Hypothesis-testing theories and information processing theories have both been used to explain the results of problem solving experiments. The theories have, in general, been applied to different claasee of problems, resulting in little overlap or interaction between them. Information processing theories have tended to emphasize problems in which the problem solver' s primary source of information concerning the correctness of his moves is the problem goal and, as a result, the goal plays a major role in the control mechanisms determining moves. Hypothesis theories have emphasized problems in which direct feedback is the primary control mechanism available. It is suggested that altering a problem solver's major source of information from a goal to feedback or vice versa will have important implications for the course of problem solving and what islearned during the solution process. Specifically, changes in the information feedback andlor goal-specificity characteristics of a problem should, in a predictable fashion, alter problem solving strategies, which should, in turn, through the presence or absence of information compression devices such as rule induction, strongly affect transfer. Experiments using the Tower of Hanoi and visual maze-tracing problems indicated that altering the density of subgoals resulted in alterations in subsequent transfer performance.A wide variety of tasks studied in the problem solving laboratory are characterized by decision-making processes. Solution is attained by aseries of decisions. Two major classes of theories have arisen to explain and predict experimental results: information processing theories and hypothesis theories. (It might be noted that although these labels are commonly used, they are not strictly descriptive, since both theories discuss information processing.) Both theories arose and developed largely in isolation from each other, and although both emphasize decision making during problem solving, there appears to be little overlap either in the theoretical statements made or in the types of problems studied.
Two Problem TypesHypothesis theory has been applied primarily to the stimulus-response-feedback type of problems, of which concept learning tasks are the best known. Subjects are presented a stimulus, such as one or more colored geometric shapes, and must decide which one of two or more possible responses to make (e.g., pointing to one of the objects) before being informed of the correctness of the response. Most stimulus-response-feedback problems require the induction of a rule for solution, and this has influenced the theories used to explain the results of experiments employing these types of problems. The most influential theory has been hypothesis-testing theory, with Levine (1975) providing the most comprehensive statement. In essence, the theory assurnes that subjects solve problems by testing aseries of hypotheses I wish to thank Marvin Levine for extended discussions and comments on an earlier version of this paper. Requests for reprints should be ...