Recent findings of dissociations between direct and indirect tests of memory and perception have renewed enthusiasm for the study of unconscious processing. The authors argue that such findings are heir to the same problems of interpretation as are earlier evidence of unconscious influences-namely, one cannot eliminate the possibility that conscious processes contaminated the measure of unconscious processes. To solve this problem, the authors define unconscious influences in terms of lack of

more »

... nscious control and then describe a process dissociation procedure that yields separate quantitative estimates of the concurrent contributions of unconscious and consciously controlled processing to task performance. This technique allows one to go beyond demonstrating the existence of unconscious processes to examine factors that determine their magnitude. A layperson might ask, "Can techniques that rely on unconscious processes be used to make me act in ways that are counter to my own purposes?" The layperson's question about unconscious processes has little to do with problems of definition, thresholds, or experimental design-problems that have occupied psychologists. We side with the layperson by treating the question of unconscious influences as a question of control over thought and behavior. Indeed, we rely on demonstrations of effects that are counter to a person's conscious intent as a methodological tool for revealing unconscious influences. Perhaps discomforting for the layperson, our research, along with research done by others, indicates that unconscious influences are very common. People sometimes consciously plan and then act, but more often behavior is influenced by unconscious processes; that is, people act and then, if called upon, make their excuses. As noted by Greenwald (1992, this issue), academic psychologists have held a skeptical view of psychoanalytic conceptions of unconscious cognition. Recent investigations of unconscious processes have been grounded in more "respectable" areas, such as neuropsychology, and in cognitive-oriented theories of memory and perception. The result has been the acceptance of a cognitive unconscious that differs in important ways from the psychoanalytic unconscious (Kihlstrom, 1987) . However, we argue that the procedures used to gain evidence of the cognitive unconscious share important similarities (and weaknesses) with those used to gain evidence of the psy-choanalytic unconscious. In this article, we describe an approach to the study of unconscious processes that centers on the question of control. Our goals here are to (a) review experiments that demonstrate the existence of (cognitive) unconscious influences, (b) highlight the importance of subjective experience in the control of thought and behavior, and (c) describe a technique that allows us to separately estimate the contributions of unconscious and consciously controlled processing to task performance. Our technique is based on the distinction between automatic and controlled processing; it allows one to go beyond demonstrating the existence of unconscious processes to examine factors that differentially influence the two forms of processing. We begin with a selective review of findings that have generated renewed interest in unconscious processes. Task Dissociations Current approaches to the study of unconscious processes have a number of similarities with older, psychoanalytic methods used to investigate the unconscious. Consider, for example, projective tests of personality, such as the Rorschach (1921 Rorschach ( /1981 . The rationale for using the Rorschach is that it is thought to reveal unconscious needs, motivations, and expectancies that would not be revealed by self-report measures. A similar logic underlies recent interest in indirect tests of memory and perception. On an indirect test, subjects are not instructed to report on a past or present event, but rather engage in some task that can indirectly reveal the influence of memory or perception of the event. In contrast, a direct test, like a selfreport measure, asks the subject to consciously recollect or identify the event in question. Dissociations between direct and indirect tests of memory and perception are analogous to dissociations between self-report measures and projective tests of personality. In both cases, the pattern of results is interpreted as showing that a source of influence unavailable to consciousness has an effect on thought and behavior.