A Connectionist Model of Instructional Feedback Effects.

Connectionist models apply various mathematical rules within neural network computer simulations in an effort, among other things, to mimic and describe human memory associations and learning. Learning involves the interaction of information provided by instruction with existing information already in the learner's memory (Ausubel, 1968; Bruner, 1990). When a learner commits to a lesson response, that response reflects the learner's immediate understanding of that instructional instance, thus initial lesson response (ILR) provides a measure of a learner's existing information. Describing what happens to memory traces of ILRs that are errors is necessary for determining whether errors interfere with attaining correct responses, and so is one key to understanding how feedback works. Among a number of connectionist learning rules, the delta rule (Shanks, 1995; Widrow & Hoff, 1960) is one of the simplest and most common that includes the effects of feedback on learning. Clariana (1999) has suggested that a connectionist approach using the delta rule can be used to predict posttest memory activation levels of ILR errors and of correct responses for immediate and delayed feedback. This experimental investigation provides empirical support for the potential of a connectionist approach to predict instructional feedback effects. Posttest data from high school students is compared to values predicted by the delta rule. Results and implications are provided. (Contains 27 references.) (AEF)