When a neuron fires, it excites or inhibits other neurons it connects with. Neural networks are precisely tuned to a narrow range between too much excitation (seizure) and too much inhibition (silence). Somehow the brain stabilises itself on this knife-edge, compensating for perturbations that disrupt the balance between excitation and inhibition. This process is called homeostatic compensation ('homeo'=similar, 'stasis'=staying still) and is thought to be impaired in disorders such as epilepsy, autism and schizophrenia. How does homeostatic compensation occur?
We have addressed this question using an ideally-suited circuit in the fruit fly, in which 'Kenyon cells', the neurons that store odour-associated memories, receive both excitation and inhibition. Altering the E/I balance dramatically changes the magnitude of Kenyon cells' odour responses, but with time the circuit compensates for the disruption.
Using state-of-the-art gene targeting techniques, we measured Kenyon cells' neural activity while altering their excitatory or inhibitory inputs, and discovered that this circuit can compensate for too much inhibition (too much activity in a neuron that inhibits Kenyon cells) but not too little inhibition (blocking signalling from the neuron that inhibits Kenyon cells). We further discovered that this compensation occurs through a combination of decreased activity in the inhibitory neuron, plus increased excitation onto Kenyon cells.
Through computational modelling, we also discovered that homeostatic plasticity can make the mushroom body more efficient at its main behavioural function, i.e. allowing flies to learn to discriminate between different odours. We further found that predictions of compensatory variability made by our models were confirmed in the fly hemibrain connectome.
