Cognitive flexibility refers to how individuals adapt their behaviour to changes in the environment. Although important for survival and wellbeing, cognitive flexibility is impaired in a wide range of neurological and neuropsychiatric disorders, including Parkinson’s Disease (PD) and Obsessive-Compulsive Disorder (OCD). Optimal flexibility is known to depend on dopamine (DA) neurotransmission in the central nervous system, but the precise mechanism and brain loci underlying the effects of DA on flexible decision-making remain unclear. In this thesis, cognitive flexibility was inferred in experimental rats by evaluating their performance on a reversal-learning task involving a simple discrimination between rewarded and non-rewarded stimuli. During reversal, subjects must adapt and respond to the formerly non-rewarded stimulus whilst ignoring the initially rewarded stimulus. Learning on this task thus requires constants shifts in behaviour in response to positive (rewarded) and negative (non-rewarded) feedback. The overarching hypothesis of my thesis is that DA modulates reversal learning performance by signaling positive and negative reward prediction errors (RPE) within the direct (rewarded) and indirect (non-rewarded) pathways, respectively.
To investigate this hypothesis, I used a range of experimental approaches to interrogate the neuromodulation of the direct and indirect pathways by DA. In chapter 4, I found dissociable effects of D1 and D2 receptor antagonists during different stages of serial visual reversal learning when administered into the nucleus accumbens shell. In chapter 3, I used a recently developed valence-probe visual discrimination task to dissociate different components of reversal learning and tested the extent to which these were dependent on D2 receptors. We found that the D2 agonist quinpirole impaired reversal learning when given systemically, an effect that depended on decreased sensitivity to negative feedback, but improved performance when given directly into the nucleus accumbens. In chapter 5, the synaptic location of D2 receptors involved in the modulation of reversal learning was evaluated. Using a post-synaptic probe compound (an adenosine 2A receptor antagonist) evidence is presented for a predominately post-synaptic locus underlying the effects of D2 agents on reversal learning. Finally, in chapter 6, an in-vivo optogenetics intervention was used to simulate activity in the mesoaccumbal and nigrostriatal circuits during reversal learning. Here, activation of the mesoaccumbal, not nigrostriatal, circuit modulated reversal learning on trials when the expected reward was omitted. Taken together, these original results provide support for a dissociable role of DA receptors and striatal sub-regions in learning from positive and negative feedback in reversal learning. These findings expand our understanding of the neural circuit mechanisms underlying cognitive inflexibility and highlight potential therapeutic targets to improve flexible decision making in PD, OCD and a range of other brain disorders.