Latent factors in economic decision making: three studies using drift diffusion modelling

Behind every decision there is a cognitive process that supersedes it. However, the process of decision making is often overlooked in conventional economic analysis that tends to exclusively focus on decision outcomes. Indeed, unlike decision outcomes that can be directly observed, factors that govern the process of decision making are usually latent. In this thesis, we conduct behavioural experiments aimed at exploring the effects of three latent factors on economic decision making: cognitive uncertainty, attentional effort, and intuitive thinking. The data generated from the experiments is analysed via the Drift Diffusion Model (DDM), a process model developed within cognitive science, which is used to describe how latent factors affect preference formation in economic decision making.

Study 1 of the thesis probes into cognitive uncertainty and attentional effort, which are two latent factors that cause inconsistency in risky decisions through distinct means. Cognitive uncertainty refers to the uncertainty in subjective judgement of risky prospects, and attentional effort refers to the use of limited attentional resources for information acquisition. We investigate whether individual differences in cognitive uncertainty and attentional effort are the underlying causes of why those with high cognitive abilities are less prone to decision inconsistency. In the experiment, the cognitive ability of subjects is measured by using standard psychometric tests, whereas decision inconsistency is assessed by detecting preference reversals in a series of risky decisions. Cognitive uncertainty and attentional effort are measured via the application of the DDM. A decision maker’s cognitive uncertainty level is represented by a DDM parameter that indexes the decision maker’s performance in processing stimulus information, and her attentional effort level is represented by a DDM parameter that indexes the balance in emphasising decision speed versus decision accuracy. The results show that cognitive uncertainty, but not attentional effort, acts as the mediator between cognitive ability and decision inconsistency. Therefore, while attentional effort is a factor that causes decision inconsistency in the population, intervention strategies that aim to enhance the decision performance of individuals with decreased cognitive abilities should focus on addressing their excessive cognitive uncertainty. We discuss the application of these results with respect to risk preference elicitation where decision inconsistency is recognised as a major issue hindering robust inferences.

Study 2 and Study 3 concern intuitive thinking (or intuition) in decision making. Intuitive thinking can be broadly defined as judgements that arise from rapid, holistic thinking with little apparent 6 effort and conscious awareness. In Study 2, we investigate what constitutes the predictive power of the Cognitive Reflection Test (CRT) for superior decision performance. The CRT has been developed to measure a person’s tendency to inhibit intuitive responses that are prone to erroneous judgement. However, it has been disputed whether the CRT indeed measures inhibition of intuitive thinking and whether superior decision performance cannot arise from intuition. Hence it is unclear what a role intuitive thinking plays in the relationship between the CRT and decision performance. Using the DDM, we directly measure and investigate the effect that intuitive thinking has on the relationship between CRT measurements and normatively superior risky decisions in a behavioural experiment. Specifically, the effect and strength of intuitive thinking in risky decision making is assessed by a DDM parameter that indexes a decision maker’s predisposition to make one type of decision prior to the valuation of available options. The results show that CRT measurements are positively correlated with the strength of intuition in risky decision making. Moreover, under the assumption that decision making does not involve a valuation process independent from task-specific conditions, we find evidence that subjects measured to have high judgement ability in terms of the CRT rely more on intuition to obtain normatively correct responses. These results challenge the conventional interpretation that high CRT measurements imply less reliance on intuitive thinking, and that intuitive thinking is detrimental to decision performance. A major implication of these results is the need to reassess the claimed associations between the CRT and decision performance in various contexts.

Study 3 aims to understand whether a decision maker’s preference for intuitive thinking is stable across decision domains. Previous literature has investigated this issue using self-report questionaries, finding negative results. However, given that decision makers have limited abilities at judging the use of intuitive thinking, it is questionable whether the self-report approach can accurately capture the preference for intuitive thinking. Study 3 thus adopts a decision modelling approach to investigate this topic. We examine whether the preference for intuitive thinking, as indicated by the strength of intuition estimated by the DDM, is stable across risky and social decision domains. The results show that intuition is stronger in the social decision domain than in the risky decision domain, and that these measures are not correlated across the two domains. These findings confirm the results with those in the extant literature, namely, that a decision maker’s preference for intuitive thinking is unstable across decision domains, and it is necessary to develop domain-specific instruments for measuring the preference for intuitive thinking.

The three studies included in the thesis advance the knowledge on decision making by producing important new findings. Firstly, while both cognitive uncertainty and attentional effort are recognised as factors for causing decision inconsistency, only the former serves as the mechanism through which cognitive ability affects decision inconsistency. Secondly, by showing CRT measurements positively relate to the strength of intuition, we provide evidence against the conventional view that high CRT measurements imply less use of intuitive thinking. Thirdly, we confirm the finding regarding the cross-domain stability of preference for intuitive thinking by using an analytical approach that provides a more accurate assessment of intuitive thinking than the self-report approach often used to investigate the topic. Lastly, the three studies in the thesis demonstrate the usefulness of computational cognitive process modelling as it pertains to the investigation of latent factors that can affect economic decision making. This thesis concludes by discussing some potentially useful avenues for future research.