Manifold Transform by Recurrent Cortical Circuit EnhancesRobust Encoding of Familiar Stimuli

Abstract

A ubiquitous phenomenon observed along the ventral stream of the primate hierarchical visual system is the suppression of neural responses to familiar stimuli at the population level. The observation of the suppression of the neural response in the early visual cortex (V1 and V2) to familiar stimuli of size that are multiple times larger in size than the receptive fields of individual neurons reflects the plausible development of recurrent circuits for encoding these global stimuli. In this work, we investigated the neural mechanisms of familiarity suppression and showed that an excitatory recurrent neural circuit, consisting of neurons with small and local receptive fields, can develop to encode specific global familiar stimuli robustly as a result of familiarity training. This Hebbian learning based model attributes the observed familiarity suppression effect to the sparsification of the population neural code for the familiar stimuli due to the formation of image-specific local excitatory circuits and competitive normalization among neurons, leading to the paradoxical neural response suppression to the familiar stimuli at the population level. We explored the computational implications of the proposed circuit by relating it to the sparse manifold transform. The recurrent circuit, by linking spatially co-occurring visual features together, compresses the dimensions of irrelevant variations of a familiar image in the neural response manifold relative to the dimensions for discriminating different familiar stimuli. The computation can be considered as a globally non-linear but locally linear manifold transform that orthogonalizes the slow modes of network dynamics relative to the subspace of irrelevant stimulus variations, resulting in increased robustness of the global stimulus representation against noises and other irrelevant perturbations. These results provide testable predictions for neurophysiological experiments. Author summaryIn this research, we explored how the brain can become more efficient at processing familiar visual information. When we repeatedly see something, our brains response to it changes. In response to familiar stimuli, neurons across the different visual areas of the mammalian visual system become more selective and their overall activities decrease. We developed a computational model to investigate why this happens and what functional advantages these mechanisms might provide. We discovered that familiarity leads to the development of a more efficient and robust neural representation of what we see. It allows us to rapidly and robustly recognize our friends face despite changes in lighting conditions, view angle, or facial expression. Our model showed that through repeated exposure, the brains neural circuits, even in the early stages of visual processing, rapidly adapt and organize themselves to focus on important and consistent features in our visual environment while becoming less sensitive to irrelevant variations, and distractions.