Semantic memory guides and is shaped by value-based decisions

该奖项是美国国家科学基金会社会、行为和经济科学（SBE）博士后研究奖学金（SPRF）项目和SBE认知神经科学项目的一部分。SPRF计划的目标是为学术界、工业界或私营部门和政府的科学事业准备有前途的早期职业博士级科学家。SPRF奖励包括在知名科学家的赞助下进行为期两年的培训，并鼓励博士后进行独立研究。美国国家科学基金会寻求促进科学界各阶层的科学家，包括那些未被充分代表的群体的科学家，参与其研究项目和活动；博士后阶段是实现这一目标的重要专业发展阶段。每个博士后必须解决各自学科领域的重要科学问题。在哥伦比亚大学的Daphna Shohamy博士的赞助下，这项博士后奖学金支持一位早期职业科学家研究语义记忆是如何引导和塑造基于价值的决策的。通过日常经验，人们了解世界的事实（例如，了解什么水果是成熟的），然后用这些知识来指导决策（例如，选择成熟的水果吃）。这一建议旨在促进我们对这些世界知识如何指导决策和所涉及的神经机制的理解。研究结果将对教育产生重要影响，教育将受益于不仅通过单一曝光教授事实，而且通过不同内容的多次曝光获取抽象知识。这种抽象的知识特别有用，因为它可以推广到课堂以外的新情况。似乎很明显，决策是由语义记忆引导的，但值得注意的是，还没有实证研究解决这是如何发生的问题。提出的两个实验将通过整合行为、计算建模和功能成像（fMRI）来填补这一空白，以研究语义记忆和基于价值的决策相互作用以产生适应性行为的神经机制。第一项研究检验了一个假设，即选择的价值取决于在一个语义范畴的多个经验中积累的抽象知识。参与者将学习一个类别中多个项目的平均值，以测试该平均值是否用于确定单个项目的价值并指导决策。第二项研究测试了一个假设，即语义类别的定义特征是通过跨类别的多个实例做出基于价值的决策来学习的。通过分类和价值学习相结合的任务来测试特征值对分类的影响。我们假设语义信息和价值信息分别在前颞叶（ATL）和腹侧内侧前额叶皮层（vmPFC）内的不同脑区表示。类别学习模型将用于量化语义信息的一次又一次变化，而强化学习模型将用于量化价值信息的一次又一次变化。该项目预测，来自ATL的语义信息将作为vmPFC中值计算的输入，而来自vmPFC的值信息将塑造ATL中语义表示的特征。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award was provided as part of NSF's Social, Behavioral and Economic Sciences (SBE) Postdoctoral Research Fellowships (SPRF) program and SBE's Cognitive Neuroscience program. The goal of the SPRF program is to prepare promising, early career doctoral-level scientists for scientific careers in academia, industry or private sector, and government. SPRF awards involve two years of training under the sponsorship of established scientists and encourage Postdoctoral Fellows to perform independent research. NSF seeks to promote the participation of scientists from all segments of the scientific community, including those from underrepresented groups, in its research programs and activities; the postdoctoral period is an important level of professional development in attaining this goal. Each Postdoctoral Fellow must address important scientific questions that advance their respective disciplinary fields. Under the sponsorship of Dr. Daphna Shohamy at Columbia University, this postdoctoral fellowship award supports an early career scientist examining how semantic memory guides and is shaped by value-based decisions. Through everyday experiences, people learn facts about the world (e.g., learning about what fruits are ripe) and then use this knowledge to guide decisions (e.g., choosing ripe fruits to eat). This proposal aims to advance our understanding of how such world knowledge guides decisions and the neural mechanisms involved. The results will have important implications for education, which would benefit from a focus not just on teaching facts through single exposure but on acquiring abstract knowledge across multiple exposures of varying content. This abstract knowledge is especially useful since it can be generalized to novel situations outside the classroom. It may seem obvious that decisions are guided by semantic memory, but, remarkably, there has been no empirical work addressing the question of how this happens. The proposed two experiments will start to fill this gap by integrating behavior, computational modeling, and functional imaging (fMRI) to investigate the neural mechanisms by which semantic memory and value-based decisions interact to produce adaptive behavior. The first study tests the hypothesis that the value of choices depends on abstract knowledge accrued across multiple experiences with a semantic category. Participants will learn the average value across multiple items of a category to test if this average value is used to determine individual item value and guide decisions. The second study tests the hypothesis that the defining features of a semantic category are learned by making value-based decisions across multiple instances of the category. Participants will associate higher values to stimuli with features by doing a combined categorization and value learning task to test if feature value influences categorization. We hypothesize that semantic and value information are represented in distinct brain regions within the anterior temporal lobe (ATL) and ventral medial pre-frontal cortex (vmPFC), respectively. Category learning models will be used to quantify trial-by-trial variation in semantic information while reinforcement learning models will be used to quantify trial-by-trial variation in value information. This project predicts that semantic information from the ATL serves as input to value computation in the vmPFC while value information from the vmPFC shapes features of semantic representation in the ATL.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.