Voltage-dependent multiplexing as a computational role for spiking

BBSRC strategic theme: Understanding the rules of life

The classical view of neuronal function relies on a delicate balance between excitation and inhibition, with the former driving spiking and the latter suppressing it. However, growing evidence suggests that this framework oversimplifies the complex dynamics governing spike generation. For instance, Haider et al. (2013) revealed that inhibition dominates cortical responses in awake animals, challenging the notion of a simple excitatory-inhibitory balance. Furthermore, Guo et al. (2015) demonstrated that inhibition can paradoxically trigger movement through rebound currents in the mouse motor cortex, echoing long-standing studies in the central pattern generator circuits underlying locomotion (Marder & Calabrese 1996) where inhibition explicitly causes spikes with a temporal delay.

These observations raise a fundamental question: how do synaptic inputs influence spiking if not through a direct, instantaneous competition between excitation and inhibition? The answer may lie with voltage-gated (intrinsic) conductances, which propagate signals from dendrites to the axon yet comprise a huge diversity of different ion selectivities, channel kinetics, and reversal potentials. My preliminary modelling work leverages an intrinsic current to reproduce primate reaches, by casting motor preparation as the initialisation of a rebound current. Indeed by decoupling dendritic currents from spike-generation, all synaptic inputs must be preparatory in nature, configuring intrinsic conductances for future behaviour. Under this framework, preparation can be seen as a fundamental property of neuronal function across the brain rather than a niche artefact of the motor cortex. This aligns with recent findings that the barrel cortex exhibits signs of preparation before engagement in a complex sensory discrimination task (Constantinople and Bruno 2013; Park et al. 2022; Rodgers et al. 2021).

These converging lines of evidence point towards an alternative hypothesis: excitation and inhibition cooperate to precisely control voltage-dependent intracellular processes, which in turn govern neuronal output. As a corollary, the brief timescale of the action potential may allow neurons to communicate digitally without disrupting these slower voltage-dependent processes. To explore these ideas, I propose a multi-faceted approach. I will develop and analyse a multi-compartment neuron model that decouples synaptic inputs from spike generation through a set of voltage-gated conductances. This will provide a platform for investigating the mechanisms underlying "intrinsic preparation" of neurons. I will train a network of these neurons to perform naturalistic sensory and motor tasks, such as sensory discrimination (Rodgers et al. 2021) and compound reaching (Zimnik et al. 2021). This will allow me to assess the generality of preparation as a fundamental aspect of neuronal computation. Using the trained network, I will generate testable predictions about voltage dynamics in dendrites and axons. These predictions will form the basis for collaborations with experimental groups employing voltage imaging techniques (Wong-Campos et al. 2023), facilitating a dialogue between theory and experiment.

By reframing the excitatory-inhibitory interplay and investigating the role of "intrinsic preparation", this project aims to shed new light on the principles governing neuronal computation. The insights gained may have far-reaching implications for our understanding of how the brain processes information and generates complex behaviours, potentially inspiring novel approaches in artificial intelligence and neuromorphic computing.