Conductance-based dendrites perform reliability-weighted opinion pooling

TL;DR

This paper demonstrates that conductance-based dendrites in neurons can perform reliability-weighted opinion pooling, enabling probabilistic multisensory integration and learning pathway reliabilities through error-driven plasticity.

Contribution

It introduces a biologically plausible neuron model that naturally implements Bayesian inference and learns pathway reliabilities from data.

Findings

Neurons with conductance-based dendrites perform near Bayes-optimal multisensory integration.

The model predicts specific membrane potential and conductance dynamics during multisensory tasks.

Error-driven plasticity enables neurons to adaptively learn the reliability of different information sources.

Abstract

Cue integration, the combination of different sources of information to reduce uncertainty, is a fundamental computational principle of brain function. Starting from a normative model we show that the dynamics of multi-compartment neurons with conductance-based dendrites naturally implement the required probabilistic computations. The associated error-driven plasticity rule allows neurons to learn the relative reliability of different pathways from data samples, approximating Bayes-optimal observers in multisensory integration tasks. Additionally, the model provides a functional interpretation of neural recordings from multisensory integration experiments and makes specific predictions for membrane potential and conductance dynamics of individual neurons.