Meditation involves focusing one's attention on an object or a phenomenon. It is believed to provide the ability to cognitively regulate attention and emotion. Several studies have discovered beneficial effects of meditation. However, very few scientific studies have attempted to understand the mechanisms underlying meditation. In order to make further progress, a rigorous computational modeling approach is required. The current project will approach this problem in four ways. First, advanced methods will be used for preprocessing of electroencephalogram (EEG) data of meditation and rest. Second, in order to identify the cortical correlates of meditation, a battery of advanced time-series analysis will be employed on the preprocessed data. Third, a formal computational model will be constructed to account for the cortical correlates and to provide testable theories about the underlying mechanisms. Fourth, in order to test the efficacy of modeling, correlations will be performed between performance measures in attention-related tasks and model predictions. Initial results, based on blind source separation, spectral analysis, and coarse-scale computational modeling, are promising. In future work, they will be extended to synchrony, causality and source localization, resulting in a more detailed model. The project thus aims at constructing a framework to understand meditation, using a tight interaction between data, time-series analysis, and computational modeling.

http://www.cs.utexas.edu/~mishu/