and BRUCE BROMBACHER Upper Arlington (OH) Schools
In conjunctive feature identification tasks, sequences of all positive instances are favored over mixed sequences of positive and negative instances in laboratory tasks, but in mathematical conjunctive feature identification tasks, mixed sequences are favored over all positive instances. The subjects were 100 adolescents randomly assigned to one of four treatments formed by crossing sequence condition (positive or positive/negative) with relative frequency of features of irrelevant dimensions (equal or unequal). Results support the classical favoring of positive over positive/negative for equal frequency of irrelevant features (1:1) but positive/negative over positive for unequal frequency of irrelevant features (9:1) (p < .05). Feature frequency of irrelevant dimensions is shown to be a critical factor in the effects found for negative instances. Implications for educational practice are discussed.School learning is often described as skill learning, concept learning, and problem solving. For example, in mathematics, the Stanford Achievement Tests have three subtests; Concepts, Computations, and Applications. Concept learning is an integral part of the curriculum in many disciplines (Markle, 1975).We have become interested in the role negative instances play in concept learning. Briefly, if a concept is defined over a class of objects, those objects that are examples of the concept are called positive instances, and those objects that are not examples of the concept are called negative instances. Some argue that negative instances should play a central role in concept 451 at RMIT UNIVERSITY on July 23, 2015 http://aerj.aera.net Downloaded from