The basal ganglia are intimately connected to the frontal cortex via five fronto-striatal circuits. While the role of the frontal cortex in cognition has been extensively studied, the contribution of the basal ganglia to cognition has remained less clear. In Parkinson's disease, posteroventral pallidotomy (PVP) involves surgical lesioning of the internal section of the globus pallidus (GPi, the final output pathway from the basal ganglia) to relieve the motor symptoms of the disorder. PVP in Parkinson's disease provides a unique opportunity to investigate the impact of disruption of striatal outflow to the frontal cortex on cognition. We assessed executive function and working memory after withdrawal of medication in 13 patients with Parkinson's disease before and 3 months after unilateral PVP compared to 12 age-and IQ-matched normals assessed twice with an interval of 3 months. The tests used were: Wisconsin Card Sorting (WCST), Self-Ordered Random Number Sequences, Missing Digit Test, Paced Visual Serial Addition Test (PVSAT), and Visual Conditional Associative Learning Test (VCALT). After PVP, the patients performed significantly better on the Self-Ordered Random Number Sequences and the WCST, an improvement that was also observed in the normals across the two assessment and is therefore likely to reflect practice effects. Relative to the normals, the patients showed significant differential change following PVP on the Missing Digit Test and PVSAT, on which they performed worse after compared to before surgery, while the controls performed better on the second assessment. For the patients, performance on the VCALT also indicated deterioration after PVP, but the changes approached significance. The side of PVP had no effect on the results. The pattern of change observed 3 months after PVP was maintained at 15-month follow-up. The results suggest that striatal outflow to the frontal cortex may be essential for those aspects of executive function that showed deterioration after PVP. &

It has been suggested that the dorsolateral prefrontal cortex (DLPFC) is involved in free selection (FS), the process by which subjects themselves decide what action to perform. Evidence for this proposal has been provided by imaging studies showing activation of the DLPFC when subjects randomly generate responses. However, these response selection tasks have a hidden working memory element and it has been widely reported that the DLPFC is activated when subjects perform tasks which involve working memory. The primary aim of this experiment was to establish if the DLPFC is genuinely involved in response selection. We used repetitive transcranial magnetic stimulation (rTMS) to investigate whether temporary interference of the DLPFC could disrupt performance of a response selection task that had no working memory component. Subjects performed tasks in which they made bimanual sequences of eight nonrepeating finger movements. In the FS task, subjects chose their movements at random while a computer monitor displayed these moves. This visual feedback obviated the need for subjects to maintain their previous moves “on-line.” No selection was required for the two control tasks as responses were cued by the visual display. The attentional demands of the control tasks varied. In the high load (HL) version, subjects had to maintain their attention throughout the sequence, but this requirement was absent in the low load (LL) task. rTMS over the DLPFC slowed response times on the FS task and at the end of the sequence on the HL task, but had no effect on the LL task. rTMS over the medial frontal cortex (MFC) slowed response times on the FS task but had no effect on the HL task. This suggests that a response selection task without a working memory load will depend on the DLPFC and the MFC. The difference appears to be that the DLPFC is important when selecting between competing responses or when concentrating if there is a high attentional demand, but that the MFC is only important during the response selection task.