PLACE CELLS

The hippocampus is a medial temporal lobe structure of critical importance for the encoding and retention of episodic memory in general and spatial memory in particular. A milestone in the detection of these hippocampal functions was the discovery that most of the pyramidal cells in the hippocampus exhibit location-specific activity (see Figure 1) and that the activity of such "place cells" is influenced by the training history of the animal. This chapter reviews place cells' governing mechanisms, their ensemble properties, and their possible contribution to spatial memory.

Historical Landmarks

Our understanding of hippocampal place cells rests on two important discoveries from the early 1970s. First, James B. Ranck reported that hippocampal neurones fall into two functionally different classes based on their firing patterns in freely moving rats. Complex-spike cells fired at low rates (normally < 1 Hz), but often in bursts of two to seven spikes at 150-200 Hz. These cells were likely to be pyramidal cells. Theta cells—the second class—had high spontaneous firing rates (normally > 10 Hz) and were probably inhibitory interneurons.

The second discovery was the observation by John O'Keefe and Jonathan Dostrovsky that a major proportion of complex-spike cells in the rat hippocampus had spatial correlates. These "place cells" fired when the rat was in a specific location (the "place field" of the cell) but were nearly silent in other positions (see Figure 1). Firing rates in the field were often ten to twenty times higher than the background rate. Place fields reflected location as such and not behavior emitted at specific locations. Different cells had place fields at different places, and collectively they covered the entire test environment (see Figure 1). Unlike the sensory cortices, firing correlates in the hippocampus were strongly nontopographic: place fields of neighboring cells in the hippocampus were not closer than those of distant cells.

These discoveries led to the suggestion that place cells form a distributed, maplike representation of the spatial environment that the animal can use for efficient navigation (O'Keefe and Nadel, 1978). These findings and the accompanying theory have strongly stimulated research on hippocampal function. New technology has expanded the study of place cells to neuronal ensembles (Wilson and McNaughton, 1993) and has the power to reveal how cognitive functions arise from complex interactions in defined neuronal networks.

What Factors Determine a Place Field?

Hippocampal place cells respond to multiple sources of sensory information (Best et al., 2001). Under most conditions, distal visual landmarks exert the strongest influence. When such landmarks are rotated in concert, place fields frequently follow the landmarks. When the test environment is stretched or truncated, there is often a corresponding change in the shape of the place field. Proximal landmarks such as surfaces usually exert weaker control over the place field. Place fields are also controlled by other sensory modalities. Many hippocampal neurons respond to distinct odors and are influenced by kinesthetic and vestibular cues generated by the rat's own movement. These influences are particularly powerful when visual input is unavailable or less reliable. When a rat is released from a closed start box at an unpredictable location, for example, place fields are controlled by the amount of movement for the first few seconds, before external visual landmarks take over.

Place cells, however, do not passively mirror the sensory input that the animal receives. First, place fields develop slowly. When a rat enters a new environment for the first time, place fields are weak and dispersed. Sharp and distinct place fields develop only after five to ten minutes, at a much slower rate than that at which sensory information passes to the hippocampus. Second, changes in place fields are not predictable from the amount of change in sensory input. Place fields often remain in place after significant landmarks are removed from the environment. At other times, two apparently identical environments may give rise to very different place representations, and subtle changes in sensory input may completely change the spatial firing correlates of a set of hippocampal neurons. Third, not all locations in an environment are represented by an equal number of place cells. In enclosed chambers, place fields appear to be more common near edges and walls than in the center; in environments with distinct reward locations, a larger number of cells may have place fields at the goal location than at other locations. All these observations suggest that the relation between the structure of the environment and the firing fields of hippocampal neurons is nonlinear.

Place Cells, Spatial Memory, and Synaptic Plasticity

Place cells appear to be responsible for some of the underlying memory computations of the hippocampus (Moser and Paulsen, 2001; Eichenbaum, 2001). Several observations suggest that individual hippocampal pyramidal cells express information retrieved from the animal's memory. First, as long as a rat remains in its recording apparatus, there is often no change in the location-specific firing of hippocampal place cells after the removal of surrounding landmarks or the switching off of lights. That this persistence of firing obtains even in cells that were originally under strong visual control suggests that recent experience influences firing patterns. Second, because a cell's firing or not at a given location may depend on where the animal comes from and where it is going next, it appears that recent memory can influence the activity of a place cell. Third, hippocampal area CA1 contains cells that apparently respond when experience is incongruent with predictions from memory, such as when a salient stimulus suddenly appears or disappears at a particular location. The existence of these mismatch-responsive cells means that CA1 cells may simultaneously receive information from the senses and from memory.

Once a place cell has developed a localized firing pattern in a new environment, the place field remains stable for weeks or more, as predicted if the cell contributes to a particular spatial memory. Some researchers have suggested that the formation of place fields, like spatial memory, depends on long-term potentiation (LTP) in hippocampal excitatory synapses. Blockade of the NMDA receptor abolishes both associative LTP and overnight stability of new place fields in a new environment. However, the development of place fields is not disrupted, and new place fields can be maintained for at least one to two hours in the absence of NMDA receptors, suggesting that LTP-independent mechanisms may be essential only for the long-term maintenance of place fields.

Place Cell Ensembles

Memory operations are reflected in the firing properties of individual hippocampal neurons; a more complete understanding of how place cells contribute to spatial memory requires insight into the constantly changing interaction between large numbers of place cells. Two phenomena—remapping and reactivation—illustrate the dynamic organization of place cells at the ensemble level.

Place fields of simultaneously recorded pyramidal cells exhibit all-or-none remapping. The entire population of recorded neurons may adopt new firing correlates after a change in a single but defining feature of an environment, such as the conversion of a square environment to a circular one while all other landmarks remain fixed. Some place cells start to fire at new locations, others become silent, and previously silent cells become active. The original map of place fields is accurately reinstated when the original environment is restored. This pattern suggests a linkage of place cells in functional ensembles that correspond to distinguishable test environments or test conditions. Each place cell is likely to participate in multiple ensembles that are active at different times.

A second example of ensemble coding is the striking observation that cells with overlapping place fields persist in correlated firing during sleep episodes subsequent to the behavioral session (Sutherland and McNaughton, 2000). Not only the pattern of coactivity, but also temporal firing sequences resemble those recorded during preceding behavior. This form of reactivation is temporally specific. It is strongest shortly after the behavioral session and decays with time. Reactivation is particularly associated with sharp waves, which are bursts of synchronous activity in hippocampal pyramidal cells during slow-wave sleep and awake rest. These bursts have the capacity to induce plasticity in downstream areas and may be involved in the consolidation of long-term memory in the neocortex (Buzsaki, 1989). Reactivation occurs during REM sleep, too.

There is also temporal organization of pyramidal-cell activity in the hippocampus. Hippocampal networks display characteristic oscillatory activity during spatial learning, and the timing of spikes relative to oscillations in the theta and gamma frequency bands may carry significant information. Such oscillations occur during memory processing, but the exact significance of temporal firing patterns for memory formation in the hippocampus is not yet understood.

Place Fields and Hippocampal Circuitry

Place cells exist in all subfields of the hippocampus and in the dentate gyrus, subiculum, and entorhinal cortex. The sharpest fields are in the hippocampus proper; firing fields in the subiculum and entorhinal cortex are more dispersed. Place-related activity in the hippocampus may result from sequential processing along the trisynaptic circuit of the hippocampus (dentate gyrus, CA3, CA1), or the relevant information is carried by the direct excitatory input from entorhinal cortex to each subfield. Disruption of the trisynaptic circuit by selective lesions in the dentate gyrus or CA3 does not abolish place fields in CA1, suggesting that the direct input is sufficient for establishing and maintaining spatial activity in hippocampal pyramidal neurons (Moser and Paulsen, 2001).

If place cells participate in neuronal ensembles that collectively contribute to memory of location and other episodic information, these ensembles must somehow be tied together. It has been suggested that place cells are organized as continuous attractor networks where neurons with firing fields at the current or nearby locations are mutually excited, whereas those with fields at other locations are inhibited in a distance-dependent manner. Distance between place fields of two pyramidal neurons may be encoded by the strength of the connecting synapses. Some researchers have suggested that the recurrent network of area CA3 has attractor properties, but recurrent networks in afferent structures such as the entorhinal cortex may have similar capacity. The CA1 lacks the internal excitatory connections needed to maintain ensemble structure.

Conclusion

Place cells carry strong signals that are expressed reliably in large proportions of the hippocampal neuronal population, and they are found in a part of the brain that plays clear roles in specific memory operations. With new powerful techniques that allow the study of neuronal computation at the ensemble level, the study of place cells can contribute to our understanding of the workings of memory and cognition.

Bibliography

Best, P. J., White, A. M., and Minai, A. (2001). Spatial processing in the brain: The activity of hippocampal place cells. Annual Review of Neuroscience 24, 459-486.

Buzsaki, G. (1989). Two-stage model of memory trace formation: A role for "noisy" brain states. Neuroscience 31, 551-570.

Eichenbaum, H. (2001). A cortical-hippocampal system for declarative memory. Nature Reviews Neuroscience 1, 41-50.

Moser, E. I., and Paulsen, O. (2001). New excitement in cognitive space: Between place cells and spatial memory. Current Opinion in Neurobiology 11, 745-751.

O'Keefe J., and Dostrovsky, J. (1971). The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely moving rat. Brain Research 34, 171-175.

O'Keefe, J., and Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon Press.

Ranck, J. B., Jr. (1973). Studies on single neurons in dorsal hippocampal formation and septum in unrestrained rats. I. Behavioral correlates and firing repertoires. Experimental Neurology 41, 461-531.

Sutherland, G. R., and McNaughton, B. L. (2000). Memory trace reactivation in hippocampal and neocortical neuronal ensembles. Current Opinion in Neurobiology 10, 180-186.

Wilson, M. A., and McNaughton, B. L. (1993). Dynamics of the hippocampal ensemble code for space. Science 261, 1,055-1,058.

Edvard I.Moser