The Effect of Speed Accuracy Trade-off on Associative Learning

Cognitive models have been illuminating the underlying cognitive frameworks of two-choice decision making for over three decades. This thesis investigated the impact speed pressure has on underlying decision making processes using diffusion model predictions of data from a probabilistic associative learning task. The specific hypotheses are framed in terms of whether selective influence or non-selective influence of diffusion model parameters provide the best model fit to data. Participants (N = 6) categorised mutually exclusive and nonconfusable stimuli into two novel and arbitrary categories. This removed some key limitations from previous findings that implied a non-selective influence explanation of decision making under speed pressure. Outcomes were measured using choice probabilities and response times, which were analysed initially using three-way ANOVAs to check general patterns of data for both outcome measures, before the diffusion model was applied to data for the main analysis. In light of the fact that the diffusion model is not inherently designed to measure learning effects, changes in the drift rate parameter corresponding to the quality of evidence provided by an item was used to characterise learning across blocks. This method of estimation provided a good fit to data that effectively summarised the learning process under both speed and accuracy emphasised decision making. As predicted, results supported the selective influence hypothesis, revealing that speed pressure did not have a negative effect on the quality of evidence recalled during the learning process that informs decision making. These findings provide a clear and penetrating test of whether speed pressure negatively affects the quality of item recall during decision making. In addition, this thesis expanded the ability of a learning-informed diffusion model to account for the learning process under the new context of speed pressure