Cognitive Map Theory

Lu, Leutgeb, et al. (2013)


Lu, Leutgeb, et al. (2013) Impaired hippocampal rate coding after lesions of the lateral entorhinal cortex. Nature Neuroscience, 16(8): 1085-1093

The idea that the hippocampus forms new memories by binding objects to space, the foundation of cognitive map theory, had seemed so vague and hard to imagine the first time I had read about it. Following up on Deshmukh’s chapter about lateral entorhinal cortical (LEC) processing, I read the following paper in which the mechanisms that may underlie the binding of dissociated object and spatial information in the brain are made much more clear. As mentioned by Deshmukh (2014), the segregated LEC and MEC inputs to the hippocampus converge on DG and CA3 cells. So how do these cells simultaneously process this information?

The answer: separate coding mechanisms. 

Previous studies (to be reviewed) suggested that object information appears to be encoded by rate codes in the hippocampus, while the place fields contained in the ensemble of active CA3 neurons conveys spatial information.

Lu, Leutgeb, et al. (2014) performed LEC lesions and measured ensemble activity in two environments differing in the shape of their borders. They found that LEC lesions impaired rate remapping in different environments, while control mice exhibited clear differences in firing rate. Both groups showed no change in cells’ preferred place fields, i.e. the preferred spatial location that increases activity in each cell. The size of the LEC lesion correlated with rate remapping impairment. Interestingly, only lesions of the intermediate, and not the dorsolateral or ventromedial LEC seemed to have significant impacts on rate remapping. Also, the population of active neurons overlapped more in lesioned animals and was most affected by intermediate lesions. Similar results were found when two environments differed in colors of the walls, rather than shape, which preserved spatial relationships between environments.

In a rather cool experiment, the authors how minor compared to more pronounced differences in the shape of two environments would impact rate coding. They found a gradient in remapping where minor shifts resulted in minor rate differences, whereas greater changes in environment shape resulted in greater differences in the rate codes of the two environments. Animals with LEC lesions, though demonstrating a similar gradient demonstrated shifts in the rate code that were of a smaller magnitude than those of controls. Again, LEC lesions preserved place preference of recorded neurons.

Lastly, the LEC itself demonstrated no rate remapping in response to different environments.

Thus, the authors provide evidence that the LEC is at most minimally required for spatial representations, while being necessary for CA3 rate remapping. With regards to my previous post asking where object and spatial information may be bound, the authors write: 

Although some object­related informa­tion is likely to already be integrated with spatial information at the stage of the interconnections between LEC and MEC, a stronger and more complete convergence may occur downstream in the hippo­campus, in CA3 and dentate gyrus where lateral and medial perforant path axons apparently synapse on dendrites of the same principal cells. This medial-lateral convergence, in conjunction with local network processes, may be necessary for determining whether and how much a hippocampal cell is active inside its place field on a particular occasion.

As this study relates to Deshmukh’s hypothesis about a dissociation of LEC and MEC by processing external and internal information respectively, indeed the changes in color and shape appear to be processed in the LEC in order for rate remapping to occur, however, there is no exclusion of the MEC in processing external sensory information in this study. 



Deshmukh (2014)


Deshmukh (2014) Spatial and Nonspatial Representations in the Lateral Entorhinal Cortex. In Space, Time, and Memory in the Hippocampus. Ch6 (pg. 127-152)

The primate visual system is functionally and anatomically divided into two processing streams. One processes the “what” of the visual field, e.g. the color of an object. The other processes the “where” of the visual field, such as movement of an object and distance of landmarks from each other and from the self. How exactly these two streams meet, and where exactly their information is bound, is unclear. It has been hypothesized that this segregation continues all the way through the entorhinal cortex, in which the medial entorhinal cortex (MEC) computes spatial representations, while the lateral entorhinal cortex (LEC) is responsible for object representation. There is, however, evidence that (more…)

Derdikman & Moser (2014)


Derdikman & Moser (2014) Spatial Maps in the Entorhinal Cortex and Adjacent Structures. In Space, Time, and Memory in the Hippocampus. Ch5 (pg. 107-126)

When the Cognitive Map Theory of the hippocampus was originally postulated, it was suggested that spatial representation was the job solely of the hippocampus. However, it has since been discovered that other structures, such as the entorhinal cortex, are integral for establishing spatial representations in the brain. Specifically, this chapter focuses on the medial entorhinal cortex, which, compared to the lateral entorhinal cortex, seems to be preferentially involved in spatial representation rather than object representation.

The entorhinal cortex is home to a particular cell type called grid cells. Grid cells are (more…)

Winter & Taube (2014)


Winter & Taube (2014). Head Direction Cells: From Formation to Integration. In Space, Time and Memory in the Hippocampus. Chapter 4 (pg 83-06)

This chapter focusing, on head direction cells, began the discussion on how the mammalian processes space and position in it.

Head directions (HD) cells fire most when (more…)

Derdikman & Knierim (2014)

Citation: Derdikman & Knierim (2014). Introduction: A Neural Systems Approach to Space, Time, and Memory in the Hippocampal Formation. In Space,Time and Memory in the Hippocampal Formation (pg. 1-23)

I’ve decided to take a step back from the trace studies and check out this book written by some of the most influential hippocampal researchers. This book was written specifically to allow newcomers to hippocampal research to catch up on the latest trends emerging from the systems view of the hippocampus.

Chapter 1 starts by pointing out that the Cognitive Map Theory (CMT) of the hippocampus is often misconstrued solely as a theory of spatial cognition, but is rather a theory of how the hippocampus is involved in episodic memory. Then, the authors provide an overview of the rest of the book and describe briefly what makes place cells (the physiological foundation of CMT) fire. The hippocampus receives input from the medial entohorinal cortex, an area that seems to convey information about space, and the lateral entorhinal cortex, which, for the most part, conveys object information (though whether or not this is the case will be discussed in a later chapter). Thus, the processing of the hippocampus is made possible through a complex cortical-hippocampal system that includes many areas of specialization, all of which, in the view of CMT, allow for the binding and association of spatial and object information into events which can then later be used to guide behavior in familiar situations. In addition to external sensory information regarding objects and landmarks, spatial computations involve internal input, such as from the vestibular system, to compute how an organism’s position in space changes over time in relation to the world around it.

Since most of the information in this first chapter will be discussed in detail later, I wanted to instead point out an interesting study mentioned about the role of NMDA receptor-dependent LTP and their role in place field formation. It was shown early on that NMDAR’s are necessary for spatial memory recall. However, Kentros et al. (1998) demonstrated that place fields could form in the absence of NMDAR activity, but that these fields would not bind to the cognitive map in the long run. Certainly, this would be an interesting read when revisiting Ryan et al.