Certainty-weighted integration of information in individual cortical neurons
Ben von Hünerbein, University of Bern, Switzerland; Matthijs oude Lohuis, Pietro Marchesi, Umberto Olcese, University of Amsterdam, Netherlands; Walter Senn, University of Bern, Switzerland; Cyriel Pennartz, University of Amsterdam, Netherlands; Jakob Jordan, Mihai Petrovici, University of Bern, Switzerland
Session:
Posters 3B Poster
Presentation Time:
Sat, 26 Aug, 13:00 - 15:00 United Kingdom Time
Abstract:
It is essential to take into account the uncertainty related to sensory percepts when using them for guiding behavior. While it has been shown that decision making in humans and other animals takes uncertainty into account , it remains unclear how it is represented in cortex . A recent theory suggests that individual cortical neurons can represent uncertainty implicitly by modulating their membrane potential variability. Here we investigate this hypothesis by inferring membrane potential statistics from spiking activity using simulation-based inference. We find that, neurons decrease their membrane potential variability in response to task relevant stimuli. Furthermore, we show that this decrease in membrane potential variability is stronger for reliably detectable stimuli than weak ones. Our findings suggest that individual cortical neurons track uncertainty, likely providing Bayesian benefits for downstream computations.
