Willful control and the learning of complex systems
LeCompte, Denny Charles
Watkins, Michael J.
Doctor of Philosophy
A series of experiments explores the role of willful control over the learning of complex, rule-governed systems such as language. Willful control is operationalized as the enhancement of learning by the deliberate application of cognitive strategies. Subjects studied strings of symbols generated according to the rules of a system known as an artificial grammar. They were then tested on their knowledge of the grammar's rules. In some conditions, symbols drawn from the vocabulary used in the grammar were inserted into each string, rendering the string somewhat ungrammatical. In the informed condition, the inserted symbols were identified; in the uninformed condition, the inserted symbols were not identified. If subjects were able to exert willful control by ignoring the inserted symbols, performance in the informed condition should exceed performance in the uninformed condition. If they were not able to exert willful control, there should no difference between the informed and uninformed conditions. Evidence that the subjects had at least some degree of willful control was obtained in each of 10 experiments. Experiments 1, 2, 3, and 4 included a condition in which the subjects saw strings with no inserted symbols. Performance in this condition was consistently superior to performance in the informed condition, implying that the extent of willful control was less than complete. Experiments 5, 6, 7, and 8 tested the hypothesis that the extent of willful control would depend on the number of symbols inserted. Experiment 9 tested the hypothesis that exposure to strings with no ungrammatical symbols would enhance subjects' ability to ignore such symbols in subsequent strings. Finally, Experiment 10 tested the hypothesis that willful control would be increased if the inserted symbols did not come from the same vocabulary as the grammar. None of these hypotheses was supported. The general conclusion from these experiments is that subjects can exert some degree of willful control over the learning of complex systems, but the extent of that control is substantially limited.