Soutenance de thèse de Vanessa CARNEIRO MORITA

vendredi 24 novembre 2023 à 14:00
Publié le 21/11/2023

Vendredi 24 novembre à 14h, Salle de thèse n°2, Faculté de médecine de la Timone

Vanessa CARNEIRO MORITA (Equipe inVibe)

Soutiendra sa thèse de doctorat intitulée:

Behavioural characterisation of probabilistic representations for eye movements

Mme Anna MONTAGNINI – Directeur de thèse
M. Guillaume MASSON – CoDirecteur de thèse
Mme Marie-Helene GROSBRAS – Président
Mme Celine PAEYE – Examinateur
M. Laurent MADELAIN – Rapporteur
M. Karl GEGENFURTNER – Rapporteur

Tracking with our eyes a prey or an imminent danger is a vital function for the survival of almost all animals. Our nervous system accurately estimates the target visual motion and predicts its trajectory in order to anticipate sudden changes in direction, to compensate for the neural delays that are inherent to sensorimotor transformations and to optimally coordinate fast, ballistic saccades and slow pursuit eye movements. The latter can in turn efficiently inform other body movements and cognitive functions. Several motor and oculomotor models rely on an internal model of these trajectories. However, in the context of eye movements, current oculomotor models have oversimplified the nature of these internal representations, ignoring both the complexity of the sensory inflows and the stochastic nature of natural trajectories. Our group has previously proposed that a probabilistic inferential framework can unify low-level (i.e. visual motion processing) and high-level (i.e. learning contingencies) computations in order to explain why behaviour remains optimal under many challenging natural conditions. With this thesis we extended the previous literature showing a parametric relationship between the probability of visual motion properties and anticipatory smooth eye movements to more complex probabilistic contingencies, such as in contexts with a higher global uncertainty, with conditional (second-order) probabilities, and for different target kinematics (speed, acceleration). Our results are compatible with the Bayesian integration framework and in particular they indicate that anticipatory eye movements could be viewed as a dynamic readout of internal prior beliefs.