Purchase this article with an account.
Steven Thurman, Hongjing Lu; Illusory biological motion in the periphery. Journal of Vision 2012;12(9):776. doi: 10.1167/12.9.776.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Recent theoretical computational models suggest that point-light biological motion patterns are recognized on the basis of motion- and/or form-based action templates, but there is debate as to whether these templates are derived from analysis of local spatiotemporal features (Giese & Poggio, 2003) or global body postures (Lange & Lappe, 2006). To investigate this issue behaviorally, we put local and global stimulus features into conflict using a point-light actor with the joints composed of drifting Gabor disks. We imposed joint movements and orientation signals extracted directly from a walker, but set each Gabor element to drift in the direction opposite to its joint movement. The speed of counteracting local (drifting) motion was varied. When presented in the periphery (20 deg) and with sufficient counteracting drifting speed, the walking animation perceptually stands still. If drifting speed is increased further, it remarkably reverses perceived walking direction. In Experiment 2, we show that biological motion perception is possible without changes in global body form. Nine Gabor disks were configured to the shape of a generic body posture and remained stationary throughout the animation sequence. To animate the stimulus, local motion and orientation signals derived from several different actions were introduced to the Gabor disks. When viewed peripherally, observers were very accurate at discriminating the facing direction, walking direction, and the action portrayed (walking, running, marching, dancing) solely on the basis of local spatiotemporal cues. In Experiment 3, we imposed global translation on the action stimulus used in Experiment 2. Observers can readily identify global translation directions and also simultaneously achieve good performance in biological motion tasks. Taken together, these experiments provide compelling evidence that local motion signals provide sufficient information to induce action templates that facilitate recognition of biological motion. Further, in the periphery motion-based analysis dominates when pitted against global body form.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only