About Us
The Challenge of Measuring Consciousness
Consciousness has long been considered unmeasurable - a subjective phenomenon that defies quantification. Traditional approaches struggle because consciousness cannot be directly observed; we can only infer it from external markers like behavior, brain activity, or system responses.
This creates a fundamental problem: How do you measure something you cannot see?
What We Do
At CGT Group, we've developed a framework that makes consciousness measurable by focusing on observable patterns rather than subjective experience. Our methodology identifies specific signatures that indicate conscious processing across different types of systems:
In biological systems: EEG patterns during sleep states reveal varying levels of conscious awareness
In artificial intelligence: Behavioral responses under novel conditions indicate genuine understanding versus pattern matching
In organizations: Institutional decision-making patterns predict resilience or collapse
By applying the same theoretical framework across these three domains, we can validate our measurements through cross-system consistency and prospective predictions.
Our Approach
We measure two fundamental properties that underlie consciousness across all systems: how information is integrated into coherent wholes, and how systems maintain differentiated responses to different contexts.
Neither property alone is sufficient. A system that integrates everything into sameness lacks awareness. A system with many states but no coherence lacks unity. Consciousness emerges from the balance.
Our proprietary methodology translates these principles into quantifiable metrics through:
Domain-specific operationalization — the same theoretical framework, adapted to each substrate
Rigorous validation — effect sizes, classification accuracy, and statistical significance
Prospective testing — predictions made public before outcomes are known
The formulas stay protected. The results speak for themselves.
Why It Matters
Making consciousness measurable transforms it from philosophy into science. Our research enables:
✓ Objective sleep state analysis for neuroscience and clinical applications
✓ Reliable AI consciousness assessment as artificial systems grow more sophisticated
✓ Predictive institutional analysis identifying organizations at risk of collapse
Where others see an impossible measurement problem, we see patterns waiting to be decoded.
Our Work
Institutional Collapse Predictor
Track Record: 12/12 (100%)
Retrospective Cases:
- FTX (November 2022)
- Enron (December 2001)
- WeWork (September 2019)
- Theranos (2018)
- Bear Stearns (March 2008)
- Lehman Brothers (2008)
- WorldCom (2002)
- Washington Mutual (2008)
- Countrywide Financial (2008)
- AIG (2008)
Prospective Predictions:
- NVIDIA (September 2025): Predicted stable — validated November 2025 ✓
- Saks Global (December 2025): Predicted collapse — Chapter 11 filed January 2026 ✓
Current Predictions: We have 4 featured and ~20 active predictions being validated through 2027. Some predict collapse despite market optimism. Some predict survival despite widespread bankruptcy concerns. The goal is to demonstrate the model can distinguish between temporary trouble and actual collapse, not just be pessimistic about everything.
Results will validate or disprove the approach.
Sleep Neuroscience Research
EEG Consciousness Measurement Framework
Analysed 622 sleep transitions (12 subjects, Sleep-EDF database)
Published: "Dual Architecture of Consciousness Transitions" (bioRxiv, 2025)
Key Discovery: Consciousness employs TWO distinct mechanisms:
- 75% gradual integration (IIT-style)
- 25% threshold gating via spindles (GWT-style)
Results: Cohen's d = 2.814 (p<0.001) — first empirical reconciliation of competing consciousness theories.
Additional Validation:
- 160+ subjects across 3 independent datasets
- 143,000+ epochs analyzed
- Effect sizes d > 1.3 for wake/sleep discrimination
- 100% directional accuracy
Weather Forecasting
Bias-Coherence Index (BCI)
Applied CGT framework to ensemble weather forecasts across 14 major UK storms (2021-2024)
Methodology:
- φ = inter-model agreement (ensemble spread)
- ρ = temporal forecast stability (persistence)
- BCI = √(φ × ρ) — same formula as consciousness and institutional domains
Results:
- AUC 0.78-0.86 for high-error detection
- 46% improvement in forecast error identification
- 1,218 forecasts validated across multiple locations
- F1 scores: 0.59-0.64
Key Finding: Low model agreement (φ) or low temporal stability (ρ) signals reduced forecast accuracy — the geometric mean captures their interaction.
Paper submitted to Weather and Forecasting
Cross Domain Validation?
The same theoretical approach appears to work across three different domains — biological systems, atmospheric systems, and institutional systems. If this holds up under peer review and continued testing, it suggests consciousness measurement is possible in ways previously thought impossible.
The work is ongoing. Some predictions won't be validated until 2027. We'll update these results as outcomes materialize.
