This hybrid functional/anatomical analysis resulted in clusters of voxels that showed changes with task condition within an anatomical region of interest. In the main analysis, voxels within each functional/anatomical region of interest were then collapsed for the second level ROI analysis comparing control subjects and aMCI patients on placebo using a t test. A separate paired samples t test was used to compare the placebo condition to the levetiracetam condition in aMCI

patients for both the main and confirmatory analysis. We would like to thank Dr. Jason Brandt, Dr. Paul Dash, Dr. Argye Hillis-Trupe, find protocol Dr. Majid Fotuhi, and Dr. Peter Rabins for help with participant recruitment and the staff of the F.M. Kirby Center for Functional Brain Imaging and Alica Diehl, Benjamin Drapcho, and Christina Li for their assistance with data collection. This work was supported find more by NIH grant RC2AG036419 to M.G. M.G. is the founder of AgeneBio. She is an inventor on Johns Hopkins University intellectual property with patents pending and licensed to AgeneBio, and she consults for the company and owns company stock, which is subject to certain restrictions under University policy. M.G.’s

role in the current study was in compliance with the conflict of interest policies of the Johns Hopkins School of Medicine. G.L.K. is an investigator and received research support from UCB Pharma. “
“Recent studies in monkeys have shown that neurons in the lateral habenula (LHb) become active when an animal fails to receive an expected reward or if the animal receives a signal

indicating a negative outcome (Matsumoto and Hikosaka, 2007), i.e., these Vasopressin Receptor neurons encode “antireward” conditions and compute reward prediction errors—the difference between the amount of reward expected and the amount of reward received, a computation that is thought to drive reinforcement learning (Sutton and Barto, 1998). LHb neurons have also been shown to inhibit dopaminergic neurons in the ventral tegmental area (VTA) (Ji and Shepard, 2007), which encode “reward” conditions (Schultz, 1997; but see Matsumoto and Hikosaka, 2009). These findings are consistent with the view that an antireward (LHb) nucleus inhibits a reward (VTA) center and drives negative reward signals in dopamine neurons. However, the nature of the inputs that drive aversive responses in the LHb, as well as their possible modulation by other neurotransmitters, is poorly understood. The globus pallidus internus (GPi), an output region of the basal ganglia, and its nonprimate homolog, the entopeduncular (EP) nucleus, are major sources of input to the primate (Kim et al., 1976 and Parent et al., 2001) and rodent (Herkenham and Nauta, 1977) LHb, respectively, as well as the thalamus (Filion and Harnois, 1978, Harnois and Filion, 1982 and Parent et al., 2001).