Neural basis of spatial memory

Spatial memory is an essential part of our ability to function as an individual within our environmemt. For instance, if I were to walk into a darkened room that was very familiar to me, I would have a good idea where all the individual items in the room were so that I didn't walk into them. In order for me to be able to do this, I need to be able to remember what is in the room and where they all are in relation to my entry point, the door. In short, I need an internal map of the room. The map consists of three elements; what items are present, where are they in relation to each other and where am I in relation to both those items and the room in general. The generation and operation of such maps require the integration of multiple neural systems.

In order to recognise and recall what is present in the room, two neural systems come into operation. A recognition system based in the perirhinal cortex tells me that the items present are familiar to me. A second system that deals with the arrangements of individual items based in the hippocampus tells me how these items are arranged in relation to each other.

So how do we know where we are? How can we tell where we are going? How can we tell where everything else is in relation to where we are? The answer to these questions come down to a set of cells known as place cells.

What are place cells?

Place cells are cells that are tuned so that they are activated in response to the location within a particular environment. The basic characteristics of place cells can be demonstrated with the animation on the right. As an animal moves around any given space, a neuron becomes active in a given location. In this case, a mouse is moving around in it's circular cage, but it could be any animal in any space. A neuron is activated and fires action potentials when in the bottom right quadrant. This cell is a place cell and has a place field that is defined by the target. The closer to the centre of the place field the mouse is, the higher the rate of neuronal firing. In this way, the mouse can tell how close to the centre of the place field it is.

Different place cells have different place fields

There are many thousands of different place cells, each of which is tuned to a different place field, as shown in the animation on the left. The different neurons fire when the mouse is in different quadrants of the cage.

In addition to being tuned to different locations, the shape and extent of each place cell can also be different. The place fields shown so far are all circular, but this is only really true of cells that have place fields in open space, such as in the centre of the cage. Place cells that define the edge of a space, such as a wall, are more usually shaped to mirror that edge, in this case they will be crescent shaped. Place cells can also have more than one subfield, and can be large, filling much of the available space, or very small.

Using the firing of multiple, overlapping place cells, an internal spatial map of this particular environment can be constructed in which the pattern of firing of particular place cells gives the animal a location for its position. If the mouse is in the middle right section of the cage, neurons 1 and 3 will fire weakly, but if in the bottom left near the side, neuron 4 will fire strongly.

Place fields are not fixed in absolute space

The position of the place field is relative to spatial cues. The importance of such cues becomes obvious if you close you eyes, spin round a few times in the centre of a room and then try to reach, say, the door. You have no idea which direction to go in. However, if you reach an object you can recognise, it is then possible to work out where the door is in relation to the object you've just identified (the cue) and walk towards it. It is also true that if the cue is moved to a different location, you will then walk to the wrong place to find the door! This is because spatial information is processed using fixed reference points - move the reference points and you think you've also moved round. Exactly the same thing happens with place cells. A given place field is positioned relative to fixed cues. If these fixed cues are rotated by 90 degrees - the place field rotates with them.

Place fields are plastic but long lasting

If each place cell could only generate a single place field within a single spatial environment, then any animal would have a very limited number of places that it could count as familiar. This would make it difficult for an animal to fully exploit it's environment. So, each place cell has multiple place fields that are dependent on which spatial environment the animal finds itself in. For instance, you may have a place cell that is tuned to an open, central space in one room but may be tuned to an edge in another room. The place field of that cell will switch from central to edge and back to central, depending on which room you are in. The change can be as small as changing the colour of the walls, or changing the cues, as shown in the animation. This change in place field with changing environment is termed re-mapping, and vastly increases the number of spaces that can be mapped and thus viewed as familiar. The generation and expression of place fields is a very plastic process.

Place field re-mapping can be split into at least two separate processes

The re-mapping of a place field from an individual place cell can be sub-divided into at least two processes - the generation of the new place field and it's long-term maintenance (place field stability). It has been shown that NMDA receptors are involved in the latter of these two processes (Kentros et al 1998, y). In the presence of the NMDA receptor antagonist CPP, the place field of a given place cell re-maps as normal (Trial 1). However, when re-exposed to that same changed environment the following day (Trial 2), the same place cell re-maps to a different location. In other words, the cell cannot 'remember' being exposed to that environment previously and is treating it as novel. It thus generates a different place field.

Spatial memory is dependent on NMDA receptor-mediated processes

From work such as this, it is clear that the plasticity inherent in place field stability involves cellular mechanisms that are dependent on NMDA receptors, in particular those that contain NR2A and/or NR2B subunits, as receptors containing these subunits are preferentially blocked by CPP.

Further evidence of the involvement of NMDA receptor mediated processes in spatial memory comes from transgenic studies. The NR1 subunit is essential for NMDA receptor function and a number of models have been generated in which the NR1 subunit has been knocked out. For instance, the NR1 subunit has been knocked out in just the CA1 region of the hippocampus (CA1 conditional knockout). While these mice develop normally and are healthy, they have disrupted place cell function and find it difficult to remember the location of hidden items, such as a hidden platform in the Morris water maze. This is simply a tank of water in which mice are asked to swim to a platform that is just under the surface. Mice in which the NR1 subunit has been knocked out in CA1 cells take much longer to learn where the platform is than control mice. Pharmacological blockade of NMDA receptors with antagonists has a similar effect. However, in CA3 conditional knockouts, in which the NR1 subunit has been knocked out in CA3 hipocampal cells (these are adjacent to the CA1 cells), spatial memory is essentially normal, although there is some loss of complex firing patterns. So, specific neural circuits are involved in different aspects of spatial memory systems.

The specific molecular mechanisms that underlie these processes are still unclear, but it is likely that mechanisms like LTP and LTD are involved, as they may also be in many other forms of memory (Martin & Morris 2002, y).