Optimising neuronal plasticity for associative memory

Why are brains the way they are? Understanding the computational function of neuronal plasticity rules and circuit architectures is fundamental to understanding how brains work.  Yet, it remains unclear whether or how these 'design features' are optimal for their behavioural purpose. We will address these questions using the problem of stimulus-specific associative memory. In Drosophila, olfactory associative memories are stored by weakening the synapses from odour-encoding Kenyon cells (KCs) onto action-encoding mushroom body output neurons (MBONs). Our recent computational and experimental discoveries have led us to two novel, independent hypotheses for how the fly's neuronal plasticity rules might optimise the odour-specificity of memories given other constraints in the circuit:

1. Homeostatic plasticity optimises sensory coding for stimulus-specific associative memory by compensating for inter-neuronal variability.

2. The reason learning occurs by synaptic depression (not potentiation) is that, given the constraints of downstream circuitry, this strategy makes overlap in sensory representations less detrimental to odour discrimination.

We will test these hypotheses and investigate the molecular mechanisms underlying them, by a combination of two-photon imaging, electrophysiology, behaviour, genetic manipulations and computational modelling. Revealing computational functions underlying neuronal plasticity rules will shed light on how synaptic plasticity and memory work in general.