Krembil Neuroscience Program, Toronto Western Hospital, University Health Network, Toronto, ON, Canada.
Neuropsychologia. 2010 Sep;48(11):3272-81. doi: 10.1016/j.neuropsychologia.2010.07.010. Epub 2010 Jul 15.
Encoding and retrieval of relational information requires interaction between the hippocampus and various neocortical regions, but it is unknown whether the connectivity of hippocampal-neocortical networks is different at input and output stages. To examine this, we conducted a network analysis of event-related fMRI data collected during a face-recognition, remember/know paradigm. Directed analyses in the medial temporal lobe identified a small region in the left hippocampus that showed differential activation for encoding and retrieval of recollected versus familiar items. Multivariate seed partial least squares (PLS) analysis was used to identify brain regions that were functionally connected to this hippocampal region at encoding and retrieval of 'remembered' items. Anatomically based structural equation modeling (SEM) was then used to test for differences in effective connectivity of network nodes between these two memory stages. The SEM analysis revealed a reversal of directionality between the left hippocampus (LHC) and left inferior parietal cortex (LIPC) at encoding and retrieval. During encoding, activation of the LHC had a positive influence on the LIPC, whereas during retrieval the reverse pattern was found, i.e., the LIPC activation positively influenced LHC activation. These findings emphasize the importance of hippocampal-parietal connections and underscore the complexity of their interactions in initial binding and retrieval/reintegration of relational memory. We also found that, during encoding, the right hippocampus had a positive influence on the right retrospenial cortex, whereas during retrieval this influence was significantly weaker. We submit that examining patterns of connectivity can be important both to elaborate and constrain models of memory involving hippocampal-neocortical interactions.
关系信息的编码和检索需要海马体和各种新皮质区域之间的相互作用，但尚不清楚海马体-新皮质网络的连接在输入和输出阶段是否不同。为了研究这一点，我们对在面孔识别、记住/知道范式中采集的事件相关 fMRI 数据进行了网络分析。在颞叶内侧进行的有向分析确定了左海马体中一个小区域，该区域在编码和检索回忆与熟悉项目时表现出不同的激活。多元种子偏最小二乘法（PLS）分析用于识别在编码和检索“记住”项目时与该海马区域功能连接的大脑区域。然后使用基于解剖的结构方程模型（SEM）来测试这些两个记忆阶段之间网络节点的有效连接的差异。SEM 分析揭示了在编码和检索时，左海马体（LHC）和左顶下皮层（LIPC）之间的方向逆转。在编码过程中，LHC 的激活对 LIPC 有积极的影响，而在检索过程中则发现相反的模式，即 LIPC 的激活对 LHC 的激活有积极的影响。这些发现强调了海马体-顶叶连接的重要性，并突出了它们在关系记忆的初始绑定和检索/再整合过程中的相互作用的复杂性。我们还发现，在编码过程中，右海马体对右后扣带回皮质有积极的影响，而在检索过程中，这种影响明显较弱。我们认为，检查连接模式对于详细阐述和约束涉及海马体-新皮质相互作用的记忆模型都很重要。
