ここまで来たら、中途半端にまとめるよりも、本当に Nature Human Behaviour を狙う設計図まで一気に引き上げます。

以下に、

1. Systematic Alignment Section（具体的引用付き）
2. メタ解析レベルの統合理論図式
3. シミュレーション数理モデル（明示式）
4. Grant Proposal レベルの実証研究計画

を、トップジャーナル投稿を前提に提示します。

---

I. Systematic Alignment With Empirical Literature

1. Predictive Processing Foundation

The hierarchical predictive brain model is grounded in the free energy principle proposed by
Karl Friston (Friston, 2010).

Clark (2013) further articulated predictive processing as a unifying cognitive architecture.

These frameworks define:

• Prediction error minimization
• Precision weighting (inverse variance)
• Hierarchical message passing

Our model extends this framework into psychopathology via systematic precision dysregulation.

---

2. Schizophrenia: Aberrant Salience and Dopamine

Empirical Anchor 1: Aberrant Salience

Shitij Kapur (2003) proposed that psychosis arises from dopaminergic misattribution of salience.

PET studies show:

• Increased striatal dopamine synthesis capacity in schizophrenia.

Alignment:
→ Hyper-precision of low-level prediction errors.

---

Empirical Anchor 2: Hierarchical Dysconnectivity

fMRI studies (e.g., predictive coding paradigms) demonstrate:

• Impaired top-down modulation from prefrontal cortex
• Increased sensory cortex response variability

Alignment:
→ Bottom-up hyperprecision + top-down instability.

---

Empirical Anchor 3: Hallucination Paradigms

Perceptual expectation tasks show:

• Overweighting of prior beliefs in hallucination-prone individuals
• Increased false percepts under noise

Alignment:
→ Precision imbalance across levels.

---

3. Bipolar Disorder: Reward Learning and Volatility

Empirical Anchor 4: Reinforcement Learning

Computational psychiatry studies show:

• Mania → increased reward learning rate
• Increased inverse temperature (β) parameter
• Risk-seeking bias

Alignment:
→ Excessive prior precision for exploratory policies.

---

Empirical Anchor 5: Depression

Meta-analyses demonstrate:

• Blunted ventral striatal activation
• Enhanced sensitivity to negative feedback

Alignment:
→ Excessive precision of negative prediction error.

---

4. Volatility Estimation Studies

Hierarchical Gaussian Filter (HGF) studies show:

• Schizophrenia: misestimation of environmental volatility
• Bipolar disorder: increased volatility parameter during manic states

Alignment:
→ Dynamic precision instability across hierarchical levels.

---

II. Meta-Analytic Level Integrative Hypothesis Diagram

For Nature Human Behaviour level clarity, the unifying schema must show:

Three axes:

1. Hierarchical level (sensory → cognitive → self-model)
2. Precision allocation
3. Temporal stability

Disorders mapped in 3D space:

• Schizophrenia: low-level hyperprecision + global instability
• Mania: high-level hyperprecision + oscillatory dynamics
• Depression: global rigidity + negative error fixation
• Schizoaffective: cross-level oscillation

The key novelty:

Psychiatric disorders occupy distinct regions within a continuous precision-state space.

This reframes nosology dimensionally.

---

III. Formal Simulation Model

We formalize a simplified 2-level hierarchical Bayesian model.

Level 1: sensory state
Level 2: belief state

Let:

μ₂ = higher-level prior mean
μ₁ = predicted sensory input

Prediction error:

ε₁ = y − μ₁
ε₂ = μ₁ − μ₂

Precision weights:

Π₁ = precision sensory
Π₂ = precision prior

Update equations:

μ₁ ← μ₁ + Π₁ ε₁ − Π₂ ε₂
μ₂ ← μ₂ + Π₂ ε₂

---

Disorder Parameterization

Schizophrenia:

Π₁ ≫ Π₂

→ Sensory domination

Mania:

Π₂ ≫ Π₁
and policy precision γ ↑

Depression:

Π₁ moderate
Π₂ rigid negative prior
learning rate α → near 0

Bipolar:

Π₂(t) = Π₂₀ + A sin(ωt)

---

Simulation outputs:

• Policy entropy
• Belief volatility
• Error variance
• Oscillation amplitude

These can be matched to empirical RL parameters.

---

IV. Grant Proposal Level Empirical Study Design

Title

Hierarchical Precision Dysregulation Across the Psychosis–Bipolar Spectrum

---

Aim 1: Estimate Precision Parameters Across Disorders

Participants:

• 100 schizophrenia
• 100 bipolar (mania, depression, euthymia)
• 100 controls

Tasks:

• Probabilistic reversal learning
• Volatility estimation task
• Perceptual inference task

Modeling:

• Hierarchical Gaussian Filter
• Active inference model fitting

Primary hypothesis:

Schizophrenia → elevated Π₁
Mania → elevated Π₂
Depression → reduced flexibility of Π₂

---

Aim 2: Dopamine-PET Correlation

Subset (n=60):

• Dopamine synthesis capacity (PET)
• Correlate with computational precision parameters

Hypothesis:

Striatal dopamine ∝ estimated precision gain.

---

Aim 3: Longitudinal Oscillation Study

Follow bipolar participants monthly for 12 months.

Estimate:

• Time-varying precision parameters
• Mood ratings

Test oscillatory instability model.

---

V. Why This Could Reach Nature Human Behaviour

To reach NHB level, the contribution must:

1. Bridge computational neuroscience + psychiatry + existential theory
2. Reframe nosology dimensionally
3. Offer testable computational biomarkers
4. Integrate dopamine biology mechanistically

This model does all four.

Novelty claim:

A unified precision-state space model explaining psychosis and bipolar disorder within a single computational-evolutionary framework.

---

The Real Question

This is now at the level where:

• It can become a high-impact theoretical + empirical hybrid paper
• Or remain a sophisticated but unpublished framework

If we truly go NHB-level, the next steps are:

1. Reduce existential language to computational clarity
2. Produce 3 high-quality simulations
3. Pre-register hypotheses
4. Identify collaborators for PET imaging
5. Draft cover letter positioning against RDoC

---