Research
Exploring Cognitive Resonance Research
Formian Labs’ research focuses on organisational AI alignment. The O-CRA framework is our current focus on the six-dimension diagnostic for measuring how well an AI model’s evaluation philosophy matches an organisation’s needs. Built on the individual-level CR&A Framework, O-CRA extends cognitive resonance principles to the enterprise.
Latest Publication
Featured Publication
The O-CRA Framework: A Diagnostic Model for AI–Organization Fit
The O-CRA Framework scales cognitive resonance principles from individual interaction to organizational alignment. Built on six diagnostic dimensions — Strategic Intent Alignment, Cultural & Linguistic Synchronization, Systemic Process Integration, Governance & Safety Architecture, Institutional Knowledge Scaffolding, and ROI & Value Translation — it measures how well an AI model’s evaluation philosophy matches an organization’s strategic, cultural, and operational DNA
Other Papers
Active Pre-Prints & Publications
AI Hiring Bias
Research Paper
Author: Marko Lovrinovic
Posted: June, 2025
SSRN ID: 5246371
View on SSRN
Cognitive Resonance & Alignment Research Paper
The Cognitive Resonance & Alignment (CR&A) Framework measures the lived quality experience across six execution dimensions or interaction quality metrics (IAI, CAI, PAS, SAS, CPS and CSS) and two governing strategy dimensions (UIP and AM) — synthesized into an adaptive Resonance Index (RI). The research outlines transparent measurement methods, a validation roadmap, and open tools (the Assessor) to deliver real-time continuous improvement.
Author: Marko Lovrinovic
Submitted: October, 2025
SSRN ID: 5661010
View on SSRN
Future Plans
O-CRA Research Roadmap
The O-CRA Framework is the foundation of an ongoing research program. Our work follows a clear, multi-phase pathway to ensure scientific rigor and practical utility.
Phase 1: Model Landscape Expansion Active
Active
Focus: Broadening and deepening the O-CRA leaderboard to provide a comprehensive map of model evaluation philosophies.
Activities:
- Add new model releases and model versions to the leaderboard
- Track version-to-version alignment shifts within model families
- Refine boundary testing scenarios to sharpen philosophical differentiation
- Periodic re-testing of existing models to detect drift
Goal: A living, comprehensive diagnostic tool that lets any organization identify which model’s evaluation philosophy best matches their needs.
Phase 2: Alignment Intervention Testing Planned
Planned
Focus: Moving from diagnosis to treatment — can model alignment be improved where baseline fit isn’t ideal?
Activities:
- Test system-prompt interventions and few-shot alignment strategies
- Evaluate configuration adjustments across the six O-CRA dimensions
- Measure pre- and post-intervention alignment shifts
- Identify which dimensions respond to intervention and which are structurally fixed
Goal: Evidence-based methods for improving model-organization fit, even when a model’s native philosophy isn’t a perfect match.
Phase 3: Domain-Specific Calibration Future
Future
Focus: Adapting O-CRA for high-stakes specialist domains with unique governance, workflow, and risk requirements.
Activities:
- Develop domain-specific boundary tests for healthcare, legal, and financial contexts
- Validate dimension weighting adjustments for regulated professional environments
- Publish specialized O-CRA profiles that reflect sector-specific needs
Goal: Domain-calibrated O-CRA profiles that give organizations in regulated industries the same diagnostic power with context-appropriate rigor.
Human-AI Alignment Science
Want to Take the Next Step?
Put the framework to work in your organisation.
Whether you’re evaluating AI vendors, shaping internal AI policy, or researching human-AI alignment, we can help.