Probability learning and confidence – computational modeling and neural mechanisms

Supervisor and contact: Dr Florent MEYNIEL

Project description

Humans have the ability to estimate probabilities that characterize their environment. They also have the
ability to assess the confidence, or uncertainty, associated with these estimates, which underlies adaptive
learning (i.e. the ability to adjust how much is learned from each observation). Intelligent behavior in
changing environments, in turn, critically depends on adaptive learning. This ERC project aims at
characterizing computationally, psychologically and neurally the estimation of this confidence and its role in
adaptive learning. The postdoc will contribute to one or more sub-projects. All of them rely on already
collected data (behavior, MEG, fMRI at 3T and 7T), and some sub-projects will involve data collection
(behavioral data, pupillometry data, possibly MRI, MEG or EEG).

PhD in neuroscience, psychology or machine learning with good programming skills (ideally Python).
Previous experience with fMRI, MEG, pupillometry, or computational modeling. Depending on the project,
you will be responsible for conducting experiments (behavior, pupillometry, fMRI, MEG or EEG), analyzing
data, and disseminating the results at conferences and in journal articles.
The language of the lab is English. French is not required.

Work place and environment
Dr Florent MEYNIEL leads the Computational Brain team (more info here), which is located in two places.
1) Institute for NeuroModulation (INM), Saint Anne Hospital, Paris, France. The INM is part of the GHU
Paris, Psychiatry & Neuroscience. The INM combines clinical activity and innovative clinical research in
psychiatry with basic research in computational neuroscience. Team members spend most of their days here.
2) NeuroSpin, Paris-Saclay campus, France. NeuroSpin is part of the CEA (Commissariat à l’Energie
Atomique). NeuroSpin is directed by Prof. Stanislas DEHAENE, it is a world-class brain imaging center,
equipped with a MEG system (Elekta, Neuromag), and several human MRI scanners (3T Prisma, 7T, and
11.7T), all research-only. The community at NeuroSpin is very stimulating, combining MRI physicists,
machine learning experts and cognitive neuroscientists. Team members go there to collect data and
collaborate with their colleagues in the Cognitive NeuroImaging Unit.

Duration and dates
Two years, plus a possible 1-year extension. Full time.
Preferred starting date: January 2024, but sooner or later is also possible.

Application process

Please send your CV via email to Florent Meyniel (florent.meyniel AT, a research statement (what you like and want to do) and
the contact details for two reference letters.

Application will be assessed on a rolling basis (positions open until it they are filled)

Salary: According to the CEA standards. Commensurate to experience.