The Development of Categorization

项目摘要 建议的计划旨在加深我们对类别学习发展的了解- 是人类智力的基本组成部分。这项拟议的研究基于以下假设 多个机制为分类学习服务：(A)早期发展机制(机制1)是 基于对类别内统计的分布式注意和学习以及(B)后来的发展机制 (MECHANISM2)是基于对类别相关信息的选择性注意。在机制1下，两者都 对相关度和相关度较低的维度进行编码，学习的表示类似于输入 刺激结构(即没有基于学习的表征变化)。在MECHANISM2下学习 表征与输入刺激结构不同(我们将这种差异称为基于学习的 代表性变化)。机制1和机制2之间的差异将在许多方面显现出来 性能指标，包括学习过程中的注意力分配、对特征的记忆、学习的形状 曲线，以及训练和测试过程中的准确性和响应时间。为了检验这些假设，我们将 对4-9岁的儿童和成年人进行一系列实验。拟议的项目有以下几个方面 明确的目标。具体目标1是进行横断面实验和纵向研究检查 跨发展的类别学习机制。机械论1和机械论2定性预测 注意分配、分类和记忆的不同模式以及(2)学习前后的不同模式 类别的表示法。具体目标2是将计算模型应用于在研究1中收集的数据 识别两种假设的类别学习机制之间的差异。要做到这一点 目标、眼睛凝视和分类以及记忆选择和反应时间数据将提交给一套 捕捉假想机制的模型。拟议的项目将增进我们对链接的理解 在典型认知发展的关键方面--选择性注意、类别学习和记忆。

Project Summary The proposed project is designed to advance our understanding of the development of category learning – a fundamental component of human intelligence. The proposed research is based on the hypothesis of multiple mechanisms sub-serving category learning: (a) an early developing mechanism (Mechanism1) is based on distributed attention and learning of within-category statistics and (b) later developing mechanism (Mechanism2) is based on selective attention to category-relevant information. Under Mechanism1, both relevant and less relevant dimensions are encoded, with learned representations being similar to the input stimulus structure (i.e., no learning-based representational change). Under Mechanism2, learned representations are different from input stimulus structure (we refer to this difference as learning-based representational change). Differences between Mechanism1 and Mechanism2 will transpire in many performance indicators, including attention allocation during learning, memory for features, shapes of learning curves, and accuracy and response times during training and testing. To test these hypotheses, we will conduct a series of experiments with 4-9-year-old children, and adults. The proposed project has the following Specific Aims. Specific Aim 1 is to conduct cross-sectional experiments and a longitudinal study examining mechanisms of category learning across development. Mechanism1 and Mechanism2 predict qualitatively different patterns of (1) attention allocation, categorization, and memory and (2) pre- and post-learning representations of categories. Specific Aim 2 is to apply computational models to data collected within Study 1 to identify the differences between the two hypothesized mechanisms of category learning. To achieve this goal, eye gaze and categorization and memory choice and response time data will be submitted to a suite of models capturing the hypothesized mechanisms. The proposed project will advance our understanding of links among critical aspects of typical cognitive development – selective attention, category learning, and memory.