Cue-induced uncertainty and prediction error: effects on nucleus accumbens dopamine and behavioral responses to self-administered cocaine and saline

Understanding the process of associative learning between environmental stimuli and cocaine is essential for the prevention of drug-use relapse and long-term treatment of cocaine dependence. Based on contemporary learning theories, empirical studies using natural rewards have shown that cognitive factors, such as uncertainty and prediction errors, play an important role in the process of reward associative learning. Uncertainty is the lack of an accurate predictor for reward while prediction error is defined as the discrepancy between expected and received reward. In this dissertation, we focused on the role of uncertainty and prediction error in cocaine-associative learning. Olfactory and visual cues during self-administration/conditioning sessions were used to induce cocainereward expectation and uncertainty in operant trained catheterized Sprague Dawley rats. The influence of cue-induced uncertainty and prediction error on nucleus accumbens dopamine (NAcc DA) following self-administration of cocaine and saline in these conditioned animals was then measured using in-vivo microdialysis. Results showed that cocaine-stimulated NAcc DA was enhanced in the presence of cues signaling cocaine reward uncertainty (Uncertainty) as compared to animals expecting to get cocaine (Certainty). Also omission of expected cocaine reward (Prediction Error) resulted in a significant depression of NAcc DA levels below baseline. Recently diazepam (a positive GABAA modulator) has been shown to disrupt cocaine-induced LTP and it has been suggested that this disruption can block the acquisition of drug-associated memories. We therefore hypothesized that diazepam-pretreatment during conditioning sessions would disrupt the learned responses to cocaine and saline in the presence of cue-induced uncertainty and prediction error. Our results show that diazepam pretreatment duringconditioning sessions, blocked the differential cocaine-stimulated NAcc DA response to cue-induced certainty and uncertainty. Moreover, on omission of expected cocaine reward (Prediction Error) there was no significant depression of NAcc DA below baseline. The findings of this dissertation thus highlight the importance of cognitive factors (uncertainty and prediction errors) in the process of cocaine-associative learning. They also provide a platform to further explore the influence of these factors on other neuroadaptations during cocaine-associative learning, which will help us develop effective behavioral and pharmacological therapies to prevent drug-use relapse.