Learning in affective contexts The impact of incidental anxiety and outcome valence on decision-making and confidence

The impact of contexts on decision-making is observable in field and laboratory experiments. The mechanisms of context-induced changes on choice patterns have been widely investigated at both behavioral and neural levels. For instance, reference points in prospect theory explain why the same set of options is evaluated differently. However, it remains less clear how affect and prior performance are involved in decision formation over time. Reinforcement learning provides a framework to address this issue by quantifying this updating process. The present thesis combines a modified reinforcement learning task with methods from affective neuroscience, computational modelling and brain imaging to systematically investigate mechanisms underlying the effects of two affective framing manipulations (i.e., choice-irrelevant affect: incidental anxiety; choice-relevant affect: outcome valence) on belief (i.e., option value and confidence judgments) updating. Our results demonstrate that incidental anxiety (Chapter 2) and outcome valence (Chapter 2-4) have limited impact on learning per se. Importantly, the impact of outcome valence on reaction time and confidence judgments is robust across six behavioral experiment (Chapter 3) and fMRI study (Chapter 4). These findings not only imply that confidence, option value and context are updated simultaneously, but the robust valence-induced changes of confidence indicate metacognitive function might be more sensitive to affective framing manipulations. Besides the empirical evidence concerning the role of anxiety and outcome valence in reinforcement learning, we also established useful methodological and analytical approaches for future studies investigating emotion-cognition-metacognition interactions in the field of judgment and decision-making.