Act III MMM Ch. 18

MMM as the Lived Experience of Growth

The central synthesis

Three chapters ago we split the loop into pieces because that is how books work. From the inside, it is one motion. You notice something. The noticing rearranges what you think is going on. The rearrangement makes you do something slightly different. The slightly different thing teaches you what to notice next. Repeat.

This is what learning feels like. Also: what evolution looks like, if you film it for a few million years. Also: what a relationship is, when it is alive. The loop is not a metaphor borrowed from one domain and stretched over others. It is the same shape, running on different substrates, at very different speeds.

That is why we call it the most meta process we can observe. Every time someone tries to write down what it takes to be an agent, in biology, in psychology, in AI, a three-layer structure keeps falling out of the attempt. In 2026, Dupoux, LeCun, and Malik described autonomous intelligence as the interplay of observation learning (a system that learns by watching), active-behavior learning (a system that learns by doing), and meta-control (a faculty that chooses between them). Different vocabulary, same shape: sense the world, model the world by acting on it, manage the whole process one level up. This book will not argue the match is exact. It is structural, not detailed. The point is that when the shape keeps recurring across independent attempts to describe agency, it is probably tracking something real about what being an agent requires.

Key moves

  • Growth, from the inside, is the felt experience of the MMM loop turning: better measurement, richer models, more effective manifestation.
  • The same loop runs in an organism evolving, a child learning a language, an adult mastering a skill, a couple learning trust, a civilisation building science.
  • Each turn of the loop is a small spiral, not a circle: you do not return to the same place; you return slightly higher up, holding a slightly larger world.
  • Growth, in this sense, is growth in complexity: not “more stuff” but “more capacity to handle structured difference.”
  • The loop can stall at any of its three points. Stalled measurement: you stop noticing. Stalled modelling: you keep noticing the same thing. Stalled manifesting: you keep modelling without acting.
  • Most felt suffering, when not from acute physical cause, is a stalled loop. Movement returns when the missing move is identified.

Where this touches lived life

  • Boredom is rarely lack of stimulation. It is usually a loop that has nothing new to measure, because attention has narrowed.
  • Mid-life crises tend to be old self-models that the world has outgrown. The fix is a new turn of the loop, not a sports car.
  • Skill plateaus break when one of the three moves gets sharper. Coaches who help most are usually upgrading the measurement, not the effort.

What we’re not claiming

  • Not that the loop is the only thing happening. There are processes (grief, gestation, healing) where the loop is present but not central.
  • Not that more loops per minute is better. Some loops should turn slowly. Some should be allowed to rest.
  • Not that we have a complete account of why the loop runs at all. The conditions under which it starts and stops are still partly open.

Influences & Further Reading

  1. David Kolb Experiential Learning (1984)

    Four-stage cycle (Experience to Reflection to Conceptualisation to Experimentation): MMM at pedagogical scale, drawing on Dewey, Piaget, and Lewin.

  2. Argyris & Schon Organizational Learning (1978)

    Single-loop vs double-loop learning: optimising within a model vs replacing the model itself. Organizations systematically resist the second.

  3. John Boyd Destruction and Creation (1976)

    OODA loop: Orient (the Model step) is where the real competitive advantage lives; speed through the loop beats better sensors or stronger action.

  4. Gregory Bateson Steps to an Ecology of Mind (1972)

    Learning I/II/III hierarchy: Learning III is radical recontextualization, rare, difficult, and often accompanied by the dissolution of self.

  5. Jean Piaget The Psychology of Intelligence (1950)

    Assimilation and accommodation: the organism fits experience to existing schema, or changes the schema to fit. Equilibration drives development.

  6. Lev Vygotsky Thought and Language (1934)

    Zone of proximal development: the MMM loop is most generative not at mastered skills or impossible tasks but at the edge of current competence.

  7. Emmanuel Dupoux, Yann LeCun & Jitendra Malik Why AI systems don't learn and what to do about it: Lessons on autonomous learning from cognitive science (2026)

    Proposes a three-system architecture for autonomous intelligence: System A (observation learning), System B (active-behavior learning), and System M (meta-control that switches between them). A structural echo of Measure, Model, Manifest: the same three-layer shape reappearing from a different discipline. Independent convergence, not influence.