Speaker
Description
Most studies of neural encoding still rely on patch-clamp noise injection, an invasive method that poorly replicates the conductance-based, dendritic inputs neurons receive in vivo. Optogenetics offers a non-invasive alternative, but turning noisy light into realistic synaptic conductances requires new strategies.
We present a framework that uses Gaussian-noise illumination of channelrhodopsins to generate Ornstein–Uhlenbeck conductances — the standard model of in-vivo input. By characterizing the opsins’ light-to-conductance transfer functions and designing inverse filters, we produced conductance fluctuations that closely match desired statistics across a wide range of timescales. In cortical neuron models, these inputs drove naturalistic, fluctuation-driven spiking under both somatic and dendritic expression.
This approach enables flexible, non-invasive reproduction of realistic input statistics, opening the door to optogenetic studies of neuronal dynamics with true in-vivo fidelity.
