Jean-Michel Hupe

13 avril 2012

CERCO, Toulouse

Integration of motion signals : lessons of ambiguous stimuli

invité par Anna Montagnini

Abstract : The formation of perceptual objects out of complex sensory scenes requires the binding of spatially or temporally distributed local signals. Ambiguous displays can highlight the neural mechanisms involved in the integration and segmentation of those signals. According to Burr & Thompson (2011), one of the big questions in 1985 was "how are local-motion signals integrated into the coherent motion of objects (solving, amongst other things, the aperture problem) ?" Following Adelson & Movshon’s Nature paper in 1982, proposing a two-stage model of motion perception, psychophysics research used a lot moving plaid stimuli (two gratings sliding over each other) to answer the question of when we should integrate components into a single motion and when we should segregate motions. However, the answer varies with time, since moving plaids are in fact ambiguous stimuli that can be perceived either as a coherent pattern moving rigidly or as two gratings sliding over each other (transparent motion). Prolonged observation leads to bistable alternation between coherence and transparency, and even tristable perception when one considers which grating is perceived in front when the transparent percept is experienced. The dynamic competition between motion integration and segmentation provides a quite different picture of the mechanisms involved than its resolution at motion onset. In particular, when put in the ambiguous range, the first percept corresponds to the integration of motion signals, suggesting a reversed 2-stage model. But the interpretation of multistable plaid perception for motion mechanisms requires knowing the mechanisms of multistable perception. By comparing the dynamics of moving plaid perception with those of other ambiguous stimuli (apparent motion, rotating diamonds, as well as ambiguous stimuli in the auditory modality), we progressed in the understanding of both multistable perception and motion integration and segmentation.
CNRS logo université Aix Marseille logo | plan du site | mentions légales | contact | admin | intranet | intcloud |