Technical Program

Paper Detail

Paper:	PS-2A.65
Session:	Poster Session 2A
Location:	H Lichthof
Session Time:	Sunday, September 15, 17:15 - 20:15
Presentation Time:	Sunday, September 15, 17:15 - 20:15
Presentation:	Poster
Publication:	2019 Conference on Cognitive Computational Neuroscience, 13-16 September 2019, Berlin, Germany
Paper Title:	Confirmation Bias is explained by Descending Loops in the Cortical Hierarchy
Manuscript:	Click here to view manuscript
License:	This work is licensed under a Creative Commons Attribution 3.0 Unported License.
DOI:	https://doi.org/10.32470/CCN.2019.1433-0
Authors:	Vincent Bouttier, Sophie Denève, Ecole Normale Supérieure, France; Renaud Jardri, Université de Lille, France
Abstract:	In order to carry decision-making based on several pieces of evidence, one must integrate information over time. An optimal Bayesian observer would simply use Bayes' rule to combine the past knowledge with the new evidence. In this work, we tackle malfunctioning of inference, motivated by biological considerations and the recurrent structure of the brain. Allowing for loops of information when excitation and inhibition are unbalanced, we derive a functional Bayesian model of suboptimal inference, where the likelihood is corrupted by the prior knowledge. We show that, depending on the level of reverberation of the prior information, this "circular inference" model can explain cognitive biases often observed experimentally as the recency effect, the primacy effect, and the confirmation bias. The model is able to fit behavioural data on a task where healthy subjects were injected low doses of ketamine, a hallucinogenic drug thought to modify the E/I balance in favor of excitation. This work could allow to relate the microscale anomalies (E/I imbalance), the mesoscale anomalies (anomalies in frequency bands) and the macroscale anomalies (behavioural suboptimality and cognitive biases) observed in the psychotic state and under hallucinogenic drugs.