Mechanisms of spatial memory

在爬行动物、鸟类和哺乳动物中，海马体是空间认知（以及其他事物）的关键大脑结构。然而，经过几十年的研究，它的确切作用仍然是相当有争议的主题。我们的实验室一直在研究海马体在空间记忆和导航中的作用，使用各种技术，包括计算（神经网络）建模和虚拟现实（VR）环境中的人类记忆实验。我们开始了解人们依赖认知地图的条件，以及他们绘制这些地图的线索。提出的研究将进一步加深我们对这些过程的理解。我们的海马体记忆系统的计算模型将被扩展，以考虑到人们使用多种策略进行导航的事实。有些人倾向于认知地图策略，而另一些人则倾向于基于运动记忆而不是空间记忆的地标策略。“在超市左转”)。我们的模型将有助于阐明可能服务于这些复杂功能的神经机制。作为理论工作的补充，实验方法将用于测试模型的预测，并进一步阐明控制认知地图使用的条件。我们过去的研究表明，在空间记忆测试中，位于导航相关位置（如虚拟城镇的主要十字路口）的物体往往比其他物体更容易被记住，尽管在识别记忆测试中，其他物体的记忆力也同样好。如果物体被编码为认知地图的一部分，而不是简单地作为“视觉快照”或运动记忆的一部分被记住，那么当人们从一系列不熟悉的视角进行测试时，应该能够可靠地定位这些物体。此外，他们应该能够在长时间的延迟后记住物体的位置，而不是依赖于空间工作记忆和心理意象。从长远来看，我们的工作可能不仅对理解空间认知的基本机制有重要意义，而且对恢复空间记忆和导航缺陷的个体也有重要意义，这是与衰老有关的常见问题。

The hippocampus is a crucial brain structure for spatial cognition (among other things), in reptiles, birds and mammals. However, after many decades of research, its precise role is still the subject of considerable controversy. Our lab has been investigating the role of the hippocampus in spatial memory and navigation using a variety of techniques, including computational (neural network) modelling and human memory experiments in virtual reality (VR) environments. We are beginning to understand the conditions under which people rely on cognitive maps, and the cues they draw upon to form those maps. The proposed research will further our understanding of these processes. Our computational models of the hippocampal memory system will be extended to take into account the fact that people use multiple strategies for navigation. While some people tend to favor a cognitive map strategy, others favor a landmark strategy, based more on motor memory than spatial memory (e.g. "turn left at the supermarket"). Our models will help to shed light on the neural mechanisms that may subserve these complex functions. Complementing this theoretical work, experimental methods will be used to test predictions of the model and shed further light on the conditions governing the use of cognitive maps. Our past research indicates that objects at navigationally relevant locations such as major intersections in a virtual town tend to be remembered better than other objects on tests of spatial memory, even though other objects are remembered equally well on tests of recognition memory. If objects are encoded as part of a cognitive map, rather than simply being remembered as part of "visual snapshots" or motor memories, people should be able to reliably locate these objects when tested from a range of unfamiliar viewpoints. Moreover, they should be able to remember the locations of objects after long delays, as opposed to relying upon spatial working memory and mental imagery. In the long term, our work could have important implications not only for understanding the basic mechanisms underlying spatial cognition, but for rehabilitating individuals with spatial memory and navigation deficits, as are common with ageing.