MATERIALS AND METHODS

Subjects: Eighteen right-handed subjects (nine males, 19±33 years old) with normal or corrected-to-normal vision participated as paid volunteers. All gave their informed consent to participate in this study.

Stimuli and procedure: A total of 400 grey-scale digitally-scanned photographs of human faces, and 800 computer-generated two-dimensional shapes were used. Half of the faces were of men and the other half of women, and similarly half of the shapes were hatched and the other half were steadily grey. Among the 400 hatched shapes, 200 were wide-hatched and the other 200 were thinhatched. All the stimuli were matched for luminance and contrast, and were presented foveally, subtending a visual angle of 6.98, for 300 ms at a rate of one every 1400 ms. A fixation cross was present at the center of the screen between each presentation. No stimulus was repeated.

The experiment consisted of three sessions of three runs each. In the first session, we presented male and female faces (face session); in the two other sessions (shape sessions), counterbalanced across subjects, we presented either grey shapes and wide-hatched shapes (wide-hatched session), or grey and thin-hatched shapes (thin-hatched session). In all the sessions and runs, the subjects performed an oddball detection task reporting verbally, at the end of each block of 50 stimuli, the correct number of target items (20%) delivered randomly among non-target items. In the face session, the targets were faces with eyeglasses delivered among faces without glasses, while in each of the shape sessions the targets were shapes with a hole delivered among shapes without holes. In the three sessions, runs 1 and 2, referred to as one-class runs, included 100 stimuli each of one category (male in one run and female in the other run in the face session; grey shapes in one run and hatched shapes in the other run in the shape sessions). The order of runs 1 and 2 was counter-balanced across subjects. Run 3 included 200 stimuli: in the face session, there were 100 male and 100 female faces randomly mixed, and in the shape sessions, 100 grey and 100 hatched shapes. The rationale was that the level of processing required in run 3 was the same as in runs 1 and 2 but, in addition, subjects could also incidentally categorize the stimuli according to their gender (male/female) in the face session, or according to their visual characteristics (grey/hatched) in the shape sessions. Run 3 was referred to as two-class run, and always followed runs 1 and 2. This order was designed to prevent any interference of a possible incidental categorization process in the one-class runs.

EEG recording: The EEG was recorded continuously via 32 electrodes mounted on a custom-made cap with a nose reference, sampled at a rate of 1 kHz (0.10±200 Hz analog filter bandwidth), and stored for off-line analysis. Twenty electrodes were placed according to the international 10-20 system (Fz, Cz, Pz, Iz, Fp1, F7, F3, C3, P3, T3, T5, O1 and their counterpart on the right hemiscalp), and two at the left and right mastoids (Ma1 and Ma2). The remaining electrodes were placed midway between two standard positions: TP3 (between T3 and P3), CP1 (between C3 and Pz), P13 (between O1 and P3), POz (between Pz and Oz), IMa (between Iz and Ma1), and their counterpart on the right hemiscalp. The EOG was monitored from the outer canthus of the right eye and from Fp2. Electrode impedance was kept below 5 kΩ. ERPs were averaged separately for each stimulus type over a 500 ms period, including a 100 ms prestimulus baseline, and digitally filtered (0.10±20 Hz). Trials with EEG or EOG exceeding ± 150 μV were excluded from averaging.

Data analysis: Only the responses to non-target stimuli were analyzed. Within each session, the ERPs elicited in runs 1 and 2 were grouped (one-class condition, 160 stimuli) and compared with the ERPs elicited in run 3 (two-class condition, 160 stimuli). ERP analysis was restricted here within the first 100 ms post-stimulus. Student t-tests comparing the amplitude of the differences to zero were computed for each sample at each electrode. We considered as significant differences, the spatio-temporal patterns having a stable topography with a significant amplitude ( p ,0.05) for ≥15 consecutive time samples [6,12]. Scalp potential maps were generated using a two-dimensional spherical spline interpolation [13] with color scaled normalized to the peak voltage value at the considered latency.