Stochastic neural fields as gradient dynamical systems.

Continuous attractor neural networks are used extensively to model a variety of experimentally observed coherent brain states, ranging from cortical waves of activity to stationary activity bumps. The latter are thought to play an important role in various forms of neural information processing, including population coding in primary visual cortex (V1) and working memory in prefrontal cortex. However, one limitation of continuous attractor networks is that the location of the peak of an activity bump (or wave) can diffuse due to intrinsic network noise.

Laminar Neural Field Model of Laterally Propagating Waves of Orientation Selectivity.

We construct a laminar neural-field model of primary visual cortex (V1) consisting of a superficial layer of neurons that encode the spatial location and orientation of a local visual stimulus coupled to a deep layer of neurons that only encode spatial location. The spatially-structured connections in the deep layer support the propagation of a traveling front, which then drives propagating orientation-dependent activity in the superficial layer. Using a combination of mathematical analysis and numerical simulations, we establish that the existence of a coherent orientation-selective wave relies on the presence of weak, long-range connections in the superficial layer that couple cells of similar orientation preference.

Near-optimal decoding of transient stimuli from coupled neuronal subpopulations.

Coupling between sensory neurons impacts their tuning properties and correlations in their responses. How such coupling affects sensory representations and ultimately behavior remains unclear. We investigated the role of neuronal coupling during visual processing using a realistic biophysical model of the vertical system (VS) cell network in the blow fly.