Self-tuning of neural circuits through short-term synaptic plasticity

D. Sussillo, T. Toyoizumi, and W. Maass

Abstract:

Circuits of neurons in the cortex have a remarkable capability to maintain functional and dynamic stability in spite of changes in the level of external inputs, synaptic plasticity and changes in the circuit structure that occur during development and adult learning. The source of this characteristic stability of cortical circuits has remained a mystery, especially since even stronglysimplified models of such circuits do not exhibit similar stability properties. One simplification that is usually made in such models is that the empirically found nonlinear and diverse inherent short term dynamics (paired-pulse facilitation and depression) of biological synapses is replaced by static and uniform linear synapse models. We show in this article that this is a mistake, since the complex and diverse nonlinear dynamics of biological synapses supports the implementation of powerful control principles that endow circuits of spiking neurons with almost in-vivo like stability properties.

Reference: D. Sussillo, T. Toyoizumi, and W. Maass. Self-tuning of neural circuits through short-term synaptic plasticity. Journal of Neurophysiology, 97:4079-4095, 2007.