Research Platform

Open science for emotional technology research

SYSTEM ARCHITECTURE

System Overview

How the parts fit together

TEG-Blue is organized as a layered system. Each layer has a different job, a different evidence status, and a different kind of researcher who can help.

1Layer 1

Measurement

Four-Mode Gradient

2Layer 2

Explanatory

12 Frameworks

3Layer 3

Tools

Applied Instruments

4Layer 4

AI Safety

Structured Schemas

1

Layer 1The Measurement Layer

Four-Mode Gradient

A structured model of regulated and dysregulated states expressed through behavior and language.

Connection → Protection → Control → Domination

These are not personality types. They are nervous system states — regulatory positions that shift in response to perceived threat, shaped by attachment history, social context, and current capacity.

Status: Proposed model with early evidence

See Publications for what is currently tested and what still needs replication.

Four-Mode Gradient →
2

Layer 2The Explanatory Layer

12 Frameworks (F1–F12)

A connected set of frameworks that explain why these patterns emerge, how they escalate, why they look paradoxical, and how repair becomes possible again.

The frameworks are organized as a connected arc:

  • F1–F3: Internal regulation, identity adaptation, inner organization
  • F4–F7: How individual protection scales into social systems and harm
  • F8–F10: The return path — self-awareness, repair, and re-integration
  • F11: Why humans look contradictory until you see the full state logic
  • F12: The integrative lens connecting inner biology to social outcomes

Each framework draws on established theories. The originality is in the connections between them.

Status: Proposed synthesis

Grounded in established theories, with connections that need structured review and testing.

Frameworks →
3

Layer 3The Emotional Tools

Applied Instruments

16 gradient-based scales and assessments derived from the frameworks. They translate the system into practical instruments for individuals, practitioners, and clinical settings.

Each tool maps a dimension of human behavior across the gradient — from healthy to harmful — with clear markers at every point.

Examples: Empathy Gradient (genuine → selective → performed → weaponized), Accountability Gradient (genuine → performed → absent → protective), Integrity Scale (value-aligned → conditional → performed → remorseless).

Status: Designed, not yet psychometrically validated

Available for exploration on teg-blue.com, awaiting formal validation studies.

Research need: Scale design support, factor structure exploration, convergent and discriminant validity plans, bias and fairness evaluation.

4

Layer 4The AI Safety Layer

Structured Schemas for AI Systems

A structured, computationally legible schema layer that translates the emotional pattern logic into formats AI systems can consume.

This includes JSON-LD structured data, consistent terminology, semantic HTML, and gradient-based classifications designed to replace binary safe/unsafe models with nuanced assessments.

Status: Proposed architecture with early implementation

See AI Safety for the full application case.

Research need: Schema design feedback, evaluation protocols, risk analysis, misuse prevention, alignment with existing safety research.

The four core functions

The system is designed to serve four functions. Each represents a research lane where collaboration is needed.

1. Measurement and recognition

Turning subjective states into observable, trackable patterns using gradients, modes, and markers.

2. Prediction and prevention

Mapping how states shift and escalate across the gradient. Harm often follows predictable progressions from Protection into Control and Domination.

3. Navigation and intervention

Identifying interventions that help systems move from Control back toward Protection and Connection. Routes back to safety exist.

4. Pattern breaking

Understanding how entrenched patterns — individual, relational, institutional — can be interrupted and restructured.

Ethical constraint: Trauma-informed data architecture

The system assumes many difficult behaviors started as Protection Mode survival responses. Data systems built on this framework should not be designed to shame, profile, or exploit.

This is not a tagline. It is an architectural constraint that applies to every tool, schema, and classification derived from TEG-Blue.

Intent–Impact–Pattern logic

TEG-Blue maps each behavior across three dimensions:

  • Intent: What the nervous system is trying to do
  • Pattern: Which gradient pattern or tool family it belongs to
  • Impact: What it does to safety, power, and connection

This prevents reducing behavior to either "good person" or "bad person" — the same behavior can serve different functions depending on state and context.

Where to go next

If you want to…Go here
See the measurement layer in detailFour-Mode Gradient →
See the 12 frameworksFrameworks →
Review evidencePublications →
Understand methodsMethodology →
Explore AI applicationsAI Safety →
CollaborateCollaborate →

TEG-Blue Research Consortium · Open Science · CC BY-NC-SA 4.0