The Role of Synchronization of Neural Modules in Pattern Processing in the Visual System

Abstract Key operations in modeling visual processing in living systems are those that process a pattern as a holistic object, for example, in pattern recognition to ensure invariance to translation and rotation. In this article, we describe a neural module we synthesized for moving a pattern along a cortical layer without losing its shape. The module has the form of a three-layer neural network with two types of inhibitory neurons, one of which is a universal “prohibition” logic element. Modeling showed that such operations can be implemented using the proposed neural module by additionally synchronizing its operation using endogenous pacemakers with a gamma rhythm frequency. The mechanism of such synchronization is similar to the clock sweep mechanism in electronics. We believe we have clearly demonstrated the possible role of gamma-based brain activity as a mechanism for synchronizing computations in living neural networks for the first time. This opens up new possibilities for constructing neuromorphic architectures for real-time visual recognition.