Hélène (QUESTE) DEVILLEZ
PhD, Cognitive Sciences
Research

Load my thesis      View my defend




Thus far, my research has focused on high-level and low-level information influences on attention during the exploration of natural scenes. I collected behavioral data as well as recorded either eye-movement data or co-recorded eye-movement and EEG data. During these experiments, participants viewed natural scenes under different task instructions. I independently used statistical modeling and interpreted the results in order to further develop an understanding of gaze control when exploring natural scenes. The co-registration of eye movements and EEG activity offers a unique opportunity for insight into visual processing. On the one hand, EEG allows for the study of brain activity linked to specific visual events (referred to as visual evoked potentials), which can reflect visual processing with an extremely high temporal resolution. On the other hand, eye tracking allows us to know exactly where observers are looking and for how long.

Currently, I work as a postdoc on an ONR-funded grant focusing on top-down influences on visual object recognition in Curran’s lab (University of Boulder Colorado), in collaboration with Randy O’Reilly (UC Boulder, computational modeling), David Sheinberg (Brown, primate electrophysiology), and Tor Wager (UC Boulder, fMRI). The primary objective of the research is to further understand, and capture in a computational model, the powerful bidirectional constraint satisfaction processing, both within the object recognition system (i.e., the ventral What visual pathway) and between this system and spatial attention supported by the dorsal Where pathway in order to answer three major overarching questions:
1) What is the nature of bidirectional processing in the neocortex?
2) How does bidirectional processing support figure-ground organization in cluttered scenes?
3) How do the What and Where pathways interact to enable effective search, figure-ground processing, and object recognition in cluttered visual scenes?


During my thesis, I focused on the effect of task on scene exploration using simultaneous recording of eye- movements and EEG experimental design. This study aimed to better understand the neural components underlying the effects of various cognitive tasks (free-viewing, scene categorization, single-target visual search and a dual-target visual search) on the processing of natural scenes. In a follow-up analysis, I compared the patterns of results from different tasks on the processing linked to the identification of a target object. The results show a specific EFRPs, called P300, which was elicited when observers fixated the search target.

Expertize

Experimentation:

- designing experimental protocols in eye-tracking and conjoint recordings of eye-tracking and EEG

- submitting a proposal to the ethic committee

- running experimentations: recruitment and recordings with EyeLink1000 and Gtec devices

Data analysis:

- processing data of the eye-tracker with Matlab (raw data, fixation and saccade characteristics)

- statistical analysis (ANOVA) and statistical modeling (Expectation-Maximization)

- evoked potentials analysis with EEGlab (evoked-related potentials and eye-fixation related potentials)

- EEG time frequency analysis (EEGlab) and EEG source localization (sLoreta)

Computer literacy:

- object-oriented language: Java, C/C++

- Matlab (included EEGlab)

- Statistics: Statistica, R

- Prolog

- Office software: Microsoft Office, Open Office, LateX, image editing software (The Gimp)