Frontostriatal circuitry is critical to learning processes, and its disruption may underlie maladaptive decision making and the generation of psychotic symptoms in schizophrenia. However, there is a paucity of evidence directly examining the role of modulatory neurotransmitters on frontostriatal function in humans. In order to probe the effects of modulation on frontostriatal circuitry during learning and to test whether disruptions in learning processes may be related to the pathogenesis of psychosis, the authors explored the brain representations of reward prediction error and incentive value, two key reinforcement learning parameters, before and after methamphetamine challenge.
Healthy volunteers (N=18) underwent functional MRI (fMRI) scanning while performing a reward learning task on three occasions: after placebo, after methamphetamine infusion (0.3 mg/kg body weight), and after pretreatment with 400 mg of amisulpride and then methamphetamine infusion. Brain fMRI representations of learning signals, calculated using a reinforcement Q-learning algorithm, were compared across drug conditions.
In the placebo condition, reward prediction error was coded in the ventral striatum bilaterally and incentive value in the ventromedial prefrontal cortex bilaterally. Reward prediction error and incentive value signals were disrupted by methamphetamine in the left nucleus accumbens and left ventromedial prefrontal cortex, respectively. Psychotic symptoms were significantly correlated with incentive value disruption in the ventromedial prefrontal and posterior cingulate cortex. Amisulpride pretreatment did not significantly alter methamphetamine-induced effects.
The results demonstrate that methamphetamine impairs brain representations of computational parameters that underpin learning. They also demonstrate a significant link between psychosis and abnormal monoamine-regulated learning signals in the prefrontal and cingulate cortices.