INTER-AREAL LATERAL COMPETITION DRIVES EXTRA-CLASSICAL EFFECTS IN A LARGE-SCALE SPIKING MODEL

Extra-classical receptive field (eCRF) effects, such as surround suppression, demonstrate that local lateral processing within the primary visual cortex (V1) is insufficient for global integration. While hierarchical models incorporating feedback from higher areas (V2) have been proposed, simple feedback architectures often fail to replicate the strong suppression observed physiologically. This study proposes that effective feedback modulation requires lateral competition within higher-order areas throughout the sensory hierarchy.

Employing the NEST simulator for large-scale modeling, we developed a spiking neural network (LIF neurons) based on the canonical Potjans & Diesmann cortical microcircuit. The architecture integrates layered circuits representing the biological complexity of V1 and V2. We evaluated eCRF emergence across three configurations: (1) V1 with lateral connections only, (2) V1 with a single V2 feedback module, and (3) a hierarchical architecture where V1 interacts with two V2 modules exhibiting lateral competitive dynamics.

Simulations showed that neither lateral V1 connections nor non-specific feedback from a single V2 module generated physiological surround suppression. Crucially, robust suppression (~75%) that is highly consistent with established experimental electrophysiological data was only achieved by integrating vertical feedforward/feedback pathways with lateral and spatial competition in both V1 and, most critically, between modules in V2.

These findings demonstrate that contextual modulation is an emergent product of network configurations both within and between areas. This mechanism highlights the importance of inter-areal communication. We conclude that integrating vertical dynamics with lateral competition, specifically in higher-order areas, is essential for the emergence of robust extra-classical effects within the primary visual hierarchy.