认知与发展心理学研究室知识库

当生活中经历的某些事件存在重复的特征时，对这种相似记忆的成功辨别需要对该事件特异性细节信息的成功再现。经典的计一算理论模型指出海马的模式分离功能可以将相似的记忆表征彼此正交化，降低混淆和干扰，从而避免虚假记忆。最近，大量研究都一致的发现神经激活模式再现在对随后记忆的预测上，是一种较传统的激活强度更为有效和可靠的神经指标。然而，对于神经激活模式再现是如何支持相似记忆辨别的具体神经机制我们目前的理解还很少。此外，有研究发现重复学习相比单次学习，虽然在一般性的再认记忆上可以提高其行为表现，但是对相似的干扰材料再认辨别会出现更高的虚假再认。本研究采用功能磁共振成像技术慢速事件相关设计并结合表征相似性分析方法对上述问题进行探索。

我们的研究一采用编码一提取一致性作为神经激活模式再现的度量指标，发现它可以有效的支持相似记忆辨别，并且在排除了编码和提取阶段共同的知觉信息输入，提取阶段模式分离，全局匹配信号，以及编码和提取阶段各自的单变量的激活强度及激活强度变异等各种可能因素的影响后，包括海马，腹侧视觉皮层等脑区的项目特异性的神经激活模式再现仍然可以有效的支持相似记忆辨别，揭示了项目特异性的神经激活模式再现在相似记忆辨别中的核心作用。

研究二，首先我们发现了重复抑制会造成项目特异性神经激活模式再现的降低。通过偏相关和混合效应模型两种不同的分析方法，我们的研究揭示了前部海马重复抑制导致了该脑区的神经激活模式再现的降低，而在后部海马我们并没有发现类似的现象。在此基础上，我们首次从神经表征的角度探索了学习经验和海马长轴不同结构是如何与支持相似记忆辨别的神经机制交互作用的。我们的研究发现前部海马测验阶段对于编码阶段新异材料的神经激活模式再现可以支持相似记忆辨别，但无法支持重复学习的神经激活模式再现的相似记忆辨别;而后部海马在提取阶段对于重复学习材料的神经激活模式再现可以支持相似记忆辨别，但对材料新异时的神经激活模式再现并无法支持相似记忆辨别，我们的结果表明支持相似记忆辨别的神经表征会随着学习经验而沿海马长轴迁移。为了系统的探讨学习经验对于海马和皮层表征关系的影响，我们采用了一种新近发展的多元表征模式连接技术。我们的研究结果表明，后部海马和额顶高级脑区如左侧额下回的神经表征模式连接受到学习经验的显著的调控，表征连接关系随学习经验发生了显著的改变，我们进一步的分析表明，后部海马第一次学习的表征结构和左侧额下回第三次的表征结构计算的表征模式连接和随后相似记忆辨别个体差异成绩表现出显著的负相关。我们的研究揭示了重复学习经验不仅会导致海马内部表征的改变，同样也会导致海马和额顶高级脑区表征模式连接发生显著的变化。

研究三，我们揭示了不同脑区间的协同作用在学习和提取阶段对于相似记忆辨别的各自的不同影响。我们的研究提示了学习阶段后部海马和腹侧视觉皮层更强的功能连接预测了更好的随后相似记忆辨别成绩。此外，学习阶段内侧前额叶(以往研究中图式相关的经典脑区)在学习阶段和视觉皮层的更高连接强度同样可以支持更好相似记忆辨别成绩。而在测验阶段图式相关脑区内侧前额叶和角回分别为种子点的功能连接分析则提示对于图式相关脑区的依赖越高，则该材料越可能被错误的再认，揭示了在精细性的相似记忆辨别任务中更多的依赖于图式/主旨相关的抽象信息会对该任务造成显著的损伤。

The successful lure discrimination requires reinstatement of the specific details of the event. The pattern separation of the dentate gyrus in the hippocampus can be orthogonal to the similar representations, thus reducing the confusion and avoiding the memory interference. Extensive and converging evidence suggests that neural pattern reinstatement is a more effective and reliable predictor of subsequent memory than traditional univariate activation. However, we have little understanding of how neural pattern reinstatement supports similar memory discrimination. In addition, previous studies have found that compared to single learning, repetitive learning enhanced the general recognition and recall performance, but impaired similar memory discrimination.

This study is to explore the above problems by using slow event-related functional magnetic resonance techniques and combine the methods of representational pattern similarity analysis. Our study found that encoding-retrieval pattern similarity can effectively support similar memory discrimination, after eliminating many confounding factors, for example: the shared perceptual information input in the encoding and retrieval stages, pattern separation during recognition, global matching signal, and the activation intensity and its variance in the encoding and retrieval stages, separately. The encoding-retrieval similarity on the individual item level in distributed regions including the hippocampus and the ventral visual cortex, can still effectively support the successful lure discrimination, which reflects the critical role of the item-specific neural pattern reinstatement in the discrimination of similar memories. We found that repetition suppression could reduce the fidelity of neural pattern reinstatement on the individual item level. Via two different methods, the partial correlation analysis and the mixed effect model, we have verified that the anterior hippocampal repetition suppression leads to the reduced fidelity of neural pattern reinstatement in the anterior hippocampus, but similar phenomena are not found in the posterior hippocampus. Furthermore, we first explored how short-term learning experience and different hippocampal structures along long axis interact with the neural mechanisms supported similar memory discrimination from the perspective of neural representations: Our study found that learning experience modulates neural mechanisms underlay lure discrimination along the long axis of the hippocampus differently: the novel exposure' reactivation in the anterior hippocampus and the following exposures' reactivation in the posterior hippocampus underlie lure discrimination, respectively. Moreover, cross-region and repetition pattern connectivity analysis suggested that learning experience also induces posterior hippocampus and left inferior frontal cortex informational connectivity changes as a function of learning and the informational connectivity could predict subsequent lure discrimination performance.

We revealed the different effects of functional connectivity with different brain regions on the similar memory discrimination in the learning and recognition stages. Our study suggested that a stronger functional connectivity in the posterior hippocampus and ventral visual cortex during encoding phase was predicted better subsequent similar memory discrimination. In addition, the medial frontal lobe which was a schema processing region demonstrates that its stronger connectivity with the visual cortex during encoding also support better similar memory discrimination. In the recognition phase, using the medial prefrontal lobe and angular gyrus as a seed region separately, the functional connectivity analysis suggested that the higher connectivity with the above two seed regions, the more likely the material will be falsely recognized, our results revealed that relied more on the schema/gist related abstract information could cause significant damage to similar memory discrimination.