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

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

Deep Neural Networks Provide Insights into Distinct and Shared Selectivity for Faces and Bodies in Human Visual Cortex

Leonard E van Dyck¹, Katharina Dobs¹; ¹Justus Liebig Universität Gießen

Presenter: Leonard E van Dyck

Faces and bodies are key social stimuli processed by distinct functional networks in human visual cortex. However, growing evidence suggests systematic overlap between these networks, raising an important question: How segregated or integrated are face and body processing? Competing hypotheses propose fully segregated pathways, varying levels of integration, or a single multiplexed system. Here, we test these hypotheses using deep convolutional neural networks trained on object recognition. A functional localizer identified face- and body-selective units, as well as mixed-selective units that respond to both categories. Decoding analyses revealed that face- and body-selective units specialize in their respective domains, while mixed-selective units encode detailed information from both, supporting an integrative role. Finally, using fMRI encoding analyses, we found that these units account for unique variance in neural responses within both distinct and overlapping face- and body-selective cortical areas. Our findings suggest that face and body networks balance segregation and integration, supporting both fine-grained recognition and whole-person perception.

Topic Area: Object Recognition & Visual Attention

Extended Abstract: Full Text PDF