著者

小牧 純爾

出版者

日本動物心理学会

雑誌

動物心理学研究 (ISSN:09168419)

巻号頁・発行日

vol.41, no.1, pp.4-16, 1991-10-05

---

The short history of learning set (LS) research was reviewed for the purpose of elucidating the present state of our knowledge of the subject. The following remarks were made : (1) Disproof of the uniprocess theory and the revised understanding of the transfer suppression greatly reduced the plausibility of Harlow's error-factor theory of LS. (2) Discovery of the equivalence of successive-reversal training and LS training, also disclosing another weakness of the error-factor theory, led to the acceptance of Levine's model as an alternative theory of LS. (3) Bessemer and Stollnitz elaborated basic ideas of the model and, introducing a distinction between error-factor effects and hypothesis learning, formulated a new version of hypothesis theory of LS. (4) Subsequent experimental studies examining transfer effects of overtraining and successive informed-reversal training tested implications of the theory and revealed flaws and restrictions of the theory. (5) These studies also provided evidence for believing the occurrence of attention learning during LS training. Both hypothesis and attention learning seem to contribute to the formation of LS.