# AGENTS.md — Dynamic Attentional Agents in Focused Attention Meditation

**Paper**: Dynamic Attentional Agents in Focused Attention Meditation: Hierarchical Computational Modeling of Expert-Novice Differences (2026)  
**Area**: Active Inference · contemplative neuroscience · hierarchical modeling · predictive processing  
**Authors**: P. C. Kavi, Daniel Ari Friedman, G. Patow  
**Venue**: Active Inference — IWAI 2025. Communications in Computer and Information Science, vol 2857. Springer.  
**DOI**: 10.1007/978-3-032-16955-6_11

---

## Agent Roles

### 📖 Research Agent

**Focus**: Deep analysis of the three-level hierarchical framework and its theoretical grounding.

**Tasks**:

- Analyze core argument: meditation as predictive processing; thoughtseeds as Markov-blanket agents minimizing variational free energy via bidirectional coupling with attentional networks.
- Evaluate the three-level hierarchy: superordinate neuronal ensembles → thoughtseed agents → attentional networks (DMN, VAN, DAN, FPN) and Global Workspace competition.
- Map the Neuronal Packet Hypothesis onto the Free Energy Principle; evaluate how meta-cognitive precision weighting stabilizes attention.
- Assess theoretical novelty relative to the companion Thoughtseeds paper (2025_Thoughtseeds) and extensions toward AIF formalisms.

### 🔬 Methods Agent

**Focus**: Simulation methodology and parameter sensitivity.

**Tasks**:

- Document and assess simulation design: parameter sweeps over precision weighting (0.5 vs. 0.4), complexity penalties (0.2 vs. 0.4), and learning rates (0.02 vs. 0.01).
- Evaluate reported expert–novice differences: 49% lower free energy during breath focus, DMN suppression (0.18 vs. 0.31), distraction recovery speed.
- Assess bidirectional message-passing implementation and its correspondence to neuroimaging findings.
- Identify open methodological questions for future AIF rigorous implementation.

### 📚 Citation Agent

**Focus**: Citation context and scholarly impact.

**Tasks**:

- Track citations and impact of this chapter (DOI `10.1007/978-3-032-16955-6_11`).
- Map connections to IWAI 2025 proceedings and the broader Active Inference conference literature.
- Link to the Thoughtseeds companion paper (`2025_Thoughtseeds`, DOI `10.3390/e27050459`) and other meditation-relevant entries in the Friedman corpus.

### 🔗 Synthesis Agent

**Focus**: Cross-paper and cross-domain connections.

**Tasks**:

- Connect to related works in the docxology corpus: 2025_Thoughtseeds, 2025_CEREBRUM, 2024_SharedProtentions, 2020_EmergentTeams, 2018_WoodliceAndMen.
- Identify bridges to computational psychiatry, cognitive training, and contemplative neuroscience literature.
- Map overlap with cognitive security (attention regulation, precision control) and educational/training applications.

### 💼 Consultant Agent

**Focus**: Practical application and knowledge transfer.

**Tasks**:

- Relate the hierarchical framework to meditation training design, biofeedback systems, and attention-regulation interventions.
- Support teams developing AIF implementations for contemplative or psychiatric applications using this framework.
- Guide users in applying precision-weighting and Markov-blanket concepts to attentional modeling tasks.

## Extraction Log

- **Source PDF**: Not freely available
- **PDF Status**: Accessible via Springer (DOI `10.1007/978-3-032-16955-6_11`) or institutional library
- **Documentation Quality**: Hand-curated from full abstract and citation metadata

## Related Papers

See [BIBLIOGRAPHY.md](../../pages/BIBLIOGRAPHY.md) for the full catalog. Key related entries:

- 2025_Thoughtseeds — companion Thoughtseeds Framework paper (Entropy 2025)
- 2025_CEREBRUM — Case-Enabled Reasoning Engine with Bayesian Representations
- 2024_SharedProtentions — Shared Protentions in Multi-Agent Active Inference
- 2018_WoodliceAndMen — Bayesian account of cognition with Karl Friston
- 2023_VariationalSynthesis — Variational synthesis of evolutionary and developmental dynamics