🧘 Dharma Visualizer

An interactive dashboard demonstrating how cognitive rigidity and suffering emerge from Bayesian belief formation when experiences are censored to fit existing beliefs—and how meditation provides liberation by reducing certainty of the "meaning (predictive value) of past beliefs".

🌟 Core Concept

The Model of Suffering:

1. Experiences arrive with objective values (reality)
2. 95% CI Test: If experience falls outside your belief's 95% confidence interval, it gets squeezed to the boundary
3. Precision Accumulates: Each experience (even squeezed ones) increases certainty (tau), narrowing the CI
4. Trap Forms: Narrower CI → more squeezing → even narrower CI → self-reinforcing rigidity
5. Meditation Breaks Free: Reduces precision (this is the mathematical term for the inverse of the standart deviation, so: higher precision → more narrow curve) of past experiences, widening CI to accommodate reality

🚀 Quick Start

Prerequisites

• Python 3.8+
• Node.js 16+

Start bash script (recommended)

Just run

./start.sh

This will start both backend and frontend servers. You can then open your browser to http://localhost:5173 to access the dashboard.

Manual

Backend (FastAPI)

cd backend
pip install -r requirements.txt
python3 main.py

Backend runs on http://localhost:8000

Frontend (React + Vite)

cd frontend
npm install
npm run dev

Frontend runs on http://localhost:5173

Open browser to http://localhost:5173 and start adding experiences!

📊 Dashboard Components

1. Posterior Distribution (Top Plot)

• Blue curve: Your current belief P(μ|data)
• Shaded region: 95% confidence interval (the "acceptance window")
• Experience distributions:
  • Solid curve: Stored (squeezed) value used in belief
  • Dashed curve: Objective (true) value from reality
  • Arrow: Amount of squeezing (suffering)
• Color coding:
  • Green = Accepted as-is (within CI)
  • Red = Squeezed (outside CI, censored)
  • Yellow = Being meditated on

2. Add Experience Controls

• Value Slider (-10 to +10): The objective mean of new experience
• Sigma Slider (0.1 to 3.0): Uncertainty of the observation
  • Low σ (0.1) = High precision, strong belief about this value
  • High σ (3.0) = Low precision, uncertain observation
• Add Experience Button: Adds experience and automatically squeezes if outside 95% CI

3. Meditation Controls

• Select Experience: Click experience in list to meditate on it
• Meditation Level Slider: How much to reduce precision (0-100%)
• Meditate Button: Apply meditation to selected experience
• Effect: Reduces tau → widens distributions → may unsqueeze if CI now covers objective value

4. Brahma Viharas (Cultivation Level)

• Single counter (0-100%) tracking your meditation skill
• Increases with each manual meditation session
• Higher cultivation = stronger meditation effects
• Represents long-term practice benefits

5. Suffering Metrics

• Total Squeezing Cost: Accumulated suffering from all censored experiences
• Squeezing Cost per Experience: Bar chart showing which experiences cause most suffering
• Precision Over Time: Line chart showing how tau accumulates (rigidity increasing)
• Current CI Width: How narrow/wide your acceptance window is

6. Time-Based Meditation (Automatic)

• Passive meditation: Recent experiences automatically get mild meditation
• Decay: Meditation naturally fades after ~5 seconds
• Manual override: Click "Meditate" button for stronger, cultivation-building meditation

🧠 The Mathematics

Gaussian-Gaussian Bayesian Inference

We model beliefs and experiences as Gaussian distributions:

Prior (initial belief):

P(μ) = N(0, 10²)    # Very weak prior: σ₀=10, τ₀=0.01

Likelihood (each experience):

P(xᵢ | μ) = N(xᵢ; μ, σᵢ²)    # Experience has mean xᵢ, variance σᵢ²

Posterior (updated belief) via Bayes' theorem:

P(μ | x₁,...,xₙ) = N(μₙ, σₙ²)

Precision-Weighted Update (Conjugate Prior)

Why Gaussian-Gaussian? It's a conjugate prior—posterior has same form as prior, giving closed-form updates:

Precision accumulation:

τₙ = τ₀ + Σᵢ τᵢ        # where τ = 1/σ²

Precision-weighted mean:

μₙ = (τ₀μ₀ + Σᵢ τᵢxᵢ) / τₙ

Posterior standard deviation:

σₙ = 1/√τₙ            # Higher precision → lower uncertainty

This IS Bayes' theorem! For Gaussians:

P(μ|data) = P(data|μ) × P(μ) / P(data)

simplifies to these precision-weighted formulas.

The 95% Confidence Interval Rule

Acceptance window:

CI = [μₙ - 1.96σₙ, μₙ + 1.96σₙ]

Squeezing rule:

if x_objective < CI_lower:
    x_stored = CI_lower
elif x_objective > CI_upper:
    x_stored = CI_upper
else:
    x_stored = x_objective    # No squeezing

What this means: We modify the likelihood in Bayes' theorem:

Normal:   P(μ | x_true) ∝ P(x_true | μ) × P(μ)
Squeezed: P(μ | x_squeezed) ∝ P(x_squeezed | μ) × P(μ)

We're using fake data in the likelihood, creating a self-fulfilling belief.

Squeezing Cost (Suffering)

C_squeeze = (x_objective - x_stored)² × √τ_posterior

Interpretation:

• Large shift (|x_objective - x_stored|) = major reality distortion
• High precision (τ) = rigid belief → harder to update → more suffering
• Cost grows quadratically with distortion distance

Meditation Effect (Precision Reduction)

Without meditation:

τ_effective = τ_base

With meditation at level m ∈ [0,1]:

τ_effective = τ_base × r(m)

where r(m) is calculated so that at m=1:

CI₉₅% ⊇ {all objective means}

Effect on posterior:

τₙ(m) = τ₀ + Σᵢ τᵢ × r(m)    # Reduced precision sum
σₙ(m) = 1/√τₙ(m)              # Wider uncertainty

Result: CI widens → previously squeezed experiences may now fit → unsqueezing!

The Feedback Loop (Why It Traps)

1. Initial experiences form posterior with some tau
2. Outlier arrives outside CI → gets squeezed
3. Tau increases (standard Bayesian accumulation)
4. CI narrows (σ = 1/√τ decreases)
5. Next outlier must also squeeze
6. Loop repeats → accelerating rigidity

Mathematical progression:

Squeeze → τ↑ → σ↓ → CI shrinks → More squeezing → τ↑↑ → ...

🎯 Try This Journey

1. Empty Mind (Start)

• Reset model
• Observe: Wide prior (σ=10), almost no beliefs

2. First Experiences (Formation)

• Add experience at value = 0
• Add experience at value = 1
• Add experience at value = -1
• Observe: Posterior forms around mean ≈ 0, CI narrows slightly
• All experiences GREEN (accepted)

3. Cognitive Dissonance (First Squeeze)

• Add experience at value = 5
• Watch it turn RED! Squeezed to CI boundary (~2.0)
• Suffering metric increases
• CI gets even narrower

4. The Trap Tightens

• Try adding value = 4
• Also squeezed! (CI now excludes even moderate values)
• Try adding value = 6
• Also squeezed!
• Observe: You're stuck, can only accept values near 0

5. Liberation Through Meditation

• Click on the squeezed experience at value = 5
• Slide meditation to 100%
• Click "Meditate"
• Watch:
  • Yellow glow appears
  • CI widens
  • Arrow shortens (less squeezing)
  • May turn GREEN if CI now covers objective value!
• Suffering decreases

6. Building Skill

• Observe "Brahma Viharas" counter increased
• This cultivation persists across sessions
• Higher cultivation = more effective meditation