How implicit is implicit learning?
Dianne C. Berry
In everyday life there are many examples of our learning to respond in some rule-like way without being able to state the rules or regularities that govern our behaviour. For example, most of us learn to recognize and produce grammatical utterances without ever being able to describe the underlying rules of the grammar. The term 'implicit learning' has been used to characterize such situations, and has been the subject of increasing interest and debate in recent years. In fact, understanding the processes involved in implicit learning, and its relationship to explicit learning, have become central goals in current cognitive psychology.
Implicit learning has been investigated in a wide range of experimental paradigms including artificial grammar learning, control of complex systems, and sequence learning. What these situations have in common is that a person typically learns about the structure of a fairly complex stimulus environment, without necessarily intending to do so, and in such a way that the resulting knowledge is difficult to express. In terms of controlling complex systems, for example, people can learn to reach and maintain specified levels of target variables without being aware of the basis on which they are responding (e.g. Berry and Broadbent 1984; Stanleyet al. 1989). Similarly, people can learn to classify exemplars of an artificial grammar, and can acquire knowledge about the sequential structure of stimuli, without adopting explicit code-breaking strategies and without being able to articulate any rule they might be using or the basis on which they are responding (e.g. Reber 1989; Nissen and Bullemer 1987). In addition, recent studies have provided convincing demonstrations of implicit learning in neuropsychological patients (e.g. Squire and Frambach 1990; Knowltonet al. 1992; see Berry and Dienes 1993 for a review).
Despite these wide-ranging empirical demonstrations, however, a number of researchers have argued against the existence of implicit learning. They have suggested that reported discrepancies between measured performance and explicit verbalizable knowledge can be accounted for without resorting to the notion of implicit or unconscious learning