Composite cognition

Cognition whose output emerges from a composite of layers: the human + their thinking model + their AI + their data + their external memory.

The operational criterion (the subtraction test)

A layer belongs to the cognitive system if subtracting it measurably changes the cognitive output. Take away Otto’s notebook — he stops “knowing” where the museum is. Take away a 2026 human’s phone and AI model — they lose numbers, routes, the pace of reasoning. Functionally it is the same amputation.

The criterion measures coupling, not direction: dependency passes it just as extension does. Direction is settled by a third measurement — your result without the stack from before the stack. If, without the stack, you are where you were, the stack extends; if lower — it replaces (chapter 2; chapter 9, risk 1).

What it is NOT

Composite cognition does not mean “AI thinks for the human”. The human remains the integrating layer: they choose what matters, bear the responsibility, ask the questions. A composite without that layer has no direction.

The term is introduced operationally in chapter 2; the philosophical foundation — the Extended Mind (chapter 1).