Representational momentum and the human face : an empirical note

Abstract

Recent evidence suggests that observers may anticipate the future emotional state of an actor when viewing dynamic expressions of emotion, consistent with the notion of representational momentum. The current paper presents data that conflicts with these previous studies, finding instead that memory for the final frame of an emotional video tends to be shifted back in the direction of the first frame. While simple methodological issues may explain this difference (e.g., the use of morph sequences in previous studies versus naturalistic expressions here) a more theoretically interesting possibility is also considered. Specifically, recent studies of ensemble representations have shown that observes can rapidly extract the average expression from a display of up to 20 faces. It is suggested that the need to predict versus the need to maintain a stable estimate of the current state often compete when we interact with dynamic stimuli. Our memory for the final expression on an emotional face may be particularly sensitive to task demands and response timing, thus coming to reflect different solutions to this anticipation-averaging conflict depending on the precise experimental scenario.