Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session B: Wednesday, August 13, 1:00 – 4:00 pm, de Brug & E‑Hall

The role of motion cues in object representation: A study of visual area specializations using deep learning

Nastaran Darjani¹, Maryam Vaziri-Pashkam², Shahab Bakhtiari¹; ¹Université de Montréal, ²University of Delaware

Presenter: Nastaran Darjani

The human brain processes both static and motion-defined visual cues for object representation, yet most computational models emphasize static information. We investigated neural responses to motion-defined object stimuli ("object kinematograms") by comparing brain activity with a dual-pathway artificial neural network that separates slow- and fast-varying visual information. Our findings show that while this dual-stream network captures aspects of biological motion processing, integration of slow and fast information improves similarity to brain in some regions but not others. These results highlight the functional diversity across visual areas in dynamic object representation.

Topic Area: Visual Processing & Computational Vision

Extended Abstract: Full Text PDF