Electrophysiological evidence for a natural/artifactual dissociation

The aim of this work was to test whether the differences usually found between the processing of visual stimuli corresponding to natural and artifactual domains reflect the different ways in which these domains are organized in the brain or are rather due to varying tasks demands. For this purpose, we designed two tasks in which subjects had to classify a series of line drawings. In one task (semantic categorization), the subjects were asked to categorize the stimuli as corresponding either to the natural or the artifactual class, and in the other (gender decision), the subjects had to decide if the names of the stimuli corresponded to either the masculine or the feminine gender. Event-related potentials (ERPs) and reactions times (RTs) were registered during the two tasks. We found both quantitative and topographical differences between ERPs elicited by natural stimuli and those by artifactual stimuli. In the 50- to 200-ms period, ERPs were more positive for the natural stimuli in the categorization task, but more positive for the artifactual stimuli in the gender decision task. In addition, natural stimuli elicited larger P600 and were associated with shorter RTs than artifactual stimuli in the categorization task. These results likely reflect differences concerning the relative difficulty of processing the stimuli of each domain in each task. In the N400 range, in contrast, there were differences between the two domains which were independent of task. In the two tasks, natural and artifactual stimuli elicited ERPs with a different scalp distribution: ERPs were more positive at posterior (parietal and temporal) sites for the natural stimuli and in the frontal areas for the artifactual ones. This finding is consistent with the hypothesis that semantic knowledge associated with natural and artifactual domains is represented in separate subsystems with presumably different anatomical bases.