Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster C in Poster Session C: Wednesday, August 5, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Mid-level Motion Statistics as a Model for Dynamic Visual Perception in the Lateral Stream

Ming Zhou¹, Leyla Isik¹; ¹Johns Hopkins University

Presenter: Ming Zhou

The lateral visual stream supports dynamic social perception. While motion energy models have successfully characterized neural representations in early and mid-level motion-sensitive visual regions, interpretable computational models for higher-level lateral stream areas are still lacking. Here, we present an image-computable model, derived from first principles, to characterize social motion responses across the lateral stream. We derive second-order motion features inspired by work in static textures to construct mid-level motion statistics. Visualizations revealed that these features captured meaningful visual patterns, such as periodicity, curved contours, and biological motion. We next fit this model to existing fMRI data from participants viewing short social video clips, and found that second-order motion features explained significantly more variance in mid- and high-level lateral regions (e.g., EBA, pSTS) than first-order motion energy features. These findings suggest that second-order motion statistics derived from first-principles in computational neuroscience can capture complex motion information, analogous to their static image counterparts, and may support social perception along the lateral stream.

Topic Area: Computational Models of Vision & Visual Cortex