Computational Models of Implicit Learning

Implicit learning – broadly construed as learning without awareness – is a complex, multifaceted phenomenon that defies easy definition. Frensch (1998) listed as many as eleven definitions in an overview, a diversity that is undoubtedly symptomatic of the conceptual and methodological challenges that continue to pervade the field forty years after the term first appeared in the literature (Reber, 1967). According to Berry and Dienes (1993), learning is implicit when an individual acquires new information without intending to do so and in such a way that the resulting knowledge is difficult to express. In this, implicit learning thus contrasts strongly with explicit learning (e.g., as when learning how to solve a problem or learning a concept), which is typically hypothesisdriven and fully conscious. Implicit learning is the process through which one becomes sensitive to certain regularities in the environment: (1) without trying to learn regularities, (2) without knowing that one is learning regularities, and (3) in such a way that the resulting knowledge is unconscious. Over the last twenty years, the field of implicit learning has come to embody ongoing questioning about three fundamental issues in the cognitive sciences: (1) consciousness (how one should conceptualize and measure the relationships between conscious and unconscious cognition); (2) mental representation (in particular, the complex issue of abstraction); and (3) modularity and the architecture of the cognitive system (whether one should think of implicit and explicit learning as being subtended by separable systems of the brain or not). Computational modeling plays a central role in addressing these issues.