Measuring perception with confidence

A debate has waged in psychology research since the middle of last century. How much, if at all, do our thoughts, emotions, and desires impact on what and how we see? The classic view is that the mind is organised into encapsulated and impenetrable modules, with ‘seeing’ and ‘thinking’ being functionally distinct mental processes. In this view, the perception of red, or the sound of a bell, is entirely unimpacted by the individual’s satisfaction, or belief, or hunger. This modular view is contrasted with the so-called new look psychology in which all mental faculties are continuous with each other. This latter view suggests that our imaginations and thoughts can directly change the brain’s processing of sensory information. If this is true, then cognitive states should change what and how we see. If it is false, then cognitive states might alter behaviour only as a consequence of changing how sensory information is judged or weighted.This thesis introduces a novel approach to this long-standing question. The thesis comprises three key themes: 1) first-person subjective experiences, 2) the capability and limits of behavioural measurements, and 3) self-awareness or ‘meta’-cognition. The first theme highlights that an individual’s perspective of the world (i.e., perception) is private and cannot be measured directly. The second theme highlights that empirical measurement depends on observable behaviours and actions that a person takes when engaging with the world (i.e., choices). The third theme is that human participants in psychological experiments can not only make choices, but can additionally report on their own awareness or competence in these choices. This “meta” cognitive representation of one’s own mind provides an additional dimension of psychological data (i.e., confidence) from which to draw conclusions about the nature and behavioural impact of first-person subjective experiences. A principal argument of this thesis is that different mental processes, if they are distinct, ought to produce distinct signatures in behaviour. The question addressed in the following chapters is whether decisions and confidence track different aspects of sensory processing and cognitive judgements, and if they do, it suggests that confidence might be a key tool in understanding the structure of psychological processes.In behavioural experiments, choices are often taken as proxies for perception. This assumption is critiqued in the opening chapter with examples, which leads to the development of the empirical hypotheses. These hypotheses are expanded in the following chapter in which decision making is simulated. These simulations are a useful way to explore different models of decisions and confidence that might explain behavioural effects that persist across sequential choices (known as serial dependence). This chapter addresses a question prominent in the current psychology literature: is serial dependence a perceptual or cognitive phenomenon? The next two chapters provide empirical context of the simulation; participants are presented with physical stimuli and asked to make judgements about their appearance. Additional metacognitive judgements are also provided in the form of confidence. On the basis of these chapters, it becomes clear that confidence has diagnostic value in separating the behavioural effects of perception from the behavioural effects of changes in cognitive judgements. The concluding chapter contextualises the modelling and experimental findings and offers critical next steps in advancing this thesis into several contemporary questions in psychological science.The empirical work presented here leverages tools commonly used in the perceptual and cognitive sciences, including sensory adaptation, perceptual illusions, computational modelling, and simulation, all in order to explore the empirical link between subjective experience and behaviour. This thesis speculates that the debate between modular and continuous mental faculties could be resolved in many situations as a result of measuring both a person’s choices and their appraisal of their own confidence in that choice. Overall, this thesis offers a proof-of-concept that confidence judgements have diagnostic utility for separating the effects of perception and cognition. Potential challenges to, and extensions of, the present thesis are also discussed.