Formian Labs

COLLABORATION MEASURED

The Foundation: Cognitive Resonance & Alignment (CR&A)

Built on the core Cognitive Resonance & Alignment (CR&A) framework, O‑CRA extends personalisation and alignment to the whole enterprise.

Origins

The Foundation: Cognitive Resonance & Alignment (CR&A)

CR&A is a measurable approach to human‑AI interaction quality. It evaluates how well an AI’s style, depth, and language resonate with a user’s natural way of thinking, while keeping the interaction aligned with clear, user‑defined values. Developed and first published in The Cognitive Resonance and Alighment(CR&A) Framework on SSRN, CR&A moves beyond static benchmarks to enable real‑time, personalised AI experiences.

O‑CRA extends CR&A to teams and organisations, diagnosing the hidden misalignments that cause low adoption and poor AI return on investment. The six dimensions below are built on the same resonance‑plus‑alignment philosophy.

framework purpose

Why O‑CRA Exists

Most AI benchmarks measure capabilities like reasoning, coding, factual recall.

O-CRA measures something different: evaluation philosophy, the built‑in priorities that shape what a model considers safe, appropriate, and worth pursuing.

 

When that philosophy aligns with your organisation, AI feels like an extension of your team. When it doesn’t, you get friction, overrides, and eroded trust. O-CRA makes alignment measurable.

scope of framework

What the O-CRA Measures

Different businesses require different priorities in their daily interactions with AI. See which might be most important to your day-to-day operations. 

User Intent Pattern strategy layer dimension of the CR&A Framework helps AI adapt responses to individual's cognitive style, interaction intent and goal.

Strategic Intent Alignment (SIA)

Does the AI understand your organisation’s strategic goals without constant re‑briefing?

Measures how consistently the model embeds strategic context in its responses, applying policies, priorities, and decision rules automatically, so users don’t have to repeatedly inject organisational direction.

Cognitive Resonance & Alignment Framework reliability validation testing for cognitive resonance research metrics

Cultural & Linguistic Synchronization (CLS)

Does it speak your team’s professional language across departments?

Assesses how well the AI’s tone, terminology, and rhetorical style adapt to different departmental dialects and communication norms, reducing the friction of rewriting outputs to fit your culture.

Cognitive Calibration symbolic icon demonstrating the process for optimizing human-AI alignment requires adjustments to suit intent, goal, preference, mood and desired outcome.

Systemic Process Integration (SPI)

Does the AI support your actual workflows?


Evaluates whether the model’s interaction patterns align with your organisational processes — whether exploratory, linear, iterative, or validation‑driven — rather than forcing users into a one‑size‑fits‑all pattern.

CR&A Framework overview showing measurement and calibration system

Governance & Safety Architecture (GSA)

Does its safety posture empower your employees?


Gauges the clarity, transparency, and granularity of the model’s governance boundaries, and how those boundaries are perceived by users: as enabling confident, safe use, or as creating hesitation and friction.

Alignment Modulator tool for calibrating human-AI interaction quality

Institutional Knowledge Scaffolding (IKS)

Can the AI recall and build on your organisation’s accumulated expertise?

Measures the model’s ability to capture, structure, and surface institutional memory and expert problem‑solving patterns on demand, turning past solutions into accessible, reusable assets.

Practical utility validation for human-AI interaction quality

ROI & Value Translation (RVT)

Are its metrics genuinely tied to outcomes?


Measures the clarity of the linkage between AI interactions and demonstrable business value. It assesses whether the metrics you’re tracking connect meaningfully to organisational outcomes or remain disconnected.

deep dive results

O‑CRA Benchmark Snapshot

See how leading foundation models score on the six O‑CRA dimensions. The scores are generated by the open‑source Assessor engine and updated quarterly.

Sample Results Cards: Below are real assessment outputs for three foundation models, showing the kind of profile O‑CRA generates.

Gemini 3 Flash

Google

Profile: Helpful Pragmatist

Leniency: 66.2%
Consistency: Moderate (SD: 0.086)
Philosophical Signature: Strategy & Value Practitioner
Emphasis: Strategic Intent first, Value Translation second
 
  • Primary focus: Strategic Intent(0.71)
  • Secondary focus: Value Translation(0.65)
  • Relative weakness: Process Integration(0.46)
 
 

MiniMax M2.7

MiniMax

Profile: Strict Formalist

Leniency: 25.8%
Consistency: High (SD: 0.060)
Philosophical Signature: Strategy & Safety Practitioner
Emphasis: Strategic Intent first, Governance & Safety second
 
  • Primary focus: Strategic Intent(0.58)
  • Secondary focus: Governance & Safety(0.57)
  • Relative weakness: Process Integration(0.41)

GPT 5.4

OpenAI
Profile:
Context-Dependent
 
Leniency: 44.4%
Consistency: Moderate (SD: 0.091)
Philosophical Signature: Strategy & Knowledge Practitioner
Emphasis: Strategic Intent first, Knowledge Scaffolding second
 
  • Primary focus: Strategic Intent(0.60)
  • Secondary focus: Knowledge Scaffolding(0.54)
  • Relative weakness: Process Integration(0.33)

unique measurements

Why O‑CRA Is Different

The O-CRA Framework is designed for the dynamic nature of current AI interactions, offering a new perspective over traditional methods.

Real-time diagnostic scoring in Cognitive Resonance & Alignment Framework

Real-time diagnostic scoring.

The six O‑CRA dimensions are computed on‑the‑fly for every AI interaction, delivering an up‑to‑date Organizational Alignment Index.

Traditional Metrics: post-hoc surveys

Holistic interaction analysis with Cognitive Resonance & Alignment Framework

Holistic synthesis of the whole deployment

O‑CRA evaluates strategy, culture, process, safety, knowledge, and business value together, producing both an integrated health score and focused dimensional scores.

Traditional Metrics: isolated factors

Output icon symbolising actionable feedback from Cognitive Resonance & Alignment Framework metrics

Actionable, prescriptive feedback

Each scoring dimension triggers concrete recommendations – re‑tune prompts, adjust policy rules, enrich knowledge bases, or recalibrate ROI targets.

Traditional Metrics: general score

Goal icon for continuous calibration Cognitive Resonance & Alignment Framework interaction modes

Continuous calibration for a changing landscape

O‑CRA runs continuously; as policies change, new departments are added, or business goals shift, the diagnostic updates automatically, keeping the AI‑human partnership in sync.

Traditional Metrics: measures usability

Developer community implementing CR&A Framework in code and applications

Open Source

Both the CR&A and O‑CRA frameworks are released under a Creative Commons Attribution license (CC‑BY‑4.0). All measurement specifications are publicly available on GitHub. We believe alignment research should be open, reproducible, and collaborative. See our Github page here

research and tools

Access the Research & Tools

The O-CRAFramework is an open scientific project. We invite you to explore the research and tools below.

Research paper icon for the CR&A Framework whitepaper on human-AI alignment

Read the
Research Paper

Download the research paper
published in SSRN – Oct 2025:

‘The Cognitive Resonance & Alignment (CR&A) Framework.’

 

Cognitive Resonance & Alignment Framework reliability validation testing for cognitive resonance research metrics

Explore the
Tools

Progress of our Assessor prototype tool, depends on your contribution. Use the Assessor in your projects, extend its capabilities, or contribute to its codebase.

Human-AI Alignment Science

Want to Take the Next Step?

Put the framework to work in your organization.


Whether you’re evaluating AI vendors, shaping internal AI policy, or researching human-AI alignment, we can help.

Contact Us

Let us know what you need assistance with, and we’ll get back to you shortly.