Participants fixated a central cross (3° diameter) for 1000 msec and made saccades as quickly as possible to a target, find more 10° to the left or right (50% probability). Saccades to targets on only one side were rewarded depending upon reaction time (with a discounting function as for the TLT), and the rewarded side (RS) was altered, without warning, after a series of trials. Rewards were acknowledged by the display of a pound coin and a number representing the reward magnitude in pence. Reward value was dependent on latency using a function

similar to that in the TLT. The RS changed every 10–14 trials. Participants performed two blocks of 120 trials. The difference in SRTs to the RS and unrewarded sides (US) was the measure of reward-sensitivity. KD received a single dose of Madopar 125 mg (100 mg l-dopa with a peripheral dopa-decarboxylase inhibitor, benserazide 25 mg), directly after the baseline tests. He was ABT-888 reassessed an hour later when peak l-dopa levels are reached.

To assess whether any effects on l-dopa were due to simply more experience on the tasks, six controls were also tested an hour after performing their first session. A second group of controls (N = 12) also received the same dose of l-dopa but in double-blind randomized fashion, receiving placebo/drug one week apart. KD was then given slowly increasing doses, reaching Madopar CR (long-acting preparation) 125 mg three times daily after eight weeks. Although there was moderate improvement in apathy, it was decided that there might be better response with a direct dopamine receptor agonist. l-dopa was therefore slowly discontinued and KD was off medication for 4 weeks (‘drug holiday’) before starting on the dopamine agonist ropinirole, initially .25 mg three times a day for 1 week, then increasing by .25 mg every week eventually Atorvastatin to reach 1 mg thrice daily after three weeks. After a further four weeks he was established on 4 mg once daily of the long-acting formulation of ropinirole (Requip XL). KD’s lesions (Fig. 1) involved the GPi bilaterally,

with greater involvement on the left. These lesions were not complete and it is important to note that part of the GPi was spared. Using a recently validated atlas of the pallidum (Prodoehl et al., 2008) we found only modest damage to GPe (external segment of the GPi) on the left. There was no involvement of the habenula, STN, septum, medial hypothalamus, midline thalamic nuclei, and bed nucleus of stria terminalis, verified using a MR adapted version (Krauth et al., 2010) of a histological atlas (Morel, 2007). Probabilistic diffusion tractography (Fig. 2) was used to examine the topography of pallidal connections to three cortical regions (Draganski et al., 2008). The region of GPi which is most strongly connected to LOFC and VMPFC was particularly affected, compared with projections to primary motor cortex (M1), more so on the left: VMFC > M1 left Z = 5.41, right Z = 3.