Use this URL to cite or link to this record in EThOS:
Title: Good things come to those who weight : evidence integration and decision termination in human choices
Author: Tickle, Hannah
ISNI:       0000 0004 7429 1947
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Perceptual decision-making describes the processes by which sensory information is recognised, evaluated and combined before making a commitment to a course of action. The goal of this thesis is to understand the neural and computational mechanisms underlying human perceptual decisions. Good decisions are made when all the available evidence is taken into account, and allowed to influence choice in proportion to its reliability. The first experimental chapter describes a categorisation task employed to investigate how information is integrated and employed according to its reliability during sequential sampling. It is observed that humans weight information approximately optimally. A subsequent experiment involving electroencephalographic (EEG) recordings elucidates a neurobiologically plausible mechanism that could give rise to this effect. However, reliability-based evidence integration may only be possible in relatively simple decisions, when task demands are lower. Previous work investigating more challenging decisions has shown that when two alternatives are viewed in series, locally preferred alternatives are processed with higher gain (“selective integration”). Experiment 2 asks (at both the behavioural and neural level) whether this selective integration happens at the level of attributes - i.e. category A versus B - or features - i.e. sub-dimensions of each of the attributes. Finding that it occurs at the level of features, we discuss the optimality of this strategy. We show, interestingly, that whilst selective integration at the feature level is not harmful to performance, only attribute-level selectivity is actively beneficial in this context. In everyday settings, the choice to stop integrating evidence and commit is often determined by the agent, rather than an external deadline. Experiment 3 uses a self-paced categorisation task to investigate what factors predict when decisions are made. The results show that decisions and their latencies are described by a quasi-optimal model, that times commitment in a way that depends on the evidence consistency. We show that an approximation based on normalisation can account for these findings at the computational level. This model predicts neural signals observed in humans.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available