DISCUSSION

In a previous study based on a similar paradigm, we had found that the ERPs elicited by task-irrelevant male and female stimuli differed during very early epochs (45±85 ms) between a condition where the stimuli of both genders were presented randomly mixed in the same run, and a condition where they were delivered in separate runs. These ERP differences were observed for both faces and hands, and were tentatively interpreted as re¯ecting an early, automatic process of rapid but coarse categorization between two wide classes of stimuli (male and female) [11]. In the present experiment, we still found similar ERP differences between conditions where male and female face stimuli were randomly intermixed or presented separately, although the differences were of smaller amplitude and shorter duration than those previously described. The fact that these differences were observed for two types of biological stimuli in the previous study (faces and hands) and are replicated here for faces, emphasizes the robustness of the effect whatever its interpretation.

In addition, the present results show that early differences (40±80 ms) can also be observed in the ERPs evoked by geometrical figures of two categories (grey and hatched shapes), according to whether the stimuli of the two categories are presented randomly mixed in the same run, or in separate runs. The effect, however, was significant only when the two stimulus categories were separated by salient visual features (wide hatched vs grey shapes). Such findings would suggest that these early ERP effects are not specific for biological stimuli, and, in line with our initial interpretation, might reflect the existence of coarse and automatic categorization mechanisms for rapid distinction between two wide stimulus classes.

Other explanations can however be proposed to account for these early ERP effects. First, they could be due to different attentional expectancies from the subjects (and thus to different physiological responses) according to the mode of stimulus presentation, with the runs including two classes of task-irrelevant stimuli attracting more attention than those including only one stimulus class. This hypothesis, however, is not supported by the absence of effects for geometrical shapes separated by finer visual characteristics (thin-hatched session).

Another hypothesis is that these early ERP differences at posterior sites may be related to neuronal habituation or refractoriness effects in the visual cortex. Such effects would be more marked in the wide-hatched session than in the thin-hatched session because the physical differences between the grey shapes and the hatched shapes are more conspicuous in the former than in the latter case. While this interpretation may explain the results of the shape session, it seems less likely that they may account for the effects in the face session. This is because no stimulus was repeated, and the physical features separating male faces (or hands) from female faces (or hands) are probably as complex and varied as the physical features separating male or female faces within gender. Therefore, similar refractoriness effects, or similar absence of effects, should have been observed in the processing of the faces within and between the two gender categories.

We therefore suggest that the early ERP effects reported here and in our previous experiment possibly reflect the existence of a coarse, low-level categorization process between two wide classes of biological (face) stimuli. Such rapid visual processes could be related to previous findings on the organization of the visual system in both human [6,8±11] and monkey [5,7], suggesting the existence of feed-forward pathways reaching directly higher-order visual cortical areas and acting massively in parallel [6,7].

These observations may reflect the expertise that humans have developed to process, in far more detail and depth [14,15] and/or more globally and rapidly, stimuli with strong psychosocial or biological significance, as faces or hands, compared to other common object categories.