The relationship between conscious and unconscious influences: Independence or redundancy?

Abstract

Determining the relationship between conscious and unconscious influences is essential for obtaining valid estimates of the 2 types of influence. S. Joordens and P. M. Merikle (1993) recently argued that a redundancy relationship provides a plausible alternative to the independence model proposed by L. L. Jacoby, J. P. Toth, and A. P. Yonelinas (1993). In this article, the authors address Joordens and Merikle's concerns and still find the independence model preferable: First, the redundancy model requires the questionable assumption that a direct test (inclusion) is process pure. Second, results obtained with the independence model, but not with the redundancy model, converge with results from indirect tests. Finally, conclusions drawn from the independence model are in accordance with the theorizing that surrounds the concept of automaticity. Findings of task dissociations have been important for theorizing about both unconscious perception and unconscious influences of memory (implicit memory). People show evidence of perception and memory on indirect tests although they are seemingly unaware of the event that gave rise to those effects when given a direct test (e.g., Roediger & McDermott, 1993). However, Reingold and Merikle (1990), as well as we (Jacoby, Toth, & Yonelinas, 1993) have argued against the practice of equating unconscious influences with performance on indirect or implicit tests and conscious influences with performance on direct or explicit tests. The difficulty is that conscious processes might contaminate performance on indirect tests and, less obviously, unconscious processes might contaminate performance on direct tests. Consequently, we have sought process dissociations rather than task dissociations. That is, rather than equating processes with tasks, our strategy has been to gain estimates of the contributions of conscious and unconscious processes to performance of a single task and to show dissociations of effects of variables on those estimates.