# The research behind the reasoning. Canonical: https://axiomcx.dev/research/ Publisher: Axiom Cortex / TeamStation AI Content reviewed: 2026-09-26 Axiom Cortex applies neuro-psychometric alignment intelligence to the interviews your team already runs. Explore the behavioral axioms, reasoning domains, and mathematical methods behind the analysis. We use science to align tomorrow’s IT talent with the work ahead. ## The person behind the answer. AI can help produce an answer. Real engineering work gets messy. Engineering work still needs people who can explain the problem, weigh tradeoffs, catch errors, and take responsibility. Axiom Cortex studies that demonstrated reasoning in the context of a specific role. ### The answer What did the candidate explain, and which parts are supported by their own words? ### The role What does this job require when the system fails, the scope changes, or a decision affects other teams? ### The next question Which missing detail would help the hiring team understand alignment before deciding? ## Five checks. Each answer. B-Axiom checks keep the analysis tied to what the candidate actually demonstrated. ### Accuracy Technical correctness against the question and ideal-answer criteria. ### Mental Model The explanation of mechanisms, dependencies, and cause and effect. ### Procedural Knowledge The steps used to implement, diagnose, test, and recover. ### Clarity Whether the technical explanation communicates the relevant reasoning. ### Cognitive Load How the answer handles interacting constraints and technical complexity. Clarity concerns the technical explanation. Cognitive Load concerns the handling of interacting constraints. Neither is a brain, health, stress, or intelligence measurement. ## How reasoning meets the role. Mental shape is Axiom’s name for an evidence-bound profile of demonstrated work reasoning. It describes the available interview evidence. It doesn’t describe the whole person. ### Conceptual Fidelity Does the explanation preserve the concepts the system actually depends on? ### Architectural Instinct How are boundaries, dependencies, failure paths, and tradeoffs handled? ### Problem-Solving Agility How does the approach change when new evidence changes the problem? ### Collaborative Mindset How are decisions shared, corrections handled, and handoffs made? ### Learning Orientation How is new evidence used to revise an incomplete or outdated model? ### Metacognitive Conviction Does expressed confidence match the evidence, including what remains unknown? ### Four foundational traits and six research domains The 2025 report combines five B-Axiom checks into four role-fit traits. The 2026 study represents mental shape using six work-reasoning domains. These are related research layers, not interchangeable score definitions. The foundational traits are Architectural Instinct, Problem-Solving Agility, Collaborative Mindset, Learning Orientation. ## What the mathematics examines. Different methods answer different questions. A result needs the right source material, configured inputs, and a record of which calculation actually ran. Axiom’s public method registry describes 44 governed methods across 6 mathematical families: 23 formula methods, 16 logic methods, and 5 measurement methods. The topics below organize the explanations by what they examine. They are not a new count of mathematical families. ### Meaning and conceptual distance Methods: Conceptual Fidelity, Fréchet semantic distance. Compare the concepts expressed in the answer with the concepts required by the ideal-answer blueprint. Shows where the technical meaning aligns, diverges, or lacks support. Correct paraphrases can preserve meaning without repeating the blueprint. Distance methods require approved embeddings, comparison inputs, and a calibrated mapping. Semantic similarity alone does not establish technical correctness. ### Reasoning structure and connections Methods: Discourse analysis, Optimal Transport, Wasserstein distance. Compare represented relationships among concepts and steps, including the connection between a claim, its mechanism, and its consequence. Helps distinguish a connected explanation from adjacent technical terms and locate differences from the expected reasoning. A transport distance describes the supplied representations and cost model. It does not prove a reasoning sequence unless those relationships are explicitly represented. ### Five behavioral axioms Methods: Accuracy, Mental Model, Procedural Knowledge, Clarity, Cognitive Load. Examine correctness, causal understanding, execution steps, explanation clarity, and handling of technical complexity within each answer. Shows which part of an answer is supported and which needs more evidence. A correct definition and a demonstrated implementation address different criteria. Measurements stay tied to the approved role rubric. Cognitive Load is an evidence-bound rubric dimension, not a measurement of brain activity, stress, or health. ### Depth, adaptation, ownership, and collaboration Methods: Latent Trait Inference, Logical Knowledge Depth, Problem-Solving Trajectory Analysis, Context Setting, Metacognitive Calibration. Connect answer-level findings to mechanisms, problem decomposition, changing constraints, clarification, contribution, stakeholder impact, and acknowledgment of unknowns. Builds the role-specific mental-shape view: architectural instinct, problem-solving agility, learning orientation, and collaborative mindset. Trait synthesis needs an approved mapping from answer evidence. The separate six-domain human-task-agent research model must not be substituted for a production scoring policy. ### Uncertainty and the next useful question Methods: Bayesian evidence update, Expected Information Gain. Model how additional evidence changes an estimate and which approved follow-up could reduce the remaining uncertainty. Directs attention to the gap that matters before a decision, including evidence that could change the current interpretation. These documented methods require calibrated priors, question mappings, and measurement-error inputs. Missing inputs cannot become an invented confidence interval. ### Relationships across demonstrated skills Methods: Gaussian Graphical Models, Partial correlation. Study associations among measured skill dimensions while accounting for other represented dimensions. Research can examine which skill measurements connect and where the available evidence is fragmented. Requires an adequate dataset and approved model. The V4 specification keeps this in offline validation or shadow research unless those requirements are met; association does not establish causation. ### Language calibration, reliability, and bias analysis Methods: Translation invariance, Inter-rater reliability, Generalizability analysis, Expected Calibration Error, Differential Item Functioning. Test whether equivalent meaning receives consistent treatment, whether evaluators agree, and whether measurements vary across questions or relevant study groups. Examines the quality of the measurement itself, including unwanted sensitivity to language form and subgroup differences. Reliability, calibration, and fairness statistics need suitable study data. Accent, pauses, pronunciation, and protected traits cannot stand in for job-related evidence. ### Role alignment and critical requirements Methods: Configured aggregation, Critical-requirement gates, Human-task-agent alignment distance. Combine approved measurements according to the role and preserve the effect of must-have criteria, assessment coverage, and unresolved requirements. Shows the supported fit to a particular delivery role and the gaps an overall average could otherwise conceal. One complete versioned policy governs a result. The published six-domain distance model studies human-task-agent alignment separately from production aggregation. If required evidence or parameters are missing, the affected result must remain unavailable. A method listed in the research is not proof that it ran in a particular evaluation. - [Read the documented calculation and evidence contract](/data/processing-engine.json): Public method purposes, input conditions, and execution boundaries. ## Read the source work. These links distinguish working papers from company explanations. Publication on SSRN does not, by itself, mean peer review. - [AxiomCortex: Scientific R&D Report](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5433476): FOUNDATIONAL WORKING PAPER · SSRN. Defines Answer Evaluation Units, B-Axiom checks, trait synthesis, Conceptual Fidelity, L2-aware calibration, aggregation, reliability, fairness, and monitoring. - [Human-Task-Agent Alignment Across Software Team Topologies](https://ssrn.com/abstract=7256278): WORKING PAPER · SYNTHETIC STUDY. Tests six reasoning domains, weighted alignment distance, team topology, agent autonomy, measurement sensitivity, queue pressure, and synthetic coefficient recovery. - [Axiom Cortex for LATAM Agentic Engineering](https://teamstation.dev/research/articles/axiom-cortex-latin-america-agentic-engineering-alignment): COMPANY ARTICLE · TEAMSTATION AI. Explains how interview evidence connects to role fit, engineering loops, governance, and delivery alignment. - [CTO Guide to Agentic Workflow Fit Signals](https://teamstation.dev/research/articles/how-ctos-can-align-the-right-mental-shape-in-their-agentic-ai-dev-workflows): COMPANY ARTICLE · TEAMSTATION AI. Connects work-reasoning signals, team topology, and human-plus-agent engineering workflows. - [Telemetry Predicts Team Performance](https://teamstation.dev/research/articles/how-telemetry-finds-the-right-mental-shape-and-predicts-team-performance): COMPANY ARTICLE · TEAMSTATION AI. Explores how post-hire delivery signals can test whether interview evidence stays aligned with real work. ### What the current evidence establishes The 2025 report documents the evaluation framework. The 2026 study stress-tests an outer alignment model on synthetic data. Both are working papers, not peer-reviewed validation. Company articles are not independent scientific validation. The published six-domain alignment study uses synthetic profiles to test model behavior. Independent predictive validity for future job performance is not publicly established. A repeatable calculation and a useful example are different from a validated hiring outcome. The complete author, revision, and publication metadata should be read on the linked source record. No missing paper metadata is inferred here. ## Keep exploring - [how the method uses interview evidence](/processing-engine/) - [see the illustrative report](/evaluation/) - [read baseline and language limits](/guides/interview-integrity/) [See the evaluation workflow](/processing-engine/)