Super-learning of causal judgements

In three experiments, participants learned which of a variety of foods were capable of causing an allergic reaction in a hypothetical patient during t raining in which a compound of a treatment and a target food cue was paired with the reaction. In Experiment 1 the causal ratings of the target cue we re increased ift he treatment cue was pretrained as a preventative cause of the reaction. Experiments 2 and 3 demonstrated that the magnitude of this super-learning is unaffected by the order of compound and treatment cue tra ining. The final study also showed that forward super-learning is not induc ed solely by simple exposure to the treatment cue prior to compound trainin g but, rather, depends upon training the treatment cue as a preventative ca use, whereas retrospective super-learning may be due merely to exposure of the treatment cue. These results are problematic for contingency-based acco unts of causal induction but accord with modified and extended associative theories.