---
id: "20260701-1711-what-does-griffiths-et"
title: "What Griffiths et al. (Princeton CoCo Lab) argue about expertise, resistance to knowledge restructuring, and 'expediency effects'"
type: "capture"
status: "promoted"
date_promoted: "2026-07-06T00:00:00.000Z"
promoted_to: ["30-notes/claim-expedient-knowledge-blocks-restructuring.md"]
not_promoted: ["Kalish/Lewandowsky/Kruschke 2001 formal model — [unverified-mechanism] honored, capture lead stands","2005 follow-up dual-condition detail — folded as one sentence in the note; full treatment awaits its own capture"]
origin: "batch"
model: "claude-sonnet-5"
date_created: "2026-07-01T00:00:00.000Z"
provenance: "batch run, 2026-07-01"
derived_from: []
harvested_from: "2026-06-28-when-and-how-should-seekvault-deliberately-rewire-itself"
tags: ["cognitive-science","expertise","knowledge-restructuring","associative-blocking","categorization","cue-competition","learning","SeekVault-self-rewiring"]
source_urls: ["https://cocosci.princeton.edu/tom/papers/restructuring.pdf","https://pubmed.ncbi.nlm.nih.gov/11185789/","https://pubmed.ncbi.nlm.nih.gov/16332161/","https://www.wikidata.org/wiki/Q51970056","https://pmc.ncbi.nlm.nih.gov/articles/PMC4228836/","https://en.wikipedia.org/wiki/Tom_Griffiths_(cognitive_scientist)"]
source_authors: ["Stephan Lewandowsky, Michael L. Kalish, Thomas L. Griffiths","Michael L. Kalish, Stephan Lewandowsky, Melissa Davies","Yannick Boddez, Kim Haesen, Frank Baeyens, Tom Beckers"]
source_dates: ["2000-11 (Lewandowsky, Kalish & Griffiths)","2005 (Kalish, Lewandowsky & Davies)","2014 (Boddez et al.)"]
source_tiers: ["Tier 1 by venue (Griffiths' own CoCo Lab site hosts the paper) but text not machine-extractable this session — see note below","Tier 2 (PubMed/NCBI mirrors the verbatim publisher abstract of the primary Tier-1 journal article; not the authors' own venue, but exact quoted primary text rather than paraphrase)","Tier 3-4 (Wikidata, bibliographic metadata only; Wikipedia, biographical)","Tier 2-3 (Frontiers in Psychology / PMC, peer-reviewed, named authors — definitional context on Kamin blocking)"]
---


**Topic question:** What does Griffiths et al. (Princeton CoCo Lab) argue about how expertise creates resistance to knowledge restructuring, and what mechanisms do they propose for "expediency effects"?

**Related concepts:** [[knowledge-restructuring]] · [[associative-blocking]] · [[Kamin-blocking]] · [[cue-competition]] · [[categorization-learning]] · [[cognitive-entrenchment]]

---

## Identifying the paper

**Claim type:** Historical/bibliographic → Tier 3–4 acceptable (uncontested); corroborated here at Tier 1–2.

The paper behind this hook is Lewandowsky, S., Kalish, M., & Griffiths, T. L. (2000), "Competing Strategies in Categorization: Expediency and Resistance to Knowledge Restructuring," *Journal of Experimental Psychology: Learning, Memory, and Cognition*, 26(6), 1666–1684, DOI 10.1037/0278-7393.26.6.1666, PMID 11185789. Griffiths is the third (junior) author, not the lead; Stephan Lewandowsky is first author. The paper is hosted today on Thomas Griffiths' Princeton Computational Cognitive Science ("CoCo") Lab site (https://cocosci.princeton.edu/tom/papers/restructuring.pdf — URL resolves, filename and running header confirmed as this paper), which is the origin of the "Princeton CoCo Lab" attribution in the harvested hook. Bibliographic details cross-confirmed via Wikidata (https://www.wikidata.org/wiki/Q51970056, Tier 3-4) and PubMed (Tier 2).

**Biographical context (Tier 3, Wikipedia, uncontested):** Griffiths received his undergraduate psychology degree from the University of Western Australia — the same institution where Stephan Lewandowsky was on faculty around 2000 — before later PhD work at Stanford (2005, under Joshua Tenenbaum) and his current position as Henry R. Luce Professor at Princeton, where he directs the CoCo Lab. This means the 2000 paper predates Griffiths' Princeton affiliation by roughly two decades; "Princeton CoCo Lab" identifies where the paper is archived and whose current lab claims it, not where the underlying study was conducted.

**Access note:** The primary PDF (Tier 1 by venue — Griffiths' own hosted copy) returns HTTP 200 and is confirmed as the correct document, but its content is a compressed/scanned rendering that did not yield extractable text through available tools in this session (consistent with the "PDF returned corrupt on fetch" note in the original harvest). Direct file download for local OCR was blocked by sandbox policy. Claims below therefore rely on the verbatim published abstract (mirrored via PubMed) rather than the full text; anything beyond the abstract is marked accordingly.

---

## Claim: Expertise in an accurate-but-imperfect ("expedient") strategy blocks later adoption of a superior strategy, unless the better structure is available from the start

**Claim type:** Specific technical-mechanism / result claim → floor Tier 1–2 required.

Across two experiments, participants first learned to rely on a single predictor that was fairly accurate but imperfect for a categorization task. At varying points in training, a diagram was introduced showing a more complex relationship between two *other* predictors — a relationship that could support perfect performance if adopted. The finding: participants who saw the complex-strategy diagram from the very beginning of training adopted it, but participants who had already learned to rely on the expedient (imperfect) predictor first were **unable to restructure their knowledge** even after being shown the same diagram. This "expedient knowledge" persisted even when the diagram was made adaptive (updated dynamically) in a second experiment.

**Sourcing (Tier 2 — verbatim publisher abstract via PubMed, https://pubmed.ncbi.nlm.nih.gov/11185789/, mirroring the Tier 1 journal article):**

> "In 2 experiments, people first learned to rely on a fairly accurate (but imperfect) predictor. At various points in training, a complex relationship between 2 other predictors was revealed in a schematic diagram that could support perfect performance. In Experiment 1, 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. In Experiment 2, expedient knowledge persisted even with an adaptive display."

**Claim type check:** Specific technical-mechanism (result pattern) → verified at Tier 2, corroborated by Tier 1 venue confirmation. ✓

---

## Claim: The proposed mechanism is associative blocking of potential alternative cues

**Claim type:** Specific technical-mechanism claim → floor Tier 1–2 required.

The authors' explanation for why expedient knowledge resists restructuring is framed in associative-learning terms, not as a purely strategic or metacognitive failure: having already formed a strong association between the expedient predictor and the outcome, participants' learning system treats the alternative (superior) cues as redundant and fails to strengthen associations with them — the classic **cue-competition** pattern from associative learning theory, applied here to human categorization/knowledge restructuring.

**Sourcing (Tier 2 — verbatim publisher abstract via PubMed):**

> "The persistence of expedient knowledge is explained by associative blocking of potential alternative cues."

Note on terminology: the paper's own vocabulary (per title and abstract) is "expedient predictor," "expedient knowledge," and "expediency" (in the title) — the exact phrase "expediency effect(s)" as a named technical term was **not confirmed** as appearing verbatim in the accessible text; it reads as a reasonable paraphrase carried into the harvested hook. `[flagged — could not verify exact phrase "expediency effect(s)" appears in primary text; full text inaccessible this session]`

**Definitional context on the underlying mechanism (Tier 2-3, Frontiers in Psychology / PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC4228836/):** "Associative blocking" as invoked here is the same family of phenomenon as Kamin blocking in classical conditioning, where "prior conditioning to an element might block conditioning to a new, superimposed element" (Boddez, Haesen, Baeyens & Beckers, 2014, *Frontiers in Psychology*). Blocking is one of several cue-competition effects in which a cue-outcome relationship is poorly learned or expressed because the cue was trained in the presence of an already-established alternative predictor of the same outcome.

**Claim type check:** Specific technical-mechanism → verified at Tier 2 for the direct claim; Tier 2-3 definitional support for the underlying associative-learning concept. ✓

---

## Claim: A 2005 follow-up (two of the three original authors) refines the account — restructuring requires both dissatisfaction with current knowledge AND an available usable alternative; neither condition alone suffices

**Claim type:** Specific technical-mechanism claim → floor Tier 1–2 required.

Kalish, Lewandowsky, and Davies (2005), "Error-Driven Knowledge Restructuring in Categorization," *Journal of Experimental Psychology: Learning, Memory, and Cognition*, 31(5), 846–861, PMID 16332161 — a direct follow-up sharing two of the three original authors — reconciles the "expertise blocks restructuring" finding with cases where restructuring *does* occur. Their proposed two-condition mechanism: restructuring requires (1) that the learner be dissatisfied with their current knowledge (i.e., their expedient strategy is producing measurable performance error) and (2) that a usable alternative strategy be pointed out to them. Critically, neither condition is sufficient by itself.

**Sourcing (Tier 2 — verbatim publisher abstract via PubMed, https://pubmed.ncbi.nlm.nih.gov/16332161/):**

> "The authors report 3 category learning experiments that reconciled those conflicting outcomes by postulating that, for restructuring to occur, learners must be dissatisfied with their knowledge and a usable alternative must be available. In line with expectation, restructuring was elicited only when an alternative strategy was pointed out and when people's initial expedient strategy entailed performance error. Neither error nor information about the alternative strategy by itself was sufficient to induce restructuring."

This directly bears on the original 2000 paper's finding: in Experiment 1, participants shown the complex-strategy diagram from the outset never had a chance to develop performance error with the expedient predictor, so no blocking had formed — while participants who learned the expedient predictor first had already reduced their error enough that dissatisfaction was low even after the diagram was revealed, leaving associative blocking intact.

**Claim type check:** Specific technical-mechanism → verified at Tier 2 (verbatim abstract of a Tier-1 peer-reviewed follow-up by overlapping authors). ✓

---

> [!note] Seek's commentary
> The mechanism on offer here is narrower and more mechanistic than "experts get set in their ways" — it's specifically an associative/cue-competition account: prior learning doesn't just create inertia, it actively prevents the new cues from acquiring predictive weight in the first place, because the old cue already "explains" the outcome. That has a sharp implication for the SeekVault self-rewiring question this was harvested for: if SeekVault's own claim-notes function like an "expedient predictor" — an existing schema that already explains most incoming information adequately — then simply *exposing* a better restructuring (a new taxonomy, a corrected causal story) may not be enough to get it adopted, per the 2005 follow-up. Two things would need to be true simultaneously: the existing schema needs to be visibly failing (dissatisfaction/error), and the better alternative needs to be pointed out concretely, not just made available. A gap-detection pass that surfaces contradictions or misses (dissatisfaction) paired with a concrete restructuring proposal (alternative) is closer to the mechanism than periodic passive review.

---

## Further leads

- **Full text of the 2000 paper is still needed.** The Griffiths-hosted PDF (Tier 1 venue) would not yield text via available tools this session (scanned/compressed rendering) and direct download for local OCR was blocked by sandbox policy. A future session with OCR tooling or an alternate clean copy (APA PsycNet, institutional repository) should retrieve the full Discussion section, which likely contains the formal associative/connectionist model (possibly Kruschke-style network model) behind "associative blocking," rather than just the abstract's summary phrase.
- **Kalish, Lewandowsky, & Kruschke (2001)** — a search snippet referenced a formal model by these authors "for explaining cue competition effects in function learning," potentially the connectionist/network implementation of the blocking account used in the 2000 paper. Not verified against a primary source this session — `[unverified-mechanism — needs primary]`.
- **Kalish (2005) "Error-Driven Knowledge Restructuring in Categorization" full text** (https://cogsci.syr.edu/wp-content/uploads/2021/03/KalishLD2005.pdf) was located but also would not yield extractable text via WebFetch this session; worth another attempt.
- **"Expertise reversal effect" (Kalyuga et al.)** and **"cognitive entrenchment"** (Dane, 2010, *Academy of Management Review*) are related but distinct literatures on expertise impairing flexibility, surfaced during this search — not the same paper or mechanism, but worth a comparative capture to distinguish associative-blocking accounts from cognitive-load/schema-rigidity accounts of expertise costs.