Modelling dynamics of ruminative thought through a Bayesian sampling framework - Dr. Maria M. Robinson w/collaborators Prof. Adam Sanborn, Prof. Nicole Tang, Lucas Marti

The project bridges research on decision-making, Bayesian models of human cognition and clinical psychology to examine how individual differences in mental health affect cognitive processes in core decision-making tasks. Previous studies have shown that individuals often rely on memory-based samples to estimate probabilities when making decisions because such a strategy is computationally feasible. However, human performance in tasks such as random generation suggests that people do not always sample independently; instead, they follow sampling runs where sequences have similar patterns. This project builds on this research to the novel clinical area of quantitatively characterizing ruminative tendencies, which have been associated with perturbed executive function. Our goals are to examine how individual differences in rumination influence the accuracy of reproducing target distributions in random sequence generation and compare advanced (multi-chain and recycled momentum) Bayesian sampling algorithms to elucidate dynamics of rumination.

So far, we have implemented the experiment and collected data from 200 participants. Data consisted of standard survey data as well as verbal responses in a random number generation task. Data from the random number generation task were transcribed by research assistants. Funds from the Spotlight small grant scheme were used to compensate research assistants for transcription. We have started preliminary data analyses.

We have considered using these data to motivate a larger grant on modelling dynamics of rumination, however, we have yet to pursue this.

Parts of this research were presented at the Warwick Department of Psychology research away day.