Using functional neuroimaging to determine local neuronal activity, we have found that normal subjects performing tasks involving working memory use a cortical network including dorsolateral prefrontal cortex, inferior parietal lobule, and inferior temporo-occipital cortex. In other tasks related to prefrontal cortex, we have shown, with word generation (verbal fluency) tasks that semantic and phonologic cues activate similar brain regions including anterior cingulate, left frontal cortex, thalamus and cerebellum, but subtle differences exist between them that are consistent with the lesion literature. In a study of cognitive activation in normal aging we found that neurophysiological changes were context dependent. That is, apparent changes over the life span differed in different tasks, depending on the role of the particular neural system for the particular task. In regions where physiological activity is normally suppressed when young subjects perform the tasks older subjects activate more, and, moreover, the more they activate (or fail to suppress), the worse they perform on the tasks. In other areas, where physiological activity is normally increased in young people performing the tasks, older subjects activate less; and the less they activate these regions, the more impaired their performance. We have also confirmed that large-scale changes in brain activity at the systems level (accompanying nonlinear changes in behavior) follow predictions from nonlinear dynamical theory, and we have defined discriminable frontal lobe regions (lateral prefrontal cortex and anterior cingulate cortex) that subserve different cognitive components in switching between tasks (overcoming the residual inhibition of previous task demands and initiating a new task, respectively). We have also shown that the neural system patterns activated in individual healthy subjects reflect their levels of performance and cognitive strategies (verbal versus spatial) on a working memory task.