To study the beginning stages of expertise, 14 students. who were inexperienced with ballads, heard and recalled a series of 5 ballads over the course of 5 weeks. Compared with their first recall of the first ballad, their first recall of the fifth ballad had one and a half times as many words, two times as many rhyming words, and three times as much line structure evident in the written recall protocols. Compared with novices, the 14 beginning experts more often filled in blank spaces in novel ballads with words of the correct number of syllables and more often chose the original stanza of a novel ballad that was paired with a changed version of the stanza. The beginning experts were also able to compose, in 20 min, ballads about two thirds as long as the 10-stanza ballads they learned. Thirty characteristics were identified in the set of the five learned ballads. The ballads composed by the beginning experts used over half of these. The beginning experts also explicitly stated about one quarter of these 30 characteristics, but there was no statistical relationship between the characteristics used and the characteristics stated. Memory expertise is viewed as a pervasive aspect of cognition in which people make use of a variety of regularities in the material to be learned.In 1949, Harlow, reviewing this work on repeated choice tasks with monkeys a n d children, noted thatThe variety of learning situations that play an important role in determining our basic personality characteristics and in changing some of us into thinking animals are repeated many times in similar form. The behavior of the human being is not to be understood in terms of the results of single learning situations but rather in terms of changes which are affected through multiple, though comparable, learning problems.. . .This learning to learn transforms the organism from a creature that adapts to a changing environment by trial and error to one that adapts by seeming hypothesis and insight. (p.51)A l l reported effects are significant at the .05 level unless otherwise noted.We wish to thank John Bransford, Arnold Glass, Keith Holyoak and Ian Hunter for their comments, the two professional singers, Margaret and Wayne Martin, and above all, the students for their help. Support was provided by NSF grants number BNS-8410124 and number BNS-9010174.Correspondence and requests for reprints should be sent to David C. Rubin. Department of Experimental Psychology, Duke University, Durham, NC 27708-0086. RUBIN, WALLACE, AND HOUSTONResearch on human memory has concentrated on "the results of single learning situations," but not exclusively. The problem of memory expertise has had a small but long and continuous history in mainstream memory research. Ericsson (1985) (Holding, 1985;Intons-Peterson & Smyth, 1987;Noice, 1991Noice, , 1992Thompson et al., 1991). Similarly, Neisser (1982) reprinted several articles on memory experts. In the last two decades the detailed investigation into possible mechanisms that allow expertise has expanded, including studies...