A functional magnetic resonance imaging (fMRI) study was conducted to determine whether prefrontal cortex (PFC) increases activity in working memory (WM) tasks as a specific result of the demands placed on WM, or to other processes affected by the greater difficulty of such tasks. Increased activity in dorsolateral PFC (DLPFC) was observed during task conditions that placed demands on active maintenance (long retention interval) relative to control conditions matched for difficulty. Furthermore, the activity was sustained over the entire retention interval and did not increase when task difficulty was manipulated independently of WM requirements. This contrasted with the transient increases in activity observed in the anterior cingulate, and other regions of frontal cortex, in response to increased task difficulty but not WM demands. Thus, this study established a double-dissociation between regions responsive to WM versus task difficulty, indicating a specific involvement of DLPFC and related structures in WM function.

The anterior cingulate cortex (ACC), on the medial surface of the frontal lobes of the brain, is widely believed to be involved in the regulation of attention1,2. Beyond this, however, its specific contribution to cognition remains uncertain. One influential theory has interpreted activation within the ACC as reflecting ‘selection-for-action’3,4,5, a set of processes that guide the selection of environmental objects as triggers of or targets for action. We have proposed an alternative hypothesis, in which the ACC serves not to exert top-down attentional control but instead to detect and signal the occurrence of conflicts in information processing6,7,8. Here, to test this theory against the selection-for-action theory, we used functional magnetic resonance imaging to measure brain activation during performance of a task where, for a particular subset of trials, the strength of selection-for-action is inversely related to the degree of response conflict. Activity within the ACC was greater during trials featuring high levels of conflict (and weak selection-for-action) than during trials with low levels of conflict (and strong selection-for-action), providing evidence in favour of the conflict-monitoring account of ACC function.

Although recent neuroimaging studies suggest that prefrontal cortex (PFC) is involved in working memory (WM), the relationship between PFC activity and memory load has not yet been well-described in humans. Here we use functional magnetic resonance imaging (fMRI) to probe PFC activity during a sequential letter task in which memory load was varied in an incremental fashion. In all nine subjects studied, dorsolateral and left inferior regions of PFC were identified that exhibited a linear relationship between activity and WM load. Furthermore, these same regions were independently identified through direct correlations of the fMRI signal with a behavioral measure that indexes WM function during task performance. A second experiment, using whole-brain imaging techniques, both replicated these findings and identified additional brain regions showing a linear relationship with load, suggesting a distributed circuit that participates with PFC in subserving WM. Taken together, these results provide a “dose–response curve” describing the involvement of both PFC and related brain regions in WM function, and highlight the benefits of using graded, parametric designs in neuroimaging research.