Recognition of objects and actions

Abstract

Human observers can readily recognize point-light defined versions of locomoting people and animals. Is this ability restricted to the analysis of simple animal motions? To answer this question, we created point light displays of complex mechanical and biological actions. The stimuli included point-light defined vehicles, toys, animals, random motion foils, and humans. The human stimuli ranged from simple actions such as picking up an object to extremely challenging yoga moves. Two second clips of each movement were presented in randomized order to naive subjects. The first experiment involved a free report procedure in which subjects wrote a description of what they had seen after each clip. The second experiment involved a multiple choice task in which observers were asked to identify the term that best described what they had seen. Subjects selected one term from a list of five possibilities: animal, human, mechanical, plant-like, and other. Data from the free report procedure were categorized according to whether the observers reported local or global levels of analysis. The results indicated that the human and animal motions appeared to have been more globally analyzed than the mechanical motions. Moreover, physically challenging yoga poses as well as complex object motions were described with terms that reflected more localized levels of perceptual analysis than the terms used to describe relatively simplistic human actions. The results of the second experiment suggest that observers were most accurate in their classification of simple human motions but still performed at above chance with all types of stimuli. Interestingly, challenging yoga moves were much less likely to be categorized as “human” than were simple human actions. Thus, with some interesting twists, recognition of point light displays extends to a wide variety of complex, embedded motions.

Footnotes

 Supported by NEI grant R01 EY12300