Inhibitory synaptic plasticity generates global and detailed balance of excitation and inhibition
Institute of Neuroinformatics,
Inhibitory synaptic plasticity generates global and detailed balance of excitation and inhibition.
The balance of excitatory and inhibitory membrane currents recorded in neurons during stimulated and spontaneously active network states has been the focus of many recent experimental and theoretical studies. The function of such balance states has been hypothesized to enable fast, stable and diverse network responses, to amplify the response to certain stimuli, or to allow the establishment of functional network architectures by means of dynamically controlling this balance in distinct groups of cells. Despite the recent interest in these phenomena, no mechanism has been brought forward that would allow the establishment of such balanced networks. Using networks of integrate and fire neurons, we show that spike timing-dependent inhibitory learning rules can succeed in establishing globally balanced networks.
Additionally, we show that in a feedforward architecture, the same learning rules establish a detailed balance in each cell. As a result, cells with stimulus-tuned excitatory input adjust their inhibitory synapses until inhibition and excitation have similar stimulus tuning.
This effect is largely independent of the absolute number of synapses, the input correlation structure and the firing rate statistics of the input signals. We conclude that inhibitory plasticity can establish stable network configurations, in which perturbations (memories) in the excitatory tuning are quickly and automatically balanced out by inhibitory plasticity. Given that the balance of excitatory and inhibitory activity in the brain has recently emerged as a powerful mechanism to govern neuronal dynamics, our results could be most helpful in understanding how such balanced systems could establish and maintain themselves naturally and without intelligent control mechanisms.