Track genomics, neuroscience and molecular biology.
Search papers, labs, and topics across Lattice.
We propose a recurrent neural network framework that models perceptual grouping in natural scenes by integrating local boundary signals with top-down feedback about object identity. This mechanistic model, instantiated with recurrent interactions between simulated cortical areas, successfully learns to propagate enhanced neuronal activity within object representations, mimicking biological mechanisms. The network's emergent dynamics accurately predict human reaction times, bridging computational principles, neural circuitry, and perceptual psychology.
Recurrent neural network dynamics, not just local features, explain the temporal progression of perceptual grouping in complex natural scenes.
How the brain groups image elements into coherent objects in natural scenes remains unclear. Existing models for grouping in biological vision rely on simplified stimuli with explicit boundaries and cannot explain human behavior in naturalistic settings. We propose a mechanistic framework in which recurrent interactions propagate enhanced neuronal activity within and between cortical areas. The framework links computational principles, neural circuitry and perceptual psychology. Local boundary signals govern early grouping, whereas later stages integrate top-down feedback carrying information about object identity, grouping features across internal edges into coherent object representations. We instantiated this framework as a brain-inspired recurrent neural network trained to group features in natural images. The network learned to propagate enhanced activity across a cued object's representation, mirroring the brain's perceptual grouping mechanisms. The network's dynamics also predicted human reaction times. The framework links cortical dynamics to human vision and generates testable predictions for neuro-physiological and psychophysical experiments.