Semantic Cognition in Alzheimer's Disease: Neuroimaging Investigations of Representational Integrity and Controlled Retrieval

虽然情景记忆衰退是阿尔茨海默病（AD）的标志性症状，但语义知识的缺陷通常在疾病进展的早期就很明显。AD中这些语义缺陷背后的神经机制仍然知之甚少。事实上，它是未知的，是否语义表示本身是退化的AD，或是否有一个损害的访问，否则完整的表示。神经成像分析技术的最新进展-即代表性相似性分析（RSA; Kriegeskorte等人，2008; Charest等人，2014）和动态因果建模（DCM; Friston等人，2003年）-给了我们直接探索这些机制的工具。RSA评估神经激活模式的相似性，以揭示大脑区域的代表性几何形状。如果语义表征在AD中退化，那么RSA应该揭示已知存储语义表征的皮层区域中的表征几何学的系统差异。DCM评估了大脑区域之间的有效连接，以服务于任务表现，重要的是，这些连接是如何被任务需求调节的。如果AD中的语义缺陷是由语义表征的访问受损引起的，那么这在与语义控制相关的区域和与表征相关的区域之间的任务驱动的连接中应该是显而易见的。该项目将研究（i）在AD中支持语义控制和表示的脑区域之间的连接性如何受到影响;（ii）在AD中语义表示几何（行为，神经）如何受到影响;（iii）语义表示和检索的时空动态;以及（iv）这些措施在区分AD亚型中的应用（Stopford等人，2008年）或识别AD风险最高的人群。为了实现这些目标，我们将使用多模式神经成像（功能磁共振成像，脑电图）和认知行为的方法在老年人和非AD。学生将有机会设计具体的实验（参与者组、刺激、任务、成像模式、分析策略），以实现上述目标;然而，预计分析将包括评估代表完整性的RSA，评估控制区和代表区之间有效连通性的DCM，和多模态分析的功能磁共振成像和脑电图，以评估时空动态的语义检索。

Whilst episodic memory decline is the hallmark symptom of Alzheimer's disease (AD), deficits in semantic knowledge are often evident early in the disease progression. The neural mechanisms behind these semantic deficits in AD remain poorly understood. Indeed, it is unknown whether semantic representations themselves are degraded in AD, or whether there is an impairment of access to otherwise intact representations. Recent advances in neuroimaging analysis techniques - namely representational similarity analysis (RSA; Kriegeskorte et al., 2008; Charest et al., 2014) and dynamic causal modelling (DCM; Friston et al., 2003) - have given us the tools to probe these mechanisms directly. RSA assesses the similarities of patterns of neural activation in order to reveal a brain region's representational geometry. If semantic representations are degraded in AD, then RSA should reveal systematic differences in representational geometry in cortical regions known to store semantic representations. DCM assesses the effective connectivity between brain regions in service of task performance, and importantly, how these connections are modulated by task demands. If semantic deficits in AD are caused by impaired access to semantic representations, then this should be apparent in the task-driven connections between regions associated with semantic control and those associated with representation. This project will investigate (i) how connectivity between brain regions supporting semantic control and representation is affected in AD; (ii) how semantic representational geometries (behavioural, neural) are affected in AD; (iii) the spatio-temporal dynamics of semantic representation and retrieval; and (iv) application of these measures to differentiate subtypes of AD (Stopford et al., 2008) or to identify those most at risk for AD. To achieve these aims, we will use multimodal neuroimaging (fMRI, EEG) and cognitive behavioural methods in older adults with and without AD. The student will have the opportunity to design the specific experiments (participant groups, stimuli, task, imaging modalities, analysis strategy) in service of the objectives presented above; however, it is expected that analyses will include RSA to assess representational integrity, DCM to assess effective connectivity between control and representation regions, and multimodal analyses of fMRI and EEG to assess spatio-temporal dynamics of semantic retrieval.