Image generation and processing of generated images in the cerebral hemispheres

Recent computational models describing the contribution of the cerebral hemispheres to visual imagery have suggested an exclusive capacity of the left hemisphere to generate multipart images. A brief review of relevant findings indicates that the evidence presented in support of this suggestion is not entirely compelling; this prompted a reexamination of this issue in a lateral tachistoscopic study on normal adults. Sixteen subjects participated in two experiments in which they had to decide whether or not a lowercase letter contained a segment extending above or below the main body of the letter. This decision was made directly on lowercase letters in one experiment (perceptual task) and on their generated images in the other experiment (imagery task). The quality of the letters (clear or blurred) and the retinal eccentricity of stimulus presentation (small or large) were orthogonally manipulated. The perceptual task yielded no main effect of visual field but a significant interaction of visual field and letter quality. By contrast, the imagery task resulted in a left visual-field superiority but no interaction involving the visual fields--a departure from predictions based on current models of visual imagery. In addition, the pattern of results in the imagery task corresponded to that obtained with blurred letters in the perceptual task, suggesting limitations in spatial resolution of visual images. Implications of these results for models of cerebral lateralization and visual imagery are discussed.