Thoughtseeds as Latent Causes: A Dual-Process Computational Phenomenology of Focused-Attention Meditation

Summary

A computational model of focused-attention meditation simulates four mental states: focusing on the breath, mind-wandering, becoming aware of distraction, and redirecting attention. The model uses a three-layer architecture inspired by active inference and the Global Neuronal Workspace theory. The deepest layer represents brain network activity; a middle layer encodes mental content and action tendencies; the top layer acts as a metacognitive monitor that gates these tendencies. Meta-awareness, the signal that ignites this workspace, arises from a competition between mental contents and is used to select actions that minimize expected free energy. Simulations of both expert and novice meditators produce patterns consistent with empirical findings, offering a bridge between subjective experience and brain measurements.

Study at a glance

Abstract

Meditative expertise involves sustained attention, rapid recovery from distraction, and coordinated dynamics of large-scale brain networks. We present a computational phenomenology of focused-attention meditation traversing four attractor states: breath focus, mind-wandering, meta-awareness, and redirect attention. Within a dual-process active inference formulation, the model implements a three-layer nested Markov-blanket architecture: (L1) a high-dimensional physiological neuronal substrate modeled as a stochastic multivariate Ornstein--Uhlenbeck process over attentional Yeo networks; (L2) a low-dimensional generative model (System 1) that encodes latent mental content as thoughtseeds and evaluates autonomic action tendencies; and (L3) an agentic metacognitive monitor (System 2) that implements a Global Neuronal Workspace (GNW) capacity bottleneck to selectively gate these tendencies. In L3, meta-awareness functions as the GNW ignition signal, derived from policy-prior divergence and dynamically gated by direct competition between orchestrator and distractor thoughtseeds. Policy selection actively minimizes expected free energy, and L2 actions furnish descending predictions over network activity to close the enactive perception--action cycle. Training uses variational Expectation-Maximization (EM) across expert and novice phenotypes. Simulations reproduce behavior consistent with empirical observations and findings in contemplative neuroscience, providing a tractable link between first-person phenomenology and objective neurophysiological measures.