Proceedings of the Annual Meeting of the Cognitive Science Society

When descriptive sentences are linked by disjunctive logical connectives (or, xor), they may be true for multiple possible world configurations, unlike sentences linked by the conjunctive which reference a single possibility. Moreover, some of the component propositions may be false in some of the possible world configurations, while the compound sentence is still true. Determining the truth values of such sentences requires, therefore, dealing with propositions that cannot be grounded because they are false and evaluating sentences relative to multiple possible outcomes of the grounding process. Building on Dynamic Field Theory (DFT), we propose a neural process account that (1) represents affirmative and negative relational propositions linked by logical connectives, (2) autonomously drives perceptual grounding via sequential attention shifts and evaluations, and (3) determines the truth values of each component proposition and of the compound sentence by matching grounding outcomes to the possibility representations implied by each connective. The model reproduces the negation-by-truth-value interaction effect at the proposition level and extends it to the compound level.