Abstract
Research indicates that persons do not employ a constant unit of perception in processing social information. An experiment investigated effects of varying the predictability of an actor's behavior upon the manner in which observers partitioned it into units of action and made attributions of behavioral causality. Subjects viewed a tape of an actor who behaved either predictably throughout, predictably at first and then unpredictably, unpredictably at first and then predictably, or unpredictably all the time. Subjects divided the actor's behavior into more units (a) when the behavior began unpredictably than when it began predictably and (b) when the pattern of behavior changed (from predictable to unpredictable or vice versa) than when no change occurred. Number of units was positively correlated with dispositional attributions of causality.

This publication has 2 references indexed in Scilit: