Body Knowledge

In the AI Story

The concept traces to Gilbert Ryle's The Concept of Mind (1949), which attacked what Ryle called the 'intellectualist legend' — the idea that all intelligent action is preceded by an act of theoretical reasoning. Ryle argued that knowing how to ride a bicycle is not a matter of applying propositions about cycling; it is a capacity that can only be exercised, not stated, and that resists communication in language. Subsequent thinkers extended the insight: Michael Polanyi's concept of tacit knowledge, Hubert Dreyfus's phenomenology of skilled coping, and the enactivist synthesis that made embodied know-how central to cognition itself.

For Noë, body knowledge is not merely one category among others. It is the developmental ground of all knowledge. Infants develop practical mastery of reaching, grasping, and locomotion long before they develop any propositional understanding of physics or space. The practical mastery is not preparation for cognition; it is the medium through which cognition develops. The body knows first; the mind articulates later. This developmental priority has profound implications when tools enter the picture that bypass the body's participation.

The Orange Pill's geological metaphor — every hour of debugging deposits a thin layer of understanding, accumulating into something solid — describes body knowledge precisely. The senior developer's architectural intuition is not a set of rules stored in memory; it is a distributed capacity across her cognitive-bodily system, built through decades of friction with resistant systems. When Claude handles implementation, this deposition does not occur. The output exists; the body knowledge does not. And body knowledge is what enables judgment in situations the training data did not anticipate.

The neuroscience of embodied cognition supports this analysis empirically. Motor areas of the brain activate during purely perceptual tasks; mirror neuron research suggests perception and action are integrated aspects of a single sensorimotor system; bodily states influence cognitive processes in ways the computational model struggles to accommodate. These are not quirks. They are evidence that cognition is constitutively embodied, and that the body's participation in cognitive work is not ornamental but constitutive.

Origin

Gilbert Ryle, The Concept of Mind (1949), introduced the knowing-that/knowing-how distinction. Michael Polanyi's Personal Knowledge (1958) and The Tacit Dimension (1966) developed the concept of tacit knowledge. Hubert Dreyfus's work on skilled coping (What Computers Can't Do, 1972; Being-in-the-World, 1991) brought the tradition into direct engagement with AI.

Key Ideas

Know-how over know-that. Practical bodily mastery is the foundational form of knowledge; propositional knowledge is derivative.

Non-transferable through language. Body knowledge cannot be communicated by description alone; it must be developed through the body's engagement with resistant material.

Deposited through friction. Each hour of struggle with a system that pushes back deposits a layer of embodied understanding that accumulates over years.

Invisible in the output. The product of skilled work does not display the body knowledge that produced it, making the erosion of that knowledge undetectable by output metrics.

The condition of judgment. The capacity to handle situations the training data did not anticipate depends on body knowledge; its absence shows up only in the crisis.

Debates & Critiques

Critics of strong embodiment claims argue that what Ryle and Polanyi called tacit or practical knowledge could in principle be made explicit through sufficiently detailed description, and that the appearance of non-transferability reflects current limits of our descriptive vocabulary. Defenders respond that some forms of knowing are constitutively bodily and cannot be exhausted by any propositional specification, however detailed.