The Case of Implicit Learning

Is nonconscious goal pursuit useful in novel environments? The prevalent view of automaticity and control implies that an unconscious mode of goal pursuit can only reproduce formerly learned actions, and therefore that its usefulness in novel environments is very limited. Our results demonstrate that this conclusion is not always warranted, as nonconscious goal pursuit facilitated participants’ learning of the structure of completely novel environments. Specifically, two experiments, using markedly different implicit-learning paradigms, demonstrated facilitation of implicit learning when the goal of achievement was primed. We propose that nonconscious goal pursuit can facilitate not only reproductive operations, but also productive ones, and that implicit learning is sensitive to the organism’s nonconscious goals. To what degree can people count on unconscious processes to pilot them toward the future they desire, that is, toward their goals? Successfully attaining one’s goals (defined here as desired end states) depends critically on one’s sensitivity to the ever-changing environment and its affordances. Because of the environment’s complexity, however, the crucial task of identifying goal-relevant structures is far from trivial. In the study reported in this article, we examined whether nonconscious goal pursuit (NCGP) can increase the likelihood of learning goalrelevant structures in the environment, thereby boosting the likelihood of goal attainment. One answer to the question we have posed may come from recent dual-system models of reasoning. These models share an all-encompassing assumption that there are two qualitatively different mental systems, one that is associative and one that is rule based (e.g., Evans, 2003; Sloman, 1996; Strack & Deutsch, 2004; but see Keren & Schul, 2007; Osman, 2004). The rulebased system has the ability to reason, deliberate, formulate strategies, and pursue them, and is generally assumed to require consciousness and mental resources. The associative system, in contrast, is assumed to operate automatically, triggering elements from past experience by way of spreading activation. Given these pervasive models, it seems natural to conclude that consciously controlled goal pursuit is delegated to the rulebased system, and hence this form of goal pursuit is viewed as productive, deliberative, and flexible, and as involving intention and conscious monitoring (e.g., Locke & Latham, 2002). Similarly natural is the conclusion that NCGP is delegated to the associative system, and hence this form of goal pursuit is viewed as reproductive, unintentional, and inflexible. Indeed, this is the spirit of recent models of NCGP (e.g., Bargh, 1990; Kruglanski et al., 2002), which adopt the theoretical view that NCGP is based on unintentional and nonconscious reproductions of past associations. Put differently, these models assume that a nonconscious goal can trigger an action (broadly defined) if that action has formerly been chosen to attain that goal. Accordingly, the lion’s share of past research on NCGP has concentrated on overlearned (i.e., strongly associated) means and goals. For example, Bargh, Gollwitzer, Lee-Chai, Barndollar, and Troetschel (2001) found that people who had been primed with the goal of achievement invested more effort than those who had not been so primed (see also Stajkovic, Locke, & Blair, 2006); Aarts, Gollwitzer, and Hassin (2004) found an increase in the help given to people of the opposite sex when the goal of obtaining sex was activated; and Fishbach, Friedman, and Kruglanski (2003) found that people were more likely to refrain from sweet temptations when the goal of dieting was indirectly triggered than when no goal was triggered. Still, given the dynamic nature of human environments (whether in the physical, the social, or the mental realm), strictly Address correspondence to Baruch Eitam, Ran R. Hassin, or Yaacov Schul, Hebrew University, Department of Psychology, Mt. Scopus Campus, Jerusalem 91905, Israel, e-mail: [email protected], [email protected], or [email protected]. PSYCHOLOGICAL SCIENCE Volume 19—Number 3 261 Copyright r 2008 Association for Psychological Science at The Hebrew University Library Authority on January 25, 2011 pss.sagepub.com Downloaded from reproductive NCGP is bound to be of limited use. For a system that is very limited in its conscious processing ability (Kahneman, 1973) yet hasmany concurrent goals, such a limitationmay even be detrimental. Our study, then, examined whether this restriction on NCGP can be rejected. Are nonconscious goals beneficial in situations that require mastering novel environments? THE PRESENT RESEARCH Based on the arguments we have just discussed, our general hypothesis was that NCGP can help people achieve their goals even in novel environments, that is, in environments with which they have no prior experience. In particular, we investigated the hypothesis that NCGP can facilitate learning the structure of novel environments. To do this, we examined how NCGP influences performance on implicit-learning tasks. Research on implicit learning has established that humans can learn and use complex patterns of information without intending to learn them and without being aware of the patterns (e.g., Nissen & Bullemer, 1987; Reber, 1967, 1993; for a recent review, see Frensch & Rünger, 2003). In fact, humans’ ability to learn relations implicitly seems to exceed their ability to process this type of information consciously (cf. Halford, Baker, McCredden, & Bain, 2005). Traditionally, implicit learning was assumed to be a bottomup, unselective, data-driven process (e.g., Cleeremans, 1993; Hayes & Broadbent, 1988). Therefore, researchers assumed it was not influenced by an individual’s goals (cf. Lewicki, Hill, & Czyzewska, 1992). Recent evidence, however, suggests that selective attentionmay play a role in implicit learning (Jiang&Chun, 2003; Jiménez & Méndez, 1999; Turk-Browne, Junge, & Scholl, 2005). These findings, together with the conjecture that attention is influenced by goal-driven prioritization of information processing (e.g., Eriksen & Hoffman, 1972; Posner, 1980), suggest to us that implicit learning could be affected by the organism’s NCGPs. We hypothesized that NCGP can enhance implicit learning and conducted two experiments to test this hypothesis. In both, we examined whether priming an achievement goal affected performance on an implicit-learning task. To prime this goal, we used the ‘‘unrelated experiments’’ paradigm (Chartrand & Bargh, 1996). Briefly, following completion of a first ‘‘experiment’’ (in actuality, a goal-priming procedure), participants were presented with ‘‘the second experiment’’—an implicit-learning task. EXPERIMENT 1: LEARNING TO CONTROL A DYNAMIC SYSTEM