The hippocampus is one of the most easily recognizable structures in the brain owing to its characteristic seahorse-like shape. Brain imaging studies in the 1990s famously showed the hippocampus to be larger in London taxi drivers than in other people, suggesting that it plays a role in spatial navigation. This was consistent with previous findings in rodents, which had shown that the hippocampus is active when animals find their way through mazes.

Electrode recordings have revealed that whenever an animal is in a specific location of a particular environment (for example, in the back left-hand corner of a small room with white walls) one or a small number of cells within the hippocampus will fire to encode that location. When the animal moves to a new location within the same environment, other cells will fire to encode the new location. In this way, the population of cells—which are known as place cells—can together construct a virtual ‘map’ of the environment.

It is generally assumed that this hippocampal map represents space in terms of the absolute distances and angles between locations, rather like a street map. However, this type of geometric map appears inconsistent with the results of certain experiments. Dabaghian et al. proposed instead that the hippocampal map is based on topology, or the relative order of locations and the connections between them, rather like a subway map. Subsequently, computer models demonstrated that virtual simulations of place cells could effectively ‘learn’ the topological features of different environments.

Now, Dabaghian et al. provide their own empirical data to support the existence of a hippocampal ‘subway-style’ map by recording the electrical impulses from place cells in the rat hippocampus as the animals ran through a U-shaped maze. The maze was constructed so that its arms could either be straight or folded into zigzags. Changing the maze in this way does not alter its topology because the relative order of its various components—such as the positions of food wells in the arms—is unchanged, but it does alter the maze's geometry. Notably, as rats ran through different conformations of the maze, the activity of the place cells in their hippocampi remained largely unchanged, consistent with a map based on topology rather than geometry.

By providing evidence that hippocampal maps have more in common with subway maps than street maps, the work of Dabaghian et al. offers an explanation for previously challenging results and provides a framework for further experiments into hippocampal memory function.

When O'Keefe and Dostrovsky discovered that the firing of hippocampal place cells correlates strongly with the position of the rat with respect to discrete locations within the environment (the cell's place field) (O'Keefe and Dostrovsky, 1971), they launched decades of intensive research into how place cells contribute to an internal map of the environment (Best et al., 2001). That there is a considerable gulf between place cell firing and a cognitive spatial map, however, has become only clearer with time. Place cells appear to respond to a perplexing array of stimuli, from visual cues to head direction, goal planning, color changes, shape changes, and olfactory, vestibular and kinesthetic inputs, and discrepancies between the expected and actual location of a target (Harnad, 1994; Frank et al., 2000; Wood et al., 2000); recent work has shown that both spatial and nonspatial cell types in the entorhinal cortex provide myriad inputs to the hippocampus (Ideker et al., 2011). The fascination with place cell promiscuity, however, has tended to distract from the fact that cognition is the work not of individual neurons but of large ensembles of cells (Ludvig, 1999; Fenton et al., 2008). The nature of the information embedded in the place cell ensemble code and transmitted to downstream neurons has been largely ignored, along with the consequences for the type of spatial properties that might form the basis for the cognitive map.

Whatever the identity of these downstream neurons, they clearly have no direct access to the physical environment or to the place fields mapped by experimentalists. The only information they receive is contained in the temporal pattern of the spike trains of the thousands of place cells that are active in a given environment. What sort of information about a space might be constructed from such signals? The dominant assumption within the field of neuroscience has been that the hippocampal spatial map—if there is one (Eichenbaum et al., 1999)—is an allothetic, Euclidian representation of the local environment that integrates detailed information about distances and angles arising through self-motion, head direction, and speed in addition to visual and other cues (Wills et al., 2010; Erdem and Hasselmo, 2012; Chen et al., 2013; Rubin et al., 2014). The compilation of such geometric data would seem to require considerable computational power for the spatial map to be formed quickly enough to be useful under normal conditions of animal navigation (Wilson and McNaughton, 1993; Frank et al., 2004). Geometry is not the only aspect of space that can form the basis of a map, however. From a practical standpoint, the topological qualities of a space—relations between locations such as continuity, enclosure, sequence (Poincaré, 1895; Aleksandrov, 1965)—are arguably even more important (Poucet, 1993). Yet scant attention has been paid to the possible role of topological features in the cognitive map, despite the fact that any movement through space is describable in topological terms (locations A and B must be somehow connected if movement between them is possible). This neglect is even more curious in light of Piaget's early work indicating that children first represent space topologically and only later learn to incorporate metrical details into their mental representations of space (Poucet, 1993; O'Keefe and Nadel, 1978). A few studies have shown that place cell firing is responsive to topological changes in the environment such as the placement or removal of a barrier across a previously learned route (Poucet and Herrmann, 2001; Alvernhe et al., 2011), but as noted above, place cells seem to be responsive to an enormous variety of inputs. Our interest ultimately is not whether place cells can respond to topological features of a space, but whether the information conveyed downstream by place cell spiking might encode topological information, which would offer considerable advantages over geometry in terms of computational speed and flexibility.

There are two compelling reasons to consider such a radical departure from the dominant assumptions about the spatial component of the cognitive map. First is the nature of place cell firing. If, in fact, place cell co-firing implies spatial overlap of place fields (McNaughton et al., 1983; Brown et al., 1998; Zhang et al., 1998; Jensen and Lisman, 2000; Barbieri et al., 2001, 2005; Guger et al., 2011), then a map derived from place cell spiking will necessarily emphasize contiguities between locations as well as the temporal sequence in which they are experienced. (The reason we include temporal sequence along with more conventional spatial relationships is because movement through space takes place over time; sequence thus embodies connectivity.) This would seem to be more compatible with the broader role of the hippocampus in representing and storing experience. Second, how geometric changes would be transmitted downstream as metrical changes per se is unclear; precisely because place cells respond to so many inputs, a change in firing rate within a period of overlap might just as well reflect a change in the animal's speed or some other input as it would a change in distance. Muller and Kubie (1987) found that enlarging their open field set-up caused place fields to enlarge while maintaining the same shape and relative order (i.e., the connectivity of the environment remained invariant). Similar observations were made by Gothard et al., 1996a on a linear track, by O'Keefe and Burgess (1996) in rectangular environments, and by other studies in morphing environments (Lever et al., 2002; Leutgeb et al., 2005; Touretzky et al., 2005; Wills et al., 2005; Colgin et al., 2010). O'Keefe et al. (O'Keefe and Burgess, 1996) reasoned that the responses of the place fields could be described in terms of the geometric characteristics of the deforming environment, such as the altered distances to the walls of the enclosure. If we consider the information available to the downstream neurons that ‘read’ spiking coactivity, however, this explanation becomes a bit less plausible. As adjacent place fields enlarge in response to an expanded environment, the place cells will still be coactive, and the readout to the downstream neurons will be much the same. Thus, even if the place field layout stretches significantly, the corresponding geometrical information will not necessarily be captured by the downstream networks unless the change is quite large. Note that this phenomenon of place field stretching has a topological quality to it, since the fields stretch with the space but maintain their relative positions and order. In this context it is interesting to note that a more detailed analysis of place cell activity in a stretching 1D environment (Diba and Buzsaki, 2008) showed that the temporal gaps between the spikes of the respective place cells remain fixed throughout the stretch, despite the change in the overall firing rate (rate remapping, Dupret et al., 2010). Mathematically speaking, this suggests that spatial information contained in the hippocampal map may be invariant with respect to a certain range of geometric transformations, so that the flexible arrangement of place fields does not represent specific locations on a Cartesian grid, but rather a coarser framework of spatiotemporal relationships.

Another advantage to the topological model is that it implies that questions of spatial learning might be amenable to recently developed approaches in the field of computational topology. The way place fields cover a space actually calls to mind a fundamental theorem in algebraic topology that holds that a given space, if covered by a sufficient number of discrete sub-regions, can be reconstructed by the pattern of overlaps between those sub-regions (a simplicial complex) (Hatcher, 2002). This theorem is key to the power of topological approaches to resolve otherwise overwhelmingly complex problems in high-dimensional data analysis (Lum et al., 2013), and it lends itself to the problem at hand. We recently developed a computational algorithm based on algebraic topology to simulate the behavior of a rat exploring several different environments (Dabaghian et al., 2012; Arai et al., 2014). We created a wide range of testable values for three parameters—place cell firing rate, number of place cells, and place field size—and showed that simulated groups of neurons accurately encode spatial information from topologically distinct environments (they ‘learn’ the space) when the parameter values happen to closely parallel biological values derived from animal experiments. Since we made no a priori assumptions about what values would work, this was a promising validation of our model. We subsequently included theta precession in the model and showed that theta oscillation affects the behavior of the neuronal ensemble in ways that augment its performance to enhance learning (Arai et al., 2014).

Still, there has been no experiment designed explicitly to examine how place cells respond to significant geometric changes when the topology is held invariant. Here we report such an investigation. Since the location-specific firing of place cells is thought to be produced by a path integration mechanism (accurate idiothetic counting of the elementary displacements and changes of orientation, speed and acceleration produced during movements [McNaughton et al., 1996; Etienne and Jeffery, 2004; McNaughton et al., 2006]), we used morphing linear tracks whose geometry limits the pool of possible confounding motions. If place cells are primarily concerned with encoding geometric information, then their firing activity should reflect the dynamic planar geometry of the track. If, on the other hand, place cells are primarily concerned with topological information, then variations of the track's configuration should have little effect on where the place cells fire.

We devised two elevated U-shaped tracks with a flexible construction that allowed the geometry of one arm to be significantly altered while the rat was on the other part of the track. On the first track, Track A, each arm of the U could bend at four joints to transition from a straight runway in the expanded configuration to a compressed zigzag (Figure 1; ‘Materials and methods’). The dynamic configurations of the track were mapped against a stable set of spatial bins, which allowed us to dissociate two frames of reference: the effects produced on the CA1 place cells by changes in the track configuration can be described either in the reference frame associated with the room (the 2D planar reference frame) or with the track itself (the 1D linear reference frame). If path-integrating the 2D displacements, velocities and acceleration produces place fields that from an allocentric, Cartesian coordinate map of the space, the place fields should be relatively stable in the 2D plane, remaining in the same spatial bins despite changing track segments, and variable in the 1D frame of reference (See Figure 1, Figure 1—figure Supplement 1). In contrast, if the place cells attend to the sequence and connectivity of places experienced on the track irrespective of the track's geometry and the directions and angles of the rat's movements, then the place fields will be stable in the linear reference frame, but variable in the planar (2D) reference frame.

Figure 1 with 1 supplement see all

Download asset Open asset

Three animals were exposed to the initially novel Track A. We recorded a total of 557 pyramidal cells in CA1 and analyzed the pattern of their activity in response to the expansions and contractions of the track in both reference frames (‘Materials and methods’) and compared the spatial firing rates across track configurations. We found that, for the CA1 place cells active on the mobile sections of the track, spiking was stable in the linear but not in the 2D reference frame. Figure 2A,B shows the spiking pattern of one representative neuron in two different track configurations, each of which is plotted in both the linear and planar reference frames. As evident in the overlay of the two plots (Figure 2A+B), the place cell's spiking was widely dispersed in the 2D reference frame but localized in the linear frame. We quantified this tendency by computing the correlation between the firing rates in the planar and linear reference frames for all pairs of configurations. Because any two linearized track configurations are geometrically identical, whereas only a few parts of the 2D track configurations overlap, we computed the correlations using only the spatial bins that were visited and where at least one spike was fired in both 1D and 2D configurations (‘Materials and methods’).

Figure 2

Download asset Open asset

Since the track is divided into uniformly spaced vertical sections, the geometric dissimilarity between two configurations can be expressed simply as the difference between the section indices (See explanation in Figure 1D, ‘Materials and methods’). Therefore, all the correlation coefficients with the same difference between the first and the second index, C_i,i+d, will describe the correlations between ‘equally dissimilar’ track configurations separated by d bins and appear on the d-th subdiagonal of the correlation matrix. Their average value, r_d = < C_i,i+d >, thus characterizes the mean correlation for a given degree of dissimilarity, and the decay of r_d as a function of d characterizes the decay of correlations as a function of the difference between track geometries. The matrices in Figure 2C show that the correlations computed in the 2D reference frame decayed rapidly as more geometrically dissimilar configurations were compared. In contrast, the linear correlations remained high even for the fully expanded configuration. The difference in the mean correlation is highly significant (p < 0.001, two sample t tests) for each point (all values of d) on both curves (Figure 2D). The high values of the correlation coefficients for individual neurons in the linear case also show that the place fields were stable, despite the fact that active place cells could produce only a few spikes during the animal's single run through the corresponding place field in each track configuration, the effect of which was to create greater variability in our estimates of the firing rate and reduce the value of correlation coefficients (Figure 2C). To this end, we also examined cells that were active on the static portions of the track (Figure 2E–F) and found that they demonstrated uniformly high correlation values for both reference frames (Figure 2G–H). For these cells, the average values of the inter-configuration correlation coefficients in the1D and the 2D case did not diverge: the place fields of these neurons were stable despite the motion of other track elements. This demonstrates that place fields on the moving segments shift in 2D not because of some general instability but because they are, in fact, responding to changes in the configuration of the track.

The same place field stability in the linear frame is clearly evident in the ensemble averages for all place cells recorded over all days of recording for each animal (Figure 3A–C) and in the combined data across all cells from all animals (Figure 3D,E). As is clear from Figure 3A–C, the correlations were stable in the linear reference frame but decayed rapidly as a function of track configuration changes in the 2D reference frame. At the same time, the correlations for cells active only on the static portion of the track showed no tendency to decay as a function of the distance between vertical sections as the configurations changed (compare Figure 3D,E). The place fields in the linear frame of reference were thus largely invariant with respect to changes in track geometry.

Figure 3

Download asset Open asset

The observed stability of place cell activity patterns cannot be attributed to the animals' lack of awareness of the 2D shape of the environment or to their habituation to the environment (‘Materials and methods’). The animals were avidly curious about the space around the elevated track, and their deft performance on the constantly changing (if faintly glowing) track in a dark room suggests that the proprioceptive and idiothetic information required for such behavior was available to the animals because of the anatomical connections to the hippocampus (e.g., with head direction cells [Calton et al., 2003; Navratilova et al., 2012] and other inputs [Sharp et al., 1995; Muir and Bilkey, 2003; Lu and Bilkey, 2010]). Thus, it is likely that the recorded activity in area CA1 was stable despite the availability and the variability of these inputs.

It might be argued that the above results could be explained by a 1D path integration mechanism that somehow excludes vestibular and some proprioceptive input so that the place cells respond simply to the distance that the animal traveled from/to the food wells (Gothard et al., 1996a; Gothard et al., 1996b). Such a reduced path integration mechanism is sometimes expanded ad hoc to include place cell binding to local features of the track, such as joints or segments (Chen et al., 2013; O'Keefe and Burgess, 1996). To evaluate these possibilities, we carried out an additional experiment in two animals on a U-shaped track that consisted of 11 segments, three of which could form a protrusion on either of the two sides (Track B, see Figure 4A, ‘Materials and methods’). The geometric transformation in this case consisted of keeping the protrusion on one side of the track during the first session and then flipping it to the other side during the second session. The starting side for the protrusion alternated each day. On days 1, 3, and 5 during the first 20-min running session, the protrusion was formed by segments 3, 4 and 5; during the following 20 min running session, segments 7-8-9 formed the protrusion (Figure 4B). On days 2, 4 and 6 the protrusion first appeared in position 7-8-9 and then flipped to the opposite side for the second session (Figure 4C,D). In order to enter the protrusion area, the animal turned to the right on its inbound journeys and to the left on its outbound journeys during both runs, before and after the flip of the protrusion. The protrusion switch thus did not affect the sequence of turns but did produce a local deformation of space, which altered the relative arrangement of the track's segments and changed the distance relationships between locations on it. As before, the deformation did not violate the track's topological structure. Unlike the Track A experiment, however, the geometry of track B changed discontinuously, as the animal was physically taken off the track between the two running sessions.

Figure 4

Download asset Open asset

The geometric transformation provided by Track B allowed us to conduct analyses that complemented our observations on Track A. First, we asked whether place cell activity was bound to the local features on the track (individual segments and joints). If so, place fields would tend to move from one side of the track to the other with the protrusion. Second, we could verify whether place cell activity was determined by the animal's linear distance along the track. If this were the case, a population of place cells would remain active at the same distances from the ends of the track before and after the protrusion switch, that is, they would shift along the track as a result of the protrusion's flip. In contrast, our topological model would predict that the place field in the region of the protrusion should not flip over or shift; instead, it should simply stretch and shrink as the protrusion is inserted and removed. So if place cell spiking activity represents connectivity, the place fields on the section of track replaced by the protrusion should extend continuously into the space provided by the protrusion.

We tested these hypotheses by recording from each animal for 7 days: a total of 114 putative excitatory pyramidal cells were recorded while animals traversed Track B. As shown in the examples in Figure 4B–D, individual place cells were generally not bound to the protrusion but instead tended to stretch flexibly along the track when the protrusion was moved. To quantify the changes induced by the movement of the protrusion on the population level, we computed the ensemble averages for several types of correlations between the firing rates on the two protruding regions before and after the protrusion flip. All of these correlations were computed using the linearized data to ensure fair comparisons and included all cells that had a peak occupancy normalized firing rate above a 2 Hz threshold. This low threshold was chosen to be as inclusive as possible of different place cell firing patterns.

In our analysis, we saw no tendency of CA1 place fields to remain at fixed distances from the ends of the second track (Track B). Our findings cannot be explained by place cells responding purely to local features (such as junctions) or measuring distances from those features, since the place fields showed no tendency to be bound to specific track sections on the protrusion. Instead, place fields stretched or shrank to reflect the insertion or removal of the protrusion. The place fields were best described as remaining in their relative locations on the track. The bar graph in Figure 4E shows the mean of the resulting values of the correlation coefficients for four cases.

First, we considered the possibility that the place fields remained bound to specific features of each segment of the track. The first bar, marked ‘Flip’, shows the average correlations of the firing rate over the protrusion across the two configurations. Since this value is negative and close to zero, it is clear that the place cells on average showed no tendency to be bound to individual segments of the protrusion. The second bar, ‘Distance’, shows the value of the firing rate correlations between the sections occupied by protrusion in the first configuration (sections 3–5 or 7–9, depending on which way the protrusion was facing) and the sections at the same distance from the wells from the second configuration. Similarly to the previous case, the correlations for the distance-based comparison were close to zero; moreover, the difference in the mean correlation values between cases 1 and 2 was not statistically significant. The third bar, ‘Segment’, shows the average firing rate correlations over the protruding section in the first configuration (i.e., sections 4 or 8) and the section onto which the center segment of protrusion projects directly after the flip (sections 3 and 9, respectively). Clearly this correlation is significantly higher than in the previous two cases. The low protrusion-to-protrusion ‘Flip’ activity correlation values compared to the protruding ‘Segment’ correlations demonstrate that most or all place cells that were active on the protrusion in Configuration I did not follow the protrusion when it was moved to the other side but instead remained in the same relative positions on the track.

Lastly, the bar ‘Rescaled’ shows the firing rate correlations following rescaling of the protrusion length (and the coordinates of the corresponding spikes) to compensate for the spatial deformation produced by protrusion move (e.g., scaling the total length of the sections 3, 4 and 5 down to the length of the section 3 that replaces the protrusion after the flip). Remarkably, the rescaled correlates were not only significantly larger than those of the local features (case 1) and distance (case 2) scenarios, but also were larger than the protruding segment correlations, indicating that local stretching could describe a greater amount of the variability in place fields than path integration (non-parametric u-tests, which give the p values 10⁻⁵ for the differences and one way ANOVA, Tukey–Kramer post-hoc tests, p’s < 10⁻⁴). The rescaled case illustrates that rate remapping compensated for the spatial expansion in such a way as to not alter the structural framework of the map.

The average correlations on Track B were relatively low compared to those on Track A. This could reflect firing variability because the ‘flip’ of the protrusion Track B happened discontinuously, in one step, leading to some degree of disruption of the continuity of the environment (similar to [Diba and Buzsaki, 2008], where the transformation also induced partial remapping). We hypothesize that, were we able to carry out this experiment using a continuous transformation from the straight section of the track to the protrusion, as we were able to do with Track A, the hippocampal network would have retained a greater degree of similarity across the configurations. Nevertheless, the difference in the correlations between these four cases was statistically significant. The fact that both the mean and individual place cell activity across the two sessions showed the highest correlation after the rescaling transformation was applied, while the ‘Distance’ and ‘Flip’ correlations were low, suggests that place cell firing, despite its variability, served primarily to represent the connectivity of the environment through the temporal sequence of place cell activity. The CA1 spatial representation thus remained invariant with respect to a wide range of geometric transformations.

By explicitly counterposing geometry and topology in this series of experiments, we have demonstrated that hippocampal place cells show a remarkable degree of stability with respect to even dramatic geometrical changes in the environment: they retained their relative order and connectivity despite routes involving diametrically opposite protrusions and orientations. As long as the sequence of spaces experienced remains the same, in other words, and the place fields still overlap, it would appear not to matter to the hippocampus if the place fields are stretched by two centimeters or twenty. We therefore propose that place cells do not represent locations in space but rather provide a spatial context for experience (Eichenbaum et al., 1999; Moser et al., 2008).

This would seem to be more compatible with the higher-order functions of the hippocampus in representing and storing experience, and it is consistent with results from experiments that have been conducted in morphing environments, which have an implicit, if overlooked, topological aspect. Muller and Kubie (1987) found that enlarging their open field set-up caused place fields to enlarge while maintaining the same shape and relative order (i.e., the connectivity of the environment remained invariant). Similar observations were made by Gothard et al., 1996a on a linear track, by O'Keefe and Burgess (1996) in rectangular environments, and by other studies in morphing environments (Lever et al., 2002; Leutgeb et al., 2005; Touretzky et al., 2005; Wills et al., 2005; Colgin et al., 2010). This phenomenon is topological in character, since the fields stretch with the space but maintain their basic position and sequence. A more detailed analysis of place cell activity in a stretching 1D environment reported in Diba and Buzsaki (2008) shows that the temporal relationships between the firing of the relevant place cells remains fixed throughout the stretch, despite the change in the overall firing rate (rate remapping, Dupret et al., 2010).

In addition to these previous studies showing that place fields stretch (maintain their sequence and relative positions) in morphing environments, we can add several other compelling reasons to doubt that the hippocampal map is exclusively or fundamentally Cartesian, like a street map. There is our recent modeling work showing the feasibility of spatial learning based on topological information (Dabaghian et al., 2012; Arai et al., 2014); early studies by Piaget and Inhelder showing that children first conceive of spatial relations in topological terms and only later think in terms of geometrical positions (Dawson and Doddington, 1973); and previous studies on rodent navigation showing that place cells respond to topological changes in the environment such as the placement or removal of a barrier across a previously learned route (Poucet and Herrmann, 2001; Alvernhe et al., 2011, 2012). We propose that a topological framework, more like a subway map, would provide the hippocampus a more powerful, flexible, and readily formed spatial substrate for creating experiential memories or evaluating the spatial context of behaviors (Lavenex and Amaral, 2000; Banquet et al., 2005). Indeed, one study found that cells in the rat medial prefrontal cortex, recipients of place cell efferents, did not fire in response to the rat's position or head direction but rather in response to specific goal-oriented behaviors (Poucet, 1997).

All this is not to say that other brain regions do not contribute metrical information to the internal spatial map—there is abundant evidence that many regions are involved in providing sensory, idiothetic and proprioceptive information for path integration (Knierim et al., 1996, 1998; Fenton et al., 2000a, 2000b; Quiroga et al., 2005; Yoganarasimha et al., 2006; Moser et al., 2008)—only to propose that the hippocampus serves a more general role in spatial cognition by providing a qualitative, flexible representation of the environment. We suggest that place cells emphasize the connections between portions of a given environment and the sequence in which they are experienced. Such an encoded sequence is evident in spontaneous replays of place cell spiking during quiescent (Muller and Kubie, 1989; Davidson et al., 2009; Karlsson and Frank, 2009; Carr et al., 2011; Dragoi and Tonegawa, 2011; Zeithamova et al., 2012) and sleep states (Skaggs and McNaughton, 1996; Louie and Wilson, 2001; Foster and Wilson, 2006), which recapitulate the order in which the place cells fired during navigation. This implies that when the hippocampal network is driven internally, the structure of sequences remains the same as during navigation, although the activity of place cells under these circumstances is clearly not coupled directly with spatial locations. The encoded map thus appears to serve as a template for generating replay sequences and facilitating imaginative navigation. More recently, studies of hippocampal preplay (as opposed to replay) indicate that the hippocampus has a repertoire of preconfigured temporal place cell firing sequences that can be called upon to rapidly encode multiple novel spatial experiences (Dragoi and Tonegawa, 2013). It is striking that studies from such diverse fields, from electrophysiology to algebraic topology, seem to converge on the notion that the hippocampus operates by providing a flexible, rough-and-ready framework into which spatially grounded experiences can be organized and remembered.

If our topological model is borne out, it will have several implications. It is well known that if the environment changes significantly, then place cell activity can undergo global remapping, and if the changes are small, then the place cells preserve the relative order of spiking and respond only by a regular change of their spiking rates (rate remapping) (Colgin et al., 2008; Allen et al., 2012). Our topological hypothesis predicts that, within limits (which are as yet unknown), geometric transformations should cause only rate remapping while the sequence of firing remains invariant. Our topological model would also greatly narrow the realm of meaningful correlations between changes in visual cues and responses in place cell spiking. As we have tried to emphasize in this work, a place cell response does not necessarily translate into information communicated downstream. In the absence of global remapping, it may thus not be safe to interpret correlations between changes in stimuli and place cell activity as representing information that is transmitted to other brain regions. A change in response would primarily serve to establish the scope of geometric invariance, which could depend on a variety of biological conditions and which is a question that must be left for future investigations. Moreover, since the sequence of spaces traversed is encoded in ensembles of place cells rather than individual neurons, it is place cell ensembles, not individual neurons, whose responses to changes in the environment need to be studied. This of course poses technical challenges, but such challenges are more likely to be met if we see the necessity of doing so.

The notion that the hippocampus encodes topological relationships could help clarify the spatial vs non-spatial debate that has gone on for many years (Eichenbaum et al., 1999; O'Keefe and Nadel, 1978). Indeed, it is well known that the rat hippocampus is important for learning and remembering various types of non-spatial information, including visual, odor and action sequences (Agster et al., 2002; Fortin et al., 2002, 2004; Sauvage et al., 2008). These functions, and the importance of the human hippocampus for episodic memory (Squire, 1992), have led to the suggestion that the hippocampus performs a more general associative memory function (Eichenbaum et al., 1999; Wood et al., 1999). It is not entirely clear how the same mechanism would be responsible for representing geometry (distances and angles) in the spatial domain and encoding sequences and events in the mnemonic domain. The beauty of the topological hypothesis is that hippocampal spatial maps would have the same structure as memory maps or ‘memory spaces’ (Eichenbaum et al., 1999; Eichenbaum, 2000), which would not only explain the stability of relative firing order in flexible environments, but could also provide a common framework for understanding spatial representation, spatial memory and episodic memory mechanisms.

1. 1. Agster KL
   2. Fortin NJ
   3. Eichenbaum H

   (2002)

   The hippocampus and disambiguation of overlapping sequences

   The Journal of Neuroscience 22:5760–5768.

   • Google Scholar
2. 1. Aleksandrov PS

   (1965)

   Elementary concepts of topology

   F. Ungar Pub.

   • Google Scholar
3. 1. Allen K
   2. Rawlins JNP
   3. Bannerman DM
   4. Csicsvari J

   (2012) Hippocampal place cells can encode multiple trial-dependent features through rate remapping

   The Journal of Neuroscience 32:14752–14766.

   https://doi.org/10.1523/JNEUROSCI.6175-11.2012

   • Google Scholar
4. 1. Alvernhe A
   2. Sargolini F
   3. Poucet B

   (2012) Rats build and update topological representations through exploration

   Animal Cognition 15:359–368.

   https://doi.org/10.1007/s10071-011-0460-z

   • Google Scholar
5. 1. Alvernhe A
   2. Save E
   3. Poucet B

   (2011) Local remapping of place cell firing in the Tolman detour task

   The European Journal of Neuroscience 33:1696–1705.

   https://doi.org/10.1111/j.1460-9568.2011.07653.x

   • Google Scholar
6. 1. Arai M
   2. Brandt V
   3. Dabaghian Y

   (2014) The effects of theta precession on spatial learning and simplicial complex dynamics in a topological model of the hippocampal spatial map

   PLOS Computational Biology 10:e1003651.

   https://doi.org/10.1371/journal.pcbi.1003651

   • Google Scholar
7. 1. Banquet JP
   2. Gaussier P
   3. Quoy M
   4. Revel A
   5. Burnod Y

   (2005) A hierarchy of associations in hippocampo-cortical systems: cognitive maps and navigation strategies

   Neural Computation 17:1339–1384.

   https://doi.org/10.1162/0899766053630369

   • Google Scholar
8. 1. Barbieri R
   2. Quirk MC
   3. Frank LM
   4. Wilson MA
   5. Brown EN

   (2001) Construction and analysis of non-Poisson stimulus-response models of neural spiking activity

   Journal of Neuroscience Methods 105:25–37.

   https://doi.org/10.1016/S0165-0270(00)00344-7

   • Google Scholar
9. 1. Barbieri R
   2. Wilson MA
   3. Frank LM
   4. Brown EN

   (2005) An analysis of hippocampal spatio-temporal representations using a Bayesian algorithm for neural spike train decoding

   IEEE Transactions on Neural Systems and Rehabilitation Engineering 13:131–136.

   https://doi.org/10.1109/TNSRE.2005.847368

   • Google Scholar
10. 1. Best PJ
    2. White AM
    3. Minai A

    (2001) Spatial processing in the brain: the activity of hippocampal place cells

    Annual Review of Neuroscience 24:459–486.

    https://doi.org/10.1146/annurev.neuro.24.1.459

    • Google Scholar
11. 1. Brown EN
    2. Frank LM
    3. Tang D
    4. Quirk MC
    5. Wilson MA

    (1998)

    A statistical paradigm for neural spike train decoding applied to position prediction from ensemble firing patterns of rat hippocampal place cells

    The Journal of Neuroscience 18:7411–7425.

    • Google Scholar
12. 1. Calton JL
    2. Stackman RW
    3. Goodridge JP
    4. Archey WB
    5. Dudchenko PA
    6. Taube JS

    (2003)

    Hippocampal place cell instability after lesions of the head direction cell network

    The Journal of Neuroscience 23:9719–9731.

    • Google Scholar
13. 1. Carr MF
    2. Jadhav SP
    3. Frank LM

    (2011) Hippocampal replay in the awake state: a potential substrate for memory consolidation and retrieval

    Nature Neuroscience 14:147–153.

    https://doi.org/10.1038/nn.2732

    • Google Scholar
14. 1. Chen G
    2. King JA
    3. Burgess N
    4. O'Keefe J

    (2013) How vision and movement combine in the hippocampal place code

    Proceedings of the National Academy of Sciences of USA 110:378–383.

    https://doi.org/10.1073/pnas.1215834110

    • Google Scholar
15. 1. Colgin LL
    2. Leutgeb S
    3. Jezek K
    4. Leutgeb JK
    5. Moser EI
    6. McNaughton BL
    7. Moser MB

    (2010) Attractor-map versus autoassociation based attractor dynamics in the hippocampal network

    Journal of Neurophysiology 104:35–50.

    https://doi.org/10.1152/jn.00202.2010

    • Google Scholar
16. 1. Colgin LL
    2. Moser EI
    3. Moser M-B

    (2008) Understanding memory through hippocampal remapping

    Trends in Neurosciences 31:469–477.

    https://doi.org/10.1016/j.tins.2008.06.008

    • Google Scholar
17. 1. Dabaghian Y
    2. Mémoli F
    3. Frank L
    4. Carlsson G

    (2012) A topological paradigm for hippocampal spatial map formation using persistent homology

    PLOS Computational Biology 8:e1002581.

    https://doi.org/10.1371/journal.pcbi.1002581

    • Google Scholar
18. 1. Davidson TJ
    2. Kloosterman F
    3. Wilson MA

    (2009) Hippocampal replay of extended experience

    Neuron 63:497–507.

    https://doi.org/10.1016/j.neuron.2009.07.027

    • Google Scholar
19. 1. Dawson WW
    2. Doddington HW

    (1973) Phase distortion of biological signals: extraction of signal from noise without phase error

    Electroencephalography and Clinical Neurophysiology 34:207–211.

    https://doi.org/10.1016/0013-4694(73)90051-5

    • Google Scholar
20. 1. Diba K
    2. Buzsaki G

    (2008) Hippocampal network dynamics constrain the time lag between pyramidal cells across modified environments

    The Journal of Neuroscience 28:13448–13456.

    https://doi.org/10.1523/JNEUROSCI.3824-08.2008

    • Google Scholar
21. 1. Dragoi G
    2. Tonegawa S

    (2011) Preplay of future place cell sequences by hippocampal cellular assemblies

    Nature 469:397–401.

    https://doi.org/10.1038/nature09633

    • Google Scholar
22. 1. Dragoi G
    2. Tonegawa S

    (2013) Distinct preplay of multiple novel spatial experiences in the rat

    Proceedings of the National Academy of Sciences of USA 110:9100–9105.

    https://doi.org/10.1073/pnas.1306031110

    • Google Scholar
23. 1. Dupret D
    2. Pleydell-Bouverie B
    3. Csicsvari J

    (2010) Rate remapping: when the code goes beyond space

    Neuron 68:1015–1016.

    https://doi.org/10.1016/j.neuron.2010.12.011

    • Google Scholar
24. 1. Eichenbaum H

    (2000) Hippocampus: mapping or memory?

    Current Biology 10:R785–R787.

    https://doi.org/10.1016/S0960-9822(00)00763-6

    • Google Scholar
25. 1. Eichenbaum H
    2. Dudchenko P
    3. Wood E
    4. Shapiro M
    5. Tanila H

    (1999) The hippocampus, memory, and place cells: is it spatial memory or a memory space?

    Neuron 23:209–226.

    https://doi.org/10.1016/S0896-6273(00)80773-4

    • Google Scholar
26. 1. Erdem UM
    2. Hasselmo M

    (2012) A goal-directed spatial navigation model using forward trajectory planning based on grid cells

    The European Journal of Neuroscience 35:916–931.

    https://doi.org/10.1111/j.1460-9568.2012.08015.x

    • Google Scholar
27. 1. Etienne AS
    2. Jeffery KJ

    (2004) Path integration in mammals

    Hippocampus 14:180–192.

    https://doi.org/10.1002/hipo.10173

    • Google Scholar
28. 1. Fenton AA
    2. Csizmadia G
    3. Muller RU

    (2000a) Conjoint control of hippocampal place cell firing by two visual stimuli. II. A vector-field theory that predicts modifications of the representation of the environment

    The Journal of General Physiology 116:211–221.

    https://doi.org/10.1085/jgp.116.2.211

    • Google Scholar
29. 1. Fenton AA
    2. Csizmadia G
    3. Muller RU

    (2000b) Conjoint control of hippocampal place cell firing by two visual stimuli. I. The effects of moving the stimuli on firing field positions

    The Journal of General Physiology 116:191–209.

    https://doi.org/10.1085/jgp.116.2.191

    • Google Scholar
30. 1. Fenton AA
    2. Kao HY
    3. Neymotin SA
    4. Olypher A
    5. Vayntrub Y
    6. Lytton WW
    7. Ludvig N

    (2008) Unmasking the CA1 ensemble place code by exposures to small and large environments: more place cells and multiple, irregularly arranged, and expanded place fields in the larger space

    The Journal of Neuroscience 28:11250–11262.

    https://doi.org/10.1523/JNEUROSCI.2862-08.2008

    • Google Scholar
31. 1. Fenton AA
    2. Muller RU

    (1998) Place cell discharge is extremely variable during individual passes of the rat through the firing field

    Proceedings of the National Academy of Sciences of USA 95:3182–3187.

    https://doi.org/10.1073/pnas.95.6.3182

    • Google Scholar
32. 1. Fortin NJ
    2. Agster KL
    3. Eichenbaum HB

    (2002)

    Critical role of the hippocampus in memory for sequences of events

    Nature Neuroscience 5:458–462.

    • Google Scholar
33. 1. Fortin NJ
    2. Wright SP
    3. Eichenbaum H

    (2004) Recollection-like memory retrieval in rats is dependent on the hippocampus

    Nature 431:188–191.

    https://doi.org/10.1038/nature02853

    • Google Scholar
34. 1. Foster DJ
    2. Wilson MA

    (2006) Reverse replay of behavioural sequences in hippocampal place cells during the awake state

    Nature 440:680–683.

    https://doi.org/10.1038/nature04587

    • Google Scholar
35. 1. Frank LM
    2. Brown EN
    3. Wilson M

    (2000) Trajectory encoding in the hippocampus and entorhinal cortex

    Neuron 27:169–178.

    https://doi.org/10.1016/S0896-6273(00)00018-0

    • Google Scholar
36. 1. Frank LM
    2. Stanley GB
    3. Brown EN

    (2004) Hippocampal plasticity across multiple days of exposure to novel environments

    The Journal of Neuroscience 24:7681–7689.

    https://doi.org/10.1523/JNEUROSCI.1958-04.2004

    • Google Scholar
37. 1. Gothard KM
    2. Skaggs WE
    3. McNaughton BL

    (1996a)

    Dynamics of mismatch correction in the hippocampal ensemble code for space: interaction between path integration and environmental cues

    The Journal of Neuroscience 16:8027–8040.

    • Google Scholar
38. 1. Gothard KM
    2. Skaggs WE
    3. Moore KM
    4. McNaughton BL

    (1996b)

    Binding of hippocampal CA1 neural activity to multiple reference frames in a landmark-based navigation task

    The Journal of Neuroscience 16:823–835.

    • Google Scholar
39. 1. Guger C
    2. Gener T
    3. Pennartz CM
    4. Brotons-Mas JR
    5. Edlinger G
    6. Bermúdez I Badia S
    7. Verschure P
    8. Schaffelhofer S
    9. Sanchez-Vives MV

    (2011) Real-time position reconstruction with hippocampal place cells

    Frontiers in Neuroscience 5:85.

    https://doi.org/10.3389/fnins.2011.00085

    • Google Scholar
40. 1. Harnad S

    (1994)

    Why and how we are not zombies

    Journal of Consciousness Studies 1:164–167.

    • Google Scholar
41. 1. Hatcher A

    (2002)

    Algebraic topology

    Cambridge University Press.

    • Google Scholar
42. 1. Ideker T
    2. Dutkowski J
    3. Hood L

    (2011) Boosting signal-to-noise in complex biology: prior knowledge is power

    Cell 144:860–863.

    https://doi.org/10.1016/j.cell.2011.03.007

    • Google Scholar
43. 1. Jensen O
    2. Lisman JE

    (2000)

    Position reconstruction from an ensemble of hippocampal place cells: contribution of theta phase coding

    Journal of Neurophysiology 83:2602–2609.

    • Google Scholar
44. 1. Karlsson MP
    2. Frank LM

    (2008) Network dynamics underlying the formation of sparse, informative representations in the hippocampus

    The Journal of Neuroscience 28:14271–14281.

    https://doi.org/10.1523/JNEUROSCI.4261-08.2008

    • Google Scholar
45. 1. Karlsson MP
    2. Frank LM

    (2009) Awake replay of remote experiences in the hippocampus

    Nature Neuroscience 12:913–918.

    https://doi.org/10.1038/nn.2344

    • Google Scholar
46. 1. Knierim JJ

    (2002)

    Dynamic interactions between local surface cues, distal landmarks, and intrinsic circuitry in hippocampal place cells

    The Journal of Neuroscience 22:6254–6264.

    • Google Scholar
47. 1. Knierim JJ
    2. Kudrimoti HS
    3. McNaughton BL

    (1996) Neuronal mechanisms underlying the interaction between visual landmarks and path integration in the rat

    International Journal of Neural Systems 7:213–218.

    https://doi.org/10.1142/S012906579600018X

    • Google Scholar
48. 1. Knierim JJ
    2. Kudrimoti HS
    3. McNaughton BL

    (1998)

    Interactions between idiothetic cues and external landmarks in the control of place cells and head direction cells

    Journal of Neurophysiology 80:425–446.

    • Google Scholar
49. 1. Knierim JJ
    2. Rao G

    (2003) Distal landmarks and hippocampal place cells: effects of relative translation versus rotation

    Hippocampus 13:604–617.

    https://doi.org/10.1002/hipo.10092

    • Google Scholar
50. 1. Lavenex P
    2. Amaral DG

    (2000) Hippocampal-neocortical interaction: a hierarchy of associativity

    Hippocampus 10:420–430.

    https://doi.org/10.1002/1098-1063(2000)10:43.0.CO;2-5

    • Google Scholar
51. 1. Leutgeb JK
    2. Leutgeb S
    3. Treves A
    4. Meyer R
    5. Barnes CA
    6. McNaughton BL
    7. Moser MB
    8. Moser EI

    (2005) Progressive transformation of hippocampal neuronal representations in ‘morphed’ environments

    Neuron 48:345–358.

    https://doi.org/10.1016/j.neuron.2005.09.007

    • Google Scholar
52. 1. Lever C
    2. Wills T
    3. Cacucci F
    4. Burgess N
    5. O'Keefe J

    (2002) Long-term plasticity in hippocampal place-cell representation of environmental geometry

    Nature 416:90–94.

    https://doi.org/10.1038/416090a

    • Google Scholar
53. 1. Louie K
    2. Wilson MA

    (2001) Temporally structured replay of awake hippocampal ensemble activity during rapid eye movement sleep

    Neuron 29:145–156.

    https://doi.org/10.1016/S0896-6273(01)00186-6

    • Google Scholar
54. 1. Lu X
    2. Bilkey DK

    (2010) The velocity-related firing property of hippocampal place cells is dependent on self-movement

    Hippocampus 20:573–583.

    https://doi.org/10.1002/hipo.20666

    • Google Scholar
55. 1. Ludvig N

    (1999) Place cells can flexibly terminate and develop their spatial firing. A new theory for their function

    Physiology & Behavior 67:57–67.

    https://doi.org/10.1016/S0031-9384(99)00048-7

    • Google Scholar
56. 1. Lum PY
    2. Singh G
    3. Lehman A
    4. Ishkanov T
    5. Vejdemo-Johansson M
    6. Alagappan M
    7. Carlsson J
    8. Carlsson G

    (2013) Extracting insights from the shape of complex data using topology

    Scientific Reports 3:1236.

    https://doi.org/10.1038/srep01236

    • Google Scholar
57. 1. McNaughton BL
    2. Barnes CA
    3. Gerrard JL
    4. Gothard K
    5. Jung MW
    6. Knierim JJ
    7. Kudrimoti H
    8. Qin Y
    9. Skaggs WE
    10. Suster M
    11. Weaver KL

    (1996)

    Deciphering the hippocampal polyglot: the hippocampus as a path integration system

    The Journal of Experimental Biology 199:173–185.

    • Google Scholar
58. 1. McNaughton BL
    2. Barnes CA
    3. O'Keefe J

    (1983) The contributions of position, direction, and velocity to single unit activity in the hippocampus of freely-moving rats

    Experimental Brain Research 52:41–49.

    https://doi.org/10.1007/BF00237147

    • Google Scholar
59. 1. McNaughton BL
    2. Battaglia FP
    3. Jensen O
    4. Moser EI
    5. Moser MB

    (2006) Path integration and the neural basis of the ‘cognitive map’

    Nature Reviews Neuroscience 7:663–678.

    https://doi.org/10.1038/nrn1932

    • Google Scholar
60. 1. Moser EI
    2. Kropff E
    3. Moser MB

    (2008) Place cells, grid cells, and the brain's spatial representation system

    Annual Review of Neuroscience 31:69–89.

    https://doi.org/10.1146/annurev.neuro.31.061307.090723

    • Google Scholar
61. 1. Muir GM
    2. Bilkey DK

    (2003) Theta- and movement velocity-related firing of hippocampal neurons is disrupted by lesions centered on the perirhinal cortex

    Hippocampus 13:93–108.

    https://doi.org/10.1002/hipo.10052

    • Google Scholar
62. 1. Muller R
    2. Kubie J

    (1989)

    The firing of hippocampal place cells predicts the future position of freely moving rats

    The Journal of Neuroscience 9:4101–4110.

    • Google Scholar
63. 1. Muller RU
    2. Kubie JL

    (1987)

    The effects of changes in the environment on the spatial firing of hippocampal complex-spike cells

    The Journal of Neuroscience 7:1951–1968.

    • Google Scholar
64. 1. Navratilova Z
    2. Hoang LT
    3. Schwindel CD
    4. Tatsuno M
    5. McNaughton BL

    (2012) Experience-dependent firing rate remapping generates directional selectivity in hippocampal place cells

    Frontiers in Neural Circuits 6:6.

    https://doi.org/10.3389/fncir.2012.00006

    • Google Scholar
65. 1. O'Keefe J
    2. Burgess N

    (1996) Geometric determinants of the place fields of hippocampal neurons

    Nature 381:425–428.

    https://doi.org/10.1038/381425a0

    • Google Scholar
66. 1. O'Keefe J
    2. Dostrovsky J

    (1971) The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat

    Brain Research 34:171–175.

    https://doi.org/10.1016/0006-8993(71)90358-1

    • Google Scholar
67. 1. O'Keefe J
    2. Nadel L

    (1978)

    The hippocampus as a cognitive map

    Clarendon Press; Oxford University Press.

    • Google Scholar
68. 1. Poincaré H

    (1895)

    Analysis situs

    Journal de l'École Polytechnique 2:1–123.

    • Google Scholar
69. 1. Poucet B

    (1993) Spatial cognitive maps in animals: new hypotheses on their structure and neural mechanisms

    Psychological Review 100:163–182.

    https://doi.org/10.1037/0033-295X.100.2.163

    • Google Scholar
70. 1. Poucet B

    (1997) Searching for spatial unit firing in the prelimbic area of the rat medial prefrontal cortex

    Behavioural Brain Research 84:151–159.

    https://doi.org/10.1016/S0166-4328(96)00144-1

    • Google Scholar
71. 1. Poucet B
    2. Herrmann T

    (2001) Exploratory patterns of rats on a complex maze provide evidence for topological coding

    Behavioural Processes 53:155–162.

    https://doi.org/10.1016/S0376-6357(00)00151-0

    • Google Scholar
72. 1. Quiroga RQ
    2. Reddy L
    3. Kreiman G
    4. Koch C
    5. Fried I

    (2005) Invariant visual representation by single neurons in the human brain

    Nature 435:1102–1107.

    https://doi.org/10.1038/nature03687

    • Google Scholar
73. 1. Rubin A
    2. Yartsev MM
    3. Ulanovsky N

    (2014) Encoding of head direction by hippocampal place cells in bats

    The Journal of Neuroscience 34:1067–1080.

    https://doi.org/10.1523/JNEUROSCI.5393-12.2014

    • Google Scholar
74. 1. Sauvage MM
    2. Fortin NJ
    3. Owens CB
    4. Yonelinas AP
    5. Eichenbaum H

    (2008) Recognition memory: opposite effects of hippocampal damage on recollection and familiarity

    Nature Neuroscience 11:16–18.

    https://doi.org/10.1038/nn2016

    • Google Scholar
75. 1. Shapiro ML
    2. Tanila H
    3. Eichenbaum H

    (1997) Cues that hippocampal place cells encode: dynamic and hierarchical representation of local and distal stimuli

    Hippocampus 7:624–642.

    https://doi.org/10.1002/(SICI)1098-1063(1997)7:63.0.CO;2-E

    • Google Scholar
76. 1. Sharp PE
    2. Blair HT
    3. Etkin D
    4. Tzanetos DB

    (1995)

    Influences of vestibular and visual motion information on the spatial firing patterns of hippocampal place cells

    The Journal of Neuroscience 15:173–189.

    • Google Scholar
77. 1. Skaggs WE
    2. McNaughton BL

    (1996) Replay of neuronal firing sequences in rat hippocampus during sleep following spatial experience

    Science 271:1870–1873.

    https://doi.org/10.1126/science.271.5257.1870

    • Google Scholar
78. 1. Squire LR

    (1992) Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans

    Psychological Review 99:195–231.

    https://doi.org/10.1037/0033-295X.99.2.195

    • Google Scholar
79. 1. Touretzky DS
    2. Weisman WE
    3. Fuhs MC
    4. Skaggs WE
    5. Fenton AA
    6. Muller RU

    (2005) Deforming the hippocampal map

    Hippocampus 15:41–55.

    https://doi.org/10.1002/hipo.20029

    • Google Scholar
80. 1. Wills TJ
    2. Cacucci F
    3. Burgess N
    4. O'Keefe J

    (2010) Development of the hippocampal cognitive map in preweanling rats

    Science 328:1573–1576.

    https://doi.org/10.1126/science.1188224

    • Google Scholar
81. 1. Wills TJ
    2. Lever C
    3. Cacucci F
    4. Burgess N
    5. O'Keefe J

    (2005) Attractor dynamics in the hippocampal representation of the local environment

    Science 308:873–876.

    https://doi.org/10.1126/science.1108905

    • Google Scholar
82. 1. Wilson MA
    2. McNaughton BL

    (1993) Dynamics of the hippocampal ensemble code for space

    Science 261:1055–1058.

    https://doi.org/10.1126/science.8351520

    • Google Scholar
83. 1. Wood ER
    2. Dudchenko PA
    3. Eichenbaum H

    (1999) The global record of memory in hippocampal neuronal activity

    Nature 397:613–616.

    https://doi.org/10.1038/17605

    • Google Scholar
84. 1. Wood ER
    2. Dudchenko PA
    3. Robitsek RJ
    4. Eichenbaum H

    (2000) Hippocampal neurons encode information about different types of memory episodes occurring in the same location

    Neuron 27:623–633.

    https://doi.org/10.1016/S0896-6273(00)00071-4

    • Google Scholar
85. 1. Yoganarasimha D
    2. Yu X
    3. Knierim JJ

    (2006) Head direction cell representations maintain internal coherence during conflicting proximal and distal cue rotations: comparison with hippocampal place cells

    The Journal of Neuroscience 26:622–631.

    https://doi.org/10.1523/JNEUROSCI.3885-05.2006

    • Google Scholar
86. 1. Zeithamova D
    2. Schlichting ML
    3. Preston AR

    (2012) The hippocampus and inferential reasoning: building memories to navigate future decisions

    Frontiers in Human Neuroscience 6:70.

    https://doi.org/10.3389/fnhum.2012.00070

    • Google Scholar
87. 1. Zhang K
    2. Ginzburg I
    3. McNaughton BL
    4. Sejnowski TJ

    (1998)

    Interpreting neuronal population activity by reconstruction: unified framework with application to hippocampal place cells

    Journal of Neurophysiology 79:1017–1044.

    • Google Scholar

Author details

1. Yuri Dabaghian

   1. The Jan and Dan Duncan Neurological Research Institute at Texas Children's Hospital, Houston, United States
   2. Baylor College of Medicine, Houston, United States

   Contribution

   YD, Conceived of the idea, Conducted the experiments, Analyzed and interpreted data, Wrote and edited the manuscript

   For correspondence

   dabaghia@bcm.edu

   Competing interests

   The authors declare that no competing interests exist.

2. Vicky L Brandt

   1. The Jan and Dan Duncan Neurological Research Institute at Texas Children's Hospital, Houston, United States
   2. Baylor College of Medicine, Houston, United States

   Contribution

   VLB, interpreted data, wrote and edited the manuscript

   Competing interests

   The authors declare that no competing interests exist.

3. Loren M Frank

   1. Sloan-Swartz Center for Theoretical Neurobiology, W.M. Keck Center for Integrative Neuroscience, University of California, San Francisco, San Francisco, United States
   2. Department of Physiology, University of California, San Francisco, San Francisco, United States

   Contribution

   LMF, Acquisition of data, Analyzed and interpreted data

   Competing interests

   The authors declare that no competing interests exist.

• 3,662

  views

• 608

  downloads

• 101

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.