Abstract
While accuracy is commonly understood to be an important outcome of communicative exchanges research directed at its prediction and explanation has been pretheoretical. The literature relevant to accuracy prediction is updated beyond that provided by Mehrabian and Reed (1968) and the cause-effect relationships among predictors and criterion variables specified. The role of information-seeking is found to have special importance under three different types of tasks. Models of accuracy change for each task are specified, represented mathematically and simulated. Contrary to expectations the simulations show no long-run qualitative differences in accuracy for the three tasks. The implications for verbal versus mathematical modeling, situation, specificity, and dominate subsystems are discussed.