预印本：脑层级动态的两条可分离轴分别对应正念训练的普遍效应与人群特异性效应

arXiv:2610.20854v1 Announce Type: new Abstract: How mental training reshapes the temporal architecture of large-scale brain dynamics and how these mechanisms differ between healthy and clinically vulnerable populations remains poorly understood. Time-averaged neural measures collapse the multi-scale nature of brain activity, obscuring the dynamic processes through which behavioural interventions reorganise hierarchical brain function. Here, applying a sliding-window Temporal Irreversibility Index to source-reconstructed EEG across three complementary cohorts (n = 141), we decompose hierarchical brain trajectories into three orthogonal temporal descriptors and uncover two dissociable, mechanistically interpretable axes of intervention-induced plasticity. Cumulative meditation practice drives a universal, pan-cerebral reduction in the exploratory breadth of hierarchical dynamics, tracking dose-dependently with lifetime hours of practice. Temporal persistence is bidirectionally tuned by 8-week mindfulness-based interventions in a population-specific way: reduced in healthy meditators but increased in trauma-exposed individuals, revealing that comparable MBI protocols engage qualitatively distinct network mechanisms depending on baseline clinical status. A joint Hidden Markov Model further links these dynamical shifts to distinct microstate substrates, and longitudinal changes selectively track improvements in a single facet of mindfulness, attentional presence in daily life. Together, these findings establish a mechanistic framework for how behavioural training reorganises non-equilibrium brain dynamics, with direct implications for tailoring mindfulness-based interventions to individual clinical profiles.

中文翻译还没有生成，先显示原文。