Charting the role of variability in learning across development

项目摘要/摘要 个人从环境中学习的方式会随着年龄和经验的变化而可靠地改变。一种常见的 在广泛的认知任务中观察到的模式是，年龄较小的孩子往往表现出更大的 准确性、反应时间和其他一般反应模式的可变性，而这种可变性既减少了 随着年龄的增长，受试者内部和受试者之间的相互作用。虽然这种增加的可变性被广泛认为是一种 发育的特征，这种变异背后的神经和计算过程仍然是 它的潜在的学习功能仍然没有得到很好的描述。这项研究的目的是 建议将与学习变化和改善相关的行为和神经变异性绘制成图表 跨越发展。为了实现这一目标，我们将应用计算建模和神经成像 更好地区分儿童学习过程中观察到的变异性模式和变异性的方法 随着成熟而发生的还原过程。我们将探索两个关键的可变性衡量标准：(1) 在学习过程中探索选择的过程，以及(2)用于测试的策略的发展 关于要学习的信息的规律性的假设。比较儿童的连续样本 通过成年人，我们将测试三个核心预测。首先，在目标1中，我们预测儿童将表现出更大的 勘探开发加固中更大的勘探和选择随机性的形式的可变性 学习任务，这种探索模式会随着年龄的增长而减少。第二，在目标2中，我们假设 当形成策略以便学习如何在简单的知觉分类任务中对刺激进行最佳分类时， 儿童将通过使用更长持续时间的随机猜测策略来表现出更大的变异性 在不同策略之间切换的频率比成年人更高。第三，我们假设这些行为影响将 受较晚成熟的前额叶皮质区域的调节--探索时的额叶皮质和内侧 在策略形成期间，前额叶皮质-这些区域的激活程度将在 随年龄增长的大小。之前没有一项研究同时检查了行为和神经 在整个发展过程中学习过程中潜在的探索和策略使用。因此，我们希望 更详细、更细致入微地了解精确的可变性测量如何阐明 我们学习方式的发展差异。通过更好地阐明减少变异性的过程 在开发过程中学习，这项研究可能会为未来的翻译工作提供信息，以确定何时以及如何 儿童或青少年的认知发展可能并不遵循标准的轨迹。这一知识 可以帮助设计更好的早期诊断措施，帮助有学习障碍或 与更大的神经和行为变异性相关的神经发育障碍，如ADHD和 自闭症。

Project Summary/Abstract The way individuals learn from their environment reliably changes with age and experience. A common pattern observed across a wide range of cognitive tasks is that younger children tend to exhibit greater variability in accuracy, reaction time, and other general response patterns, and this variability decreases both within and between subjects with increasing age. While this increased variability is widely acknowledged as a feature of development, the neural and computational processes underlying such variability remain to be discovered, and its potential functions for learning remain poorly characterized. The goal of this research proposal is to chart behavioral and neural variability associated with changes and improvements in learning across development. To accomplish this goal, we will apply computational modeling and neuroimaging methods to better discriminate patterns of variability observed in children during learning, and the variability reduction process that takes place with maturation. We will explore two key measures of variability: (1) the process of exploration of choices during learning, and (2) the development of strategies used to test hypotheses about the regularities of the information to be learned. Comparing a continuous sample of children through adults, we will test three core predictions. First, in Aim 1 we predict that children will show greater variability in the form of greater exploration and choice stochasticity in an explore-exploit reinforcement learning task, and this pattern of exploration will decrease with age. Second, in Aim 2 we hypothesize that when forming strategies in order to learn how to best classify stimuli in a simple perceptual categorization task, children will demonstrate greater variability by using a random guessing strategy for a longer duration and switching between strategies more often than adults. Third, we hypothesize that these behavioral effects will be modulated by late maturing prefrontal cortical regions - the frontopolar cortex during exploration, and medial prefrontal cortex during strategy formation – and the degree of activation in these regions will increase in magnitude with age. No previous study has simultaneously examined both the behavioral and neural processes underlying exploration and strategy use during learning across development. As such, we hope to gain a more detailed and nuanced understanding of how precise measures of variability can elucidate developmental differences in the way we learn. By better clarifying the process of variability reduction in learning during development, this research may inform future translational work in identifying when and how the cognitive development of a child or adolescent may not be following a normative trajectory. This knowledge can aid in designing better early diagnostic measures for individuals at risk of developing learning disabilities or neurodevelopmental disorders associated with greater neural and behavioral variability such as ADHD and autism.