Our goal is to study the brain from the viewpoint of the computations that subtend various aspects of cognition, from higher-level perception to decision making, statistical inference and the monitoring of uncertainty. Those topics correspond to the fields of expertise of the teams’ PIs and constitute a representative sample of human faculties. Our team uses the computational viewpoint to unify different levels of description:
- Formalize computational models of various cognitive functions and account for the behavior collected during cognitive tasks.
- Uncover the corresponding algorithms, i.e. the successive steps in information processing, and relate their latent variables to brain recordings (M/EEG and fMRI)
- Identify the underlying neural codes and probe their organization with high resolution fMRI.
Florent MEYNIEL – website – publications
Are you sure about your own judgments? Does it actually matter? I am interested in understanding the computation and roles of uncertainty in the brain. More specifically:
(1) How does the brain estimate the uncertainty about its own judgments, and communicate this subjective confidence to others?
(2) How does subjective confidence regulate learning? In particular, how does it regulate the inference of trends and regularities in dynamic environments?
(3) How does subjective confidence regulate our decisions, such as the decision to explore certain options, or to seek out information by mere curiosity?
I combine several scales and approaches in order to understand how those three processes operate in the brain:
Computational level: the goal is to provide abstract, formal accounts of brain computations. I rely on Bayesian (and more generally probabilistic) inference to this end.
Macro- and meso- scales: the goal is to understand the coordination of brain-scale networks that subserves those computations (macro-level), and how the inferred variables and their uncertainty are encoded by large populations of neurons (meso-scale). I currently use fMRI and MEG to this end.
Chemicals: I am interested in the role of neuromodulators in those computations, which actually bridge gaps between single neurons (of which they modulate the activity and plasticity) and large-scale networks (in which they are released simultaneously). I use fMRI, physiological measures (pupillometry, heat beats) and pharmacological interventions to address those questions.
Evelyn EGER – publications
What are the brain representations and computations underlying our perceptual experience of high-level visual entities (object category, identity and properties of ensembles of objects, such as their number)? These properties being entangled or insufficiently specified in the retinal input, how does the brain achieve the abstractness necessary for successful recognition, through bottom up and/or top–down processing? What is the relation between the precision of neuronal representations at different levels of the cortical hierarchy and behavioral discrimination capacity (e.g., previous work on the relation between number acuity and fMRI decoding) and how is it affected in developmental disorders (e.g., dyscalculia)?
My work is based on fMRI in humans and psychophysical methods, which I would also like to relate to computational models in the future.
My use of fMRI methods attempts to access finer spatial scale information (within a single region and/or perceptual dimensions) beyond traditional macro-scale mapping. To that goal, I initially used fMRI adaptation methods and later multivariate decoding of fMRI patterns, recently moving to ultra-high-field fMRI at 7 Tesla with the aim of enhancing decoding sensitivity and later enabling direct mapping of meso-scale layouts in higher-level brain regions.
To reduce the gap between the information obtained from invasive measurements in animals and non-invasive techniques in humans, I also collaborate on and provide neuroscientific guidance for development and validation of ultra-high-field fMRI acquisition methods.