The role of perceptual processes in the recognition of social interaction

Humans typically interact with two major classes of visual stimuli in everyday life: objects and other humans. While we have a fair understanding of how objects are recognized, little is known about the recognition of social interactions (recognizing person-directed actions). Classically the recognition of social interactions was assumed to rely mainly on higher-level cognitive processes (e.g. inference), whereas more recent evidence suggests that lower-level perceptual processes may play an important role [1].
Of particular importance are cortical units that directly link the execution of a particular action and perception of the same action (i.e. mirror neurons or action-perception units) [1]. However there is little evidence that such units exist in humans. Moreover, relatively little is known about the involvement of other perceptual processes in social interaction recognition.
Our aim is to examine the involvement of perceptual processes, particularly those of an action-perception unit, in the recognition of social interactions.
We investigated the existence of an action-perception unit behaviorally and with fMRI in a human-object interaction task. We reasoned that the repeated execution of an action should lead to an adaptation effect in the action-perception unit which, in turn, should affect the perception of that same action (see also Figure 1ab). In a different set of experiments, we assessed the efficiency of social interaction recognition when the contribution of the action-perception unit was minimized. Because we know that object recognition is efficient [2], we compared the speed of recognition when viewing static images of social interactions and static images of objects.
We failed to find an action adaptation-perception transfer under a variety of experimental conditions. Moreover, we found that social interaction recognition is efficient even when the contribution of an action-perception unit is minimized. Specifically, the speed of identifying social interactions is not statistically different from identifying objects (Figure 2).
We might not have found action adaptation-perception transfer because an action-perception unit a) generally does not show adaptation effects, or b) is tuned to human-human interactions .The latter possibility will be explored in future experiments. Our second line of experiments, however, suggests that social interaction recognition is efficient even when an action-perception unit is hardly employed. Our results therefore challenge the view that an action-perception unit is critical for social interaction recognition. Future experiments will examine the perceptual processes mediating social interaction recognition behaviorally with virtual environments (see also Streuber, S.) and with imaging techniques.
Last updated: Friday, 05.10.2012