This paradigm is intuitively simpler and relies on a well-known paradigm (i.e., the same–different task; e.g., Bamber, 1969; Krueger, 1978; Thomas,1996). Its simplicity should make it easier to determine whether errors are due to the categorization process, or to processes unique to the goal-directed task. In addition, we used the randomization technique from Ashby and Gott (1988) to generate a large number of stimuli, and the possibility of indirectly learning rule-based and information-integration category structures is explored. Second,because the category epresentations that are learned directly and indirectly differ, a follow-up question is whether the representations learned in one context can be used in another context. Specifically, can a category representation learned indirectly be used in a direct classification
task (and vice versa)? This paper is an initial attempt at answering both of these questions using a new indirect categorization task, the same–different categorization task.