Abstraction Processes in Associative Learning

联想学习通常是指学习世界上的属性、对象或事件之间的关系。生成的表示在它们保留的细节级别上可能不同，我们认为非模态表示与模态表示的一个方面是不同的。非情态表征从个体刺激的细节中抽象出来，专注于它们的共同或诊断属性。相比之下，情态表征保留了所遇到的刺激的特征。据推测，非情态表示允许更强和更广泛的概括，并且比情态表示更健壮。在这个项目中，我们调查了一种假设，即学习到的关系的抽象性取决于刺激环境的结构。基于错误驱动学习和线索竞争的概念，我们考察了刺激可变性和刺激呈现顺序等表征抽象性的潜在调节因素。在我们的项目中，我们采取了一种综合的观点，比较了这些因素在两种中央联想学习范式中的作用，即评估条件反射和概念学习。

Associative learning typically refers to the learning of relationships between attributes, objects, or events in the world. The resulting representations may differ in the level of detail they preserve, which we consider one aspect in which amodal representations differ from modal ones. Amodal representations abstract away from the details of individual stimuli to focus on their shared or diagnostic properties. By contrast, modal representations preserve the features of the stimuli encountered. Presumably, amodal representations allow for stronger and wider generalization and are more robust than modal representations. In this project, we investigate the hypothesis that the abstractness of the relationships learned depends on the structure of the stimulus environment. Based on the notion of error-driven learning and cue-competition, we investigate potential moderators of the abstractness of representations such as stimulus variability and the sequence of stimulus presentation. In our project, we take on an integrative perspective, comparing the role of these factors between two central associative learning paradigms, namely evaluative conditioning and concept learning.