapping rates throughout the blocks (right panels) in the homogeneous (A) and the heterogeneous (B) conditions. Participants zapped outgroup bot-players more than neutral and ingroup bot-players, indicating a consistent intergroup bias. Participants were also more likely to zap bot-players that displayed zapping behavior than zap-avoiders and increased their likelihood to zap zappers over time, indicating learning effect on behavior. In the heterogeneous condition participants showed reduced differentiation between avoiders and zappers in the progression of zapping rates, indicating a group-level attribution effect. Lines in learning curves indicate mean zapping frequencies, shadows indicate 95% confidence intervals of the mean. Boxplots include the mean by light circles, median in bold line, interquartile range is represented by the box, minimum and maximum range by the whiskers, and outliers by black dots.