Visual features underlying neural selectivity

Overview

A magnetoencephalography (MEG) study asking which visual features actually drive neural selectivity when the brain looks at natural images.

Participants viewed natural images alongside degraded versions engineered to target specific visual properties. Neural activity was then decomposed into contributions from retinotopy, spatial frequency, orientation, shape, and texture, isolating how much each feature carries image- versus category-level information.

Finding

Retinotopy and global shape emerge as the strongest features underlying image and category information respectively, showing how the brain draws on multiple visual features in parallel during object recognition.

Reference

Schlüsener JK, Shah M, Sandhaeger F, Siegel M. Visual features underlying neural selectivity for natural stimuli. bioRxiv, 2025. DOI: 10.64898/2025.12.08.692716