179 — Mutual inhibition of lateral inhibition: a network motif for an elementary computation in the brain

Koyama & Pujala (10.1016/j.conb.2017.12.019)

Read on 15 February 2018
#neuroscience  #connectomics  #logic-gate  #circuitry  #motif  #graphs  #inhibition  #disinhibition  #cortical-motif  #brain-circuit  #retina 

Long before the advent of what we now consider a “computer,” mathematicians hypothesized that a series of simple ciruits called “logic gates” were adequate to describe complex mathematical functions. We now rely on that principle as we develop arbitrary software in Turing Complete code.

And so it makes sense that there is an analog to logic gates in neuroscience: There ought to be a fundamental building-block with a known function that evolution can leverage to simplify the process of producing large, complex computational structures. After all, building a model spaceship is a difficult task, but any ten-year-old worth her salt can build one out of legos.

This paper identifies a circuit that is found — with slight variation — in a variety of species, with a variety of functions: In zebrafish, it induces an escape motion based upon the location of an acoustic stimulus. In barn owls, it acts as a “multiple-choice” selector. In mice, it acts as a noise reduction signal processing stream.

The circuit is dubbed “mutual inhibition of lateral inhibition” which is a bit cryptic. But it simply means that $n$ inputs, $I_0$, $I_1$, … $I_n$, are able to inhibit their neighbors ($I_1$ can inhibit $I_0$ and $I_2$) and that inhibitory circuit can in turn amplify the original signal by removing its neighbors’ inhibition. So if $I_1$ is very strong, $I_0$ and $I_2$ will be nearly silenced.

That we see this same structure in a variety of organisms perhaps means that there is a collection of fundamental motifs from which the brain can choose when constructing complex circuitry. If this is true, then understanding large connectivity patterns does not require seeing every single neuron and synapse in a brain. Instead, it just means that we must identify the circuits in use, and how they modulate their inputs in order to produce their outputs.