Mechanisms of belief-logic conflict detection and early intuitive reasoning

<p dir="ltr">The interaction between automatic and controlled reasoning requires a 'switch' mechanism to monitor incompatible automatic responses and engage controlled processes to further reflect on the problem and resolve the conflict (De Neys, 2023). Previous research demonstrates that people can detect conflict between beliefs and logic, or between multiple heuristic responses, to some degree. Understanding factors that influence conflict detection is considered a central issue for understanding when people are more likely to reflect on the problem and search for an alternative response. In this thesis we consider the limitations of assuming that perceived conflict is necessarily tied to normative benchmarks and investigates the mechanism of conflict detection from a metacognitive perspective. </p><p dir="ltr">Chapter 1 provides an overview of current issues in the study of conflict detection, general background on metacognition, and outlines how it can be informative for conflict detection research. Chapter 2 presents two experiments demonstrating a dissociation between where conflict is expected to happen and where it is observed in certain conditional inferences, due to these items containing a conflict between heuristics and beliefs, rather than the expected conflict between logic and beliefs. Chapter 3 applies the self-consistency model of metacognition to examine what cues contribute to the formation of confidence judgements and how confidence and the consistency of cues to confidence affect the probability of participants changing their mind when given an opportunity to reconsider their initial response. We found that in syllogisms, several cues contribute to confidence judgements, but belief-logic conflict is not one of them. Response changes are overall uncommon, but most likely on easy items, where participants correct their initial accidental errors. </p><p dir="ltr">Chapter 4 extends the metacognitive approach to the base-rate task, and manipulates confidence experimentally to demonstrate the role of overall inconsistency between cues in determining confidence, rather than the presence of belief-logic conflict specifically. Confidence and responses in both syllogisms and the base-rate task behaved as predicted by the self-consistency model of metacognition, and in some ways consistent with the uncertainty monitoring account of conflict detection. Chapter 5 summarises the findings from the three experimental chapters, discusses the implications of the results for the models of conflict detection, and outlines some further potential applications of metacognition research to reasoning.</p>