Place Cells

Place Cells

Home
Psychology homework help
psychology paper
Neuron, Vol. 17, 979–990, November, 1996, Copyright 1996 by Cell Press

A Quarter of a Century of Place Cells Review

Robert Muller Neurons with properties similar to place cells are found in other parts of the hippocampal formation (see below),Department of Physiology but here the term “place cell” is reserved for the principalSUNY cells of CA3 and CA1.Health Sciences Center at Brooklyn

Functionally, place cells are characterized by loca-Brooklyn, New York 11203 tion-specific firing. A given place cell is intensely active only when the rat’s head is in a certain part of the envi- ronment called the cell’s “firing field” or “place field.”Background When the head is outside the field, the discharge rateWhen place cells were discovered 25 years ago by John is virtually zero. The strength of the signal is demon-O’Keefe and John Dostrovsky (1971), they came com- strated in Figure 1. At the top is a color-coded map ofplete with a theory. They were taken as evidence that the the positional firing rate of a robust place cell for a 16

rat hippocampus is the anatomical locus of a “cognitive min recording session; the circular area is an overhead

map,” a holistic neural representation of the environ- view of a cylindrical apparatus 76 cm in diameter with

ment that permits rats to efficiently solve spatial a 51 cm high wall. The firing field is the dark region near

problems. 7:30 o’clock. When the rat’s head is in the field, the By now, there is a consensus that place cells are real average firing rate is 8.3 action potentials per s (AP/s).

and that they are intimately involved with navigation. Outside the firing field, the cell fires rarely, and in yellow- Nevertheless, cognitive mapping may not be the only coded pixels the firing rate is zero. The two diagrams story. Even as the ability of hippocampal pyramidal cells at the bottom of Figure 1 show individual passes of the to represent place was being accepted, many groups rat through the field; the paths are black lines and action reported that pyramidal cells do not always act as place potentials are superimposed red dots. cells (e.g., Breese et al., 1989; Young et al., 1994) and Firing fields are cell specific. In a fixed environment, that some activities of pyramidal cells are hard to recon- each place cell has a stable field that is characteristic cile with a pure mapping system. We still do not know of the cell. The field centers are fairly evenly distributed if the rat hippocampus is best treated as a cognitive over the surface of the cylinder and similar apparatuses map with embellishments, or if cognitive mapping is a (Muller et al., 1987), although there may be some ten- specific, albeit very important example of a more general dency for fields to be more common near walls (Hether- function. ington and Shapiro, 1997). There also seems to be a

The position taken here is that cognitive mapping is tendency for fields to occur in front of prominent stimuli not the fundamental mode of function of the rat hippo- on the apparatus wall (Hetherington and Shapiro, 1997; campus, even if it is the predominant mode. Neverthe- Fenton and Muller, 1996, Soc. Neurosci., abstract). less, I focus on place cells and navigation rather than One explanation of the restriction of firing to the field

is that unit discharges occur in association with a certainon explaining how the complete range of pyramidal cell behavior, a behavior that is emitted only when the ratactivity reflects a more general kind of processing. The is in the field. The conditions of the recording in Figurereason for this is pragmatic: the strength of the signal 1 permit this hypothesis to be rejected. The recordingmakes the study of place cells by far the best approach was made as a well trained hungry rat retrieved foodto discovering the overall function of the hippocampus. pellets randomly scattered into the cylinder. Since theIt is easy to design experiments to test place cell proper- rat spent almost all its time running around, its behaviorties. Moreover, the manipulations in such experiments was quite constant in time (Bostock et al., 1991). There-are readily understood, and it is easy to see how fore, the location-specific firing cannot be attributed tochanges in place cells are associated with changes in a specificbehavior that is executed in a location-specificthe environment. In short, place cells are a good window fashion. Even if the same behavior is generated con-into the mysteries of the hippocampus, because the stantly, rats tend to spend more of their time in certainproperties of place cells are readily related to the geome- parts of the apparatus than in others. There is, however,try of the space in which the rat exists. no tendency of fields to preferentially occur in regions where the total time spent is greater or less than the

Basic Properties of Place Cells average. In addition, the discharge of individual cells Anatomically, place cells are pyramidal cells of the hip- is on the average uncorrelated with the time spent in pocampus in rats (O’Keefe, 1979) and mice (Rotenberg different parts of the apparatus. et al., 1996, Soc. Neurosci., abstract; McHugh and Wil- It is interesting that robust place cells are seen in son, 1996, Soc. Neurosci., abstract). They are found a task that imposes hardly any demand on the rat’s in both the CA3 and CA1 regions of the hippocampus navigational abilities. It appears, therefore, that a hippo- (O’Keefe, 1979). Most recordings of place cells have campal representation of the environment is formed been from the dorsal (septal) hippocampus. Recently, even when it is not needed for adequate performance. place cells were also found in the ventral (temporal) As predicted by O’Keefe and Nadel (1978), a map may hippocampus, implying that the entire structure partici- be generated during exploration, and curiosity may be

a sufficient motivation.pates in mapping (Poucet et al., 1994; Jung et al., 1994).

Neuron 814

Figure 1. Summary of the Firing Properties of a CA1 Hippocampal Place Cell

At the top is a “firing rate map” that shows the time-averaged firing rate of the cell as a function of the rat’s head position. The col- ored circular region is an overhead view of a 76 cm diameter cylinder. Each small square region (pixel) is about 2.5 cm squared. The color in each pixel encodes the firing rate in that region. The firing rate is simply the total number of spikes fired in the pixel divided by the total time spent in the pixel. The firing field is the dark region at about 8:30 o’clock, near the apparatus wall. Higher rates of firing are encoded by darker colors; the firing rate in the median pixel for a color category is given by the key to the right. In the color code, yellow indicates pixels in which the firing rate was exactly zero (i.e., no spikes were fired there). The gray pixel indicates the “field cen- ter”; this is the pixel in which the rate (aver- aged over a 3 by 3 square of pixels) was highest. The rate map was obtained by recording from the place cellas a hungry rat ran around for 16 min in the cylinder chasing small food pellets. About 90% of the rat’s time was spent run- ning; the remainder was mainly occupied by eating the food pellets. The haphazard run- ning induced by the pellet chasing ensures that a properly trained rat will visit each part of the apparatus many times during the 16 min session. The two maps at the bottom show the spike activity on two separate pas- ses through the field. The black line indicates the rat’s path and the red dots the locations at which action potentials were fired. The gray pixels indicate the location of the firing field, copied from the rate map. Note that action potentials were fired all along the second

path even though the rat turned and ran out of the field in the direction opposite to its entry; this is an indication that the firing is not directionally selective. It is also interesting that spike density is higher during the first path than the second; this is an indication of the great temporal firing variance showed by place cells (Fenton and Muller, 1995, Soc. Neurosci., abstract).

On the Signal Carried by Place Cells well known example is the calculation of movement di- rection of a monkey’s hand from an average of the direc-Since the time-averaged firing rate of a given place cell tions associated with the maximum firing rates of a setis a strong function of the position of the rat’s head in of neurons in motor cortex (Georgopolous et al., 1982).the environment, the time-averaged across-cell firing Across-cell schemes for computation of position haverate profile must be in great part determined by head been proposed by Muller et al. (1987), O’Keefe (1991),position. The quite strict mapping of location onto the and Wilson and McNaughton (1993).across-cell firing rate profile makes it natural to ask if the

This simple picture assumes that each cell has oneinverse relationship is true: can location be accurately and only one field and that the activity of each celldetermined from the conjoint firing of many place cells? signals proximity to a single place. What would happenIntuitively, the answer is yes. The discharge of a single if some place cells had two (or more) fields? It seemsplace cell is insufficient to accurately locate the animal, clear that their activity would reduce the accuracy bybut the firing of many cells could signal location with which location is computed from population activity.

great precision. How might location be computed? Cells The magnitude of the problem then depends on the

with field centers at the current location discharge most number of cells with two fields. If there are few, one can

rapidly and cells with field centers increasingly far from imagine that location is signaled accurately. In contrast, the current location fire more slowly. The current loca- if most place cells have multiple fields, a simple tion could be computed as the average position of the weighted average of field centers would usually signal field centers of active cells. The weighting of a field in an incorrect location. the average grows as the peak rate of the cell grows. There is good evidence that single place cells can Encodings of this type, in which a value is computed have two fields, but the fraction seems to depend on according to a weighted average of something signaled the apparatus. In the cylinder illustrated in Figure 1,

about 5%–10% of cells have two fields (Muller et al.,by each cell may be called “across-cell” schemes. A

Review 815

1987). In walled apparatuses with open centers much different modes of representing the environment? It is interesting to speculate that omnidirectional firing islarger than the rat, two-field cells may therefore be too

rare to seriously compromise locational computations. used to represent behaviorally important regions such as a foraging ground, the vicinity of a water source, orIn contrast, place cells are often active on two or more

arms of an 8-arm maze (McNaughton et al., 1983; Jung the home territory, and that the directionally selective mode may be used to represent trails that connect re-and McNaughton, 1993; Muller et al., 1994).

A second difficulty with the across-cell hypothesis gions. In this view, thedifference in directional selectivity reveals that rats may do two very different kinds ofarises from the temporal firing properties of place cells.

Specifically, the moment-to-moment discharge of place navigation. cells cannot be accurately predicted from the time-aver- aged firing distributions, even assuming that discharge Stability of Firing Fields occurs according to a random (Poisson) process. (In Place cell recordings are usually made in a continuous this model, the mean for the Poisson process is the “session” of, say, 16 min that starts when a rat is put time-averaged rate in the pixel the rat is currently in; the in an apparatus and ends when it is removed. Almost mean changes as a function of the rat’s position and always, firing field location is stable during sessions that firing is generated by an “inhomogenous” Poisson pro- last for minutes or even hours. Firing fields are also cess.) The number of action potentials fired on individual stable across sessions separated by hours, weeks, or passes of the rat through the firing field is extremely months (Muller et al., 1987; Thompson and Best, 1990). variable, even for passes that are closely matched for They are stable regardless of whether the rat spends all the sequence of regions along the pass and for the time its time between sessions in its home cage or some of spent in each region. It is possible for a place cell with its time in a different recording apparatus. If sessions a robust firing field to be absolutely silent on a pass are run in two or more apparatuses over extended times, through the field, even if the rat goes through the field fields arestable in each (Muller and Kubie,1987; Thomp- center (Fenton and Muller, 1995, Soc. Neurosci., ab- son and Best, 1989). The long term stability of firing stract). The extreme temporal variability of place cell fields implies that the representation is recalled and discharge implies that very many cells are likely to be not created de novo each time the rat enters a familiar necessary for the across-cell computation of position to environment. The stability of different fields in different be accurate. Alternatively, the temporal variability may familiar environments implies that many representations mean that place cells may carry an as yet undetected can be stored without interference. signal in addition to the positional signal. The variability Here is a form of memory in which each of several also implies that place cellsare not driven by the smooth representations is latent unless the rat is in the environ- summed effects of many small depolarizations; the con- ment for which the representation is tailored. Recall of version of magnitude of depolarization to firing fre- the representation of a familiar environment presumably quency carried out by alpha motor neurons is not an reflects successful recognition of the environment-spe- accurate model for firing fields. Instead, place cell dis- cific stimulus pattern. It has been proposed that recog- charge likely is caused by synchronous discharge along nition is accomplished by the hippocampus itself, a subset of the inputs to each pyramidal cell. operating on highly processed sensory information sup-

plied via entorhinal cortex (McNaughton and Morris, 1987; Rolls, 1989; McNaughton and Nadel, 1990; KubieDirectional Firing Properties of Place Cells and Muller, 1991), but see Recce and Harris (1996) forIn apparatuses like the cylinder, place cell firing depends a different notion.only on the position of the rat’s head and is virtually

independent of the direction in which the rat’s head is pointing; the firing is not directionally selective. This Quantitative Properties of Firing Fields

Field Shapesresult falsifies the “local-view” theory of place cell dis- charge, according to which place cells are triggered by In a walled apparatus with an open floor, firing fields

away from the walls are circular or elliptical, whereasthe visual stimuli that are seen at a certain place in the environment while the rat points in a certain direction those near the wall tend to follow the wall (Muller et al.,

1987; O’Keefe and Burgess, 1996). Fields on the sides(McNaughton et al., 1991). Under other circumstances, place cell firing is direc- of rectangular apparatuses are often linear; fields at the

edge of a cylinder are usually crescent-shaped. It is hardtionally selective, e.g., when a rat runs in one direction or the other along a linear track (McNaughton et al., to see how extended fields would be included in an

across cell, since there is no unique field center whose1983; O’Keefe and Recce, 1993). An individual cell can be nondirectional in the cylinder and directionally selec- position can contribute to the computed position of the

animal.tive on an arm of an 8-arm maze (Muller et al., 1994). Directional selectivity can be different in a single appara- Field Size

Although activity is strongly confined to the firing field,tus if the behavioral task for the rat is changed (Markus et al., 1995). The variable nature of place cell directional the probability that the cell will fire is greater than zero

over the whole apparatus surface. Accordingly, unlessselectivity has been reproduced in a competitive learn- ing model (Sharp, 1991). a minimum rate is chosen for a pixel to be included in

the field, the field size will always approach the sizeAre nondirectional and directionally selective firing fields different only in detail, or are they fundamentally of theapparatus. Setting the threshold to 1.0 AP/s (about

Neuron 816

the time-averaged rate for place cells over the whole al., 1994). How can we then account for the conditional apparatus), theaverage field isabout 13% of the appara- partial control exerted by sensory stimuli? That stimuli tus area; the range is between about 3% and 50%. are salient but not prepotent? Part of the answer seems Rates and Rate Contours to be that a hippocampal map of the environment has a Although fields may be elongated, well sampled fields certain integrity. Once activated, a map tends to remain generally have a single maximum rather than two or active, even when major changes are made in the envi- more local minima. Peak in-field rates range from about ronment. Two different maps may be used to represent 40 AP/s to 5 AP/s or less (Muller et al., 1987). Rate a single environment, and which one is used is deter- surfaces resemble mountains rather than plateaus. Con- mined by the recent experience of the rat. For example, tours get smoother as recording time increases, as ex- one map is used if the rat is put into a familiar environ- pected if fields are stable. In general, well sampled fields ment in the light and another if it is put in the same have a single maximum, rather than two or more local environment in the dark. Critically, if the lights are put maxima. on after the rat is introduced in the dark, the map does

The significance of variations in field properties is not revert to the light-first case (Quirk et al., 1990). unclear. The variations in field size and intensity might In this view, therefore, the map is not a collection of be just Gaussian noise around population means. Alter- independent place cells but is instead a stable unit, a natively, variations in field size might mean that the envi- sort of very large “cell assembly” (Hebb, 1949) whose ronment is represented at different resolutions, and vari- activation may persist despite strong perturbations in ations in field intensity might reflect a hierarchy of place the sensory information from the outside. Many changes cells with similar or overlapping firing fields. Variations in the environment will not cause the positional firing in field shape are of special interest, since they suggest patterns of individual cells to change independently of that boundaries are recognized and treated in special each other; the relative locations of firing fields will re- ways. main constant until the environment changes “enough.”

At this point, the positional firing patterns of all cells Sensory Control Over Place Cell Firing and their relationships to each other in two-dimensional Why do place cells fire only in certain places? As we space will change all at once. We refer to such a change have seen, the firing fields are not simply consequences as a “complete remapping,” a concept developed more of the rat tendency to do certain things in certain places.

fully below, along with the ideas of “partial remappings” An alternative explanation is that each place cell is trig-

and “null remappings.” gered by a combination of stimulus features, so that

O’Keefe and Speakman (1987) made a fascinating ob- most cells fire in only one place because the combina-

servation relating to sensory control. Rats were taught tion occurs at that place in the environment. This may

to use a set of distal cues to get food at the end of one be referred to as the “sensory hypothesis.”

arm of a 4-arm maze, regardless of which arm theyOne piece of evidence against the sensory hypothesis started on. During “memory trials,” the cues were pre-is the direction-independent firing of place cells in unob- sented and then withdrawn.Even in the absence of cues,structed regions. Place cells seem to signal position rats went to the correct arm, as if the cues were stillitself and arenot merely triggered by thestimuli available present. On some trials, however, the cues were neverat a certain place as the rat looks in a certain direction. presented,so that there was no correctgoal location. OnThere is also direct evidence against the sensory model. such trials, the location of cell firing could be accuratelyThe angular position of a single white cue card on the predicted according to where the rat went, which iscylinder wall accurately determines the angular position presumably the rat’s “goal” choice. In the absence ofof firing fields. When the card is rotated to a new posi- adequate information, the goal choice presumably indi-tion, the field rotates equally, showing that the card cates how the rat thinks its environment is oriented. Theexerts a reliable form of stimulus control over firing observation that the firing is in register with the choicefields. suggests that place cell discharge is causally associatedWhat happens if the unquestionably salient cue card with the rat’s overall behavior during spatial problemis deleted? The prediction from the sensory hypothesis solving.is that positional firing patterns will be disrupted; firing

might become annular or homogeneous or might cease. Silent Place Cells, Active Subsets, and RemappingIn fact, after cue removal the fields remain intact (Muller A robust place cell in one environment may be virtuallyand Kubie, 1987; O’Keefe and Speakman, 1987). Fields silent in a different environment (O’Keefe and Conway,often rotate to an unexpected angular position, but are 1978; Kubie and Ranck, 1983; Muller and Kubie, 1987;otherwise unchanged. Experiments of this type support Thompson and Best, 1989; Bostock et al., 1991; Wilsonthe idea that place cells are part of a stored representa- and McNaughton, 1993). In any given environment,tion of the environment that can be recalled as a whole about half the pyramidal cells act as place cells andeven if some parts of the environment are not currently about half are silent, although a careful study by Thomp-present (McNaughton and Morris, 1987; O’Keefe and son and Best (1989) reports a considerably larger frac-Speakman, 1987). Thus, pattern completion appears to tion of silent cells.be a feature of the memory used to reactivate the repre-

Just as place cells reliably have the same firing fieldsentation of a familiar environment. each time the rat enters a certain familiar environment,Other experiments also argue against the idea that silent cells are reliably silent in a given environment.place cells are triggered in an absolute fashion by con- This suggests that every “sufficiently different” environ-junctions of specific stimuli (Sharp et al., 1990; Speak-

man and O’Keefe, 1990; Quirk et al., 1990; Markus et ment is represented by an independent subset of place

Review 817

cells chosen, with replacement, from the pyramidal cell take on anyvalue. Since neighborliness is not preserved, there cannot be a topographic representation of bothpopulation (Bostock et al., 1991; Kubie and Muller, 1991).

The cells used in a given environment are called the environments (Kubie et al., 1992, Soc. Neurosci., ab- stract). In fact, assuming that no single environment is“active subset” to distinguish them from the silent cells.

The active subset is the cellular analog of the map of preferred, it is arguable that no environment is topo- graphically represented (Kubie and Muller, 1991).an environment and characterizes the environment. The

place cells that aredischarging at any moment signal the It therefore seems that the relationship between cells in the hippocampus and the mapped surface is verycurrent location in the environment. The active subset is

what is reactivated whenthe rat is putback into a familiar different than for visual, somatosensory, auditory, and motor cortices, and it is natural toask why this differenceenvironment. If about half the pyramidal cells are active

in any given environment, the number of environments exists. A possible reason comes from considering the relationship between the mapped surface and the corti-that can be independently represented is maximized.

There is an important concomitant of the silent cell cal area. For the sensory and motor areas, the mapped surface is invariant—the size of the retina or the area ofphenomenon. If two environments have different active

subsets, the firing fields of those cells that are active in the skin does not change except in development. In contrast, the hippocampus must represent parts of thebothenvironments are generallyunrelated toeach other.

If all place cells in one environment are either silent world that are very different in size. Imagine, for in- stance, a topographic mapping of a region large enoughor have unrelated fields in a second environment, one

representation may be referred to as a “complete re- that all the cells of the hippocampus are involved. There then may be two ways in which a smaller region mightmapping” of the other (Bostock et al., 1991; Kubie and

Muller, 1991). In a “null remapping,” the active subsets be represented. In one, only those cells that represent the central part of the large region might be used; cellsare the same and the direction and distance from the

field of any cell to the field of any other cell is unchanged. that represent the peripheral part of the large region would be silent. This possibility seems implausible, andThere is currently no theory for predicting if a given

environmental change will produce a complete or a null in any case is contradicted by the empirical observation that the same number of cells is used to represent re-remapping. The complexity of the problem becameclear

when it was asked if merely changing the visual appear- gions that differ in area by a factor of four (Muller and Kubie, 1987). The second way is that the same cell setance of a cylinder could cause a complete remapping

(Bostock et al., 1991). In this experiment, the only manip- is used, independent of the size of the mapped region. In this model, the resolution (firing field size) of the repre-ulation was to replace a white card with a black card.

Stable complete remappings were seen for about half sentation would be proportionally lower in larger areas. This possibility also does not hold, since firing field areathe rats and null remappings for the others. Presumably,

this means that half the rats “judged” the two environ- does not scale up equally to the scaling of apparatus size.ments to be quite similar and the other half judged the

environments to be quite different. That rats show indi- The size of the mapped region is not the only problem with imagining a topographic mapping of the environ-vidual differences with regard to remapping strongly

suggests that prediction rules will be very complex and ment onto the hippocampal surface. Similar difficulties arise if the shape of the apparatus is changed; eitherdependent on the experience of the animal. The matter

is further complicated because partial remappings are some cells must be unused or there must be topological distortions of the mapped area. There is also the prob-possible (see the next section for examples.)

Remapping has an important implication for the over- lem that there is no constant angular relationship be- tween the mapped area and the hippocampal surface;all organization of the hippocampal representation. Just

as it is attractive to imagine that place cells are high the map could be in register with the environment only at one head orientation. Certainly, mental rotations areorder sensory cells, it is also attractive to think that

there is a point-for-point topographic projection of the possible, but the increased latency for larger rotations would mean that it would take longer to use the maphippocampal surface onto the two-dimensional surface

of the environment. But, just as place cells appear not starting at certain orientations than at others. Even though topographic mappings are common into be high order sensory cells, so there is little reason

to think that proximity of cells in the pyramidal cell layer sensory and motor systems elsewhere in the brain, there bears any strong systematic relationship to the proxim- is a great difference between either a sensory receptive ity of the corresponding firing fields in the environment. surface or a set of muscles compared with a surface in

What evidence is there against a topographic map? the external world. Sensory surfaces and muscle sets One fundamental observation is that cells with widely are connected to the cortex via a set of relatively stable separated fieldscan be recorded from single electrodes. connections, and it is clear that neighborliness-preserv- Neighboring cells can have fields that are as far apart ing mappings can be used. In contrast, features in the as possible in the environment (Muller et al., 1987). An external world are extremely variable and there is no equally fundamental finding is that the distance or direc- way to match a priori an arbitrary environment to hippo- tion between firing fields of pairs of cells is not preserved campal cortex. in a complete remapping. Consider a pair of cells with overlapping fields in one environment. When the same

Partial Remappingscells are recorded in a second environment (and neither In a null remapping, the relative positions of firing fieldscell becomes silent), the fields may be any distance

apart and the direction from one field to the other may are all unchanged. In a complete remapping, the relative

Neuron 818

positions of fields (for cells that have fields in both envi- the pattern. Interest is currently focused on how firing ronments) are all different except for happenstance. It probability varies with the phase of the “theta” rhythm. is possible also to recognize intermediate circum- Theta (also called rhythmical slow activity [RSA]) is a stances, in which the relationships among fields are 5–12 Hz sine-like large amplitude (up to 2 mV) variation preserved for some but not all cells. One example arises of the voltage difference recorded across the CA1 pyra- in investigating how barriers influence place cell activity, midal cell layer; similar large amplitude signal can be barriers being of interest for navigation because they recorded with other electrode placements. Theta is syn- force the animal to detour around a region it could pre- chronous in much of the hippocampal formation, includ- viously cross. In general, discharge is strongly sup- ing the entire length of the hippocampus itself (Buzsaki, pressed when a barrier is placed so as to bisect a firing 1989). field. In contrast, firing is unaffected if the barrier did Modulation of pyramidal cell firing by theta phase was not encroach on the field, even though the barrier reported by Ranck (1973). More recently, O’Keefe and changes the views from the region of the field. Thus, Recce (1993) found a more complicated relationship the effect of barriers is local and the relationships among in which the theta phase at which a place cell fires most firing fields are preserved even though cells in the systematically shifts as the rat passes through a firing vicinity of the barrier are effectively uncoupled from the field. O’Keefe and Recce suggest that phase precession rest of the network. This barrier effect fits very nicely is a fundamental mechanism for sharpening the posi- with a topographical map model. When the model is tional resolution of place cell firing. applied to unobstructed space, it generates straight line The strength of the place cell signal is greater during paths from any starting position to any goal. When the theta than during large irregular activity (LIA); the ratio barrier effect is included, the model immediately gener- of in-field to out-of-field firing was greater because the ates efficient realistic-looking detour paths around the in-field rate is greater during theta and the out-of-field barrier with no need for relearning (Muller et al., 1996a). rate is higher during LIA (Kubie et al., 1985). The in- Partial remappings are also seen if a cylindrical or rect- creased out-of-field rate during LIA may be associated angular apparatus is scaled up in size (Muller and Kubie, with hippocampal sharp waves (Chrobak and Buzsaki, 1987) or the size or aspect ratio of a rectangular box is

1996). A critical area for future work is to understand changed (O’Keefe and Burgess, 1996).

how the functional reorganizations of the hippocampus revealed by EEG state changes are related to navigationWhy Two Sets of Place Cells? and navigational control over locomotion.

CA3 and CA1 place cells have nearly identical properties (Muller et al., 1987; Markus et al., 1995). It is instructive

Development of Place Cell Firingto consider two different reasons for which many prop- erties might be the same. One reason is that CA3→CA1 Place cell discharge develops rapidly when a new envi- synapses are unimportant in causing CA1 place cells to ronment is explored and learned. Hill (1978) recorded fire. In this view, the positional information for CA1 ar- place cells when rats were put in an unfamiliar T-maze. rives directly from layer III of entorhinal cortex and is Cells fired in the stable field the first time the rat passed separate from the positional information needed to es- through the region. Bostock et al. (1991) watched the tablish CA3 place cells. One might speculate that the development of a new discharge pattern while investi- CA3→CA1 synapses are the basis of comparator that gating remappings caused by changing a cue card from checks if two the representations are in register. Modifi- white to black. The change was rapid. There was usually cations of the synapses via long-term potentiation might a 2–3 min latency before the first discharge in thestation- operate to bring them into agreement. ary field occurred. Wilson and McNaughton (1993) re-

An alternative reason is that direct entorhinal cortex corded simultaneously from many cells when a barrier inputs to CA1 are ineffective, and that CA1 place cells was raised to admit the rat into a new section of a are driven by CA3 place cells. In this case, the group of rectangular box after running in the original section. CA3 cells that are effective in driving a certain CA1 cell Discharge began in the new section with a latency of a must have overlapping or coincident firing fields. If the few minutes, in agreement with Bostock et al. (1991). effective CA3 cells had fields scattered over the surface Discharge in the new section of the box was mainly of the apparatus, the positional firing distribution of the confined to units that were silent in the original section. CA1 cell would have to be flatter than for CA3 cell. This result is another indication that the mapping be-

The parallel arrangement of two input sources to CA1 tween hippocampus and environment is not topo- is also seen in CA3 and the subiculum. Thus, CA3 re- graphic (Wilson and McNaughton, 1993). ceives input from both dentate granule cells and directly from entorhinalcortex, and the subiculum receives input

Do Hippocampal Pyramidal Cells Signalfrom both CA1 and entorhinal cortex. A fundamental More than Place?goal of place cell research is to understand the signifi- Several groups have reported that pyramidal cells cancance of the dual input arrangements in many parts discharge in relationship to nonspatial variables. Theseof the hippocampus. Readers interested in this issue observations present a fundamental challenge to a pureshould read the extremely interesting theoretical paper mapping theory of hippocampal function. For instance,by Treves and Rolls (1992). pyramidal cell discharge may be correlated to a variety of aspects of a complex task, including approach to anyPlace Cells and the Hippocampal EEG one of the four water cups in the corners of a rectangularThe hippocampal EEG exhibits several distinct patterns

in the awake rat and place cell discharge varies with box, running in any direction toward a centrally located

Review 819

water cup and so on (Wiener et al., 1995). Wiener et al. strength of any such signal is not likely to be very great (Quirk et al., 1992). As expected if directional selectivityconclude that pyramidal cells signal virtually all aspects

of the task. It is the task, including the environment, is low, entorhinal cortex cells recorded on an 8-arm maze showed little difference in their firing rates as ratsrather than just the environment that is represented.

Similar conclusions were drawn (Deadwyler et al., 1996) ran in or out on arms (Barnes et al., 1990). The dentate gyrus is one major target of entorhinalusing a delayed match-to-sample task. Cells were ob-

served that fired in relation to each part of the task, cortex efferents. Connections from layer II of entorhinal cortex onto DG granule cells are the first part of theincluding during the delay interval. Positional firing bi-

ases were overlaid on the other correlates for most cells. “trisynaptic circuit” (Andersen et al., 1971). Recent work reveals clear positional firing correlates for identifiedIn another study, rats had to leave a box, go to a point

relative to two landmarks to obtain food, and return to granule cells (Jung and McNaughton, 1993). The posi- tional selectivity is about equal to that of CA3 place cellsthe box to get more food (Gothard et al., 1995). The

location of the box at the trial start, the location of and is therefore more precise than for entorhinal cortex cells. The units were selective for either inward or out-the landmarks, and the location of the box at the end

of the trial were varied, creating four reference frames ward movements along an arm of an 8-arm maze. The postsynaptic elements of the second and third steps in(laboratory, box-departure, landmark, and box-arrival).

The discharge of individual cells was then plotted rela- the trisynaptic circuit are the pyramidal cells of CA3 and CA1, which of course are place cells.tive to each frame. Remarkably, there were different

cells whose discharge was densest in each of the four Another class of neurons in the hippocampus are the theta cells of stratum oriens and stratum pyramidale offrames, suggesting that each frame has its own repre-

sentation. How such representations would combine to CA1; similar units are found in CA3. The theta cells fire about twice as fast during running (a theta-associatedpermit navigation is not clear. It is interesting that if the

Gothard experiment were repeated with the box and behavior) than during quiet alertness (a non-theta be- havior) (Kubie et al., 1990). There was also positionallandmarks in fixed locations, one might conclude that

only place cells exist and that they occur preferentially modulation of firing. Pairs of simultaneously recorded theta cells had distinct positional firing patterns, at oddsnear important locations. with the idea that positional rate variations are second- ary to the behavioral modulation.

Spatial Cells in Other Parts of the The subiculum is the major target of CA1 efferents. Hippocampal Formation The positional firing patterns of subicular cells resemble It is convenient to define “spatially selective discharge” those of place cells, although the confinement of dis- as discharge that depends on either head position, head charge to a small area is not as tight (Sharp and Green, direction, or both head position and direction. I now 1994). The positional firing of many subicular cells is summarize recent work that reveals the existence of modulated by head direction, although the magnitude of many classes of spatial cells in the hippocampal for- the modulation is not great. Units in the parasubiculum, a mation. major target of subicular efferents, have properties simi-

The gateway to the hippocampal formation is entorhi- lar to those in the subiculum, including crisp positional nal cortex. The hippocampal input from the septum firing and detectable modulation of discharge by head plays a critical role in controlling the hippocampal EEG direction (Taube, 1995b). So far, no recordings have pattern, and other subcortical afferents serve important been made from deep layers of entorhinal cortex, a ma- modulatory roles. Nevertheless, detailed highly pro- jor target of units from the subiculum, the parasubicu- cessed sensory information reaches the hippocampus lum, and the postsubiculum. from entorhinal cortex and specifically its superficial Many postsubicular units are “head direction cells” layers (II and III) (Steward and Scoville, 1976). Cells in (Ranck, 1985; Taube et al., 1990a, 1990b; Muller et al., superficial layers of medial entorhinal cortex show loca- 1996b). A typical head direction cell is strongly active tion-specific firing, although the firing patterns are nois- only when the rat’s head points in a relatively narrow ier than those of place cells (Quirk et al., 1992). Similar range of angles in the horizontal plane. (Head direction discharge properties are seen in lateral entorhinal cells is relative to the environment; it is not the neck angle (Fox et al., 1994, Soc. Neurosci., abstract). Cells in me- made by the head with the body). Firing is most rapid dial entorhinal cortex also differ from place cells in how at a “preferred direction” and drops off linearly for angles their positional firing patterns are affected by changing clockwise or counterclockwise to the preferred direc- the shape of the apparatus. First, none of the medial tion. The peak rate varies widely from cell to cell, but is entorhinal cortex units became silent, so that the notion characteristic of a given cell in all circumstances tested of an active subset does not apply. Second, for many (Taube et al., 1990a, 1990b). In contrast, a robust place medial entorhinal cortex cells, the transformation of the cell may fail to fire when the rat runs through the field positional firing pattern mimicked the shape change of center (Fenton and Muller, 1995, Soc. Neurosci., ab- the apparatus, so that in contrast with place cells, the stract). pattern could be predicted across apparatuses. Thus, The preferred direction for a given cell is the same the medial entorhinal cortex representation does not everywhere in an apparatus. Since the preferred direc- undergo remapping, but instead more closely reflects tion vectors are parallel, the ensemble discharge of the sensory aspects of the surroundings. No directional se- directional cell population acts as a compass. Interest- lectivity measurements were made for medial entorhinal ingly, the compass property does not arise because

head direction cells are triggered by a signal from acortex units, but direct observation suggests that the

Neuron 820

larger frame, perhaps a distant stimulus in the environ- That the hippocampal formation is involved with high level processing of spatial information is greatly rein-ment or the earth’s magnetic field. In fact, the same cue

card that controls the angular position of firing fields forced by the observation that many if not all cells un- dergo identical changes in angular discharge correlateprecisely controls the preferred direction of head direc-

tion cells. The ability of head direction cells to act as a when the cue card is removed. Specifically, firing pat- terns remain intact, despite the powerful control by thecompass even thoughthey arecontrolled by local stimuli

implies that they, like place cells, are not well described white card, and in cases where it has been tested, rota- tions are consistent from cell to cell (Taube et al., 1990b;as a sophisticated kind of sensory unit; head direction

cells are also part of an abstract environmental repre- Knierim et al., 1995). sentation.

Several features of the directional representation were An Overall View: Theories of Cognitive Mappingcharacterized by Taube et al. (1990a, 1990b). First, there With the current enthusiasm for computational modelingseemed to be no silent head direction cells. Thus, the of neurons and networks, it is hard to imagine that theactive subset is the whole directional cell population hippocampal place cells and attendant phenomenaand is not environment specific. In addition, whenever have escaped attention, and indeed this is a very activethe preferred direction of one cell is caused to rotate area of investigation. Even a cursory review of the cur-by a certain amount, the preferred direction of a second rent offerings would increase the length of this paper bysimultaneously recorded cell rotates by the same one-third or more. Accordingly, we end by summarizingamount (Taube et al., 1990b). By inference, the angular what a complete theory would explain, rather than bydistance between the preferred directions of all cell pairs considering the individual theories themselves. An up-is fixed and there is no analog of remapping. coming issue of Hippocampus, organized by MarkHead direction cells are found in structures other than Gluck, lays out the theoretical ideas of several workersthe postsubiculum, notably the anterior dorsal nuclei of interested in the hippocampal basis of navigation.the thalamus, which is reciprocally connected with the

A theory of cognitive mapping needs to explain sixpostsubiculum. Features of head direction cells in differ- features of place cell firing considered in this review:ent brain areas and thoughts about their functional sig-

(1) The origins and properties of spatial firing in anificance are given in a recent review (Muller et al., familiar environment. A key is to find mechanisms that1996b). allow a strong uncoupling of spatial firing from details of the sensory cues, even though such cues powerfully control the spatial distribution of spike activity. An asso-The Hippocampal Formation

as a Navigational System ciated issue is how place cells can show directional selectivity in some circumstances and be unselective inThe list of sites in which spatial cells have been found

is only a start in understanding the operation of a crucial others. (2) How it is possible for a representation to be stablepart of the rat navigational system. Little is known about

the mechanisms by which spatial discharge patterns over long periods of time. This is especially interesting in light of the waxing and waning of spines on CA1are organized or transformed. Nevertheless, there are

several indications that the anatomically related areas pyramidal cells with the estrous cycle (Woolley and McEwen, 1993, 1994).mentioned above constitute a functional system. First,

a positional signal can be traced along the trisynaptic (3) How are new representations generated? In the context of the hippocampus, it is hard to imagine thatpathway and beyond (Barnes et al., 1990; Muller et al.,

1991). Second, a directional signal is found in the post- long-term potentiation and long-term depression arenot involved here and at othersteps. Indeed, it is remarkablesubiculum, in related parts of the thalamus, and in the

subiculum and parasubiculum. Third, theta cells similar that virtually nothing is presently known about relation- ships between spatial firing and the known mechanismsto those in Ammon’s horn have been seen in all parts

of the hippocampal formation so far investigated (but of synaptic modification. (4)How multiple environmentscan bemapped withoutnot in the thalamus [Taube, 1995a]). It is not known if

theta cells in regions other than Ammon’s horn also crippling mutual interference. Again, one wonders how synaptic modifiability comes into play.show positional or directional discharge properties.

Fourth, lesion datasuggest that all parts of hippocampal (5) What are the relationships between the positional system and the directional system? Cells in both sys-formation are critical for normal navigational behavior

(O’Keefe and Nadel, 1978), although relationships be- tems are controlled in remarkably similar ways by sen- sory information from the environment. Does this infor-tween lesion locations and impairments in spatial versus

nonspatial tasks are quite complicated (Jarrard, 1993). mation pass from one system to the other, or do the two systems operate in parallel? Lesion and single cellThe final evidence that the hippocampal formation forms

a unit is the consistency of changes in spatially corre- data, reviewed by Muller et al. (1996b), suggest that the systems operate in parallel. Perhaps the role of thelated discharge after altering the environment. Except

for dentate granule cells, which have not been investi- positional system is to compute paths through the envi- ronment, and the role of the directional system is to putgated in the cylinder, rotations of the cue card produce

nearly equal rotations of the appropriate spatial dis- such paths into register with the environment so they can be properly executed (Muller et al., 1996b).charge correlate; so far as is known, every active cell

in every one of the anatomical regions is affected to the (6) If the hippocampal formation is the locus of the cognitive map, how is the map organized? A successfulsame extent.

Review 821

Kubie, J.L., and Muller, R.U. (1991). Multiple representations in thetheory must be able to explain how specific behavioral hippocampus. Hippocampus 1, 240–242.capacities of the rat arise from the ensemble activity of Kubie, J.L., and Ranck, J.B., Jr. (1984). Hippocampal neuronal firing,the parts of the hippocampal formation and possibly context, and learning. In Neuropsychology of Memory, L.R. Squireother brain parts. Examples of interesting navigational and N. Butters, eds. (New York: Guilford Press).

capacities are theproduction of geodesic paths inunob- Kubie, J.L., Muller, R.U., and Fox, S.F. (1985). Firing fields of hippo-structed space, taking detours and taking shortcuts. campal place cells: interim report. In Electrical Activity of the Archi-

Existing theories focus on different aspects of the cortex, G. Buszaki and C.H. Vanderwolf, eds. (Budapest: Hungarian overall problem, but no current theory is both correct Academy of Science). and comprehensive. Nevertheless, computational mod- Kubie, J.L., Muller, R.U., and Bostock, E.M. (1990). Spatial firing eling is essential. There is no other way of knowing if a properties of hippocampal theta cells. J. Neurosci. 10, 1110–1123. theory can generate adequate navigational perfor- Markus, E.J., Barnes, C.A., McNaughton, B.L., Gladden, V.L., and mance, a necessary but not sufficient condition that Skaggs, W.E. (1994). Spatial information content and the reliability

of hippocampal CA1 neurons: effects of visual input. Hippocampusmust be passed by adequate theories of cognitive 4, 410–421.mapping. Markus, E.J., Qin, Y., Leonard, B., Skaggs, W.E., McNaughton, B.L., and Barnes, C.A. (1995). Interactions between location and taskAcknowledgments affect the spatial and directional firing of hippocampal neurons. J. Neurosci. 15, 7079–7094.Work done in this laboratory is supported by NIH grant NS 20686. McNaughton, B.L., andMorris, R.G.M. (1987). Hippocampal synaptic

References enhancement and information storage in a distributed memory sys- tem. Trends Neurosci. 10, 408–415.

Andersen, P., Bliss, V.P., and Skrede, K.K. (1971). Lamellar organiza- McNaughton, B.L., and Nadel, L. (1990). Hebb-Marr networks and tion of hippocampal excitatory pathways. Exp. Brain Res. 13, the neurobiological representation of action in space. In Neurosci- 222–238. ence and Connectionist Theory, M.A. Gluck and D.E. Rumelhart, Barnes, C.A., McNaughton, B.L., Mizumori, S.J.Y., Leonard, B.W., eds. (Hillsdale, New Jersey: Lawrence Erlbaum Associates). and Lin, L.-H. (1990). Comparison of spatial and temporal character- McNaughton, B.L., Barnes, C.A, and O’Keefe, J. (1983). The contri- istics of neuronal activity in sequential stages of hippocampal pro- butions of position, direction, and velocity to single unit activity in cessing. Prog. Brain. Res. 83, 287–300. the hippocampus of freely-moving rats. Exp. Brain Res. 52, 41–49. Bostock, E., Muller, R.U., and Kubie, J.L. (1991). Experience-depen- McNaughton, B.L., Chen, L.L., and Markus, E.J. (1991). “Dead reck- dent modifications of hippocampal place cell firing. Hippocampus oning,” landmark learning and the sense of direction: a neurophysio- 1, 193–206. logical and computational hypothesis. J. Cognit. Neurosci. 3, Breese, C.R., Hampson, R.E., and Deadwyler, S.A. (1989). Hippo- 190–202. campal place cells: stereotypy and plasticity. J. Neurosci. 9, 1097– Muller, R.U., and Kubie, J.L. (1987). The effects of changes in the 1111. environment on the spatial firing of hippocampal complex-spike Buzsaki, G. (1989). Two stage model of memory trace formation: a cells. J. Neurosci. 7, 1951–1968. role for “noisy” brain states. Neuroscience 31, 551–570.

Muller, R.U., Kubie, J.L., and Ranck, J.B., Jr. (1987). Spatial firing Chrobak, J.J., and Buzsaki, G. (1996). High-frequency oscillations patterns of hippocampal complex-spike cells in a fixed environment. in the output networks of the hippocampal-entorhinal axis of the J. Neurosci. 7, 1935–1950. freely behaving rat. J. Neurosci. 16, 3056–3066.

Muller, R.U., Kubie, J.L., Bostock, E.M., Taube, J.S., and Quirk, Deadwyler, S.A., Bunn, T., and Hampson, R.E. (1996). Hippocampal G.J. (1991). Spatial firing correlates of neurons in the hippocampal ensemble activity during spatial delayed-nonmatch-to-sample per- formation of freely moving rats. In Brain and Space, J. Paillard, ed. formance in rats. J. Neurosci. 16, 354–372. (Oxford: Oxford University Press), pp. 296–333. Georgopolous, A.P., Kalaska, J.F., Caminiti, R., and Massey, J.T. Muller, R.U., Bostock, E., Taube, J., and Kubie, J.L. (1994). On the (1982). On the relations between the direction of two-dimensional directional firing properties of hippocampal place cells. J. Neurosci. arm movements and cell discharge in primate motor cortex. J. Neu- 14, 7235–7251. rosci. 2, 1527–1537.

Muller, R.U., Stead, M., and Pach, J. (1996a). The hippocampus as Gothard, K.M., Skaggs, W.E., Moore, K.M., and McNaughton, B.L. a cognitive graph. J. Gen. Physiol. 107, 663–694. (1995). Binding of hippocampal CA1 neural activity to multiple refer-

Muller, R.U., Ranck, J.B., and Taube, J.S. (1996b). Head directionence frames in a landmark-based navigation task. J. Neurosci. 16, cells: properties and functional significance. Curr. Opin. Neurobiol.823–835. 6, 196–206.

Hebb, D.O. (1949). The Organization of Behavior (New York: John O’Keefe, J. (1979). A review of the hippocampal place cells. Prog.Wiley). Neurobiol. 13, 419–439.

Hetherington, P.A., and Shapiro, M.L. (1997). Hippocampal place O’Keefe, J., and Burgess, N. (1996). Geometric determinants of thefields are altered by the removal of single cues in a distance-depen- place fields of hippocampal neurons. Nature 381, 425–428.dent fashion. Behav. Neurosci., in press. O’Keefe, J., and Conway, D.H. (1978). Hippocampal place cells inHill, A.J. (1978). First occurrence of hippocampal spatial firing in a the freely moving rat: why they fire where they fire. Exp. Brain Res.new environment. Exp. Neurol. 62, 282–297. 31, 573–590.Jarrard, L.E. (1993). On the role of the hippocampus in learning and O’Keefe, J.,and Dostrovsky, J. (1971). The hippocampus as a spatialmemory in the rat. Behav. Neural Biol. 60, 9–26. map: preliminary evidence from unit activity in the freely-movingJung, M.W., and McNaughton, B.L. (1993). Spatial selectivity of unit rat. Brain Res. 34, 171–175.activity in the hippocampal granular layer. Hippocampus 3, 165–182. O’Keefe, J., and Nadel, L. (1978). The Hippocampus as a CognitiveJung, M.W., Wiener, S.I., and McNaughton, B.L. (1994). Comparison Map (Oxford: Clarendon).of spatial firing characteristics of units in dorsal and ventral hippo- O’Keefe, J., and Recce, M. (1993). Phase relationship between hip-campus of the rat. J. Neurosci. 14, 7347–7356. pocampal place units and the EEG theta rhythm. Hippocampus 3,Knierim, J.J., Kudrimoti, H.S., and McNaughton, B.L. (1995). Place 317–330.cells, head direction cells, and the learning of landmark stability. J.

Neurosci. 15, 1648–1659. O’Keefe, J., and Speakman, A. (1987). Single unit activity in the rat

Neuron 822

hippocampus during a spatial memory task. Exp. Brain Res. 68, Woolley, C.S., and McEwen, B.S. (1994). Estradiol regulates hippo- campal dendritic spine density via an N-methyl-D-aspartate recep-1–27. tor-dependent mechanism. J. Neurosci. 14, 7680–7687.Poucet, B., Thinus-Blanc, C., and Muller, R.U. (1994). Place cells in Young, B.J., Fox, G.D., and Eichenbaum, H. (1994). Correlates ofthe ventral hippocampus of rats. NeuroReport 5, 2045–2048. hippocampal complex-spike cell activity in rats performing a non-Quirk, G.J., Muller, R.U., and Kubie, J.L. (1990). The firing of hippo- spatial radial maze task. J. Neurosci. 14, 6553–6563.campal place cells in the dark reflects the rat’s recent experience.

J. Neurosci. 10, 2008–2017.

Quirk, G.J., Muller, R.U., Kubie, J.L., and Ranck, G.B. (1992). The positional firing properties of medial entorhinal neurons: description and comparison with hippocampal place cells. J. Neurosci. 12, 1945–1963.

Ranck, J.B., Jr. (1973). Studies on single neurons in dorsal hippo- campal formation and septum in unrestrained rats, Part I. Behavioral correlates and firing repertoires. Exp. Neurol. 41, 461–555.

Ranck, J.B., Jr. (1985). Head direction cells in the deep cell layer of dorsal presubiculum in freely moving rats. In Electrical Activity of Archicortex. G Buzsaki and C.H. Vanderwolf, eds. (Budapest: Aka- demai Kiado), pp. 217–220.

Recce, M.,and Harris, K.D. (1996). Memory for places: a navigational model in support of Marr’s theory of hippocampal function. Hippo- campus, in press.

Rolls, E.T. (1989). Parallel distributed processing in the brain: impli- cations of the functional architecture of neuronal networks in the hippocampus. In Parallel Distributed Processing: Implications for Psychology and Neurobiology, R.G.M. Morris, ed. (Oxford: Clarendon Press).

Sharp, P.E. (1991). Computer simulation of hippocampal place cells. Psychobiology 19, 103–115.

Sharp, P., and Green, C. (1994). Spatial correlates of firing patterns of single cells in the subiculum of the freely moving rat. J. Neurosci. 14, 2339–2356.

Sharp, P., Muller, R.U., and Kubie, J.L. (1990). Firing properties of hippocampal neurons in a visually symmetrical environment: contri- butions of multiple sensory cues and mnemonic processes. J. Neu- rosci. 10, 3093–3100.

Speakman, A., and O’Keefe, J. (1990). Hippocampal complex spike cells do not change their place fields if the goal is moved within a cue controlled environment. Eur. J. Neurosci. 2, 544–555.

Steward, O., and Scoville, S. (1976). Cells of origin of entorhinal cortical afferents to the hippocampus and fascia dentata of the rat. J. Comp. Neurol. 169, 347–370.

Taube, J.S. (1995a). Head direction cells recorded in the anterior thalamic nuclei of freely moving rats. J. Neurosci. 15, 70–86.

Taube, J.S. (1995b). Place cells recorded in the parasubiculum of freely moving rats. Hippocampus 5, 569–583.

Taube, J.S., Muller, R.U., and Ranck, J.B., Jr. (1990a). Head direction cells recorded from the postsubiculum in freely moving rats. I. De- scription and quantitative analysis. J. Neurosci. 10, 420–435.

Taube, J.S., Muller, R.U., and Ranck, J.B., Jr. (1990b). Head direction cells recorded from the postsubiculum in freely moving rats. II. The effects of environmental manipulations. J. Neurosci. 10, 436–447.

Thompson, L.T., and Best, P.J. (1989). Place cells and silent cells in the hippocampus of freely-behaving rats. J. Neurosci. 9, 2382–2390.

Thompson, L.T., and Best, P.J. (1990). Long-term stability of the place-field activity of single units recorded from the dorsal hippo- campus of freely behaving rats. Brain Res. 509, 299–308.

Treves, A., and Rolls, E.T. (1992). Computational constraints suggest the need for two distinct input systems to the hippocampal CA3 network. Hippocampus 2, 189–200.

Wiener, S.I., Korshunov, V.A., Garcia, R., and Berthoz, A. (1995). Inertial, substratal and landmark cue control of hippocampal CA1 place cell activity. Eur. J. Neurosci. 7, 2206–2219.

Wilson, M., and McNaughton, B.L. (1993). Dynamics of the hippo- campal ensemble code for space. Science 261, 1055–1058.

Woolley, C.S., and McEwen, B.S. (1993). Roles of estradiol and progesterone in regulation of hippocampal dendritic spine density during the estrous cycle in the rat. J. Comp. Neurol. 336, 293–306.

Blog ArchiveCopyright © 2019 HomeworkMarket.com Read More
Applied SciencesArchitecture and DesignBiologyBusiness & FinanceChemistryComputer ScienceGeographyGeologyEducationEngineeringEnglishEnvironmental scienceSpanishGovernmentHistoryHuman Resource ManagementInformation SystemsLawLiteratureMathematicsNursingPhysicsPolitical SciencePsychologyReadingScienceSocial Science

"Order a similar paper and get 15% discount on your first order with us
Use the following coupon
"FIRST15"

Order Now