The cognitive functions carried out by human dorsolateral prefrontal cortex (Brodmann's areas 46 and 9 in particular) are not sufficiently understood. Alternative hypotheses have been made to suggest that this is either the seat of self-generated (willed) action (Frith et al., 1991), or the anatomical locus for working memory (Petrides et al., 1993). A problem with the working memory hypothesis of Petrides was that his language tasks involved monitoring of string of 10 digits, which exceed average short-term span. In this experiment we will use positron emission tomographic measurements of regional cerebral blood flow to study whether a distinction can be drawn in functional anatomical terms between these two hypotheses. In the same experiment on the same subjects we shall try to tease apart the anatomy of executive functions and short-term storage by comparing the pattern of activation associated with a random digit generation task and a a short-term memory task for digits which do not exceed span. Each of these tasks will be compared with appropriate baselines.