The hippocampus plays an important role in spatial processing and episodic memory (Andersen et al., 2006; Tulving, 2002). Spatial processing and navigation are supported by spatially tuned cells throughout the hippocampal formation, such as place cells in the hippocampus, which sparsely encode location within an environment (O’Keefe and Dostrovsky, 1971; Muller and Kubie, 1987). Subsequent discovery of time cells in the hippocampus (Kraus et al., 2013; Pastalkova et al., 2008; MacDonald et al., 2011; Tsao et al., 2018; Rueckemann et al., 2021), which encode time within an episode, suggests that the latter may contribute to the building blocks of episodic memory formation. Time cells and place cells share many physiological properties, pointing to a unifying concept of the role of the hippocampus in encoding features required to organize relevant information. We asked whether the encoding of hippocampal neurons is flexible, capable of changing according to behavioral context and task demand. We used previously published data by Kraus et al., 2013, from an experiment which sought to resolve an inherent ambiguity in the interpretation of time cells. Time cells were initially reported in animals running on a running wheel without control of velocity (Pastalkova et al., 2008, although time cells were also reported for immobile rats MacDonald et al., 2011). This led to a potential ambiguity between encoding of time and of distance, due to the fact that, in fixed velocities, distance may be encoded by integration of time. Kraus et al., 2013 varied the velocity of rats running in place on a treadmill, and found subpopulations of hippocampal cells that encoded time, other cells that encoded path-integrated distance and additional cells that encoded both time and distance. These experiments were composed of two types of recording sessions. In one type of session, in all the trials the running duration remained constant at different velocities, whereas in the second type, the treadmill runs accumulated up to a constant distance, at different velocities. We hypothesized that in this experiment, the task demand (i.e. constant time vs. constant distance) determined the type of activity exhibited in the corresponding session. We re-analyzed the data according to the type of behavioral session and found a direct relation between the class of most active cells and the type of session in which they were recorded. In sessions in which the rats ran for a fixed time, the cells’ population was dominated by time-encoding cells, while in sessions where they ran for a fixed distance, the population was dominated by distance-encoding cells.

To examine the dependence of hippocampal coding on task demand, we analyzed data based on experiments by Kraus et al., 2013, which aimed to differentiate between cells encoding time and cells encoding distance in the hippocampus. In these experiments, six rats were trained to run on a treadmill in the central stem of a figure-8 maze (Figure 1a). The rats were provided with a small water reward prior to the initiation of the treadmill session and upon its cessation, thereby conditioning them to maintain their snouts positioned at the water port throughout the duration of the treadmill run, and "clamping" their behavior and spatial position. In each session, consisting of 31–57 runs, the treadmill was operated either for a fixed time or for a fixed distance, where in each run the velocity was set to a randomly chosen speed in the range of 35–49 cm/s (Figure 1b and c). The rats were forced to alternate their post-treadmill turns between right and left. Three of the six rats were trained and recorded exclusively in fixed-distance or fixed-time sessions, while the remaining three rats were trained and recorded in sessions of both types.

Figure 1

Download asset Open asset

Kraus et al. reported that some cells preferentially encoded the distance the rat had run on the treadmill, while other cells preferentially encoded the time from the start of the treadmill movement. We hypothesized that the type of task employed in each session (i.e. fixed-time vs. fixed-distance) would determine the encoding of the neurons (i.e. time-based vs. distance-based). We therefore analyzed the cells on a run-by-run basis, as follows: For each neuron, we defined its response in each run according to the onset of peak firing (see Materials and methods), and examined its relation to the treadmill’s velocity. We classified time-encoding cells as those, in which the response did not systematically depend on the treadmill velocity but instead fired at a fixed time after the initiation of treadmill running. We classified distance-encoding cells as those, in which the onset time was proportional to the treadmill velocity. To examine this classification, we determined the firing onset of each cell in each run and determined the cell’s properties according to three metrics (see Materials and methods section).

We defined a metric, ${\displaystyle CellType=\frac{Var\left(distance\right)-Var\left(time\right)}{Var\left(distance\right)+Var\left(Time\right)}}$ , based on the distance and time variances (see Analysis Methods). We classified cells with positive CellType, where the distance variance is greater than the time variance as time cells, and negative CellType, where the time variance is greater than the distance variance, as distance cells. For an ideal time cell, CellType = 1 and for an ideal distance cell, CellType=-1.

Of 930 cells recorded we analyzed 679 cells with at least 10 runs showing firing peaks greater than 0.5 Hz. Only cells with peak firing rates occurring during the treadmill run were included in the analysis. As previously reported in Kraus et al., 2013, we observed both distance cells, showing a response at constant distances the animal traveled on the treadmill, and time cells, showing a response at constant times from the treadmill start (Figure 2). In line with our hypothesis, there was a clear relation between the types of experiment and the distribution of time coding and distance coding neurons. In fixed-distance sessions, the neurons exhibited a significant majority (67%) of distance cells. By contrast, in fixed-time sessions time encoding cells dominated (68%). Of the 444 neurons recorded in fixed-distance sessions, the CellType index classified 298 cells as distance cells and 146 as time cells (Figure 3a). Conversely, in the fixed-time sessions 76 of 235 neurons were classified by this index as distance cells and 159 of 235 were classified as time cells. The relation between the cell type and the experiment type is significant (χ²(1)=75.1, p>>0.001 for the total cell population). These proportions were maintained when classifying by other metrics and when using the peak of firing instead of the onset of the response (see supplementary Methods and Figure 3—figure supplement 1). In 5 out of 6 animals, the cells’ encoding depended on the session type (Figure 3—figure supplement 2). (χ²(1)>12, p<<0.001 for 4 animals, χ²(1)=5.7, p<0.02 for one animal), except in one animal (χ²(1)=2.38, p=0.12).

Figure 2 with 1 supplement

Download asset Open asset

Figure 2—figure supplement 1

Download asset Open asset

Figure 3 with 3 supplements

Download asset Open asset

Figure 3—figure supplement 3

Download asset Open asset

Figure 3—figure supplement 2

Download asset Open asset

Figure 3—figure supplement 1

Download asset Open asset

We then checked the Receiver Operating Characteristics (ROC) graph of the CellType metric (Figure 2d), using discriminating thresholds in the range of [–1,+1]. The ROC plots the True Positive Rate (TPR) defined as the percentage of cell classified as distance cells in the fixed-distance session, against the False Positive Rate (FPR) defined as the percentage of cells classified as distance cells in the fixed-timed sessions. We found that the optimal threshold (maximum Youden index) for classification of a session based on the CellType metric, was 0. This threshold classifies 67% of the cells as Distance in the Distance sessions and only 32% on the Time sessions. Accordingly, we classified a cell as a time-cell if CellType >0 and as a distance-cell if CellType <0.

To assess the power of the statistics, we compared the results to a distribution generated from shuffled session types (Figure 3c). In order to mitigate any potential biases in this distribution, we truncated all data to a common duration of 16 s, which represents the shortest duration of a treadmill run across all sessions. These results indicate that the dimension the cells encode (Time vs. Distance) is related to the session type (fixed time vs. fixed distance).

Classifying neuronal activity according to either time or distance revealed that the hippocampal population encoding strongly registered with the features of the experimental task. In experiments where the treadmill running-time was fixed, the majority of cells encoded a given time from treadmill onset. In contrast, in experiments where the treadmill running-distance was fixed, the majority of cells encoded a specific accumulated distance from treadmill onset. It is worth noting that accumulated time in fixed-time experiments and accumulated distance in fixed-distance experiments may be used as predictors for the progress of the rat towards anticipated reward, which is given at the end of the treadmill run (Whittington et al., 2020; Stachenfeld et al., 2017). As noted previously in Kraus et al., 2013 the same cells, which showed distance-encoding and time-encoding properties in the treadmill, were often selective to places outside of the treadmill as well. To summarize, CA1 pyramidal cells can encode location, distance, or time, depending on the conditions of the experiment or task demand.

Consistency with task demands has been repeatedly demonstrated in hippocampal recording for diverse parameter spaces, such as auditory linear frequency (Aronov et al., 2017), social mapping (Omer et al., 2018; Schafer and Schiller, 2018) or more abstract spaces (Constantinescu et al., 2016; Retailleau and Morris, 2018). How is task-relevant encoding achieved? The activity of place cells and grid cells is commonly modeled using continuous attractor networks (Morris and Derdikman, 2023; Samsonovich and McNaughton, 1997; Burak and Fiete, 2009; Fuhs and Touretzky, 2006; Bush and Burgess, 2014; Hasselmo and Brandon, 2012; Tocker et al., 2015; Gu et al., 2018; Geiller et al., 2022; O’Keefe and Burgess, 1996). Such networks may serve as a natural substrate for amplification of encoding of certain task features, at the expense of others, resulting in an over representation of the salient variables. Such over-representation may help the brain prioritize survival and decision making. By allocating more resources to the salient stimuli, the brain enhances the ability to process and retain the important information.

Another possible mechanism for acquiring representations that are consistent with task structure involves an associative learning process. Such learning would strengthen all connections between cells that were active in a particular trial and weaken others, but ultimately only the connections between cells that are consistently co-active will be strengthened, while others will average out. Thus, in time-fixed experiments, the connections to cells that fire in a manner that is consistent with time would be strengthened, while in distance-fixed experiments connections to cells that are consistent with distance would gain strength. Consequently, those cells will gradually encode either distance or time, depending on the type of experiment.

Irrespective of the exact mechanism explaining the results of this study, the hippocampus is adaptive in its cells’ encoding and seems to be capable to tune them to the parameters best describing the task.

We used the data provided by Kraus et al., 2013, containing the neurons firing times, the treadmill movement times and the treadmill velocity. The data was analyzed using custom Matlab scripts.

We divided the treadmill moving times into 100ms time bins (other bin resolutions between 50 ms and 500 ms were tested and provided similar results). Response onset for each neuron and run, was defined as the first bin, following at least 1 s of silence within a series of consecutive bins with firing activity which includes the peak of firing. The peak firing was defined as the bin with the highest value within a run, however only cells with an average peak firing of at least 0.5 Hz were included. This methodology was applied consistently across all runs within a session. We chose this approach to mitigate potential biases that could arise from firing rate peaks occurring near the end of the treadmill, which might have been truncated. Basing our analysis on the peak values instead yielded comparable results and levels of statistical significance (Figure 3—figure supplement 1).

Our classification method is based on the premise that for ideal time cells, firing (and hence onset time) should be independent of running speed. Conversely, an ideal distance cell would display firing onsets at times depending on the speed. We therefore performed a linear regression between the onset times ($T_i$) and the reciprocal velocity (${\displaystyle 1/V_i}$) as well as the onset distance ($S_i$) and the velocity ($V_i$), where i is an index to a specific run on the treadmill, and extracted the slopes (m and k) and offsets (n and q), as described in Equations 1 and 2. Consequently, in the case of a time-encoding cell, the velocity dependent term ${\displaystyle k\ast \frac{1}{V_i}}$ would be small in comparison to the constant q, while for a distance encoding cell, the slope k would approximate the estimated encoded distance. Similarly, in accordance with the relation stated in Equation 2, an ideal distance encoding cell would exhibit a small velocity dependent term $m*V_i$ compared to the constant n, while a time encoding cell would have an estimated encoded time equivalent to the slope m.

The CellType metric utilizes the variances of the onset times ${\left(T_i-\overline{T}\right)}^2$, where the average onset time is computed across all runs within the session, and the onset distances ${\displaystyle {\left(S_i-\overline{S}\right)}^2}$, where the average onset distance is calculated across all runs within the session.

CellType is in the range of –1 to +1. For an ideal time-encoding cell, the onset variance ${\displaystyle \sum _i{\left(T_i-\overline{T}\right)}^2=0}$, and hence CellType = 1. For an ideal distance encoding cell, the distance variance (multiplied by the respective velocity in order to match units) ${\displaystyle V_i\ast {\left(S_i-\overline{S}\right)}^2=0}$, and hence CellType=-1.

Additional metrics defined and used for classifying the cells encoding:

The “FIT” metric is defined as follows:

Where m and k are the linear fit slope coefficients (from Equations 1 and 2), $\overline{T}$ is the average firing onset time and $\overline{S}$ is the average distance the animal traveled until the onset. Fit is –1 for a distance cell and 1 for a time cell.

The ‘p-value’ metric is defined as follows:

A stricter metric, utilized the statistical significance of the linearity in Equations 1 and 2, through F-statistics. We classified a cell as distance encoding if the null hypothesis that there is no linear relation between the distance and velocity was rejected with p<0.05. We classified a cell as time encoding if the null hypothesis that there is no linear relation between the onset and the reciprocal velocity was rejected with p<0.05.

Results using these metrics are shown in Figure 3—figure supplement 1.

To ensure the activity peak is not missed, we extended the analysis to 5 s past the treadmill stop time. Otherwise, if a cell activity is concentrated towards the treadmill stop, the calculated onset may be influenced by the truncated activity time and show a false relation of the cell type activity to the time or distance. Moreover, since the truncated data time relates to the experiment type, whether time-fixed or distance-fixed, this could create a false bias of such a relation.

The relation between the type of cell classified in the above metrics and the session type was then tested by Pearson’s chi-square using two categories. The expected distribution of the cells was calculated based on the total number of cells, of each type, out of total cells number, in all sessions. The null hypothesis was defined as no dependency of the cells type distribution on the session type (either fixed-time or fixed-distance). On the per-animal analysis, for animals that were recorded only at a single type of session, we used the distribution of the cell types according to their distribution in all animals’ cells population.

We conducted additional analysis to explore potential relationships between the firing rates and the encoding properties of the cells. Our findings revealed that the distributions of peak firing rates and average firing rates, for time cells and distance cells, were similar (see Figure 3—figure supplement 3).

The current manuscript is a re-analysis of data collected for a previously published paper (Kraus et al., 2013). Data used in this paper is available as Matlab files on Dryad: https://doi.org/10.5061/dryad.ngf1vhhxp. Matlab code used for the analysis in the paper is available on https://github.com/derdikman/Abramson_code (copy archived at Derdikman, 2023).

1. 1. Abramson S
   2. Kraus BJ
   3. White JA
   4. Hasselmo ME
   5. Morris G
   6. Derdikman D

   (2023) Dryad Digital Repository

   Data for Time or distance: predictive coding of hippocampal cells.

   https://doi.org/10.5061/dryad.ngf1vhhxp

1. Book

   1. Andersen P
   2. Morris R
   3. Amaral D
   4. Bliss T
   5. O’keefe J

   (2006)

   The Hippocampus Book

   Oxford University Press.

   • Google Scholar
2. 1. Aronov D
   2. Nevers R
   3. Tank DW

   (2017) Mapping of a non-spatial dimension by the hippocampal-entorhinal circuit

   Nature 543:719–722.

   https://doi.org/10.1038/nature21692

   • PubMed
   • Google Scholar
3. 1. Burak Y
   2. Fiete IR

   (2009) Accurate path integration in continuous attractor network models of grid cells

   PLOS Computational Biology 5:e1000291.

   https://doi.org/10.1371/journal.pcbi.1000291

   • PubMed
   • Google Scholar
4. 1. Bush D
   2. Burgess N

   (2014) A hybrid oscillatory interference/continuous attractor network model of grid cell firing

   The Journal of Neuroscience 34:5065–5079.

   https://doi.org/10.1523/JNEUROSCI.4017-13.2014

   • PubMed
   • Google Scholar
5. 1. Constantinescu AO
   2. O’Reilly JX
   3. Behrens TEJ

   (2016) Organizing conceptual knowledge in humans with a gridlike code

   Science 352:1464–1468.

   https://doi.org/10.1126/science.aaf0941

   • PubMed
   • Google Scholar
6. Software

   1. Derdikman D

   (2023) Abramson_Code, version swh:1:rev:943dead3e093eea9c105ae83cebdd94b9011268c

   Software Heritage.

   https://archive.softwareheritage.org/swh:1:dir:9911cd4dc8ea759de60bcc5830e4edbc16a6af9a;origin=https://github.com/derdikman/Abramson_code;visit=swh:1:snp:8762ad64efee2d0f105bb0d94ae52ca67736eedb;anchor=swh:1:rev:943dead3e093eea9c105ae83cebdd94b9011268c
7. 1. Fuhs MC
   2. Touretzky DS

   (2006) A spin glass model of path integration in rat medial entorhinal cortex

   The Journal of Neuroscience 26:4266–4276.

   https://doi.org/10.1523/JNEUROSCI.4353-05.2006

   • PubMed
   • Google Scholar
8. 1. Geiller T
   2. Sadeh S
   3. Rolotti SV
   4. Blockus H
   5. Vancura B
   6. Negrean A
   7. Murray AJ
   8. Rózsa B
   9. Polleux F
   10. Clopath C
   11. Losonczy A

   (2022) Local circuit amplification of spatial selectivity in the hippocampus

   Nature 601:105–109.

   https://doi.org/10.1038/s41586-021-04169-9

   • PubMed
   • Google Scholar
9. 1. Gu Y
   2. Lewallen S
   3. Kinkhabwala AA
   4. Domnisoru C
   5. Yoon K
   6. Gauthier JL
   7. Fiete IR
   8. Tank DW

   (2018) A map-like micro-organization of grid cells in the medial entorhinal cortex

   Cell 175:736–750.

   https://doi.org/10.1016/j.cell.2018.08.066

   • PubMed
   • Google Scholar
10. 1. Hasselmo ME
    2. Brandon MP

    (2012) A model combining oscillations and attractor dynamics for generation of grid cell firing

    Frontiers in Neural Circuits 6:30.

    https://doi.org/10.3389/fncir.2012.00030

    • PubMed
    • Google Scholar
11. 1. Kraus BJ
    2. Robinson RJ
    3. White JA
    4. Eichenbaum H
    5. Hasselmo ME

    (2013) Hippocampal “time cells”: time versus path integration

    Neuron 78:1090–1101.

    https://doi.org/10.1016/j.neuron.2013.04.015

    • PubMed
    • Google Scholar
12. 1. MacDonald CJ
    2. Lepage KQ
    3. Eden UT
    4. Eichenbaum H

    (2011) Hippocampal “time cells” bridge the gap in memory for discontiguous events

    Neuron 71:737–749.

    https://doi.org/10.1016/j.neuron.2011.07.012

    • PubMed
    • Google Scholar
13. 1. Morris G
    2. Derdikman D

    (2023) The chicken and egg problem of grid cells and place cells

    Trends in Cognitive Sciences 27:125–138.

    https://doi.org/10.1016/j.tics.2022.11.003

    • PubMed
    • Google Scholar
14. 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.

    https://doi.org/10.1523/JNEUROSCI.07-07-01951.1987

    • PubMed
    • Google Scholar
15. 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

    • PubMed
    • Google Scholar
16. 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

    • PubMed
    • Google Scholar
17. 1. Omer DB
    2. Maimon SR
    3. Las L
    4. Ulanovsky N

    (2018) Social place-cells in the bat hippocampus

    Science 359:218–224.

    https://doi.org/10.1126/science.aao3474

    • PubMed
    • Google Scholar
18. 1. Pastalkova E
    2. Itskov V
    3. Amarasingham A
    4. Buzsáki G

    (2008) Internally generated cell assembly sequences in the rat hippocampus

    Science 321:1322–1327.

    https://doi.org/10.1126/science.1159775

    • PubMed
    • Google Scholar
19. 1. Retailleau A
    2. Morris G

    (2018) Spatial rule learning and corresponding CA1 place cell reorientation depend on local dopamine release

    Current Biology 28:836–846.

    https://doi.org/10.1016/j.cub.2018.01.081

    • PubMed
    • Google Scholar
20. 1. Rueckemann JW
    2. Sosa M
    3. Giocomo LM
    4. Buffalo EA

    (2021) The grid code for ordered experience

    Nature Reviews. Neuroscience 22:637–649.

    https://doi.org/10.1038/s41583-021-00499-9

    • PubMed
    • Google Scholar
21. 1. Samsonovich A
    2. McNaughton BL

    (1997) Path integration and cognitive mapping in a continuous attractor neural network model

    The Journal of Neuroscience 17:5900–5920.

    https://doi.org/10.1523/JNEUROSCI.17-15-05900.1997

    • PubMed
    • Google Scholar
22. 1. Schafer M
    2. Schiller D

    (2018) Navigating social space

    Neuron 100:476–489.

    https://doi.org/10.1016/j.neuron.2018.10.006

    • PubMed
    • Google Scholar
23. 1. Stachenfeld KL
    2. Botvinick MM
    3. Gershman SJ

    (2017) The hippocampus as a predictive map

    Nature Neuroscience 20:1643–1653.

    https://doi.org/10.1038/nn.4650

    • PubMed
    • Google Scholar
24. 1. Tocker G
    2. Barak O
    3. Derdikman D

    (2015) Grid cells correlation structure suggests organized feedforward projections into superficial layers of the medial entorhinal cortex

    Hippocampus 25:1599–1613.

    https://doi.org/10.1002/hipo.22481

    • PubMed
    • Google Scholar
25. 1. Tsao A
    2. Sugar J
    3. Lu L
    4. Wang C
    5. Knierim JJ
    6. Moser M-B
    7. Moser EI

    (2018) Integrating time from experience in the lateral entorhinal cortex

    Nature 561:57–62.

    https://doi.org/10.1038/s41586-018-0459-6

    • PubMed
    • Google Scholar
26. 1. Tulving E

    (2002) Episodic memory: from mind to brain

    Annual Review of Psychology 53:1–25.

    https://doi.org/10.1146/annurev.psych.53.100901.135114

    • PubMed
    • Google Scholar
27. 1. Whittington JCR
    2. Muller TH
    3. Mark S
    4. Chen G
    5. Barry C
    6. Burgess N
    7. Behrens TEJ

    (2020) The tolman-eichenbaum machine: Unifying space and relational memory through generalization in the hippocampal formation

    Cell 183:1249–1263.

    https://doi.org/10.1016/j.cell.2020.10.024

    • PubMed
    • Google Scholar

Author details

1. Shai Abramson

   Rappaport Faculty of Medicine and Research Institute, Technion - Israel Institute of Technology, Haifa, Israel

   Contribution

   Conceptualization, Software, Investigation, Methodology, Writing - original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5437-7992

2. Benjamin J Kraus

   Center for Memory and Brain, Boston University, Boston, United States

   Contribution

   Conceptualization, Data curation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

3. John A White

   Department of Biomedical Engineering, Boston University, Boston, United States

   Contribution

   Conceptualization, Data curation, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1073-2638

4. Michael E Hasselmo

   Center for Memory and Brain, Boston University, Boston, United States

   Contribution

   Conceptualization, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

5. Dori Derdikman

   Rappaport Faculty of Medicine and Research Institute, Technion - Israel Institute of Technology, Haifa, Israel

   Contribution

   Conceptualization, Supervision, Investigation, Methodology, Writing – review and editing

   Contributed equally with

   Genela Morris

   For correspondence

   derdik@technion.ac.il

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3677-6321

6. Genela Morris

   1. Rappaport Faculty of Medicine and Research Institute, Technion - Israel Institute of Technology, Haifa, Israel
   2. Tel Aviv Sourasky Medical Center, Tel Aviv, Israel

   Contribution

   Conceptualization, Supervision, Investigation, Methodology, Writing – review and editing

   Contributed equally with

   Dori Derdikman

   For correspondence

   genelam@tlvmc.gov.il

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-5417-8977

• 1,619

  views

• 297

  downloads

• 9

  citations

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