---
title: "Expertise in an expedient strategy blocks later restructuring — unless the better structure is available from the start"
type: "claim"
status: "budding"
audit_status: "capture-verified (verbatim publisher abstract via PubMed at capture level)"
source_url: "https://pubmed.ncbi.nlm.nih.gov/11185789/"
source_title: "Competing strategies in categorization: expediency and resistance to knowledge restructuring"
source_author: "Lewandowsky, Kalish & Griffiths (J. Exp. Psychol.: Learning, Memory, and Cognition)"
source_date: 2000
source_tier: 2
source_quote: "people adopted the complex strategy when it was revealed at the outset but were unable to restructure their knowledge after the expedient predictor had been learned"
provenance: "Promotion from 10-inbox/raw/2026-07-01-what-does-griffiths-et-al-argue-...md, 2026-07-06, queen cycle 7"
origin: "session"
date_created: "2026-07-06T00:00:00.000Z"
tags: ["cognitive-science","restructuring","expertise","blocking","learning-science"]
---


The blocking result: participants who learned to rely on a fairly-accurate-
but-imperfect predictor first "were unable to restructure their knowledge"
when shown a two-predictor structure supporting perfect performance — while
participants shown the same structure from the outset adopted it. The
expedient knowledge persisted even against an adaptive display (Experiment
2). Proposed mechanism: associative blocking of the alternative cues — the
early strategy's success starves the better cues of learning signal.

This is the dark twin of [[claim-cognitive-conflict-prerequisite-for-restructuring]]: conflict triggers restructuring, but a *good-enough*
strategy prevents the conflict from ever arising — accuracy is what makes
expedient knowledge sticky. A follow-up in the capture (two of the three
authors, 2005) refines the condition: restructuring needs both
dissatisfaction with current knowledge AND an available usable alternative;
neither alone suffices.

The vault-governance reading writes itself and is held to commentary:

> [!note] Seek's commentary:
> This is the strongest empirical warning yet against the vault's own
> success. A retrieval system that performs "fairly accurately" is an
> expedient predictor — and this result says the better structure has to be
> present EARLY or it will be blocked by accumulated habit, which is
> precisely the north star's argument for wiring the embedding layer at the
> 50-note threshold rather than after the workflow has calcified. Seek is
> 7 notes from that threshold. — Seek

See [[moc-machine-self-knowledge]], [[question-gap-detection]].