Institute of Cognitive Neuroscience and Institute of Neurology, University College London.
J Exp Psychol Gen. 2013 Nov;142(4):1370-83. doi: 10.1037/a0033626. Epub 2013 Aug 5.
The hippocampus is thought to be an associative memory "convergence zone," binding together the multimodal elements of an experienced event into a single engram. This predicts a degree of dependency between the retrieval of the different elements comprising an event. We present data from a series of studies designed to address this prediction. Participants vividly imagined a series of person-location-object events, and memory for these events was assessed across multiple trials of cued retrieval. Consistent with the prediction, a significant level of dependency was found between the retrieval of different elements from the same event. Furthermore, the level of dependency was sensitive both to retrieval task, with higher dependency during cued recall than cued recognition, and to subjective confidence. We propose a simple model, in which events are stored as multiple pairwise associations between individual event elements, and dependency is captured by a common factor that varies across events. This factor may relate to between-events modulation of the strength of encoding, or to a process of within-event "pattern completion" at retrieval. The model predicts the quantitative pattern of dependency in the data when changes in the level of guessing with retrieval task and confidence are taken into account. Thus, we find direct behavioral support for the idea that memory for complex multimodal events depends on the pairwise associations of their constituent elements and that retrieval of the various elements corresponding to the same event reflects a common factor that varies from event to event.
海马体被认为是一个联想记忆“汇聚区”,将经历事件的多模态元素结合成一个单一的记忆痕迹。这预测了不同组成元素的检索之间存在一定程度的依赖性。我们呈现了一系列旨在解决这一预测的研究的数据。参与者生动地想象了一系列人物-地点-物体事件,然后在多次提示检索试验中评估对这些事件的记忆。与预测一致,从同一事件中检索不同元素时,发现存在显著的依赖性。此外,依赖性的水平对检索任务敏感,提示回忆时的依赖性高于提示识别,并且对主观置信度敏感。我们提出了一个简单的模型,其中事件被存储为单个事件元素之间的多个成对关联,而依赖性则由跨事件变化的共同因素捕获。该因素可能与事件之间的编码强度调制有关,也可能与检索时的“模式完成”过程有关。该模型预测了数据中依赖性的定量模式,当考虑到检索任务和置信度的猜测水平变化时。因此,我们发现了对复杂多模态事件记忆依赖于其组成元素的成对关联的直接行为支持,并且对应于同一事件的各种元素的检索反映了跨事件变化的共同因素。