Remembering connectives: Visual-auditory consonance influences representations in time

Reasoning errors often arise when information is represented incorrectly. One example is the “conjunction effect”, a tendency to misremember OR statements as AND statements. We investigated the how the time course of consonance between visual and auditory information influenced memory for connectives. We measured memory for statements after manipulating the consonance between language and visual information. The results indicated that subjects correctly recalled more statements given language-visual consonance and were more likely to false alarm (e.g., incorrectly recalling OR statements as AND) given language-visual dissonance. We modeled performance using a Simple Recurrent Network. Our model, in which the training set was structured similarly to natural language input, provided a reasonable analog. Taken together, the results suggest that connective representations are influenced by the concordance between visual and language input and that the bias toward conjunctions arises, in part, from the relatively high frequency of conjunctions in visual and linguistic input.