A matrix method, developed to examine functional interactions between brain regions by correlating regional cerebral metabolic rates for glucose as determined by positron emission tomography in humans, was applied to regional metabolic data from 21 demented patients and 21 age-matched controls. The dementia group showed reduced frontal-parietal interactions, and a significant loss of correlations between left-right homologous regions, indicative of reduced integrative activity between these regions. The correlation approach also was applied to 14 autistic patients and 14 age- and sex-matched controls. Compared with controls, the autistic group has significantly fewer large correlations between subcortical nuclei and frontal/parietal regions. These results are consistent with the view that autism represents a dysfunction in neural systems associated with directed attention. The matrix method was applied to analyze glucose metabolism in awake Fischer-344 rats. Reduced correlations between left and right hemipheric brain regions were found in rats that had undergone corpus callosotomies, suggesting that interhemispheric interactions are mediated in part by callosal fibers. When cells in the cholinergic basal forebrain die, it was hypothesized that their cortical synaptic target sties can be reoccupied by axonal sprouting of other neurons from the basal forebrain. This neuroplasticity hypothesis leads to equations that are consistent with the quantitative data, makes specific predictions that can be tested experimentally, and suggests that the more extensive pathology of the presenile form of AD can be understood as a result of the decline in neuroplasticity with age.
