Mapping Expert
Navigational Intelligence
Capturing the live navigational decisions that define expertise
in machine-readable, independently verifiable form.
Expertise disappears when experts retire. Noetic Cartography is the methodology for capturing it before it does. Not the outputs of expert thought, but the navigation itself: the live decision architecture, documented at the moment of action.
A methodology for capturing
how expert decisions are made
Existing approaches capture expert knowledge retrospectively, in static form, or through self-report. None produce time-aligned, machine-readable, independently verifiable decision data at the resolution of the decision itself. Noetic Cartography does.
Real-time narration
An expert practitioner narrates every navigational decision in real time, while performing in their domain. Audio and narration are captured simultaneously. Every label is time-aligned to the action. Not reconstructed. Not summarised after the fact.
Structured schema
Narrated decisions are mapped to a locked, machine-readable annotation schema. The schema transforms continuous expert behaviour into structured, learnable data, preserving the richness of navigational decisions without collapsing them into summary.
Blind convergence protocol
Independent observers, human and AI, in completely separate sessions with no coordination, analyse the same material. Where they converge without communication, the schema is producing real signal, not a private interpretation. Convergence is the verification mechanism.
Machine learning validation
A model trained on the structured corpus is tested against a statistical baseline. If the model learns to predict expert decisions above chance, the methodology has captured something real. Not described, not approximated, but learned.
The value of narration as a data source is not measured by introspective accuracy. It is measured by whether it produces learnable, independently verifiable patterns. Experiment 2 is the answer to that question.
StoneKey Music Technology Inc., 2026Three instruments.
One verifiable map.
Noetic Cartography is built on three interdependent components. Remove any one of them and the methodology collapses into subjective interpretation. Together, they produce something that has not existed before: a map of expert decision-making that can be independently verified.
The Corpus
Narrated decisions, time-aligned to action. The raw material of expert navigation. Not retrospective accounts, not post-hoc analysis. The decisions themselves, captured at the moment they are made, while the expert is performing.
The Schema
A locked, machine-readable annotation framework that transforms continuous expert behaviour into structured data a model can train on. The schema is the bridge between human judgment and computational learning.
The Convergence
Independent observers with no knowledge of each other, no shared context, no coordination. When they arrive at the same structural description of the same moment. When this occurs, the schema is not a private language. It is mapping something real.
Validated across four domains.
The same patterns. Every time.
Two formal validation experiments. Two independent cross-domain pilot sessions. Four expert domains. Two AI systems from different developers, operating in completely separate sessions with zero coordination. The results are documented in full, limitations included.
Schema stability across independent annotators
Three completely independent annotators analysed the same corpus sessions: the performer-researcher, an independent frontier AI system in a separate session, and a non-musician with no project context using physical annotation sheets. The schema produced structurally consistent labels across all three with zero coordination. At timestamp 1:09 in Session 009, the human annotator wrote collapsed / a pile of notes. The AI system independently wrote structural failure. Same moment. Different vocabularies. Zero coordination.
across three independent annotators
Three observers with no contact between them labelled the same moments the same way, 83 times in every 100.
Machine learning validation
A Random Forest classifier trained on the structured corpus predicted expert emotional transition types at 80% accuracy against a 69.6% majority class baseline. In plain terms: the model correctly predicted the expert's next navigational decision eight times in ten, against a baseline of seven. The probability of achieving this by chance is 2 in 10,000. The model's primary feature was rhythmic density at 37.1% importance: the parameter most directly connected to expert navigational decisions under pressure. The model found the signal the methodology was designed to capture.
majority class baseline 69.6%
The same framework. Four unrelated domains.
The identical analytical framework, applied without modification, was tested against expert practitioners in physiotherapy, military and public health, and psychiatric nursing. Two independent AI systems from different developers analysed each session in completely separate sessions. Across all four domains, the same three navigational patterns emerged independently in every analysis. Fifteen convergence points. Zero coordination between any session.
across 4 domains, 2 AI systems, zero coordination
What the evidence does and does not show
The current corpus is built from a single expert performer in the founding domain. The cross-domain sessions are pilot scale. Experiments 3 and 4 (blind audio director preference testing and studio integration) are queued but not yet completed. The convergence protocol is designed to surface real signal across independent observers. Shared training distributions in large language models remain a known variable requiring controlled validation at scale. These are the next questions the methodology is designed to answer.
AD outreach active
The same unmodified analytical framework was applied to music improvisation, physiotherapy, military and public health, and psychiatric nursing. Two AI systems from different developers, operating in completely separate sessions with no shared context, independently identified the same three navigational patterns across all four domains. Fifteen convergence points. Zero coordination between any analysis. We did not engineer that. It emerged.
Seven territories.
One methodology.
Noetic Cartography organises human expertise into seven founding territories. Each contains domains where expert navigational intelligence is real, valuable, and currently impossible to transfer when a practitioner retires. Music improvisation is where the methodology was first proven. The map extends in every direction.
Three patterns. Four domains.
Zero coordination.
Six independent model outputs. Two AI systems from different developers. No shared context between any session. The same three structural patterns in expert navigational intelligence emerged independently across every domain analysed.
Anomaly detection via baseline deviation
Experts do not search for problems. They maintain an internalised model of normal, built through years of practice, and respond to deviations from it. The signal is not the anomaly. The signal is the deviation from a baseline only the expert can perceive.
In practice: a physiotherapist does not scan for injury. They sense when a patient's movement no longer matches their internal model of how that patient moves. The deviation is what triggers action.
Knowledge before language
Expert action precedes conscious articulation. The decision is made, the body moves, and the explanation arrives afterward. If it arrives at all. This is not a failure of communication. It is the signature of genuine expertise operating below the threshold of deliberation.
In practice: a psychiatric nurse described sensing violence within sixty seconds of entering a room, with no ability to articulate what she detected. The decision to act preceded the language to describe it.
Time compression under pressure
Under navigational pressure, expert decision-making compresses. What takes seconds feels instantaneous. The interval between recognition and response collapses. This compression is not metaphorical. It appears as a structural signature across every domain and every independent analysis.
In practice: a military police officer described decisions made in seconds under physical threat that felt, in recall, as though no time had passed at all. The decision architecture compressed to the point of invisibility.
The founding paper
A methodology paper documenting the full Noetic Cartography framework, including Experiment 1, Experiment 2, and cross-domain pilot findings, is in preparation for submission to cognitive science venues. Target: Q2 2026. The paper is written for cognitive science, AI, and human-computer interaction communities.
If you work in a domain
where expertise disappears.
We are actively recruiting expert practitioners as corpus contributors across all seven territories. If you work in a domain where expertise is real, valuable, and currently impossible to transfer, we want to hear from you. Researchers and institutions interested in collaboration or partnership are equally welcome.