Context and Classification Learning

点击翻译按钮获取中文摘要

This project aims to understand the processes underlying classification learning and performance. One focus is the question of what makes some categories natural, easy to learn and use, and other categories unnatural, difficult to learn, and awkward to use. Aside from relevance to the problem of how best to teach certain concepts (e.g., scientific concepts), the question of naturalness is a tool for evaluating categorization theories. That is, alternative theories make differing predictions about what kinds of concepts are easy to use and learn. Several of the proposed experiments will test these predictions by presenting people with different types of category structures to see how well the theories can predict difficulty of learning. A related question is how people organize examples when they are free to create their own categories. Again the idea is to see what principles determine category construction and to compare these principles with the predictions of alternative theories. Another major facet of the project will be experiments examining learning strategies and the idea that there may be distinct modes of learning. There is some evidence that direct, explicit attempts to learn difficult category structures may be less effective than indirect or nonanalytic learning. Several experiments will explore the possibility that nonexplicit learning will lead to rule-like performance even though people may be unable to verbalize their knowledge. Finally, the project will examine how people integrate abstract information about categories and concrete experience with examples. Overall, the project has the goal of developing better theories for how people learn about, use, and update concepts and categories.