Use this URL to cite or link to this record in EThOS:
Title: Decision modelling insights in cognition and adaptive decision making
Author: Pirrone, Angelo
ISNI:       0000 0004 5989 2998
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
The goal of this research is using computational models of decision making, in particular two models, the Drift Diffusion Model (DDM; Ratcliff & McKoon, 2008) and Pais et al. (2013) model, to provide insights in cognition and adaptive decision making. In the first part of this dissertation, we applied the Drift Diffusion Model to three domains: cognition in Autism Spectrum Disorder (ASD), Task-Irrelevant Perceptual Learning (TIPL) and Semantic Congruity Effect Research. Regarding ASD research, we show that differences in reaction times and accuracy in twoalternative forced-choice tasks between ASD subjects and controls, previously interpreted as enhancements or impairments, are instead due to different decision criteria and longer time to execute the motor response for ASD subjects. This result has important consequences for clinical research in which differences in response conservativeness and motor response have been interpreted as differences in information processing. In the third chapter, by applying the DDM, we show that TIPL, learning to better discriminate a stimulus that is irrelevant to a task, does not monolithically affect the sensitivity to the stimulus, but also affects the decision criterion of subjects. Our results show that an analysis only based on accuracy - that is the standard in the literature - could be potentially misleading in the interpretation of learning data, since learning affects different components of decision making, which have different effects on accuracy or reaction times. In the fourth chapter, we perform a DDM decomposition of the semantic congruity effect, the result that subjects are (i) faster in judging the bigger of two big stimuli or the smaller of two small stimuli - as opposed to the bigger of two small stimuli or the smaller of two big stimuli (ii) faster in determining whether a target stimulus is bigger or smaller than a standard stimulus when the size of the two stimuli coincides. Our DDM decomposition allows us to test different verbal theories that have been proposed for the explanation of this phenomenon and to show that this phenomenon arises as an increase in the rate at which subjects accumulate evidence in case of congruency between the magnitude of the standard stimulus and the magnitude of the target stimulus. In sum, in the first part of this dissertation, our work shows the benefits of isolating the different cognitive processes that are involved in decision making and the benefits of testing theories and generating conclusions from data by applying computational models of choice. In the second part of this dissertation, inspired by a model that describes decision making in honeybees (Pais et al., 2013), we investigate a feature of decision making that arises from this model and that cannot be accounted for by a whole family of computational models of choice, DDM included. The DDM, as many other models of choice, disregards the information regarding the overall magnitude of the alternatives, since it only focuses on the differences between alternatives. In the fifth chapter, we argue from an evolutionary perspective why we should expect decision making to take under consideration the magnitude of alternatives and why this poses a challenge to some decision making models, which are instead insensitive to such information. In the sixth chapter, we provide evidence for different species (humans and monkeys) and different domains (perceptual decision making and reward-based decision making) for the existence of magnitude sensitivity in decision making. In sum, in the second part of this dissertation, a mechanism of a computational model of decision making in honeybees, has led us to generate hypotheses in adaptive decision making and to the understanding of the limitations of some computational models of choice. Collectively, our work show the benefits of computational models of choice in the analysis of data and in the generation of hypotheses.
Supervisor: Stafford, Tom ; Marshall, James ; Gurney, Kevin Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available