Item

Abstract

Experimental studies have shown that cortical neurons often exhibit a co-tuning of their excitatory and inhibitory receptive fields (i.e. correlation of incoming E/I currents for different input signals). Such co-tuning is hypothesised to be important for efficient computations. Theoretical studies have examined how different plasticity mechanisms can create such co-tuning in feedforward settings, where distinct, uncorrelated inputs to a post-synaptic neuron allow the formation of matching excitatory and inhibitory receptive fields. Still, the cortex is characterized by high levels of recurrence which raises the question of the mechanism by which input-driven E/I co-tuning arises. We demonstrate that a possible mechanism of E/I co-tuning in recurrent settings is highly specific recurrent connectivity which produces the necessary statistics for detailed balance to emerge. We first verify that in feedforward networks a combination of triplet STDP on the excitatory synapses and symmetric homeostatic STDP on the inhibitory synapses utilizes inhomogeneities in the pre-synaptic firing rates to create tightly matching excitatory and inhibitory receptive fields. The addition of unstructured recurrent connectivity in the pre-synaptic layer can significantly hamper the ability of STDP to produce co-tuning. We use simulation based inference to uncover possible constraints on the topology of the recurrent connectivity that will allow matching receptive fields to emerge. We find that different levels of clustering within neuron groups can effectively control the statistics of the input the post-synaptic neuron receives and consequently the ability of the STDP to produce co-tuning. Specifically, clustered excitation and global inhibition is particularly beneficial to developing E/I co-tuning. Our results suggest that structured recurrent connectivity can boost information propagation and promote the development of input selectivity in higher brain areas.