恒河猴内侧颞叶在短时记忆任务中对多个物体形态-位置信息的神经表征

The medial temporal lobe (MTL) has long been considered to serve exclusively for long-term memory including the episodic memory. However, recent neuropsychological studies as well as imaging studies suggest a critical involvement of MTL in short-term retention and even visual perception, particularly for multiple items with their complex features and locations beyond the capacity of working memory. In this application, we will investigate neural mechanisms supporting short-term memory in MTL using an item-location-retention task with two visual objects as a sample (2-ILR task) for non-human primates. In the 2-ILR task, the animals encode two visual objects and locations as a sample. Following a couple seconds of delay period, a cue stimulus is presented, and then the animals remember the location where the same visual object as the cue stimulus was presented as a sample stimulus in a trial. We will record single-unit activities and local-field potentials from the macaque MTL areas including the hippocampus (HPC), entorhinal cortex, perirhinal cortex (PRC) and parahippocampal cortex as well as neocortical control regions such as visual area TE and lateral prefrontal cortex (PFC). Our preliminary results showed that HPC and PRC represent item identity, location and their integration during the 1-ILR task requiring one item-location. In this application, we will examine which areas in MTL, TE and PFC represent multiple item-location information in 2-ILR task, and further investigate their neural circuits. In addition to the encoding and short-term retention, we will examine a retrieval process which mediates a transformation from the cue item information to the associated location information. The expected findings will provide the first neurophysiological evidence indicating the representation of multiple items in a scene, which would give significant implication not only to the study of episodic memory but also to the research field of working memory and visual neuroscience.

内侧颞叶（MTL）一直被认为仅参与情景记忆在内的长时记忆，但近期的神经心理和影像学研究发现它对短时记忆和视知觉也很重要，尤其在需记住多个复杂图形及其位置从而超出工作记忆容量时作用显著。本课题以多物体-位置短时记忆范式（2-ILR）和电生理手段，探究灵长类MTL的神经表征机制。2-ILR任务中，动物需记住两个物体及其位置（编码）并保持几秒（保持），其中一个物体再次出现时需回忆其位置（提取）。我们将记录海马，内嗅、围嗅皮质、旁海马和对照脑区如视觉TE区和外侧前额叶的单细胞和局部场电位活动。前期1-ILR结果显示，海马和围嗅皮质均表征物体、位置和二者的整合信息。接下来，我们将研究MTL、TE和前额叶如何编码和保持2-ILR中的物体-位置信息及与之相关的神经环路；也将深入理解2-ILR中物体信息转变为位置信息的提取过程。本课题将为多物体场景表征提供首个神经生理学证据，对视神经学和情景记忆均有意义。

总的来说，我们使用四只猕猴成功地完成了这个项目。特别是，我们在要求受试者保留一个物品及其位置的信息的“1-物品-位置-记忆（ILR）任务”中发现了令人惊讶的结果，这些结果更新了标准视觉神经科学的理念。在传统的双流理论中，腹侧通路和背侧通路分别处理与物体有关的信息（"什么"）和与空间有关的信息（"哪里"）。在这个项目中，我们发现腹侧通路不仅处理物体信息，而且还处理其背景图像。背景图像在视网膜框架中被表征，以表明受试者正在观看的位置。这种视中心背景信号从腹侧通路扩散到包含海马在内的所有内侧颞叶区域。此外，我们发现，视中心背景信号与物体身份信号相结合，构建了 "什么 "和 "哪里 "的联合信息。一些计算神经科学和人工智能研究试图通过多层神经网络的逐步处理来检测物品的身份和位置，而我们目前的模型将其解释为两个单独处理的信号之间的关联，这种视中心背景信号和物体信号的统一模型可能为相关研究提供新的见解。虽然我们花了更多的时间用几个控制任务（如被动编码）来澄清这些关键的发现，但我们也记录了“2-ILR任务”中的数据。为了在猴子生理学之前迅速掌握2-ILR任务中的大脑机制，我们采用了人类fMRI方法。结果表明，与前额叶皮层相比，海马自动地维持最后一个事件的信息，而非主动维持事件的序列作为工作记忆。这一发现可能解决了目前海马体延迟活动是否参与工作记忆的争论。

内容获取失败，请点击重试