Computational analysis of meditation

Abstract

Meditation training has been shown to improve attention and emotion regulation. However, the mechanisms responsible for these effects are largely unknown. In order to make further progress, a rigorous interdisciplinary approach that combines both empirical and theoretical experiments is required.

This dissertation uses such an approach to analyze electroencephalogram (EEG) data collected during two three-month long intensive meditation retreats in four steps. First, novel tools were developed for preprocessing the EEG data. These tools helped remove ocular artifacts, muscular artifacts, and interference from power lines in a semi-automatic fashion.

Second, in order to identify the cortical correlates of meditation, longitudinal changes in the cortical activity were measured using spectral analysis. Three main longitudinal changes were observed in the retreat participants: (1) reduced individual alpha frequency after training, similar reduction has been consistently found in experienced meditators; (2) reduced alpha-band power in the midline frontal region, which correlated with improved vigilance performance; and (3) reduced beta-band power in the parietal-occipital regions, which correlated with daily time spent in meditation and enhanced self-reported psychological well-being.

Third, a formal computational model was developed to provide a concrete and testable theory about the underlying mechanisms. Four theoretical experiments were run, which showed, (1) reduced intrathalamic gain after training, suggesting enhanced alertness; (2) increased cortico-thalamic delay, which strongly correlated with the reduction in individual alpha frequency (found during spectral analysis); (3) reduction in intrathalamic gain provided increased stability to the brain; and (4) anterior-posterior division in the modeled reticular nucleus of the thalamus (TRN) layer and increased connectivity in the posterior region of TRN after training.

Fourth, correlation analysis was performed to ground the changes in cortical activity and model parameters into changes in behavior and self-reported psychological functions.

Through these four steps, a concrete theory of the mechanisms underlying focused-attention meditation was constructed. This theory provides both mechanistic and teleological reasoning behind the changes observed during meditation training. The theory further leads to several predictions, including the possibility that customized meditation techniques can be used to treat patients suffering from neurodevelopmental disorders and epilepsy. Lastly, the dissertation attempts to link the theory to the long-held views that meditation improves awareness, attention, stability, and psychological well-being.