CAUSAL AND NONCAUSAL INTERPRETATIONS OF REGULARITY INFORMATION

Many models of causal judgement have been based on the principle that causal relations are inferred between two things which exhibit regularity of association. The idea that regularity is necessary for causal judgement is false, as shown by several findings that people use, and often prefer, cues and rules for causal inference other than regularity. The study reported here shows that regularity is not sufficient for causal judgement either. In several scenarios, subjects preferred non-causal interpretations of patterns of association which exhibited distinctive perfect positive covariation between two things. The patterns were interpreted as representing social conventions, signs or indicators, or independent effects of a hidden cause. Choice of interpretation as causal or non-causal varied with scenario content, in the absence of variation in the degree and pattern of covariation. Regularity-based models have not explained how people choose between causal and non-causal interpretations of regularity information. The role of regularity information in judgement of any kind is not well understood at present.