Direct Instruction: More Than Meets the Eye

• In Event: Different Perspectives on the Role of Explanation and Exploration in Learning

Abstract

Theoretical Perspective: A persistent debate centers on the role of direct instruction as an instructional technique (for example, as contrasted with exploration; Hirsch-Pasek et al., 2009; Klahr & Nigam, 2004; Mayer, 2004). In science education, the primary merit of instruction (“teaching”) is said to be the ability to efficiently transmit specific facts (Gelman, 1969; Klahr & Nigam, 1999; Anderson et al., 1995). In developmental psychology, the primary benefit of teaching is to facilitate learning of abstract concepts (Csibra & Gergely, 2009; Tomasello et al., 2005). Without a formal understanding of the basis of instruction, the relative merits and effective use of the technique for science education will remain elusive.

Objective: We introduce a mathematical analysis of direct instruction (Shafto & Goodman, 2008). We contrast learning from direct instruction where the teacher is either knowledgeable or naïve. The analysis predicts that a knowledgeable teacher would choose examples that tend to maximize the learner’s belief in the correct hypothesis, while a naïve teacher, not knowing the answer, would choose examples at random. As a consequence, when learning from a knowledgeable teacher, learners may engage social learning to draw stronger inferences than would be licensed given the very same data chosen by a naïve teacher. By reducing uncertainty, teaching by a knowledgeable individual will consequently discourage children’s investigation and experimentation, in contradiction of the recommendations of state standards.

Research Evidence: We focused on children’s investigation of the causal properties of a novel machine. Four- and five-year-old children (N=85) were randomly assigned to one of three conditions: Pedagogical, Naïve, Interrupted conditions. In the Pedagogical condition, the teacher introduced the machine while telling the child that it was hers (to imply knowledge), then demonstrated a non-obvious causal property. In the Naïve condition, the teacher introduced the machine while telling the child that she had not seen it before (to imply lack of knowledge), then accidentally elicited the same causal property. The Interrupted condition was the same as the Pedagogical condition, with the exception that the teacher was interrupted after the initial demonstration, but before it was clear she that she was not going to perform subsequent demonstrations. Children were then given the machine and allowed to explore. We predicted that children in the Pedagogical condition would infer that there was nothing else to be learned, leading to decreased exploration and learning relative to the Naïve and Interrupted conditions. The results confirmed the predictions: children in the Pedagogical condition played for less time, tried fewer actions, and as a consequence discovered fewer non-demonstrated causal functions.

Significance: Given the very same evidence, children draw different conclusions depending on whether the demonstrator was knowledgeable or not. Thus, direct instruction goes beyond simply communicating facts or supporting abstraction; it invites powerful social inference, which is a malleable factor that may be harnessed to facilitate or impede learning in science education settings.

Author

• Patrick Shafto, University of Louisville