nucleation

LIT-1057: Adapting the Environment Instead of Oneself

2 min read

Properties4
StatusActive — read in full 2026-10-15 from a scan of the published article (Adaptive Behavior 4(3/4), pp. 415–452) that the author posts on his UCSD laboratory server (interactivity.ucsd.edu/articles/Adaptive/Adapting_env_00.pdf), text by OCR and cross-checked against the author's HTML version in the same directory; the PhilArchive copy refused automated requests and was not used. Worth reading as the distributed-cognition statement of environment redesign as a change to a task's state graph or to its "cognitive congeniality", with routine maintenance (keep things where they are expected so that search is unnecessary, against objects' tendency to scatter) and an investment rule for doing work before it is needed. For QUESTION-095 it shows that "amortize the cost of search" is a single parenthetical about chimps leaving nut-cracking stones in place, not developed; that the graph prices only physical transitions, so retrieval and mental cost fall outside the formalism by the paper's own account; and that the nearest thing to a cost model is a verbal "interest factor" whose footnote names the risks that the result will not be used and that the change will not still be present, with no reuse frequency, retrieval cost, staleness of a result or query predictability. Its most usable point is that labour cost varies over time, so work should be done when it is cheap, which is the cockpit's rationale. It does not cite Tarjan (that is Kirsh 1995, LIT-1076), and its value-of-information example changes outcome probabilities rather than pricing a choice made after looking. For CASE-085 it supports the labelled-compartment and standard-layout counterfactuals and their upkeep by the agent, a mechanism or the group, and says nothing on search cost, deadlines or when to stop searching.
Tagscognition, embodied-cognition
Cited as “paper” byNOTE-872 — Kirsh — Adapting the environment instead of oneself
Cited as “source” byTHEORY-355 — Agents arrange their surroundings so that perception and choice take the place of internal search, changing how much thinking a task needs without changing what can physically be done; the set-up is paid once and recovered over the later choice points the arrangement serves, so an arrangement whose scope is the whole task can repay a large set-up and a narrow one must repay within the next few actions, and the arrangement decays unless maintained; the savings are argued and piloted, not measured

Graph View

Kirsh, D. (1996). Adapting the environment instead of oneself. Adaptive Behavior 4(3–4), 415–452. DOI 10.1177/105971239600400307

Key takeaways

Read in full (NOTE-872). A conceptual paper. It models a task as a directed graph of states and actions with a cost metric on transitions, and classifies how an agent can improve its payoff by actions external to the task: pruning hard initial states, routine maintenance, scouting ahead, tools (which change the graph), and actions that make a space more “cognitively congenial” (which leave the graph intact and reduce mental operations). It names amortization once, as a parenthetical gloss on chimps leaving nut-cracking stones in place, and gives an investment rule for maintenance (future benefit against present cost times an “interest factor” for risk) in words only. There is no cost model with reuse frequency, retrieval cost or staleness; mental cost lies outside the graph, and the paper says there is “no general theory” of congeniality. The empirical figures are unpublished pilot results.

Why this is here

Turned up by the citation search on QUESTION-095 (amortized retrieval as amortized computation) and registered at the owner’s request. The closest statement of the owner’s intuition from inside distributed cognition: leaving tools in place to “amortize the cost of search”, named and, by the search’s reading, not modelled.

The reading (NOTE-872) corrects the description from the citation search. “Amortize” occurs once, in an aside about chimpanzees leaving stones where they are useful. The paper’s economics is a verbal rule: do work now if its future benefit exceeds present cost times an “interest factor” for risk, the risks being that the result is not used, that it is not “still present later”, and that the agent does not survive. No cost model.

Backlinks