Consistent and varied training in the theory of automatic and controlled information processing

Abstract

In their theory of automatic and controlled information processing, Schneider and Shiffrin (1977; Shiffrin & Schneider, 1977) assign a central role to the distinction between consistent and varied stimulus-response mapping. The present note shows that this distinction cannot be made without specification of what is to count as a unit in describing stimulus events. In any rule-governed task, including what Schneider and Shiffrin call varied-mapping search, there is a description of stimulus events which is consistently mapped on to responses. It follows that their results are concerned not with the importance of consistency per se, but with a specification of the important level of stimulus description for one particular task (visual search). Their results show directly that not all forms of consistent mapping lead to automatic processing. Attempts to generalize their work should be concerned with the issues of which forms of consistency are effective.