Meditation can deliberately alter the sense of self, allowing comparison between an active and suspended self. In 41 experienced meditators, magnetoencephalography recordings distinguished a state of reduced sense of self (self-boundary dissolution) from rest and a control meditation state. Machine learning using source band power and Lempel-Ziv complexity features predicted mental states with above-chance accuracy. The best performance, classifying self-boundary dissolution versus rest using Lempel-Ziv complexity, achieved average accuracy of about 0.64 for within-participant prediction and about 0.57 for across-participant prediction. This neural marker could support decoded neurofeedback for clinical treatments of self disorders or meditation training.
The sense of being a bounded self can be attenuated or dissolved while awareness remains. Analyzing magnetoencephalography data from 46 long-term meditators, the study found that both meditation conditions (self-boundary dissolution and maintenance) increased broadband entropy rate and directed information transfer compared to rest, driven mainly by high-frequency activity. Localized reductions in information transfer from the anterior cingulate to posterior cingulate and in high-beta entropy rate in sensorimotor and posterior-medial cortices differentiated the two meditation conditions. Reduced orbitofrontal cortex entropy rate and reduced information transfer from occipital, cingulate, limbic, and subcortical areas correlated strongly with self-boundary dissolution phenomenology. Together with a previously reported neural correlate of reduced high-beta power in the posterior-medial cortex, these two neural correlates explained over half the variance in phenomenological dissolution scores (R² = 0.52).