Social Cognition and the So-Called Conjunction Fallacy

The so-called ‘conjunction effect’, in which participants incorrectly assert that an instance from the conjunction of two sets is more probable than an instance from one of the two conjoining sets alone, has been a source of debate as to whether it is a genuine fallacy of individual thinking or not. We argue that reasoning about individuals follows a different process than reasoning about sets. 35 participants took part in 3 tasks: a) one involving blocks of different sizes and colours designed to evoke set-based reasoning, b) one where a particular block was ‘individuated’ by stating that it represented a particular person, and c) the original Tversky and Kahneman (Psychological Review 90(4):293–315, 1983) ‘Linda’ problem. As predicted, set-based reasoning was significantly more prevalent for the blocks task than for the other two tasks. Participants’ reasons for their choices suggest that some individuals correctly use set-based logic in one task and a social reasoning process for the other tasks.