Thoughtseeds Latent Causes Dual Process Computational Phenomenology Focused Attention Meditation

---
name: thoughtseeds-latent-causes-dual-process-computational-phenomenology-focused-attention-meditation
description: A skill for understanding and applying the computational phenomenology of focused-attention meditation based on the arXiv paper "Thoughtseeds as Latent Causes: A Dual-Process Computational Phenomenology of Focused-Attention Meditation" (arXiv:2607.14833v1).
---

## Overview

This skill provides a structured approach to understanding and applying the dual-process computational phenomenology model of focused-attention meditation presented in the paper. The model proposes a three-layer nested Markov-blanket architecture:
- Layer 1 (L1): High-dimensional physiological neuronal substrate (stochastic multivariate Ornstein-Uhlenbeck process over attentional Yeo networks)
- Layer 2 (L2): Low-dimensional generative model (System 1) encoding latent mental content as thoughtseeds and evaluating autonomic action tendencies
- Layer 3 (L3): Agentic metacognitive monitor (System 2) implementing a Global Neuronal Workspace (GNW) capacity bottleneck to gate these tendencies

The model explains how meditative expertise involves sustained attention, rapid recovery from distraction, and coordinated dynamics of large-scale brain networks.

## Steps for Application

1. **Understand the Three-Layer Architecture**
   - Familiarize yourself with the roles of L1 (neuronal substrate), L2 (thoughtseeds and autonomic responses), and L3 (metacognitive monitoring and GNW gating).

2. **Model the Neuronal Substrate (L1)**
   - Represent attentional networks as a stochastic multivariate Ornstein-Uhlenbeck process.
   - Use the Yeo network parcellation to define the dimensions of the state space.

3. **Implement the Thoughtseed Generator (L2)**
   - Model latent mental content as discrete thoughtseeds.
   - Define how thoughtseeds influence autonomic action tendencies (e.g., via a mapping to physiological responses).

4. **Implement the Metacognitive Monitor (L3)**
   - Implement a Global Neuronal Workspace (GNW) bottleneck that gates access to conscious processing.
   - Use policy-prior divergence to generate meta-awareness signals.
   - Allow competition between orchestrator (task-focused) and distractor (mind-wandering) thoughtseeds for access to the GNW.

5. **Train the Model with Variational EM**
   - Use variational Expectation-Maximization to fit the model to empirical data from both expert and novice meditators.
   - Optimize the parameters to minimize expected free energy.

6. **Validate with Simulations**
   - Run simulations to reproduce behavioral and neurophysiological observations from contemplative neuroscience.
   - Check for the four attractor states: breath focus, mind-wandering, meta-awareness, and redirect attention.

7. **Apply to Brain-Computer Interfaces or Meditation Training**
   - Use the model to design adaptive neurofeedback systems.
   - Apply the framework to personalized meditation training protocols.

## References

- arXiv:2607.14833v1 - Thoughtseeds as Latent Causes: A Dual-Process Computational Phenomenology of Focused-Attention Meditation

## Activation Keywords

- thoughtseeds, dual-process, computational phenomenology, focused-attention meditation, global neuronal workspace, variational EM