Introducing MEM: Multidimensional Education Mapping
A navigation framework that helps educators and learners move deliberately between three epistemological worlds - and understand where AI helps, and where it cannot.
Most conversations about AI in education begin with the tool. What can it do? What should we ban? How do we detect it? These are reasonable questions, but they start in the wrong place. They assume the hard part is the technology. It is not. The hard part is knowing what kind of learning is happening - and therefore what kind of help is appropriate.
Multidimensional Education Mapping (MEM) is the framework I have been developing to address this. It is not a curriculum, not an app, and not a policy document. It is a navigation system: a way of seeing learning that makes certain failures visible, certain AI uses obviously appropriate, and certain AI uses obviously wrong.
The architecture
MEM has four coupled layers. Each depends on the one above it.
1. Ontology - the Three Worlds. All knowledge work moves through three epistemological domains: the Mental (subjective experience, empathy, ethics, identity), the Formal (rules, models, propositions, codified knowledge), and the Physical (making, testing, embodied practice). These are not departments or subjects - they are validity regimes. What counts as evidence differs in each. I have written about this ontology in more detail in a previous post; it is the foundation everything else rests on.
2. Control system - metacognitive regulation. If the worlds are the territory, metacognition is the compass. Learners need to know which world they are in, which world they need to move to, and what tools are appropriate for each. MEM calls this world literacy. The canonical metacognitive prompt is simple: Where am I? What kind of thinking does this require? What can AI do here - and what can't it?
3. Ethical spine - agency and public purpose. Human agency is not a nice-to-have. It is the structural commitment that prevents AI from becoming a shortcut machine. Work in MEM is oriented toward real problems with real stakeholders - what I call public purpose. This is not community service bolted onto a project; it is the design constraint that forces learners into the Mental and Physical worlds, where AI cannot substitute for them.
4. AI integration - orchestration. AI is positioned as a formal intelligence: a cognitive partner whose efficacy varies systematically by world. High in the Formal world. Low in the Mental world. Indirect in the Physical world. This is not a value judgement - it is a structural observation. The role of the learner is to orchestrate AI's contributions, not to be orchestrated by them.
The diagnostic: world errors
One of MEM's most practically useful contributions is the concept of world errors - category mistakes where learners (or teachers, or institutions) apply the wrong world's tools to a problem.
Staying abstract too long when physical testing would resolve uncertainty. Building too early before the empathy work has been done. Empathising endlessly without converging on a specification. Trying to use science to answer questions about feelings, or art to answer questions about specific heat capacity.
These are not signs of weak students. They are predictable failure modes that become visible and correctable once you name the worlds. A teacher who can say "you are making a world error - you are in the Formal world when you need to be in the Physical world" has a precise diagnostic tool that cuts across subjects.
The practice: world shifting
If world errors are the diagnosis, world shifting is the treatment. The canonical MEM move is a deliberate cycle of translation:
- Use the Mental world to empathise and clarify what matters.
- Translate those insights into Physical success criteria.
- Use Formal tools - science, modelling, reasoning - to theorise designs that might achieve those criteria.
- Build, test, and return to reflection.
The quality of a learner's world shifting - its intentionality, its timing, its appropriateness - is a key indicator of metacognitive development. Expertise, in this framing, is not the ability to occupy all three worlds simultaneously. It is increasingly rapid and fluid travel between them.
The tensions
These are not flaws in the framework. They are where it becomes alive.
Fluency versus warrants. AI amplifies fluency - the ability to produce well-formed outputs. This raises the stakes for verification, grounded evidence, and epistemic humility. A student who can generate a polished essay in seconds needs stronger, not weaker, skills in evaluating whether it is true.
Sovereignty versus convenience. AI's frictionless availability tempts learners to cede judgement. Metacognitive regulation is the defence, but it requires practice and it requires that we design tasks where ceding judgement has visible consequences.
Voice versus averaging. AI's outputs are drawn from training data that under-represents marginalised perspectives. Protecting student voice while using AI is not a pedagogical preference - it is a structural commitment.
What MEM is not
MEM is not a technology framework. It does not tell you which AI tool to use. It is not a rubric, though it informs how rubrics should be designed. It is not a prohibition on AI use - it is a system for making AI use legible, so that teachers and students can reason about it together.
The ambition is a framework that works from a single lesson to a full curriculum redesign. Subsequent posts in this series will explore specific applications: assessment design, the role of disciplines as bridges between worlds, and the minimal version of MEM that a teacher can adopt in a single unit of work.