When Type 2 Processing Misfires: The Indiscriminate Use of Statistical Thinking about Reasoning Problems

Abstract

Research on dual-process theories of judgment makes abundant use of reasoning problems that present a conflict between Type 1 intuitive responses and Type 2 rule-based responses. However, in many of these reasoning tasks, there is no way to discriminate between the adequate and inadequate use of rules based on logical or probabilistic principles. To experimentally discriminate between the two, we developed a new set of problems: rule-inadequate versions of standard base-rate problems (where base rates are made irrelevant). Across four experiments, we observed conflict sensitivity (measured in terms of response latencies and response confidence) in responses to standard baserate problems but also in responses to rule-inadequate versions of these problems. This failure to discriminate between real and merely apparent (or spurious) conflict suggests that participants often misuse statistical information and draw conclusions based on irrelevant base rates. We conclude that inferring the sound use of statistical rules from normatively correct responses to standard conflict problems may be unwarranted when this kind of reasoning bias is not controlled for.