Role of implicit and explicit processes in learning from examples: A synergistic effect.

Four experiments are reported in which subjects gained extensive experience with artificial grammars in explicit and implicit processing tasks. Results indicated that (a) implicit processing was sufficient for learning a finite state grammar but was inadequate for learning another type of grammar based on logical rules, (b) Subjects were able to communicate some of their implicit knowledge of the grammars to another person, (c) Consistent with rule induction but not memory array models of learning, verbal protocols indicated there was no tendency to converge on the same set of cues used to identify valid strings, (d) A synergistic learning effect occurred when both implicit and explicit processing tasks were used in the grammar based on logical rules but not in the finite state grammar. A theoretical framework is proposed in which implicit learning is conceptualized as an automatic, memory-based mechanism for detecting patterns of family resemblance among exemplars. Explicit learning mechanisms for discovering and control-ling task variables are similar to conscious problem solving. These processes include attempts to form a mental represen-tation of the task, searching memory for knowledge of anal-ogous systems, and attempts to build and test mental models of task performance (Gentner & Stevens, 1983; Johnson-Laird, 1983). Implicit learning is thought to be an alternate mode of learning that is automatic, nonconscious, and more powerful than explicit thinking for discovering nonsalient covariance