"What does that animal eat?": The role of overhypotheses in how children and adults learn and reason about real-world categories

概念（即类别的心理表征）是思想的基本组成部分，人们以多种方式使用它们，包括在学习和推理时。令人惊讶的是，孩子们能够从一个类别的单一成员的经验中形成概念（例如，在看到一只松鼠吃坚果后，他们期望松鼠通常会吃坚果）。在这种情况下学习是具有挑战性的，然而孩子们似乎很容易做到。因此，哲学家们提出，必须有约束来指导学习。在我的研究项目中，我研究了这些约束之一——过度假设，这是一种高阶概括。学习者可能会使用对一般类别的特征特征的过度假设（例如，不同种类的动物有独特的饮食）来指导学习特定类别（例如，松鼠吃坚果）的期望。我提出了三个与使用和获取现实世界类别的过度假设有关的问题。在每一项提议的研究中，我都考察了4至8岁儿童和成年人的反应，以调查人们如何学习和使用过度假设的发展轨迹。1)过度假设在哪些方面指导人们的推理？我将系统地检查过度假设使用的范围。具体来说，我将探讨它们是否指导(1)参与者想要了解的关于不熟悉的类别的信息类型，(2)人们记住的关于类别成员的信息，以及(3)人们从单个类别成员概括到整个类别成员的特征。2)不同年龄的学习者是如何形成overhypothesis的？在这些研究中，我探讨了参与者形成过度假设的可能性是否受到所提供样本的内容和用于描述类别成员特征的语言的影响。3)过度假设能帮助学习者修正错误的信念吗？这些研究考察了人们如何轻易地修改与过度假设相关的错误信息（例如，如果一个人第一次看到松鼠吃披萨，当他们看到松鼠吃什么的准确信息时，他们会修改自己的概念吗？）该研究项目将通过显著推进我们对概念的基础知识，为认知科学领域做出贡献。研究结果将使加拿大站在许多基本认知过程研究的前沿，为人们如何学习和推理提供洞见。这些发现也将对其他领域产生影响，包括人工智能——通过了解人们如何获取和使用概念，将有可能基于对人类概念系统的了解创造尖端技术。最后，研究结果可以指导在教育环境中的应用干预。研究人员可以利用这些发现来设计适合不同年龄的人学习的教学实践。

Concepts (i.e., mental representations of categories) are the fundamental building blocks of thought and people use them in a number of ways, including when they learn and when they reason. Surprisingly, children are able to form concepts from experience with a single member of a category (e.g., after seeing just one squirrel eat nuts, they expect that squirrels, in general, will eat nuts). Learning in these situations is challenging, yet children seem to do it with ease. As a result, philosophers have proposed that there must be constraints that guide learning. In my research program, I examine one of these constraints - overhypotheses, which are higher-order generalizations. Learners may use overhypotheses about the characteristic features of general categories (e.g., different kinds of animals have a characteristic diet) to guide expectations when learning about specific categories (e.g., squirrels eat nuts). I address three questions related to the use and acquisition of overhypotheses for real-world categories. In each proposed study, I examine 4- to 8-year-olds' and adults' responses to investigate the developmental trajectory of how people learn and use overhypotheses. 1) In which ways do overhypotheses guide how people reason? I will systematically examine the scope of overhypothesis use. Specifically, I will explore whether they guide (1) the types of information that participants want to learn about an unfamiliar category, (2) the information that people remember about members of a category, and (3) which features people generalize from a single category member to category members at large. 2) How do learners of different ages form overhypotheses? In these studies, I explore whether participants' likelihood of forming an overhypothesis is influenced by the content of the samples provided and the language used to describe features of category members. 3) Do overhypotheses help learners to revise incorrect beliefs? These studies examine how easily people modify information that is incorrect if it is linked to an overhypothesis (e.g., if a person first sees a squirrel eating pizza, do they revise their concept when presented with accurate information about what squirrels eat?). This research program will contribute to the field of cognitive science by significantly advancing our foundational knowledge about concepts. The results will place Canada at the forefront of research on many fundamental cognitive processes by providing insight into how people learn and reason. The findings will also have implications for other fields, including artificial intelligence - by understanding how people acquire and use concepts, it will be possible to create cutting-edge technology based on what is known about the human conceptual system. Finally, the results may guide applied intervention in educational settings. Researchers could use the findings to design pedagogical practices that are tailored to support how people of different ages learn.