Purpose: We have recently demonstrated border ownership coding in area V2 (Zhou et al, J Neurosci 20:6594, 2000). This study focuses on the role of motion parallax and dynamic occlusion cues in figure-ground organization and their interaction with Gestalt mechanisms. Gestalt mechanisms tend to interpret the inside of a closed contour as figure; motion parallax provides local border ownership cues. Methods: Single unit responses were recorded in alert monkeys during behaviorally induced fixation. We analyzed neuronal responses to the edge of a square figure that was defined by contrast as well as relative motion between random-dot textures inside and outside the figure. The edge was centered in the receptive field at the preferred orientation. Between fixation periods the figure was flipped about this edge, contrast was reversed, and motion conditions were changed. Results: Of 100 orientation selective cells tested in V2, 24% showed selectivity for the direction of occlusion at the edge as given by motion cues; 42% were selective for the side of the figure (Gestalt factor); and 18% were selective for both. In 78% of the latter (14/18; P<0.02), the preferred side of figure was also the occluding side of the preferred motion border, indicating that these cells utilized the different sources of figure-ground information consistently. Selectivity for motion-defined direction of occlusion was more frequent in V2 than in V1 (24/100 versus 1/23; P<0.05). Conclusion: Neurons in V2 are sensitive to motion parallax (or dynamic occlusion) and combine motion cues with Gestalt cues in a manner that is consistent with 3D object perception. This result complements our previous findings on the combination of stereoscopic cues with Gestalt mechanisms in V2 (Qiu et al., Neuroscience Abstracts 2001). Together, they reveal the emergence of explicit cue-invariant border ownership coding in V2.
ASJC Scopus subject areas
- Sensory Systems