Dopamine, decision-making, and aging : neural and behavioural correlates

Sammanfattning: On any given day, we make tons of decisions. These can be as simple as deciding how to dress or what to eat, or more complex, such as whether to spend or invest money. Good decision-making involves being able to select the best alternative from a range of options, and adjust one’s preferences based on what is happening in the environment. As humans get older, their ability to do this changes. Age-related changes in decision-making ability result from changes in brain structure and function, such as the deterioration of the brain’s dopaminergic system in old age. In this thesis, we used a sample of 30 older and 30 younger participants to investigate age-related differences in neural and behavioural correlates of value-based decision-making, which involves making decisions that can result in rewards and punishments. Such decisions are known to rely on dopaminergic functioning. In the brain, we have looked at neural activity reflecting value and reward prediction errors (RPEs), the availability of dopamine D1 receptors, and integrity of white matter microstructure. For the behavioural data, we have used computational modelling to disentangle motivational biases and other parameters reflecting parts of the learning process that underlies successful decision-making. In study 1, we investigated whether performance on a value-based decision-making task differed between the two age groups. We also looked at whether performance differences could be explained by differential neural processing of RPEs and expected value in the striatum and prefrontal cortex (PFC). We used a novel computational model to estimate expected value, decision uncertainty and confidence. We found that older adults earned fewer rewards on the task. The number of rewards earned could be predicted by the strength of the neural signal reflecting expected value in the ventromedial PFC (vmPFC), which was attenuated in older adults. Beyond age, the strength of this neural signal could be predicted by dopamine D1 receptor (D1-R) availability in the nucleus accumbens (NAcc). In study 2, we showed that integrity of white matter microstructure in the pathway between the NAcc and vmPFC also predicted the neural value signal in the vmPFC, independently of age and D1-R availability in the NAcc. In study 3 and 4, we focused on dissociating the effects of action and valence on neural and behavioural correlates of decision-making. In study 3, we used com-putational modelling to characterize faster learning to act in response to rewards, and abstaining from acting in response to punishments, as being the result of biased instrumental learning. Study 3 also showed that variability in dopamine D1-R availability could be divided into cortical, dorsal striatal and ventral striatal components. Regardless of age, dopamine D1-R availability in the dorsal striatal component was related to biased learning from rewarded actions. In study 4 we investigated anticipatory value signals after learning had reached an asymptote. We observed no differences between age groups in anticipatory neural responses to action and valence, and no relationship between anticipatory neural signals and dopamine D1-R availability. Older adults did show an attenuated punishment prediction error signal in the insula, compared with younger adults. The strength of differentiation between reward- and punishment prediction error signals in the insula was related to dopamine D1-R availability in the cortex. These studies have demonstrated that the existing theoretical framework sur-rounding the role of dopamine system in decision-making and aging fits with dopamine D1-R availability data and behavioural data in older and younger adults, and partly explain why older adults show behavioural differences in value-based decision-making tasks. Collectively, the studies in this thesis provide important multimodal evidence that increases our understanding of the neural correlates that underlie value-based decision-making and how they are affected in healthy aging.