History-Dependence in the Neural Basis of Free Choice Saccadic Behaviour
Loading...
Authors
Caie, Brandon
Date
2024-10-07
Type
thesis
Language
eng
Keyword
Transcranial Direct Current Stimulation , EEG , Choice Behaviour , Decision-making , Saccadic Eye Movements , History-Dependence
Alternative Title
Abstract
Choices are formed by combining sensory information with expectations that are formed from past experience. Although progress has been made in our understanding of how these transformations occur in decision-making tasks, where outcomes are well-defined, much less is known about conditions where we are free to choose between alternatives with no objective distinction. The work in this PhD thesis combined behavioural analysis, modelling, computational theory, neuroimaging, and causal interventions to improve our understanding of how visual information is combined with expectations during free choice saccadic behaviour. We developed a saccadic choice task, wherein the asynchrony between visual stimuli was manipulated across trials such that we could elicit 'free choice' saccades. In the first paper, we combined fMRI-guided HD-tDCS with EEG recordings in human participants performing the free choice saccade task. By combining within-participant test-retesting with a state-of-the-art neuroimaging pipeline, we provided a comprehensive assessment of within and across participant variability in the effects of HD-tDCS on biasing saccadic choice through frontal eye field neural activity. We found that intra-individual variability across different sessions was a significant driver of reported effects -- the profile of the variability was such that it was unclear whether these reported differences were caused by differences in tDCS-dependent effects, or variability that arises from history-dependence in measurements typically treated as independent. In a second study, we approached the problem of history-dependence from a theoretical standpoint, asking how the assumptions we make about generative models of behaviour and neural data shape our expectations of null effects when making causal interventions on the system. In a final study, we consider saccadic choice behaviour in more detail using a combination of choice history analysis and a novel computational model. There, we found evidence that anticipatory dynamics are an important contributor to how sensory expectations are formed in saccadic choice, and that these dynamics invalidate some assumptions of the models used to explain choice behaviour as a stationary process. In combination, these studies provide an empirical assessment of sources of variability when intervening on saccadic free choice, as well as a theoretically grounded treatment of causes and consequences of history-dependence.