This paper reviews written feedback from an information-processing perspective. The first section discusses the question of feedback as a reinforcer, and describes the feedback paradigm used as a conceptual guide for the following sections. In the second section we evaluate research on the form and content o f feedback. In the last section, a model is developed that applies concepts from servocontrol theory to the feedback sequence. Finally, we report three experiments which support the major predictions of the control model.KEY WORDS: feedback; instruction; learning; cognition.
I N T R O D U C T I O NThis paper is about postresponse feedback and how it works to promote learning from written instruction. We are concerned primarily with treating feedback as a unit of information. Hence, the bulk of our discussion is framed in terms familiar to those who study information processing and its cognitive manifestations. We try to overlap as little as possible with previous reviews covering the older literature, or dealing with specific topics such as the timing of feedback presentations (e.g., Kulhavy, 1977;Kulik and Kulik, 1988).There are three objectives we hope to accomplish in this review. First, we briefly discuss the "feedback as reinforcement" position, and introduce the standard paradigm for feedback research that we use as a conceptual guide for the remainder of the paper. The next step is to evaluate studies qnstructional Science Research Facility, Payne HaI1-FPE 0611, Arizona State University, Tempe, Arizona 85287-0611. 2To whom all correspondence should be addressed.
280Kulhavy and Stock which have treated feedback within the context of information processing, and look closely at explanations stressing the informational value of the feedback itself. In this second section we also provide a classification of feedback formats into categories that should be useful for directing future research.The final sections of the paper are concerned with developing and testing a model of how feedback interacts with the learner's knowledge state to produce changes in criterion performance. We use the term "model" in the same sense as Anderson (1983), to reflect the way that a particular theory is applied to a specific problem domain. In the current case, we use concepts from the theory of servocontrol to construct a model of the standard feedback paradigm, and to generate and test various predictions concerning the accuracy of the model.
REINFORCEMENT AND THE FEEDBACK PARADIGMOur thinking about instructional feedback has matured considerably over the past 20 years. In the late 1960s, the dominant position on feedback held that the postresponse information itself acted as a type of "reinforcer," functioning to increase the probability of a correct response occurring at some later point in time. This strong position developed from the association of feedback and operant psychology through the medium of the teaching machine (cf., Skinner, 1968). There were, of course, those who questioned the wisdom of treating the writing ...