“…This process was well described by Bruner, Goodnow, and Austin (1956). This aspect of accounting for concepts has been considerably elaborated such that we now have evidence that positive instances are better than negative ones (Hovland & Weiss, 1953), that conjunctive concepts are more easily learned than disjunctive concepts (Deese & Hulse, 1967), that difficulty of the concept is related to the number of attributes defining it (Carroll, 1964), and that concepts derived from a variety of instances appear to be better learned than those based on a few (Callantine & Warren, 1955;Morrisett & Hovland, 1959;Osier & Kofsky, 1965) although there is considerable counterevidence (Adams, 1954;Amster & Marascuilo 1965;Podell, 1963;Stern, 1965).…”