The Structure of Similarity

9511757 Gentner相似性是认知加工的一个中心方面。它影响类别学习、知识转移、归纳推理、决策以及基于案例的推理和规划。由于相似性的中心性，理解相似性心理学是很重要的。这项研究所基于的方法将相似性建模为一个过程，而不是将其视为一个简单的可衡量的值。因此，该方法不同于其他人之前的研究，它强调作为相似性基础的心理过程的动态模型，并利用被比较项目的丰富表现。这项研究涉及两个相似性比较过程的计算模型，SME和SIAM，它们展示了比较过程是如何从局部对齐到全局对齐的。研究的重点是进一步探索相似性比较的时间进程，并分析其对在线加工的启示。例如，快速相似性判断显示出不同于慢速判断的敏感模式，慢判断是一种与界面设计相关的观察，在人类处于速度压力的环境中。该项目以前的工作研究了相似性处理中的两个关键因素。第一种是互联性，即特征之间的相互作用可以使某些特征在相似性判断中更加突出，甚至改变对特征的心理解释。第二个问题是对齐对两个人物知觉相似性判断的影响。在这种基于成就的更新下进行的新研究围绕两个主要问题进行：1.结构和过程如何相互作用来确定相似性；2.相似性处理的结果如何扩展到将相似性作为组成部分的其他任务。例如，新的实验将通过操纵容易或难做的对齐，来研究两个人物之间的表面相似性如何随着环境的变化而变化。另一项研究将考察相似性比较对与相似性判断、基于类别的归纳、记忆提取和决策相关的建构性归纳的影响。总体而言，该项目将使我们了解相似性加工的范围和灵活性，并考察其在其他认知功能中的作用。***

9511757 GENTNER Similarity is a central aspect of cognitive processing. It influences category learning, transfer of knowledge, inductive inference, decision-making, and case-based reasoning and planning. Because of its centrality, it is important to understand the psychology of similarity. The approach on which this research is based models similarity as a process rather than considering it as a simple value to be measured. Thus, the approach departs from prior research by others in its emphasis on dynamic models of the psychological processes that underlie similarity, and also in utilizing rich representations of the items being compared. The research involves two computational models of the process of similarity comparison, SME and SIAM, that show how the comparison process proceeds from local to global alignment. A focus of the research is further exploration of the time course of similarity comparison and analysis of its implications for on-line processing. For example, speeded similarity judgments show sensitivity patterns different from slow judgments, an observation relevant to interface design in contexts where humans are placed under speed stress. Previous work on this project investigated two key factors in similarity processing. The first was interconnectivity, the fact that interactions between features can make some features more salient in similarity judgments and even change the psychological interpretation of the features. The second issue was the effect of alignment on judgments of the perceptual similarity of two figures. The new research to be performed under this Accomplishment-Based Renewal is organized around two major questions: 1. How do structure and process interact to determine similarity; and 2. How do results on similarity processing extend to other tasks that use similarity as a component. For example, new experiments will investigate how the seeming similarity between two figures changes with context, by manipulati ng the kind of alignment that is easy or hard to do. Yet another line of research will examine the effect of similarity comparisons on constructive induction associated with similarity judgments, category-based induction, memory retrieval, and decision making. Overall, the project will inform our understanding of the range and flexibility of similarity processing and examine its role in other cognitive functions. ***