SPATIOTEMPORAL OSCILLATIONS IN NEURAL CIRCUIT

Spiral patterns exist in nature and also occur in the cortex. Multiple, complex, and interacting spirals organize spatiotemporal information for distributed, parallel neural computations at the global level. The underlying mechanisms responsible for generating such emergent dynamics are unknown. Hence, a spatio-temporal cortical field model is proposed that describes local nonlinear neural population interactions and distance-dependent axonal delays as diffusively coupled traveling waves, with global interactions occurring at multiple timescales. The model is composed of local cortical circuits consisting of excitatory pyramidal neurons and three different types of inhibitory neurons, extended to represent a two-dimensional cortical sheet. Model simulation results display mixed-mode oscillations, which might represent the coexistence of multiple rhythms and show the emergence of complex patterns such as planar, source, sink, or rotating spirals with annihilation events. These spiral waves differentially respond to the grating input strength, such as returning to the original state for a weak stimulus, whereas changing state for a strong stimulus, and also showing memory-like characteristics post stimulus. I propose a hypothesis that these local patterns might be an integral part of the global spiral wave dynamics. Possibly a speculative origin of fundamental waves seen across the brain. Lastly, these spiral waves might work as spiral processing units for performing dynamic logical operations, compute and memory simultaneously. The model may help us to understand and formulate rigorous theoretical principles for new paradigms of artificial intelligence for efficient cortical computations to develop efficient computing architectures and novel learning rules for traveling wave-based computations.