Neuro-imagerie de la reconnaissance des visages chez les sujets sains et le patient neurologique
(2010)
Files
Thèse_Dricot_30septembre_version-impression_FINAL.pdf
Open Access - Adobe PDF
- 23.03 MB
Details
- Authors
- Supervisors
- Rossion , Bruno
- Abstract
- (en) In our work, based on structural brain imaging and functional magnetic resonance imaging (MRI and fMRI), we aimed to clarify the functional neuroanatomy of face recognition and in particular the neural network involved in face perception. Rather than adhering to hierarchical models typically represented in the face literature (Haxby et al. 2000 ; Gobbini & Haxby, 2007 ; Fairhall & Ishai, 2007), we documented a reverse model (Rossion, 2008) which proposes that activation of the "Fusiform Face Area" (FFA, Kanwisher et al., 1997) is before the activation of the "Occipital Face Area" ("OFA", Gauthier et al., 2000) located nevertheless posteriorly. Inspired by the early research of Justine Sergent (Sergent, 1986) and in line with current developments in object recognition (Hochstein et al., 2002, among others) as well as recent studies in monkeys (Freiwald et al., 2009), this model assumes that a first global and coarse representation of the face in the "FFA" (composed of neurons with large receptive field) is then refined through interactions with neurons of the "OFA" (composed of neurons with smaller receptive field) for the individualization of faces. The faces, regardless of their representation -life photos, cartoons, low spatial frequency faces, etc.- are first detected and classified as a "face" in a holistic manner by the "FFA" which automatically and simultaneously collects the features AND their spatial relations (e.g., two eyes above the nose itself above the mouth). This holistic detection present in the two prosopagnosic patients (with lesions around the “OFA”) described in our studies (Dricot et al., 2008) would activate the "FFA" in fast but relatively coarse way (Goffaux et al., 2010, see also Sugase et al., 1999, Sripati & Olson 2009 for studies in monkeys), probably by inputs from early visual areas such as the inferior longitudinal fasciculus. This coarse representation would provide a "template" or guide to a holistic identification that requires the precise integration of multiple features and their spatial relationships to manage very subtle differences between faces, with the help of more posterior regions whose receptive fields are smaller. This holistic identification based on a second wave of activation at least partially from the "OFA" through a reentrant loop in the "FFA" is missing in our two patients with "OFA" no longer exist. Low residual abilities of face recognition of the patient PS did indeed rely on a strategy that focuses on the mouth (Caldara et al. Xivry 2005 Orban et al., 2008, Van Belle et al., 2010). She is unable to handle the eye area because multiple features and their spatial relationships must be finely integrated with the help of the “OFA. The face detection and individualization would be underpinnedrely on overlapping neural circuits in the right "FFA", which plays the key role in both coarse and holistic coding of faces. For face individulization, the “FFA” would need the additional help from other regions of the face network, such as the “OFA”.
- Affiliations
Citations
Dricot, L. (2010). Neuro-imagerie de la reconnaissance des visages chez les sujets sains et le patient neurologique. https://hdl.handle.net/2078.5/131092