The functional anatomy of perceptual and semantic processings for odors was studied using positron emission tomography (PET). The first experiment was a pretest in which 71 normal subjects were asked to rate 185 odorants in terms of intensity, familiarity, hedonicity, and comestibility and to name the odorants. This pretest was necessary to select the most appropriate stimuli for the different cognitive tasks of the second experiment. The second one was a PET experiment in which 15 normal subjects were scanned using the water bolus method to measure regional cerebral blood flow (rCBF) during the performance in three conditions. In the first (perceptual) condition, subjects were asked to judge whether an odor was familiar or not. In the second (semantic) condition, subjects had to decide whether an odor corresponded to a comestible item or not. In the third (detection) condition, subjects had to judge whether the perceived stimulus was made of an odor or was just air. It was hypothetized that the three tasks were hierarchically organized from a superficial detection level to a deep semantic level. Odorants were presented with an air-flow olfactometer, which allowed the stimulations to be synchronized with breathing. Subtraction of activation images obtained between familiarity and control judgments revealed that familiarity judgments were mainly associated with the activity of the right orbito-frontal area, the subcallosal gyrus, the left inferior frontal gyrus, the left superior frontal gyrus, and the anterior cingulate (Brodmann's areas 11, 25, 47, 9, and 32, respectively). The comestibility minus familiarity comparison showed that comestibility judgments selectively activated the primary visual areas. In contrast, a decrease in rCBF was observed in these same visual areas for familiarity judgments and in the orbitofrontal area for comestibility judgments. These results suggest that orbito-frontal and visual regions interact in odor processing in a complementary way, depending on the task requirements.

This article addresses three issues in face processing: First, is face processing primarily accomplished by the right hemisphere, or do both left- and right-hemisphere mechanisms play important roles? Second, are the mechanisms the same as those involved in general visual processing, or are they dedicated to face processing? Third, how can the mechanisms be characterized more precisely in terms of processes such as visual parsing? We explored these issues using the divided visual field methodology in four experiments. Experiments 1 and 2 provided evidence that both left- and right-hemisphere mechanisms are involved in face processing. In Experiment 1, a right-hemisphere advantage was found for both Same and Different trials when Same faces were identical and Different faces differed on all three internal facial features. Experiment 2 replicated the right-hemisphere advantage for Same trials but showed a left-hemisphere advantage for Different trials when one of three facial features differed between the target and the probe faces. Experiment 3 showed that the right-hemisphere advantage obtained with upright faces in Experiment 2 disappeared when the faces were inverted. This result suggests that there are right-hemisphere mechanisms specialized for processing upright faces, although it could not be determined whether these mechanisms are completely face-specific. Experiment 3 also provided evidence that the left-hemisphere mechanisms utilized in face processing tasks are general-purpose visual mechanisms not restricted to particular classes of visual stimuli. In Experiment 4, a left-hemisphere advantage was obtained when the task was to find one facial feature that was the same between the target and the probe faces. We suggest that left-hemisphere advantages shown in face processing are due to the parsing and analysis of the local elements of a face.