The learning of sequential dependencies.

Publisher Website
Google Scholar
Abstract
The 4 experiments reported here are concerned with (a) the relative difficulty of learning digram versus trigram probabilities, and (b) the effect in such situations of sequential guessing habits. Experiment I was designed to determine sequential guessing habits in a situation where it was possible to make only a chance number of successes. Twenty Ss predicted which of five events would occur, with knowledge of results provided after each response. By using a restricted random procedure, all possible engrams, digrams, and trigrams appeared equally often. The pattern of correct responses thus had maximum uncertainty. In Exp. II two groups of 20 Ss each were given patterns which had equal engram but unequal digram (and trigram) probabilities. For one group the digrams conformed to the digram guessing habits obtained in Exp. I (Concordant group); the digrams for the other group were those used infrequently in Exp. I (Discordant group). A total of 10 digrams were used for both groups practice continued for a total of 250 responses (25 repetitions of each digram). Exp. III. and IV were similar to Exp. II except that trigram instead of digram patterns were used. All possible engrams and digrams appeared equally often in the pattern of correct responses; an appropriate prediction had to be based on the two preceding events. A total of 50 trigrams was used for both groups in each of these experiments. In Exp. III a total of 500 responses (10 repetitions of each trigram) was given. In Exp. IV the principle by which such patterns are generated was explained, and a total of 1000 responses (20 repetitions of each trigram) was given. It was found that the digram patterns of Exp. II were learned, but that little or no learning occurred with the trigram patterns of Exp. III and IV. It also was found that the influence of guessing habits persisted for an appreciable number of responses, especially when there was little learning. Possible explanations for the failure to learn the trigram patterns were discussed, as were methodological implications of the finding regarding the influence of guessing sequence habits.
Keywords

This publication has 2 references indexed in Scilit: