The Dynamics of Conflict in Reasoning [PhD Thesis]


This thesis explores conflict during reasoning, using a mouse tracking paradigm that provides a measure of participants’ instantaneous attraction towards competing response options. It focuses on two kinds of conflict: conflict between competing sources of information in inductive reasoning, and conflict between fast, automatic Type 1 processes and slower, deliberate Type 2 process in dual process accounts of reasoning. Going beyond traditional analyses of participants’ responses, the mouse tracking data reveal under what circumstances conflict occurs, and at what points in time participants were influenced by different factors, as well as something of the qualitative nature of this conflict. Chapter 1 reviews previous work on conflict in reasoning, and methods used in the past to study conflict in cognition. Chapter 2 introduces the details of the mouse tracking paradigm used in subsequent chapters.
Chapters 3 and 4 explore conflict between different sources of information. In Chapter 3, two experiments are presented that pit perceptual cues (visual similarity) against conceptual knowledge (shared category membership) in an inductive reasoning task, using both real and artificial stimuli. This reveals that participants are initially driven by perceptual cues, and only later draw on conceptual knowledge, and also that they are more likely to draw on this knowledge when the properties being reasoned about were related to the distinction between the categories. In Chapter 4, two further experiments are presented that pit associative and structured knowledge against each other in an induction task. Participants were again influenced by both kinds of information, but it was less clear to what extent both forms of knowledge interacted on a single trial.
Chapters 5 and 6 explore conflict in dual process theories of reasoning. Chapter 5 reports a judgement task where participants attempted to categorise people, iiiand could rely on either descriptions of them or on background base rate statistics. Consistent with a default-interventionist dual process model, participants were strongly influenced by descriptions from early in reasoning, but only drew on statistical information, and sometimes overrode description-cued responses on the basis of it, later on. Chapter 6 presents a variation the mouse tracking paradigm, as participants’ cursor movements over up to a minute were analysed as they chose between response options on a task where Type 1 processes cue an intuitively appealing incorrect response, the Cognitive Reflection Test. Results here are again consistent with a default-interventionist perspective, as movements were apparently initially driven by Type 1 processes, and correct responses involving an initial attraction towards the heuristic option.
Chapter 7 discusses the implications of these findings, both for theories of reasoning and for accounts of conflict in cognition. It also explores how the methods used here may prove fruitful in future reasoning research.