Modeling the ruminative mind: behavioral, computational, and neurophysiological approaches

In his thesis, Anmol Gupta investigates rumination, a repetitive and self-focused form of negative thinking that plays a central role in depression. Traditional research has largely relied on self-report questionnaires, which are limited by introspective bias. To move toward more objective measures, Gupta integrates behavioural experiments, computational modeling, and neurophysiological data.

First, using behavioural studies Gupta compared two cognitive mechanisms implicated in depression, reward learning and rumination, across participants in India and the USA. While deficits in reward learning were not predictive of depressive symptoms, behaviorally defined rumination, measured through spontaneous thought content during attention tasks, strongly correlated with self-reported depression and perseverative thinking.

Second, Gupta developed a cognitive model implemented in the ACT-R architecture, simulating how rumination shapes memory recall. The model reproduced empirical patterns seen in depression, such as longer trains of negative recall, by strengthening associations among negatively valenced memory traces.

Finally, Gupta  used electroencephalography (EEG) combined with machine learning to classify both external workload levels and internal attentional states such as focused attention, worry, and future-oriented thought. Directed functional connectivity measures and deep neural networks achieved high accuracy in decoding cognitive states, indicating the potential for objective biomarkers of rumination.

Together, these studies provide converging evidence that rumination can be quantified across behavioral, computational, and neural levels. By bridging these methods, the thesis advances a multi-modal framework for understanding and objectively assessing ruminative cognition in depression.