Identifying individual differences: An algorithm with application to Phineas Gage

In many research contexts it is useful to group experimental subjects into behavioral “types.” Usually, this is done by pre-specifying a set of candidate decision-making heuristics and assigning each subject to a heuristic in that set. Such approaches might perform poorly when applied to subjects with prefrontal cortex damage, because it can be hard to know what cognitive heuristics such subjects might use. We suggest that the Houser, Keane and McCabe (HKM) robust classification algorithm can be a useful tool in these cases. An important advantage of this classification approach is that it does not require one to specify either the nature or number of subjects' heuristics in advance. Rather, both the number and nature of the heuristics are discerned directly from the data. To illustrate the HKM approach, we draw inferences about heuristics used by subjects in the well-known gambling task [Bechara, A., Damasio, A.R., Damasio, H., Anderson, S., 1994. Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 50, 7-12].