Abstract
This article demonstrates that causal judgments predicted by asymptotic values of the Rescorla and Wagner (1972) associative model and the probabilistic joint model (Van Overwalle, 1996) are mathematically equivalent under the assumptions that (a) the frequency of all target causes and all their interactions is known, (b) target causes have an accompanying context factor that can also acquire causal strength, (c) attention is selectively focused on each factor and its context, and on each order of interactions of factors. Computer simulations indicate that given these assumptions, both the probabilistic model and the Rescorla-Wagner model reach approximately the same high fit with some observed data.