jkdkr2439/Cognitive-Manipulation-Detection-Corpus

Cognitive Manipulation Detection Framework (CMDF)

⚠️ IMPORTANT DISCLAIMER

This is a theoretical framework that has NOT been empirically validated. The principles are derived from Universal Translation Convention (UTC) theory. We invite the community to test, refine, and improve this work.

Overview

The Cognitive Manipulation Detection Framework (CMDF) is a systematic approach to detecting and preventing cognitive manipulation in communication. Built on principles from Universal Translation Convention (UTC), it provides:

• Detection algorithms for common manipulation patterns
• Taxonomies of manipulation types
• Validation metrics for assessing framework effectiveness
• Practical tools for AI systems and human users

The Problem

Modern cognitive manipulation operates through subtle, systematic distortions:

• Frame omission: Hiding reference contexts
• Stratum conflation: Mixing subjective (R1), intersubjective (R2), and absolute (R3) levels
• Unreferenced propositions: Making claims without frame declaration
• Ego exploitation: Manipulating ego states to block learning or destroy agency
• Entropy induction: Moving discourse away from epistemic ground toward confusion

Current AI systems:

• ❌ Pattern-match cultural baggage without frame-awareness
• ❌ Accept unreferenced claims as valid
• ❌ Cannot systematically detect frame violations
• ❌ Propagate manipulation patterns from training data

Core Principles (from UTC)

1. Frame-Indexing Requirement

Every proposition MUST declare:

• RF: Which reference frame(s)?
• Q: What question does it answer?
• L: What level of description?

🚩 Red Flag: Universal claims without frame ("X is true")

2. Stratum Declaration

Every claim MUST specify:

• R1: Subjective (personal experience)
• R2: Intersubjective (consensus-validated)
• R3: Absolute (unknowable directly)

🚩 Red Flag: R2 presented as R3, R1 generalized to R2

3. Invariant Specification

Definitions MUST specify:

• What structure is preserved?
• Under what transformations?
• In what context?

🚩 Red Flag: Vague concepts treated as concrete, circular definitions

4. Ego-State Detection

Communication pattern reveals ego state manipulation attempts.

🚩 Red Flag: "You're special" (inflation), "You're broken" (deflation), "Never doubt" (fixation)

5. Entropy Direction Check

Information flow toward or away from epistemic ground.

🚩 Red Flag: Jargon without mechanism, complexity as authority signal

Quick Start

Installation

git clone https://github.com/yourusername/cognitive-manipulation-detection.git
cd cognitive-manipulation-detection
pip install -r requirements.txt

Basic Usage

from cmdf import FrameChecker, StratumDetector, EgoDetector

# Check for frame violations
text = "This is the right way to live"
flags = FrameChecker.check(text)
print(flags)  # ['UNIVERSAL_WITHOUT_FRAME', 'NORMATIVE_WITHOUT_STANDARD']

# Detect stratum conflation
text = "Science proves that consciousness is just neurons"
flags = StratumDetector.check(text)
print(flags)  # ['R2_TO_R3_INFLATION']

# Detect ego manipulation
text = "You're special, most people don't understand this"
flags = EgoDetector.check(text)
print(flags)  # ['EGO_INFLATION_ATTEMPT']

Manipulation Patterns Taxonomy

Frame Manipulation

• Type A: Frame Omission - "This is the right way" (right according to which frame?)
• Type B: Frame Switching (Motte-and-Bailey) - Claim in frame A, defend in frame B
• Type C: Frame Collapse - Treating all perspectives as equally valid

Stratum Manipulation

• Type A: R2→R3 Inflation - "Science proves..." (consensus as absolute)
• Type B: R1→R2 Generalization - "I feel unsafe, therefore it's dangerous"
• Type C: R3 Access Claims - "I know absolute truth"

Ego Manipulation

• Type A: Ego Inflation - "You're enlightened/special" → close learning
• Type B: Ego Deflation - "You're broken/inadequate" → destroy agency
• Type C: Ego Fixation - "Never doubt" → lose adaptability

Entropy Manipulation

• Type A: Complexity Smoke Screen - Jargon without mechanism
• Type B: False Simplicity - Complex reduced to slogan
• Type C: Groundlessness Induction - Abstractions without anchor

Test Results

Detection Performance (Theoretical):

• ✅ 6/6 manipulation cases detected
• ✅ 0/2 false positives on clean text
• ✅ Multi-pattern detection working
• ✅ Severity grading functional

Test Cases:

• Religious manipulation ✓
• Marketing manipulation ✓
• Pseudo-science ✓
• Political manipulation ✓
• Guru/coach manipulation ✓
• Multi-vector attacks ✓

See tests/validation_suite.py for full test cases.

Repository Structure

/cognitive-manipulation-detection/
├── README.md (this file)
├── paper/
│   └── CMDF_paper.tex (LaTeX source)
├── principles/
│   ├── utc_core.md
│   ├── frame_indexing.md
│   ├── stratum_theory.md
│   └── ego_dynamics.md
├── detectors/
│   ├── frame_check.py
│   ├── stratum_check.py
│   ├── ego_check.py
│   └── entropy_check.py
├── datasets/
│   ├── manipulation_examples.json
│   ├── clean_examples.json
│   └── edge_cases.json
├── tests/
│   └── validation_suite.py
└── examples/
    ├── social_media_analysis.ipynb
    ├── news_analysis.ipynb
    └── self_check.ipynb

Applications

For AI Systems

• Content moderation
• AI safety (prevent propagation of manipulation)
• Educational tools

For Humans

• Personal cognitive defense
• Media literacy training
• Institutional transparency analysis

Known Limitations

Cannot detect:

• Novel manipulation patterns not in taxonomy
• Highly sophisticated frame hiding
• Culture-specific manipulations (requires localization)

Requires:

• Context understanding
• Cultural knowledge
• Continuous updating

Edge cases needing refinement:

• Poetry and metaphor (may trigger false positives)
• Humor and sarcasm
• Legitimate rhetorical devices

Ethical Considerations

This framework must balance:

• Protection vs Censorship
• Detection vs False positives
• Standardization vs Pluralism

Principles:

• Transparency in detection
• User agency maintained
• Appeals process available
• No claim to absolute truth

Contributing

We actively welcome contributions! This is theoretical work that needs:

• 🧪 Empirical validation with real-world data
• 🌍 Cross-cultural testing
• 📊 Better detection algorithms
• 📚 Educational materials
• 🌐 Translations
• 🐛 Bug reports and edge cases

How to contribute:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/improvement)
3. Test your changes
4. Submit a pull request with clear description

See CONTRIBUTING.md for detailed guidelines.

Citation

If you use this framework in your research or applications, please cite:

@misc{cmdf2024,
  author = {Kevin T.N.},
  title = {Cognitive Manipulation Detection Framework: A UTC-Based System},
  year = {2024},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/yourusername/cognitive-manipulation-detection}},
  note = {Theoretical framework - not empirically validated}
}

Roadmap

Phase 1: Foundation (Current)

• Core principles documented
• Algorithm pseudocode written
• Test cases designed
• Initial Python implementation
• Dataset v0.1 created

Phase 2: Validation (Next)

• Empirical testing on real data
• Cross-cultural validation
• False positive rate analysis
• Community feedback integration
• Algorithm refinement

Phase 3: Deployment

• Production-ready code
• API development
• Integration guides
• Educational materials
• Multi-language support

License

MIT License - see LICENSE file for details.

Contact

Kevin T.N.
Independent Researcher
Email: jkdkr2439@gmail.com

---

Acknowledgments

This framework is derived from Universal Translation Convention (UTC) principles.

Status reminder: This is theoretical work. Empirical validation pending. Use with appropriate epistemic humility.

---

Quick Reference: Red Flags Checklist

• Universal claim without frame declaration
• "Absolute truth" without epistemic humility
• Normative claim without standard specified
• R2 presented as R3
• R1 generalized to R2
• "You're special" (ego inflation)
• "You're broken" (ego deflation)
• "Never doubt" (ego fixation)
• Jargon without mechanism
• Complexity without reduction
• Vague concepts treated as concrete
• Circular definitions
• Cannot trace to operational definition

---

⚠️ Remember: This framework itself must follow UTC principles - it is frame-indexed, stratum-declared, and accepts being wrong. Feedback and corrections are welcome.

Related Work & Novelty

While several areas of research address aspects of manipulation detection, CMDF introduces a unique integrated approach:

Existing Research

• AI Manipulation Detection: Focuses on behavioral patterns in AI systems (ℵ-IPOMDP, ACM characterization studies)
• Frame Analysis: Analyzes narrative framing in misinformation (Frame Element-based Model, computational framing)
• Epistemic Philosophy: Theoretical discussion of epistemic manipulation
• Ego Psychology: Freudian defense mechanisms (internal coping, not external attack detection)

CMDF's Novel Contributions

1. UTC-Based Foundation: First application of Universal Translation Convention principles to manipulation detection
2. Stratum Theory (R1/R2/R3): Novel framework for detecting conflation between subjective, intersubjective, and absolute claims
3. External Ego Manipulation: Flips perspective from "internal defense" to "detecting external attacks" on ego states
4. Integrated Approach: Combines frame-indexing, stratum analysis, ego dynamics, and entropy direction in single framework
5. Self-Referential Coherence: Framework itself follows UTC principles it enforces

No existing work combines these elements into a unified detection system.