Figure 2.

(A) During manual action selection, there is activation of prefrontal cortex (PFC), pre-supplementary motor area (pre-SMA), lateral premotor cortex (PMC) and primary motor cortex (M1). These activations did not differ between patients with Parkinson's disease (PD) and control subjects, by voxel-wise group comparisons. (B) Dynamic causal modeling (DCM) was used to model the interactions among these regions. Forty-eight models were compared in all, differing in terms of anatomical connections, feed-forward versus feedback, and the connections which are subject to modulation by selection of action (FvS). The two leading models are shown in detail here (E1 and E2). (C) In healthy subjects, and patients on their dopaminergic medication, model E2 was more likely (by the posterior model probability, based on the free energy estimate of the log of model-evidence, adjusted for model complexity) in which the selection of action (FvS) was associated with greater connectivity of PFC to pre-SMA. When withdrawn from medication, to a relative “off” state, the connectivity pattern in PD patients changed to a state in which the selection of action was associated with greater connectivity between PFC and the PMC, model E1. This confirmed the hypothesis of a functional disconnection of the pre-SMA, and an enhanced role of the lateral PMC in action selection in PD. From Rowe et al. (2010).