Deduction runs on iconic mental models, not formal rule application

Correct: B — Constructing iconic mental representations of possibilities consistent with the premises and inspecting them for what must be true

Johnson-Laird (2010) reviewed 30+ years of evidence that ordinary deductive reasoning is supported not by formal-rule application but by the construction of small mental models — iconic, structure-preserving representations of the situations the premises permit. Reasoners build a 'preferred' model first, often miss alternative models, and mistake what is true in their preferred model for what must be true. This predicts systematic errors (illusory inferences, belief bias) that rule-based theories cannot easily explain. Modern variants integrate probability — possibilities can be weighted — and the framework remains a leading account of human deduction and counterfactual thinking.