关键词: 认知地图, 网格细胞, 位置细胞, 海马体

Abstract: The concept of “cognitive map” was first proposed by Tolman, who believed that cognitive map is the systematic knowledge organization of cross-domain behavior and is the basis of mental function. The key structure of cognitive map is thought to be the hippocampus, in which place cells fire when the organism is in a specific location, different place cells fire at different locations, and grid cells fire at multiple locations in physical space, providing precise spatial coordinates and the ability to represent vector relationships and distances between different locations. Place cells and grid cells provide the biological basis for the cognitive map of physical space.
Although researchers have reached a consensus on physical space mapping and its mechanisms, it remains to be confirmed whether abstract information can be represented in the form of cognitive spatial maps. Researchers have studied on different species (rats, monkeys, bats, or humans) in which different information (perception, memory, concept, or social) is constructed as cognitive spatial maps from different levels of study (cellular or brain regions). These studies suggest that cognitive space mapping has the nature of cross-information domains, and can process different levels of abstraction and types of cognitive information to form cognitive maps. However, it remains to be clarified whether these studies can support and corroborate each other. In addition, the current exploration of the mechanism of cognitive space mapping is still limited to the stage of analogy with physical space, and there is lack of in-depth exploration of the real mechanism of cognitive space mapping.
According to the degree and type of abstraction of information, cognitive space is divided into perceptual space, episodic memory space, conceptual space and social space. These findings support and corroborate each other, validating Tolman's conceptualization of a cross-domain cognitive map.
This paper also focuses on the neural mechanisms of cognitive space mapping, including the hippocampus and its coordination with other brain regions. On the one hand, the hippocampus abstracts and generalizes the hidden structure of cognitive space, which helps to generate distributed location maps and bundle stimuli with hidden contextual structures. On the other hand, the hippocampus works in concert with other brain regions. Specifically, the hippocampus integrates information from the sensory cortex. The information exchange between the hippocampus and the medial prefrontal cortex supports efficient and flexible cognitive spatial mapping. The functional synergies between the hippocampus and orbitofrontal cortex form different levels of information representation.
Finally, in view of the existing problems and limitations of existing studies, such as the possibility of other neural computation and encoding modalities in place and grid cells, the inability of cognitive space to reflect the precision and hierarchy of information, and the inability to track between the micro and macroscopic studies of cognitive space mapping, future research can deepen the understanding of cognitive space mapping and its underlying mechanisms from four aspects. First, future research should consider the possibility and rationality of applying the hypothesis of predictive cognitive map to cognitive space mapping, and explore how common computational and coding schemes for cognitive space mapping can compete and cooperate with models such as successor representation. Second, it is unclear whether place cells and grid cells at different scales in the hippocampus can support the representation of cognitive information with different precision and hierarchy. The information representation of the hippocampus can be explored by designing research paradigms for different cognitive spaces, improving the accuracy of fMRI signals, and exploring the gradient topological features of the hippocampal axis. Thirdly, more attention should be paid to the cognitive spatial mapping process of normal people, and attempts should be made to explore the activities and neural oscillations of spatial cells in the process of cognitive spatial mapping by using cellular recording, intracranial EEG recording, and magnetoencephalography and other technologies. Fourth, a comprehensive analogy between cognitive spatial mapping and physical spatial mapping should be made to clarify the commonality and specific mechanism of the two, and the overall research framework of cognitive mapping should be constructed.

Key words: cognitive map, grid cell, place cell, hippocampal formation