Judged versus perceived causality in visual and action-outcome sequences

Abstract

Previous research showed a dissociation between judgments of causality and perceived causality for Michotte's launch events: perceived causality shows no contingency effects. In standard action-outcome sequences, however, contingency effects appear both in causality judgments for event sequences and in responses to individual trials. We investigated what could be responsible for this difference. A first group saw sequences of launch events, with subjects controlling move- ment of shape A via a robot arm, so that subject's own action launched shape B. A second group of yoked subjects merely saw the resulting collisions, as in standard perceived-causality studies. In a third group, subjects made A launch B, but saw only the beginning and final motion; the collision itself was occluded. In a fourth group, subjects set B in motion by keyboard press, as in standard action-outcome studies. Contiguity effects were larger in groups seeing collisions than groups without this visual experience. In all four groups, contingency effects appeared only for judgments of causality, but not perceived causality. Accordingly, such dissociations are not limited to purely visual (rather than action-based) or natural (rather than arbitrary) events. The results are discussed in relation to current theories of perceived causality and causal judgment.