In the continuous flow of sensory evidence, cognitive systems must provide rapid behavioral choices across different time scales. For instance, seeing a moving object may result in various responses such as catching or avoiding collision depending on the trajectory and the nature of the object, but also depending on the recent experience and the expectations associated with that object and its motion properties. The principal goal of the larger scientific project in which this PhD thesis is inscribed (see ANR-REM project) is the analysis of reinforcement learning processes in the domain of voluntary eye movements (saccades and smooth pursuit eye movements) in humans. Within this PhD project we will use a dual approach, based on behavioural experiments on human subjects and on computational modelling of the experimental data, in order to address this important question, with a particular emphasis on the time course of learning effects and on the hypothesised role of probabilistic inference as underlying mechanism. <
> Visually driven eye movements provide an ideal experimental preparation to probe sensorimotor behavior across different time-scales, processing levels (from sensory encoding to the final categorical choice) and movement repertoire (e.g. smooth pursuit and saccades). In addition, a remarkable flexibility of oculomotor behaviors has been highlighted by manipulating the expectancy for sensory features or the outcome associated to particular motor responses.
Phd in Computational Neuroscience, 2017