Preferences for simple explanations in diagnosis and prediction

People sometimes prefer a simple explanation, such as a common-cause diagnosis that two symptoms are caused by one disease, rather than a more complex separate-causes diagnosis, that they are caused by two different diseases. We tested whether they also prefer a simple common-effect prediction, that two health conditions lead to one symptom, rather than a more complex separate-effects prediction that the two conditions lead to two different symptoms. Three experiments showed that participants preferred common-effect predictions to separate-effects ones, just as they preferred common-cause diagnoses to separate-causes ones; however, their preference for common-effect predictions was significantly attenuated compared to their preference for common-cause diagnoses. The preference for a simple common-effect explanation was eliminated when frequency information indicated that the two separate effects occurred far more often than the common effect, just as for common-cause explanations. The preferences occurred in a medical domain about diseases and symptoms, for judgments of satisfactory diagnoses and predictions (Experiment 1) and judgments of satisfactory explanations (Experiment 2); they also occurred for a more complex river pollution domain about bacteria and toxins (Experiment 3). The experiments indicate that the preference for simple explanations arises because of the ease of simulating causal relations in common-cause and common-effect models.