Neural correlates of human time perception

Abstract

In order to coherently perceive and effectively interact with the world around us, we must integrate sensory signals from various modalities (e.g., audition and vision). One of the key factors determining whether sensory signals are integrated is their relative timing. Despite over a century of study, it remains unclear what processes underlie our brain’s capacity to extract and evaluate the timing of, and between, sensory signals. Many proposals have been made regarding the computational processes that might be implemented by the brain to facilitate timing perception. However, no consensus has emerged regarding even the most basic assumptions we ought to build these models on. The purpose of this thesis is to address three contemporary issues in the field of time perception, with a focus on testing model assumptions and predictions, so that we might work toward a unified and biologically plausible model of the neural processes underlying multisensory timing perception.A computational model of some perceptual or decisional process is only as good as the validity of its predictions. The first of the three issues addressed in this thesis concerns the surplus of models in the timing perception literature, and the few methods of verifying them. Chapter 2 of this thesis provides a novel and simple tool with which existing and future models of timing perception can be evaluated. This chapter demonstrates that participants’ subjective confidence in their timing judgements can be used to evaluate the predictions of models of timing perception.Chapter 3 takes a broader approach, and tests an assumption made by a whole category of model. It is commonly assumed that the perceived timing of physical events is, at least partially, determined by the timing of corresponding neural events. This brain-time assumption has provided a foundation upon which many models have been proposed, but there exists very little evidence to suggest this is a reasonable assumption. The validity of this assumption is the second issue this thesis addresses. Chapter 3 presents an investigation of whether individual differences in sensitivity to timing relationships can be attributed, even in part, to variance in the time-course of neural processing. Analysis of EEG data provided no evidence to support this assumption, prompting consideration of alternative assumptions upon which we might build our models.Chapter 4 addresses a contemporary debate regarding the flexibility of timing perception. It has recently been reported that timing judgements vary systematically from one moment to the next. This was interpreted as a rapid adaptation of sensory processing, suggesting that these early sensory processes contribute to the perceived timing of sensory events. Chapter 4 presents two experiments, each failing to identify any systematic rapid adaptation of timing perception in participants’ responses. Analysis of EEG data collected during the task provides weak support for a rapid recalibration effect. However, the effect was later than would be expected of sensory processing. I concluded that rapid recalibration of timing judgements is a small effect, unreliable between tasks and samples, but it is more likely to reflect momentary changes in decisional processes than sensory processing.The three empirical chapters presented in this thesis address three contemporary issues in the field of time perception. First, I have provided a novel method to further refine models of timing perception. Second, I have investigated the sources of individual differences in sensitivity to timing relationships, while also testing a common assumption of models of timing perception. Finally, I have provided evidence that addresses a contemporary debate in the literature regarding the malleability of timing perception. The theoretical implications of this work are discussed in the final chapter, along with suggestions for future work.In three Appendix sections I discuss practical and theoretical issues I have encountered and considered during my candidature, and the solutions I have developed to overcome them. First, I describe hardware and software I developed to ensure that stimuli are presented at intended timing relationships both precisely and accurately. Second, I present a potential flaw in pseudorandom sampling of trial order, where participants could improve task-performance by keeping track of trials (like counting cards). Third, I outline a novel method I have developed for automatically identifying and removing blink-artefacts from EEG data, making EEG processing much faster and less susceptible to researcher bias and error.