Positron emission tomography study of letter and object processing: empirical findings and methodological considerations.

The study of functional-anatomical correlations of higher-order cognitive processing has benefited from recent advances in brain imaging techniques such as positron emission tomography (PET) measurements of regional cerebral blood flow (CBF). Comparisons of CBF changes by paired image subtraction provide the opportunity to isolate cerebral areas participating in the realization of the processes that differentiate two tasks. However, the subtraction method is based on assumptions that are not entirely compatible with cerebral cognitive processing, and the derived pattern of activation specifically associated with the processes that differentiate two tasks is relative to the activation associated with the subtracted task and may therefore vary as a function of the processes actually performed in this subtracted task. To examine the implications of this procedure, a PET study with the 15O water bolus technique was carried out on normal adults. Subjects performed three tasks that made nonoverlapping cognitive processing demands: a semantic categorization of visual objects, a spatial discrimination of visually presented letters, and a phonological decision on visually presented single letters. Each task produced distinct patterns of activation consistent with evidence from neurological patients, specifically in the left occipital cortex in the semantic categorization of objects, in the parietal cortex of both hemispheres in the letter-spatial task, and in the left frontal and superior temporal cortex in the letter-sound task. However, the comparisons between the two letter tasks did not result in the expected CBF changes even though these two tasks make distinct processing requirements and are dissociable by brain injury. In addition, the phonological task resulted in activation of areas of the frontal cortex that earlier PET studies had identified as participating in semantic operations, whereas letters have no semantic property. These results suggest that the interpretation of patterns of activation is confronted with difficulties due to the automatic, and uncontrolled, processing of verbal stimuli that raises the threshold for significant CBF changes between two conditions that use the same stimuli but different task demands.