Cognitive Mechanisms of Learning

5Implicit Learning

5.1. Presentation

Two articles are widely considered as fundamental in the field of implicit learning. A.S. Reber (1967) introduced the expression “implicit learning”, while D.C. Berry and D.E. Broadbent (1984) worked on the way in which control is learned in dynamic environments; the most famous example covered in the article is the sugar production factory task.

Reber and Allen (1978) used a finite state automaton represented by a state transition diagram, with nodes (states) containing numbers (S1, S2, etc.) and arcs labeled with letters. For example, the arc connecting states Si and Sj is labeled P, and the arc from Sj to Sk is letter X. Figure 5.1 is taken from Reber and Allen’s publication.

Figure 5.1. State transition diagram

This automaton can be used to create letter strings, such as VXV, VXRR, MTTVRXRRM, etc. These strings are said to be grammatical; changing just one letter in a grammatical string is sufficient to render it agrammatical. The string MTTRRXRRM, for example, is agrammatical.

In Reber et al.’s earliest work, experimenters asked subjects to memorize grammatical strings. Following a given “learning” period, the experimenter informed the subject that these memorized strings were constructed following the rules of a grammar; more letter strings were then shown, some constructed using the grammar, others constructed without the grammar. ...