Stenning & van Lambalgen model Wason-task 'fallacies' as closed-world, non-monotonic reasoning in their own words

reasoning-psychologylogic-programmingnon-monotonic-logicclosed-world-assumptionwason-selection-taskcognitive-science

Two primary papers by Keith Stenning and Michiel van Lambalgen — the 2001 Journal of Logic, Language and Information case study and the 2004 Cognitive Science paper extending it — account for the "irrational" readings people give conditionals in Wason's selection task as products of closed-world reasoning rather than failed classical logic. The commitment strengthens between the two papers, and the note's claim is about that arc, not about a single settled statement.

In 2001 the proposal is offered as a hedged conjecture. Explaining why an isolated conditional ("If the switch is up, the light is on") is heard biconditionally, they 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." The doubled "might", and the "something like", are theirs; the word "non-monotonic" does not appear anywhere in that paper. By 2004 the same move is stated flatly and folded into the paper's formal apparatus: "The classical notion of validity may also give way to a non-monotonic notion of validity... one assumes (roughly speaking) that all statements are false which are not forced to be true by the premises." Crucially, the 2004 paper offers 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'" — so conversion, the move underlying affirming-the-consequent behaviour, is presented as a direct instance of closed-world inference, not merely something it resembles.

This confirms, from the primary, the characterization the vault previously held only at second hand through Sakama's paraphrase (claim-logic-programming-account-rivals-probabilistic-account-of-conditional-reasoning, claim-conditional-reasoning-fallacies-as-closed-world-inference-in-asp). It makes the closed-world account a genuine sibling of the rival Bayesian rehabilitation of the same data (claim-oaksford-chater-recast-wason-task-as-optimal-data-selection, claim-oaksford-chater-redefine-rationality-as-reasoning-about-uncertainty): both say the yardstick was wrong, not the reasoner, but they change different things — one the objective function, one the logic.