72 — Complementary action of chemical and electrical synapses to perception

Borges et al muse on the evolutionary drivers for cortical organization; one possible optimization function is that of improving the discriminability of a neuronal network to external stimuli (increasing dynamic range) — the logarithm of the difference between the smallest and largest stimuli a system can uniquely respond to. (For example, touching a 300º stove is not dramatically different from touching a 400º stove; the DR of heat reception does not extend into the 300s range.)

This paper inspects a neuron-like node graph in which each node can be either excitatory or inhibitory ($E$ vs $I$) and inspects how a graph of this nature responds to stimuli (i.e. inputs to certain nodes) when the graph is altered. This graph is simulated using a cellular automaton model with a simple reduction of the synaptic transduction model.

The simulation suggested that there was an “optimal” branching ratio ($\frac{inputs}{outputs}$) and an optimal firing rate in order to maximize this dynamic range measure. Interestingly, electrical synapses (undirected) had a greater effect on DR when placed between excitatory cells in the $E$ layer than between inhibitory cells in the $I$ layer. (Electrical synapses were not placed across these two layers, though I imagine that this would yield interesting results as well.)

This suggests that dynamic range may be a good proxy for understanding evolutionary optimizations for the design and structure of sub-nets in a complex cortical structure. This would also point to the brain favoring smaller neighborhood design over large densely-connected design, in order to maintain this branching ratio. That makes a strong argument for the existence of small-ish local cortical motifs.