A specialized perceptual mechanism for processing information related to age or gender from faces

An important result of our previous study (Mouchetant-Rostaing et al., 2000a) was the observation, around 145-185 msec, of ERP activities associated with the processing of gender from faces, but not from hand stimuli - irrespective of whether gender extraction was task-relevant (’intentional’) or -irrelevant (’incidental’). In the present experiment, we found ERP effects in the same latency range with a similar fronto-central topography during explicit (intentional) discrimination of gender and age, and during implicit (incidental) discrimination of age.

These results have several implications: they replicate our previous findings for gender perception, they show that similar brain mechanisms can be engaged in age perception, and they indicate that age - like gender - may be implicitly processed irrespective of its relevance to the task. This set of observations supports the idea that age and gender of faces are analyzed by specific brain mechanisms, and is in accordance with Bruce and Young’s hypothesis according to which these physiognomic features are processed through a common module.

The Bruce and Young model, however, stipulates a strict sequential organization between the structural analysis of facial features and the processing of physiognomic information, an hypothesis that does not fit with our observations. Indeed, the time-course of ERP activities associated with age and gender processing (145-185 msec) indicates that physiognomic jugdements can begin before the structural encoding of faces is completed, and suggests parallel, rather than sequential, activations of the Structural and the Directed Visual encoding modules.

The neuroanatomical localization of the neural systems that mediate face perception has been largely examined in previous neuroimaging (review in Haxby et al., 2000; Kanwisher, 2000) and electrophysiological (Allison et al., 1999; McCarthy et al., 1999; Puce et al., 1999) studies which have identified discrete regions in the occipito-temporal visual extrastriate cortex, namely the lateral fusiform gyrus, the superior temporal sulcus and the inferior occipital gyri. In addition, these studies have generally shown that cognitively distinct aspects of face perception are mediated by distinct neural representations (review in Haxby et al., 2000). However, while PET experiments have shown the involvement of extrastriate visual regions including the lingual/fusiform gyri during the processing of gender (Sergent et al., 1992; Kapur et al., 1995; Andreasen et al., 1996), gender categorization in these experiments was considered either as a matched control condition or as an experimental condition contrasted with non-facial stimuli (e.g. gratings in Sergent et al., 1992), thereby not allowing for dissociating gender processing from basic visual analysis of faces (“structural encoding”).

On the other hand, neuropsychological studies on prosopagnosic patients with uni- or bilateral posterior temporo-occipital lesions have reported contradictory observations: some patients showed a preserved ability to process physiognomic information (Tranel et al., 1988; Sergent and Villemure, 1989), while others were found to be moderately impaired in judging the age or gender of faces (de Renzi et al., 1989; Bruyer et Schweich, 1991; Carlesimo and Caltagirone, 1995).

It is therefore difficult to know from neuroimaging and neuropsychological literature whether and to which extent age and gender are processed through neural systems separate from those involved in visual analysis of faces.

While the 145-185 msec ERP effect found in the present study for age and gender processing clearly differs from N170, it has been shown that activation of face-specific regions in extrastriate cortex may be manifested by several ERP components at the scalp surface (Bötzel and Grüsser, 1989; Jeffreys, 1989; Bentin et al., 1996; George et al., 1996). Particularly, the “Vertex Positive Potential” (VPP, Jeffreys, 1989) may represent the positive counterpart of a dipolar source located in the fusiform gyrus, the infero-temporal cortex or in deep limbic structures, such as the hippocampus (Allison, et al., 1994b; Bötzel et al., 1995; George et al., 1996; Jeffreys, 1996). It is therefore possible that our age/gender effects similarly represent volume-conducted potentials from these face-responsive regions, with a resulting dipolar source configuration globally different from that involved in the analysis of facial features (N170).

An alternative, although not exclusive, hypothesis refers to recent neuroimaging and neuropsychological findings that have shown the involvement of the amygdala and the ventro-medial prefrontal cortex in processing aspects of social knowledge (Chee et al., 2000; Phelps et al., 2000; Milne and Grafman, 2001; review in Allison et al., 2000). It is possible that the age/gender effects reflect activities from these cortical or subcortical structures. Indeed, facial age and gender information may be considered as relevant visual cues of social stimuli, and may therefore involve brain regions known to play an important role in automatic processing and evaluation of psychosocial perception. These regions are included in the physiological model of the distributed human neural system for face perception recently proposed by Haxby et al. (2000).