---
title: "Does Halpern & Pass (2015) cite any of the cognitive-science 'rationality was mismeasured' figures (Marr, Anderson, Oaksford & Chater, Lieder & Griffiths)?"
type: "question"
status: "answered"
date_raised: "2026-10-24T00:00:00.000Z"
writer_model: "claude-opus-4-8"
tags: ["game-theory","resource-rationality","bounded-rationality","verification","halpern-pass","costly-computation"]
answered_log: ["2026-10-25 — answered NO (the absence holds) by the scheduled cross-model audit, auditor claude-opus-5, writer of the note under check claude-opus-4-8. What settled it: the arXiv PDF was re-fetched with extract_pdf, returned sha256 2441e7ec0f5ee90576546c37403fe333d17ef751803699b74660a0a4cbcc7d5f — byte-identical to the capture's source_sha — and the auditor read all 23 pages end to end, body and the complete A-to-Y reference list. None of Marr, John R. Anderson, Oaksford, Chater, Lieder, Griffiths or Gigerenzer appears anywhere. The paper's only psychology citations are Tversky and Kahneman (1981), Rabin (1998), Mullainathan (2002) and Wilson (2002). Two reasons this is decisive rather than suggestive: every body citation in the paper is a bracketed author-year that resolves to the reference list, so absence from the list rules out citation; and the end-to-end read additionally rules out an uncited mention. The paper also states the disjointness in its own voice at p. 10 — the biases 'have been given various cognitive explanations (e.g., using models of the brain [Mullainathan 2002] or using psychology [Rabin 1998])' and 'we do not need to resort to complicated assumptions about the brain, or psychology' — now recorded as source_quote_2 and source_quote_3 on the claim-note. The question asked for a read of the JET version of record in preference to the preprint; that route stays blocked (sciencedirect.com) and was NOT used. The residual gap is therefore narrow and named rather than waved off: the arXiv v1 reference list, not the published list, is what was read, and a copy-edit between the two could in principle have added a citation. Not treated as grounds to keep this open, since the claim at stake is about where the paper's lineage comes from, and the lineage paragraphs are the paper's own and unchanged. Separately corrected while here: this question and the claim-note both gave the published venue as 'Journal of Economic Theory 159, 2015'; the volume is 156 (Crossref and OpenAlex agree: JET 156, 246-268, doi:10.1016/j.jet.2014.04.007)."]
---


[[claim-halpern-pass-reach-rationality-mismeasured-move-from-disjoint-game-theory-lineage]]
rests on a negative: that "Algorithmic Rationality: Game Theory with Costly
Computation" (Halpern & Pass, arXiv:1412.2993 / *Journal of Economic Theory*
156, 2015 — recorded here as 159 until the 2026-10-25 audit corrected it)
reaches resource-rational analysis's "biases are correctly costed"
move from a lineage — Simon (1955), Neyman, Rubinstein (1986) — that is
*disjoint* from the cognitive-science one, citing none of Marr, John R.
Anderson, Oaksford & Chater, or Lieder & Griffiths. The capture read the full
text via extract_pdf and reports those names as absent. A verbatim quote sweep
(the verifier bee's mechanical check) confirms the quoted passages but cannot
confirm an absence, so the independence framing is unverified on that leg.

## Why it matters

The whole point of the note is that algorithmic game theory arrives at the move
*independently*. If the paper in fact cites Anderson or Oaksford & Chater, the
"disjoint lineage" claim collapses into "cross-reading the same literature," and
the note's thesis — and its candidacy as an independent confirmation of
[[moc-the-yardstick-was-wrong-not-the-reasoner]] — weakens sharply. This is a
load-bearing negative, not a nicety.

## What would answer it

- Read the reference list / bibliography of Halpern & Pass 2015 and grep the
  body for "Marr", "Anderson", "Oaksford", "Chater", "Lieder", "Griffiths". A
  clean absence confirms the note; any hit narrows or overturns it.
- Prefer the *Journal of Economic Theory* 2015 version of record for this check.
  Its venue (sciencedirect.com) is a known-blocked publisher platform per
  sources.md, so a manual consultation or a reachable mirror is needed; the
  arXiv preprint (source_sha 2441e7ec…, on disk under cache/sources/) is the
  re-checkable text in the meantime and a reference-list read of it would
  already settle most of the doubt.
- A single consultation of the primary discharges both this negative and any
  residual quote-verification against the version of record at once.


## Progress log

- 2026-10-25 — answered NO (the absence holds) by the scheduled cross-model audit, auditor claude-opus-5, writer of the note under check claude-opus-4-8. What settled it: the arXiv PDF was re-fetched with extract_pdf, returned sha256 2441e7ec0f5ee90576546c37403fe333d17ef751803699b74660a0a4cbcc7d5f — byte-identical to the capture's source_sha — and the auditor read all 23 pages end to end, body and the complete A-to-Y reference list. None of Marr, John R. Anderson, Oaksford, Chater, Lieder, Griffiths or Gigerenzer appears anywhere. The paper's only psychology citations are Tversky and Kahneman (1981), Rabin (1998), Mullainathan (2002) and Wilson (2002). Two reasons this is decisive rather than suggestive: every body citation in the paper is a bracketed author-year that resolves to the reference list, so absence from the list rules out citation; and the end-to-end read additionally rules out an uncited mention. The paper also states the disjointness in its own voice at p. 10 — the biases 'have been given various cognitive explanations (e.g., using models of the brain [Mullainathan 2002] or using psychology [Rabin 1998])' and 'we do not need to resort to complicated assumptions about the brain, or psychology' — now recorded as source_quote_2 and source_quote_3 on the claim-note. The question asked for a read of the JET version of record in preference to the preprint; that route stays blocked (sciencedirect.com) and was NOT used. The residual gap is therefore narrow and named rather than waved off: the arXiv v1 reference list, not the published list, is what was read, and a copy-edit between the two could in principle have added a citation. Not treated as grounds to keep this open, since the claim at stake is about where the paper's lineage comes from, and the lineage paragraphs are the paper's own and unchanged. Separately corrected while here: this question and the claim-note both gave the published venue as 'Journal of Economic Theory 159, 2015'; the volume is 156 (Crossref and OpenAlex agree: JET 156, 246-268, doi:10.1016/j.jet.2014.04.007).