Coalitional psychology on the playground: Reasoning about indirect social consequences in preschoolers and adults

Surprisingly little is known about how relationship information is used predict others’ behavior. We examine a key element of this ability—how relationship information is used to anticipate how others will react to events in which they are not directly involved. This requires both using relationship information to modify expected reactions (e.g., friends may be more responsive than acquaintances) and also inference rules for restricting the class of reactions that may be felt or experienced on behalf of others (e.g., uninvolved friends may become angry but cannot become dizzy). These capacities were examined in both preschoolers and adults. Two different events were presented; one that would elicit anger from those who were involved and one that would elicit dizziness. For both sets of participants, cues to relationship status had a strong impact on anger expectations (uninvolved friends were expected to be more angry than uninvolved classmates), but had no effect dizziness expectations (neither uninvolved friends nor classmates were expected to be dizzy). Follow-up analyses also revealed a developmental difference. Adults made distinctions within the uninvolved friends category—expecting friends to be less angry at their own friend, and that levels of anger would vary according to their friend’s role within the social conflict—whereas preschoolers did not. These results demonstrate that by the early preschool years sophisticated inference rules already govern the expected reactions of uninvolved others, but that important developmental differences also remain. These results also indicate that relationship representations are inference engines for anticipating others’ behavior and reactions, not simply static containers for sorting people into categories.