Department of Psychology and Center for Brain Science, Harvard University, 33 Kirkland Street, Cambridge, MA 02138, United States.
Department of Psychology and Center for Brain Science, Harvard University, 33 Kirkland Street, Cambridge, MA 02138, United States; Department of Psychology and Neuroscience, Boston College, United States.
Neuroimage. 2021 Aug 1;236:118033. doi: 10.1016/j.neuroimage.2021.118033. Epub 2021 Apr 6.
Flexible retrieval mechanisms that allow us to infer relationships across events may also lead to memory errors or distortion when details of one event are misattributed to the related event. Here, we tested how making successful inferences alters representation of overlapping events, leading to false memories. Participants encoded overlapping associations ('AB' and 'BC'), each of which was superimposed on different indoor and outdoor scenes that were pre-exposed prior to associative learning. Participants were subsequently tested on both the directly learned pairs ('AB' and 'BC') and inferred relationships across pairs ('AC'). We predicted that when people make a correct inference, features associated with overlapping events may become integrated in memory. To test this hypothesis, participants completed a final detailed retrieval test, in which they had to recall the scene associated with initially learned 'AB' pairs (or 'BC' pairs). We found that the outcome of inference decisions impacted the degree to which neural patterns elicited during detailed 'AB' retrieval reflected reinstatement of the scene associated with the overlapping 'BC' event. After successful inference, neural patterns in the anterior hippocampus, posterior medial prefrontal cortex, and our content-reinstatement region (left inferior temporal gyrus) were more similar to the overlapping, yet incorrect 'BC' context relative to after unsuccessful inference. Further, greater hippocampal activity during inference was associated with greater reinstatement of the incorrect, overlapping context in our content-reinstatement region, which in turn tracked contextual misattributions during detailed retrieval. These results suggest recombining memories during successful inference can lead to misattribution of contextual details across related events, resulting in false memories.
灵活的检索机制使我们能够推断事件之间的关系,但当一个事件的细节被错误归因于相关事件时,也可能导致记忆错误或扭曲。在这里,我们测试了成功推理如何改变重叠事件的表示,从而导致错误记忆。参与者编码重叠的关联('AB'和'BC'),每个关联都叠加在不同的室内和室外场景上,这些场景在联想学习之前已经预先暴露。参与者随后在直接学习的对('AB'和'BC')和对之间的推断关系('AC')上进行测试。我们预测,当人们做出正确的推断时,与重叠事件相关的特征可能会在记忆中整合。为了验证这一假设,参与者完成了最后的详细检索测试,他们必须回忆最初学习的'AB'对(或'BC'对)相关的场景。我们发现,推理决策的结果影响了在详细的'AB'检索过程中引发的神经模式反映与重叠'BC'事件相关联的场景恢复的程度。在成功推理后,前海马体、后内侧前额叶皮层和我们的内容恢复区域(左侧下颞叶)中的神经模式与重叠但错误的'BC'上下文更相似,而在不成功推理后则不相似。此外,在推理过程中更大的海马体活动与我们的内容恢复区域中错误重叠上下文的更大恢复相关,这反过来又在详细检索过程中跟踪了上下文的错误归因。这些结果表明,在成功推理过程中重组记忆可能导致相关事件之间的上下文细节的错误归因,从而导致错误记忆。