Skip to main content

Thesis defence : Léa ENTZMANN

Thesis defence

On 22 September 2022

The influence of perceptual and emotional characteristics of facial expressions on saccade programming

Facial expressions are complex visual stimuli, characterized by both a specific spatial configuration and the emotion they communicate. Their detection is essential, whether in the context of survival or in the generation of adapted social behaviours. Some models of emotional processing suggest that faces with emotional expressions, especially fearful ones, are detected rapidly, regardless of the observer's goal. This rapid detection is thought to involve a subcortical pathway, which connects the superior colliculus to the amygdala and processes only coarse visual information, transmitted through low spatial frequencies. However, not all research is in agreement. The aim of this thesis work was to clarify the processes by which emotional facial expressions (happy or fearful) influence the programming of saccadic eye movements, compared to neutral facial expressions. In particular, we tested the hypothesis of a rapid (< 100 ms) and task-independent detection, which would favour gaze orientation towards emotional, particularly fearful, faces. We hypothesised that this effect originated from low spatial frequency processing within the subcortical pathway. Through a series of saccadic choice experiments, the results of a first experimental chapter show a privileged processing of emotional faces, in particular happy ones, which captures the gaze more efficiently than neutral faces. However, this effect was not automatic, but rather dependent on the observer's task. We also systematically observed differences between saccade endpoints on happy and fearful faces. Specifically, saccades landed closer to the mouth for happy faces than for fearful faces. The results of a second experimental chapter demonstrate the importance of high spatial frequency information in the detection and initiation of saccades toward neutral and emotional faces. Using a convolutional neural network, we were able to identify the most diagnostic region for this task: the mouth. The results of a third experimental chapter, based on a neuroimaging study, show sensitivity to facial expressions in cortical regions, independent of spatial frequencies. However, we did not observe an effect of facial expressions in regions that constitute the subcortical pathway. In most of the regions studied, neural responses were stronger to faces presented at high rather than low spatial frequencies. The results of this third experimental chapter are discussed in relation to the statistics of our stimuli. Thus, the work conducted in this thesis, which combines behaviour, neuroimaging and modelling, suggests that emotional faces can attract the gaze more effectively than neutral faces. However, this effect would not be automatic. We propose that, particularly in tasks that require a quick response, emotional faces will not attract attention and gaze more than neutral faces. However, the physical characteristics of the expressions will modulate attention automatically, resulting in shifts in saccade endpoints. Finally, whether at the behavioural or neural level, our results did not reveal a privileged processing of fearful faces, based on the rapid extraction of low spatial frequencies. They thus questionned the involvement of the subcortical pathway in the processing of facial expressions.

- Directeur de thèse : Martial MERMILLOD - (martial[dot]mermillod[at]univ-grenoble-alpes[dot]fr)
- Co-encadrant : Nathalie GUYADER - (nathalie[dot]guyader[at]gipsa-lab[dot]grenoble-inp[dot]fr)

Thesis keywords: Emotional facial expressions, Eye movements, Visual attention, Spatial frequencies, Artificial neural networks, fMRI,


On 22 September 2022


01/10/2018 - 22/09/2022

Submitted on 20 November 2023

Updated on 20 November 2023