Proceedings of the Annual Meeting of the Cognitive Science Society

People rely on mechanism knowledge when making causal inferences that involve multiple causal variables. In particular,mechanism knowledge can influence whether people use linear or alternative integration rules to predict how multiplecauses will interact to produce an effect. We examine whether general beliefs about mechanism types whether two causesoperate by the same or different mechanisms might mediate such inferences. Experiment 1 demonstrates that when acausal interaction yields non-linear positive effects, people are more likely to infer that the two causes work via differentmechanisms. Experiment 2 investigates the converse of this inference, showing that people also predict non-linear positiveinteractions more often when they know that two causes have different mechanisms.