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

Poster B in Poster Session B: Tuesday, August 4, 2:00 – 3:45 pm, Kimmel Center, Shorin & Rosenthal Rooms

Testing theories of segment-based neural representation in macaque V1

Tridib Kalyan Biswas¹, Amirhossein Farzmahdi², Anna Ivic Jasper³, Aravind Krishna³, Adam Kohn¹, Ruben Coen-Cagli³; ¹Einsteinmed, ²Columbia University, ³Albert Einstein College of Medicine

Presenter: Tridib Kalyan Biswas

Segmentation is the process of grouping and separating image features to form perceptual objects or, segments. Segments influence single neuron firing, as shown in primary visual cortex (V1) neurons (Cavanaugh et al., 2002; Gilbert & Li, 2013; Roelfsema, 2023). However, segment–specific effects on population V1 firing remain unknown. To understand population encoding of segments, we began with existing frameworks that explain population firing through probabilistic inference about latent causes (Haefner et al., 2024). We then extended those frameworks using a recently proposed probabilistic model of human segment perception where the inference of image features and segments is coupled and proceeds iteratively (Biswas et al., 2026). We evaluated V1 predictions for that model with data from one awake macaque. The model explained single–neuron modulation by segments, and successfully predicted how image segments modulate pairwise covariability, including their rich spatiotemporal dynamics. Thus, we offer a new theory of segment–based image encoding in V1 and initial experimental support for its prediction of flexible, segment–dependent pairwise covariability. This is an important first step towards understanding the distributed neural population representation of natural scenes.

Topic Area: Computational Models of Vision & Visual Cortex