Bayesian inference optimally estimates probabilities from limited and noisy data by taking into account levels of uncertainty. In line with this principle, probability estimates are accompanied in humans by rational confidence levels that denote the precision of probability estimates. For this reason, Florent Meyniel has made the proposal that the human sense of probability is Bayesian. This Bayesian nature constrains the estimation, neural representation and use of probabilities, which he aims to characterize by combining psychology, computational models and neuro-imaging.

We aim to characterize the Bayesian sense of probability computationally and psychologically. Human confidence as Bayesian precision is our starting point, we now investigate the human algorithms that approximate Bayesian inference. We are interested in whether confidence depends on explicit reasoning whether it is trainable and domain-general.

We are also interested in the neural codes of Bayesian probabilities, leveraging encoding models for functional magnetic resonance imaging (fMRI) and goal-driven artificial neural networks to propose new codes. 

Last, we investigate a key function of confidence in the sequential estimation of probabilities: the regulation of learning. We test the implication of neuromodulators such as noradrenaline in this process, using the activity of key neuromodulatory nuclei (with fMRI) and test for causality with pharmacological intervention.