Educated Attention (Ingold)

In the AI Story

James J. Gibson's ecological psychology demonstrated that affordances are real properties of the environment, perceivable by organisms with appropriate perceptual apparatus. Ingold extends Gibson by adding the temporal dimension: the perceptual apparatus itself evolves through practice. It is not a fixed endowment but a capacity continuously reshaped by engagement. The cabinet maker does not see optically more than the novice — the photons are identical. She sees semantically more: the grain pattern is not merely visual texture but a guide to internal structure, mechanical properties, likely behavior under tools. This meaning is perceived in the material by a system trained through feedback from thousands of boards handled, cut, finished, and observed in use.

Large language models possess a form of educated attention — attention shaped by training on vast text corpora, perceiving patterns that untrained perception would miss. The model detects structural parallels between domains, identifies logical inconsistencies, generates connections between concepts. These are real perceptual capacities. But the two forms of educated attention differ in what educated them and what they perceive. The cabinet maker's attention was educated by wood — by specific resistance, weight, visual and tactile character of material objects. It perceives material affordances. The model's attention was educated by text — by statistical regularities in language. It perceives linguistic affordances. Neither is superior in the abstract; each is powerful within its domain. The substitution of one for the other changes what knowledge is produced.

The software architect's decisions about system design were historically informed by years of direct material engagement — writing code, debugging, watching systems behave under load, encountering ways that computational substrates do not behave as documentation promises. This engagement educated her attention: she perceives architectural affordances and vulnerabilities that less experienced engineers miss. In AI-mediated workflows, her decisions are informed by evaluating AI-generated output — a representational operation assessing descriptions rather than a perceptual operation grounded in material encounter. The affordances she perceives are affordances of the specification, not of the system's actual behavior under conditions the specification did not anticipate. The representation may be accurate, but perceiving the representation is not the same as perceiving the system.

Origin

The concept synthesizes Gibson's ecological realism (affordances are real, perception is direct) with phenomenological accounts of skill (Merleau-Ponty's Phenomenology of Perception, Dreyfus's skill acquisition model) and anthropological fieldwork. Ingold developed it across The Perception of the Environment (2000), Being Alive (2011), and Making (2013). The term itself is Ingold's coinage, distinguishing his position from both Gibsonian direct perception (which underemphasizes development) and constructivist accounts (which overemphasize mental representation). Educated attention is direct perception by a perceptual system that has been educated — perception shaped by practice, not mediated by concepts, yet transformed by experience.

Key Ideas

Perception is educated by practice. The perceptual system is not fixed but continuously reshaped by the body's engagement with materials — what you can see depends on what your hands have done.

Affordances become progressively legible. The master perceives what the novice cannot, not through superior information but through a perceptual system trained to read material meaning — grain as structural guide, vibration as diagnostic signal.

AI possesses linguistic educated attention. Large language models perceive patterns in text that humans miss — but their attention was educated by language, not by material, and they perceive linguistic rather than material affordances.

Substituting one attention for another changes knowledge. When decisions once informed by material perception are increasingly informed by linguistic pattern-matching, the character of professional knowledge shifts from enacted to representational.

Debates & Critiques

The debate is whether AI-mediated work can educate attention in new ways — whether the iterative refinement of prompts and evaluation of outputs constitutes a new form of perceptual training that produces its own educated attention. Defenders of AI collaboration argue it does, pointing to practitioners who report perceiving patterns in AI output they would not have noticed in their own work. Ingold's framework predicts this is real but categorically different: it is attention educated by representations rather than by materials, and the knowledge it produces will differ accordingly. The long-term question is whether a profession can sustain expert judgment when the material engagement that historically educated practitioners' attention is systematically replaced by the evaluation of machine-generated specifications.