Do Stenning & van Lambalgen actually model the Wason-task 'fallacies' as closed-world reasoning with negation-as-failure?

reasoning-psychologylogic-programmingnon-monotonic-logicclosed-world-assumptionnegation-as-failurewason-selection-taskcognitive-science

This capture resolves question-verify-stenning-van-lambalgen-closed-world-wason, which flagged that the vault's existing claim (claim-logic-programming-account-rivals-probabilistic-account-of-conditional-reasoning) rested on a secondary paraphrase (Sakama's citation of "Stenning and Lambalgen (2008)") rather than a primary read. Two primary Stenning & van Lambalgen papers were read directly: the 2001 Journal of Logic, Language and Information paper and the 2004 Cognitive Science paper that extends it (both predate the 2008 MIT Press book the original question named, but are the same authors' own account of the same task, developed across the same argument). The book itself and the 2005 Cognitive Science "suppression task" paper (which appears to be where the fullest non-monotonic-logic formalization sits) were not obtained this run -- see Further leads.

Claim: Stenning & van Lambalgen do explicitly model "irrational" Wason-task readings as closed-world, non-monotonic reasoning, in their own words, across both papers

Claim type: technical mechanism. Floor: Tier 1-2. Met: both sources are Tier 1 primary, quoted directly.

In the 2001 paper, discussing why an isolated conditional like "if the switch is up, the light is on" is often read biconditionally, Stenning & van Lambalgen write: "It might be that something like closed-world assumption reasoning might operate to generate this interpretation in experimental conditions. The very fact than no other rule is known might generate the inference that this is the only explanation." In the 2004 paper this becomes a formal part of their apparatus: "The classical notion of validity may also give way to a non-monotonic notion of validity... One concrete instance of this is so-called 'closed world reasoning', in which one assumes (roughly speaking) that all statements are false which are not forced to be true by the premises." They give the "fallacy" itself as the worked example: "One example of such closed world reasoning is the often observed conversion of the conditional: 'if A then B' implies 'if B then A'" -- i.e., affirming-the-consequent-style behavior is offered as a direct instance of closed-world inference, not merely something closed-world reasoning happens to resemble.

Claim: In these two primary papers, the authors' own term is "closed-world reasoning" / non-monotonic logic -- the specific AI/logic-programming phrase "negation as failure" does not appear in either text

Claim type: technical mechanism (a negative/scope-limiting finding). Floor: Tier 1-2. Met: absence checked directly against both full primary texts.

Both papers were read in full (2001: 45 pages; 2004: 49 pages, including the reference list). "Closed-world reasoning" and "non-monotonic logic" appear repeatedly and are load-bearing terms; the literal phrase "negation as failure" does not occur in either. The 2004 paper's reference list also does not cite Clark (1978) or Reiter (1978/1980) -- the two founding papers usually cited for negation-as-failure and the closed-world assumption in AI/logic programming (compare claim-conditional-reasoning-fallacies-as-closed-world-inference-in-asp, where Sakama cites Reiter directly for exactly this lineage). This means the vault's existing citation of "negation-as-failure" as Stenning & van Lambalgen's own framing is not confirmed by these two primaries: it is Sakama's 2024 translation of their account into ASP/logic-programming vocabulary, layered on top of an account these two S&vL papers state in the vocabulary of non-monotonic logic and closed-world reasoning rather than the specific negation-as-failure operator. Whether the term appears in the fuller, later development (the 2005 suppression-task paper or the 2008 book) is [unverified-mechanism -- needs primary] and is the open half of the original question.

Claim: The authors' own two-step framework is "reasoning for an interpretation" and "reasoning from an interpretation" -- not "reasoning to an interpretation, then from it" as the vault's inherited paraphrase has it

Claim type: technical mechanism / definitional. Floor: Tier 1-2 (technical mechanism; met directly).

The 2004 paper states the distinction explicitly: "we are thus led to the important distinction between reasoning from an interpretation. and reasoning for an interpretation." The text glosses "the former" (reasoning from an interpretation) as "what is supposed to happen in a typical inference task: given premises, determine whether a given conclusion follows" -- i.e. derivation once a logical form is fixed -- with the latter, "reasoning for an interpretation," being the (often unacknowledged) prior work of settling which logical form applies. claim-logic-programming-account-rivals-probabilistic-account-of-conditional-reasoning currently states this as "reasoning first to an interpretation and then from it" (attributing the "to... from" wording to Stenning & van Lambalgen's own frame) -- the primary's actual preposition is "for," not "to." A small wording point, but the note's [unverified-claim] flag specifically named getting this phrase right as part of what a primary read needed to confirm.

Claim: the concrete non-monotonic mechanism offered for exception-tolerant ("fallacious" by classical-logic lights) conditional reasoning is a two-clause exception scheme, not a single material conditional

Claim type: technical mechanism. Floor: Tier 1-2. Met: quoted directly from the 2004 primary.

For subjects who treat the Wason rule as admitting exceptions rather than being falsified by a single counterexample, Stenning & van Lambalgen propose the pair "1. p ∧ ¬e → q" and "2. p ∧ ¬q → e", where e is a proposition letter for "exception." They gloss this as: "Condition 1 then says that the rule applies only to nonexceptional cards." This is structurally a negation-as-failure move even where the label isn't used: e is assumed false (no exception applies) unless the data forces it to be believed true, which is exactly the "believe false what cannot be proved" pattern the vault's ASP-formalization note (claim-conditional-reasoning-fallacies-as-closed-world-inference-in-asp) describes via Sakama's independent, later (2024) formalization of the same family of "fallacies." The two accounts converge on the same underlying move without the earlier authors naming it as such in these texts.

Further leads

• Stenning & van Lambalgen, "Semantic Interpretation as Computation in Nonmonotonic Logic: The Real Meaning of the Suppression Task," Cognitive Science (2005) — cited by the 2004 paper as "submitted" at the time; likely the fuller formalization, and the most probable primary home for an explicit "negation as failure" statement. Abstract only seen (Wiley, paywalled): https://onlinelibrary.wiley.com/doi/abs/10.1207/s15516709cog0000_36 — not fetched this run.
• Stenning & van Lambalgen, Human Reasoning and Cognitive Science (MIT Press, 2008) — the book the original routed question named; still unread in primary. A Studia Logica review exists at https://link.springer.com/content/pdf/10.1007/s11225-011-9309-3.pdf (Springer, unread, secondary).
• Robert Kowalski, Computational Logic and Human Thinking (2011) — the second name Sakama cites alongside Stenning & van Lambalgen for the logic-programming lineage; the routed question named this as a cross-check target and it remains unread.
• The 2001 paper's only available copy this run resolves through a Utrecht University (OZSL) archive rather than either author's own site or the publisher (Kluwer/Springer) — worth checking for an author-hosted copy on Stenning's or van Lambalgen's own pages before treating the OZSL host as the venue of record.

Entity candidates

• Peter Wason — person — creator of the selection task itself; the empirical phenomenon both this closed-world account and the rival Oaksford–Chater Bayesian account (claim-oaksford-chater-recast-wason-task-as-optimal-data-selection, entity-mike-oaksford, entity-nick-chater) are competing to explain. The foundational figure both later formalisms measure themselves against.
• Raymond Reiter — person — originator of default logic (1980) and the closed-world assumption for incomplete databases (1978), the AI lineage that the vault's Sakama-based note (claim-conditional-reasoning-fallacies-as-closed-world-inference-in-asp) traces "closed-world reasoning" to; notably, Stenning & van Lambalgen's own 2001/2004 papers do not cite Reiter directly, developing "closed-world reasoning" through a different (linguistics/philosophy-of-language) route — worth a note on the two independent routes into the same terminology.
• Keith Stenning — person — co-author; Human Communication Research Centre, University of Edinburgh.
• Michiel van Lambalgen — person — co-author; ILLC, University of Amsterdam (now Professor of Logic and Cognitive Science there).
• Robert Kowalski — person — logic-programming founder cited by Sakama alongside Stenning & van Lambalgen as a parallel non-Bayesian account of conditional reasoning; unread primary, flagged as a further lead above.
• closed-world assumption — concept — the specific non-monotonic-logic mechanism this whole capture turns on; already load-bearing for two existing vault notes.
• negation-as-failure — concept — the AI/logic-programming operationalization of closed-world reasoning; this capture's finding is that it is Sakama's term for the account, not literally Stenning & van Lambalgen's own term in the two papers read.