Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
This Week’s Citation Classic® E~ Scandura J M. Deterministic theorizing in structural learning: three levels of empiricism. J. Srucf. Learn. 3:21-53, 1971. r (Instructional Systems. Graduate School of Education. University of Pennsylvania. Philadelphia. PAl This article describes a deterministic theory of structural learning designed to explain and predict the behavior of individuals in specific situations. It includes three increasingly precise partial theories: (1) structured knowledge— tested via generative adequacy, (2) idealized behavior—tested under memory-free conditions, and (3) non-idealized behavior—including memory and processing capacity. [The SSCI® indicates that this article is the most-cited paper for this journal.] ~ thereafter at Psychonomics and the American Psychological Association. In 1970 I gave an invited address at the American Educational Research Association based on a draft of my paper with Wittrock as discussant. It covered everything from rationale and methodology to comprehensive theory and empirical data, my first overall presentation of what has come to be known as the “Structural Learning Theory.” I felt it was a tour de force, and I had put a tremenJoseph M. Scandura dous amount of working energy into it. Structural Learning, Instructional Nonetheless, both Wittrock’s formal Systems, and Intelligent CBI reaction and the following questions University of Pennsylvania were sobering: the significance of what Philadelphia, PA 19104 I was trying to say simply was not getMarch 30, 1987 ting across. Subsequent submission of the This paper grew out of a long-term formal paper to the Psychological Review concern: specifically, the inability of fared no better even though I considered then-current theory and methods to ac- it superior to a paper I had published in 4 commodate both generalized psycholog- that journal the previous year. “What ical considerations and specific knowl- is Scandura talking about?” is a reasonedge. Moreover, my own empirical re- able paraphrase. search kept showing that predicting huHappily (for me), reaction at one of our man behavior on complex tasks was pri- first structural learning conferences (that marily a matter of knowing exactly what included philosophers, mathematicians, it was that the learner did and did not computer scientists working with Al, and linguists, as well as cognitive and instrucknow. My theoretical concerns, initially, cen- tional psychologists) differed sharply. Retered on how to represent human knowl- jected by the establishmenUand given the edge (something that has become the cor- obvious enthusiasm at the conference, I nerstone of both cognitive psychology submitted my paper to the then-new, reland artificial intelligence [Al]) and how atively low-circulation Journal of Structo assess it (i.e., how to find out what the tural Learning. Founded by the gifted individual knows). In several papers, for mathematician and mathematics educaexample, I challenged the use of tor, Z.P. Dienes, this journal fortunately stimulus-response (S-R) representations stressed ideas over orthodoxy. (and hence theories) as inadequate. Since my article appeared I have conAt an invited meeting in Philadelphia tinued to publish; one of my most recent 1 in 1968, 2~3 Jim Greeno and Merle Wit- summary updates is a 1985 paper from 5 trock argued that construct itself was the Journal of Structural Learning of inadequate—that unlike S-R theory, there which I am now the editor. However, I was no theory behind my conceptualiza- believe my 1971 paper is still cited octions. This motivated me to deal with casionally because it represents a good these issues in presentations shortly introduction to the field. I. Greenu j. Psychological representation of structured knowledge. (Scandura J M. ed.) Structural learning II: issues tnt) approaches. New York: Gordon and Breach. 1976. p. 73-95. 2. Wittrock M C. Comments on Scanduras approach to rule-governed behavior. (Scandura I M. ed.) Structural learning II: issues ant) approaches. New York: Gordon and Breach. 1976. P. 65-7. 3. Scandura J M. A reply to Wittrock. (Scandura I M. ed.) Structural learning 11: issues and approaches. New York: Gordon and Breach. 1976. P. 69-70. (Cited 10 times.) 4. . Role of rules in behavior: toward an operational definition of what (rule) is learned. Psycho!. Rev. 77:516-33. 1970. (Cited 50 times.) 5. .— . System issues in problem solving research. I. Struct. Learn. 9:l-l3, 985. 14 i CC/S -~ ~-~4 ~1987 by SI® CURRENT CONTENTS®