A CORTICAL MODEL OF WINNER-TAKE-ALL COMPETITION VIA LATERAL INHIBITION

Simulations were performed of physiological interactions among excitatory and inhibitory neurons in anatomically realistic local-circuit architectures modeled after hippocampal field CA1. The simulated circuitry consists of several excitatory neurons jointly innervating and receiving feedback from a common inhibitory interneuron. Excitatory cells in the simulation receive input during a cycle of naturally-occurring rhythmic activity (the hippocampal theta rhythm), and the neuron receiving the most input activation is the first to reach its spiking threshold. Spiking excites the inhibitory cell, which in turn prevents other cells from responding. The result is the natural generation of a simple competitive or "winner-take-all" (WTA) mechanism, allowing only the most strongly-activated cell in a group or "patch" to respond with spiking activity. Formal mathematical characterization of the mechanism reveals specific physiological characteristics of the input to the network, which enable it to closely approximate an ideal winner-take-all mechanism. Unlike other, more abstract WTA mechanisms that have been proposed, the parameters of this biologically-derived WTA mechanism can be directly related to specific physiological and anatomical features of particular cortical circuits.