Thesis Eva APRILE

Our visual system is particularly good at recognizing a scene at a quick glance. To achieve this ultra-rapid recognition, the human visual system relies on a sparse and global representation of the visual scene, called the gist. The rapid extraction of the gist of a scene would also make it possible to make predictions on the objects of the scene, thus facilitating their perception. However, this theoretical proposition is based on studies that present the scenes very rapidly and where only one ocular fixation is then possible. However, visual perception is a dynamic phenomenon that alternates between ocular fixations on an object of interest and saccades towards the periphery to fix new objects of interest. To be rapid and efficient when scanning our visual environment, we postulate that the human visual system would also use predictions from the gist.
The aim of this thesis project is to characterize how predictions based on our knowledge of the visual environment can orient future ocular fixations in peripheral vision, through the development of new experimental paradigms that will involve measures of eye movements. To pursue this research aim, we identified three research operations. Operation 1 aims to study how predictions from the context (gist) of a scene influence gaze orientation in peripheral vision when visual processing needs to be rapid. Operation 2 concerns the free and more natural exploration of scenes and aims to study how the predictions that can be generated during an eye fixation influence the following eye fixation. Operation 3 aims to study how real-time changes in knowledge about a scene (and therefore the resulting predictions) influence scene exploration. For each operation, we will conduct several behavioral studies where we will measure eye movements (duration of eye fixations, latency/duration and amplitude of saccades, etc.) using an Eyelink 1000 eye tracker (SR Research).
This project is based on the collaboration between Carole Peyrin (DR2 CNRS at LPNC) and Nathalie Guyader (MCF at GIPSA-lab). It will contribute to improving the cognitive models of visual perception developed by the LPNC, and the models of signal processing and eye movements developed by the GIPSA-lab.

Supervisor :
Carole PEYRIN, DR CNRS au LPNC carole.peyrinuniv-grenoble-alpes.fr (carole[dot]peyrin[at]univ-grenoble-alpes[dot]fr)
Nathalie GUYADER, MCF UGA au GIPSA-Lab nathalie.guyadergipsa-lab.grenoble-inp.fr (nathalie[dot]guyader[at]gipsa-lab[dot]grenoble-inp[dot]fr)

Keywords : Eye movements,Visual perception,Predictive mechanisms,