Williams, Gwydion; (2022) Hierarchical Influences on Human Decision-Making. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Williams_10153535_Thesis.pdf Download (4MB) | Preview

Abstract

Deciding how to act is complicated because people often hold simultaneous intentions to meet multiple goals. These many goals can be arranged in a hierarchy of goals and sub-goals, and a hierarchy of behaviours can be established to attain them. The hierarchical structure of human behaviour is well established, but the precise form of that hierarchical structure remains unclear. Further, we do not know whether and how this hierarchical organisation of action influences the cognitive processes of deciding between candidate actions. This thesis aims to address these two open questions. In Chapter 2, I tackle the first of these two questions. Using behavioural experiments in combination with hierarchical reinforcement learning models of behaviour, I demonstrate that people can learn entirely novel sequences of action without practice, and that this ability requires a hierarchical organisation of action built from two distinct operations. First, the brain must sequence low-level components into higher-level routines of action. Second, the brain must have a method of abstracting the relational structure of a sequence away from its content. In sum, this chapter provides evidence for a theoretical framework which can be used to understand hierarchically structured action more deeply. In Chapters 3 and 4, I tackle the second question: does hierarchical structure influence decision-making? I begin (in Chapter 3) by investigating how hierarchical structure and self-efficacy interact to influence choice between candidate actions. I find that higher level actions are associated with lesser self-efficacy and therefore a lesser willingness to commit to them. This effect arises not only because higher-level actions are more difficult to carry out due to their length, but also because the restrictions that they place on future choices represent a cost. I then (in Chapter 4) investigate whether there are any subjective biases in how outcomes at high or low hierarchical levels are evaluated. I find no overall subjective bias in the evaluation of such outcomes, but I find that social context can prompt strong biases to weight evaluation of outcomes according to their hierarchical level. In sum, I find that hierarchical structure can and does influence decision-making, and I provide evidence for two distinct processes that play a part in this. These findings establish both a novel theoretical framework for future investigations of hierarchically structured action, and a novel set of interactions between the structure of behaviour and how people make action decisions.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as JSONCSVXML

Archive Staff Only

View Item