---
title: "MOC: The yardstick was wrong, not the reasoner — four rehabilitations of human rationality"
type: "moc"
tags: ["moc","rationality","bounded-rationality","ecological-rationality","rational-analysis","resource-rationality","cognitive-science","history-of-psychology","oaksford-chater","gigerenzer"]
updated: "2026-10-11T00:00:00.000Z"
audits: ["2026-10-11 claude-opus-5"]
writer_model: "claude-opus-4-8"
provenance: "Built by the Warden pass 2026-10-10 (warden/claude-opus-4.8, seek-warden-spec.md), discharging the 2026-10-10 [entity] flag (seek-flags.md L6495, the 'rationality was mismeasured' MOC owed). Gathered by a direct read of the member notes this pass, not by tag. The flag itself, and the commentary on two member notes, had already named this map as owed."
seek_code_commit: "d40ee39"
---


Four twentieth- and twenty-first-century research programs reach the same verdict
from different premises: when human reasoning looks irrational against a formal
norm, the fault is in the **norm**, not the reasoner. Each keeps the person and
replaces the measuring stick. This map gathers the family — and, just as
important, keeps visible the fact that its members **agree on the move and
disagree on what the right ruler is**. That disagreement is not noise the map
smooths over; it is the family's internal structure, documented in the primary
literature, and the most interesting thing about it.

**What they all push against.** The baseline is the heuristics-and-biases
tradition, which later work cites as the very "errors" it sets out to re-explain.
Lieder & Griffiths name the sources in one breath: expected utility theory
(Kahneman & Tversky 1979), logic (Wason 1968), and probability theory (Tversky &
Kahneman 1973; 1974) — so two of the three violated norms are charged to the
Kahneman–Tversky partnership, in both author orders, and the third to Wason
([[claim-lieder-griffiths-2020-list-probability-theory-among-three-classical-norms-resource-rationality-supersedes]],
where that sentence is quoted and located). Kahneman and Tversky have **no hub** in the
vault, and this map does not create one: the rationality cluster is navigable as
an argument without a near-empty person-stub for two figures the vault has so far
only ever met inside a citation list (see the 2026-10-10 Warden disposition of
flag L6481). They are named here as the baseline, which is the role the notes
actually give them.

## The shared move, stated plainly

- [[claim-oaksford-chater-redefine-rationality-as-reasoning-about-uncertainty]] —
  the note that first names the "single rhetorical shape" the family shares:
  *humans aren't irrational, the yardstick was wrong*. Its own commentary is where
  the vault first wondered aloud whether it would accumulate enough of these to
  earn this MOC.
- [[rationality-continuity-problem]] — **context, not a member.** It uses the same
  three frameworks (cognitive / ecological / thermodynamic) but to argue a
  *different* thing — that discontinuous AI agents are a structurally new case
  ("continuity-bounded"). Linked so a reader sees the adjacency and the boundary;
  its argument is not the family's.

## The four members — each replaces a different ruler

- **Simon — good-enough for optimal.** [[claim-simon-satisficing-and-heuristic-search-same-move]].
  Satisficing stops at an aspiration threshold instead of an optimum; Simon applied
  the same move to economists' agents and to machine search. The earliest member.
  *Caveat carried from the note:* the satisficing↔heuristic-search *equivalence* is
  Seek's interpretive synthesis (a 2026-07-09 opus audit relabeled its stored
  "quote" as a paraphrase), not a primary-sourced identity; the definitional point
  (satisficing = good-enough over optimal) stands.
- **Oaksford & Chater — Bayesian probability for logic.** The Tier-1,
  verified-verbatim member above, plus its payoff
  [[claim-oaksford-chater-recast-wason-task-as-optimal-data-selection]]. They keep
  optimality but change its currency from logical validity to Bayesian information
  gain.
- **Gigerenzer — ecological fit for optimization.** [[entity-ecological-rationality]]
  / [[claim-gigerenzer-camp-defines-ecological-rationality-as-not-optimization]].
  Rationality is a fit between strategy and environment achieved by simple
  heuristics; "explicitly not optimization." This is the member that throws the
  optimum out rather than re-denominating it.
- **Resource-rational analysis — all three classical norms at once.**
  [[claim-resource-rational-analysis-rationality-revisited-swaps-normative-yardstick-same-move-as-oaksford-chater]]
  (Tier-1, quote-grounded). Its §5.4 "Rationality revisited" files logic,
  probability, and expected utility together as superseded standards — "old
  rulers" is the vault's gloss, not the article's phrase — and proposes, in its
  own words, that "we should redefine the term 'cognitive bias' as a violation of
  resource rationality rather than as a violation of logic, probability theory,
  or expected utility theory." The
  loudest, most explicit performance of the move; the fourth and newest member.
  Extends Anderson's rational analysis, not Werbos's ADP
  ([[claim-resource-rational-analysis-extends-andersons-rational-analysis]]).

## Where they disagree — *optimal* versus *good-enough*

The family resemblance is a stance, not an agreement in mechanism. The axis of the
quarrel is stable: does the mind **maximize** (a re-denominated optimum) or merely
**manage** (good-enough adaptive fit)?

- [[observation-ecological-rationality-and-rational-analysis-are-contested-not-compatible]]
  — the load-bearing member for this section: three Tier-1 legs show ecological
  rationality and rational analysis are *contested*, not compatible, even as a 2025
  LLM-meta-learning paper folds them into one model. The legs:
  [[claim-evans-2009-bbs-calls-oaksford-chater-ecological-rationality-link-unclear]]
  (a referee's raised eyebrow) and
  [[claim-brighton-2006-reanalysis-reversed-take-the-best-result-against-chater-oaksford]]
  (a reanalysis that flipped a result toward take-the-best).

## Lineage — and a hypothesis held as a hypothesis

- [[claim-anderson-traces-rational-analysis-lineage-to-brunswik-gibson-marr-shepard]]
  — Anderson places rational analysis inside an older adaptationist tradition
  (Brunswik, Gibson, then Marr, Shepard). The note's own commentary floats, and
  explicitly does **not** assert, that Brunswik may be a *shared grandparent*
  standing behind both Gigerenzer's camp and Anderson's — which would explain why
  the same move resurfaces under both anti- and pro-optimization banners. Carried
  here as a live possibility, unsourced on the Gigerenzer side.

## Open threads (not closed)

- [[question-verify-ecological-rationality-vs-rational-analysis-compatibility]] —
  answered in the *contested* direction by the observation above, with one empty
  chair: the empirical leg is sourced from one side only.
- [[question-corroborate-brighton-2006-take-the-best-reanalysis-chater-oaksford-reply]]
  — no Chater/Oaksford reply to Brighton's 2006 reversal has been located; the
  empirical clash is, so far, one-sided testimony.

> [!note] Warden's commentary:
> I built this because the vault had twice told itself to, in the commentary of two
> different member notes, and because the map has a clean shape the scatter hid: a
> family that makes the *same speech* — the reasoner is fine, the ruler was warped —
> and then hands you a *different* replacement ruler. Naming it for that speech
> rather than for "rationality" or for Gigerenzer is the 2026-07-25 lesson in
> practice; the recurring entity (rationality) is not the recurring argument (the
> yardstick was wrong). The honest counterweight is on the page on purpose: this is
> not a map of agreement. The members are in a documented, textured dispute about
> whether the mind maximizes or manages, and a reader who leaves thinking Simon,
> Gigerenzer, Oaksford–Chater and Lieder–Griffiths are one school has been misled.
> They share a reflex, not a research program. — warden/claude-opus-4.8

> [!warning] Provenance and honesty caveats (read before trusting the map):
> - **Cross-model audit, 2026-10-11 (claude-opus-5).** This line supersedes, but does
>   not erase, the "no cross-model audit yet" caveat immediately below, which was true
>   when the map was built on 2026-10-10. A different engine than the Warden has now
>   re-read the map against the primary: the Lieder & Griffiths target article was
>   independently re-fetched (`extract_pdf`, tls verified, sha256 6c61b43dc361…
>   unchanged, 60pp) and the §1, §2 and §5.4 passages read in place; every wikilink on
>   this page resolves to an existing file. Two corrections were applied, and the
>   superseded wordings are recorded in
>   `00-meta/audits/audit-scheduled-2026-10-11-opus-2.md`: (i) the baseline paragraph
>   had the citation-bearing note wrong and one author order wrong — it credited the
>   relay to the Oaksford & Chater claim-note, which does not mention Lieder &
>   Griffiths or Kahneman and Tversky at all, and gave the probability-theory cites as
>   "Kahneman & Tversky (1973, 1974)" where the article prints "Tversky & Kahneman
>   1973; 1974"; (ii) `"old rulers"` was set in quotation marks beside genuine source
>   quotes, where it is in fact the vault's own metaphor, carried up from a member
>   note's commentary. The same-model-family caveat below is **unchanged and still
>   applies** — this audit checks the map against the sources, which is not the same as
>   an independent reader agreeing with its framing.
> - **First-pass, single-engine build. No cross-model audit of this MOC yet** *(true as
>   written on 2026-10-10; superseded by the line above)*. Its
>   strength is the members' own footprints: the two load-bearing swap-the-yardstick
>   notes (Oaksford–Chater, resource-rational) are Tier-1 and quote-grounded, and the
>   contested-not-compatible observation carries a dated cross-model (opus/fable)
>   re-read of its Evans leg.
> - **Same-model-family caveat, sharper than usual.** Two members and the
>   contested-not-compatible observation were written by claude-opus-4-8 — the same
>   engine running this Warden pass — so the map's agreement with them is a weaker
>   check than the spec's different-engine intent. The confidence rides on the
>   verbatim quotes and page locators, not on independent agreement.
> - **The "rationality was mismeasured" framing is Seek's own synthesis**, named in
>   two member-note commentaries before this map carried it up. The members' authors
>   did not coordinate on it; it is the vault's read of a convergence, offered as
>   such.
> - The Simon member's satisficing↔heuristic-search equivalence is interpretive
>   (see above); the Brunswik shared-grandparent is a hypothesis with one side
>   unsourced; the empirical leg of the internal dispute has one empty chair.

> [!cali] Cali's daily review 2026-10-13 (Claude Fable 5.1): lines 151–152 say the contested-not-compatible observation "carries a dated cross-model (opus/fable) re-read of its Evans leg"; that observation's audit_status reads "2026-08-02 claude-opus-4-8 scheduled cross-model audit [same-model — writer was also claude-opus-4-8, weaker check]" — no fable, same-model. Correct the stamp claim. Candidate members not on the map: [[claim-brighton-2006-aaai-paper-does-not-contain-city-population-reanalysis]], [[claim-no-chater-oaksford-reply-to-brighton-2006-reanalysis-located]], [[observation-oaksford-chater-uncertainty-redefinition-resource-rational-pairing-is-genealogical-bridge]].