To guide spatial behavior, the brain must retrieve memories that are appropriately associated with different navigational contexts. Contextual memory might be mediated by cell ensembles in the hippocampal formation that alter their responses to changes in context, processes known as remapping and realignment in the hippocampus and entorhinal cortex, respectively. However, whether remapping and realignment guide context-dependent spatial behavior is unclear. To address this issue, human participants learned object–location associations within two distinct virtual reality environments and subsequently had their memory tested during functional MRI (fMRI) scanning. Entorhinal grid-like representations showed realignment between the two contexts, and coincident changes in fMRI activity patterns consistent with remapping were observed in the hippocampus. Critically, in a third ambiguous context, trial-by-trial remapping and realignment in the hippocampal–entorhinal network predicted context-dependent behavior. These results reveal the hippocampal–entorhinal mechanisms mediating human contextual memory and suggest that the hippocampal formation plays a key role in spatial behavior under uncertainty.
Subscribe to Journal
Get full journal access for 1 year
only $4.92 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
The unthresholded group-level statistical brain map depicted in Fig. 5b is available on NeuroVault (collection: VABINDPD). Source data for the remaining figures are available at Mendeley Data (https://doi.org/10.17632/jvm3fhpjwn.1). Raw data contained in this manuscript (Figs. 2–6 and Extended Data Figs. 1–10) are available from the corresponding authors upon reasonable request.
Task code is publicly available at https://github.com/jbjulian/squircle_task. fMRI analyses were performed using the publicly available software packages FSL and FreeSurfer. MATLAB code for other analyses is available from the corresponding authors upon reasonable request.
El Haj, M. & Kessels, R. P. Context memory in Alzheimer’s disease. Dement. Geriatr. Cogn. Diso. Extra 3, 342–350 (2013).
Chun, M. M. & Phelps, E. A. Memory deficits for implicit contextual information in amnesic subjects with hippocampal damage. Nat. Neurosci. 2, 844–847 (1999).
Colgin, L. L., Moser, E. I. & Moser, M.-B. Understanding memory through hippocampal remapping. Trends Neurosci. 31, 469–477 (2008).
Mizumori, S. J. Y. A context for hippocampal place cells during learning. in Hippocampal Place Fields: Relevance to Learning and Memory (ed. Mizumori, S. J. Y.) 16–43 (Oxford Univ. Press, 2008).
Eichenbaum, H. The hippocampus and declarative memory: cognitive mechanisms and neural codes. Behav. Brain Res. 127, 199–207 (2001).
Myers, C. E. & Gluck, M. A. Context, conditioning, and hippocampal rerepresentation in animal learning. Behav. Neurosci. 108, 835 (1994).
Nadel, L. The hippocampus and context revisited. in Hippocampal Place Fields: Relevance to Learning and Memory (ed. Mizumori, S. J. Y.) 3–15 (Oxford Univ. Press, 2008).
O’Keefe, J. & Nadel, L. The Hippocampus as a Cognitive Map (Clarendon Press, 1978).
Ranganath, C. Binding items and contexts: the cognitive neuroscience of episodic memory. Curr. Dir. Psychol. Sci. 19, 131–137 (2010).
Hirsh, R. The hippocampus and contextual retrieval of information from memory: a theory. Behav. Biol. 12, 421–444 (1974).
Julian, J. B. & Doeller, C. F. Context in spatial and episodic memory. in The Cognitive Neurosciences (eds Poeppel, D., Mangun, G. R. & Gazzaniga, M. S.) 219–234 (MIT Press, 2019).
O’Keefe, J. & Dostrovsky, J. The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res. 34, 171–175 (1971).
Alme, C. B. et al. Place cells in the hippocampus: eleven maps for eleven rooms. Proc. Natl Acad. Sci. USA 111, 18428–18435 (2014).
Leutgeb, S., Leutgeb, J. K., Treves, A., Moser, M.-B. & Moser, E. I. Distinct ensemble codes in hippocampal areas CA3 and CA1. Science 305, 1295–1298 (2004).
Kyle, C. T., Stokes, J. D., Lieberman, J. S., Hassan, A. S. & Ekstrom, A. D. Successful retrieval of competing spatial environments in humans involves hippocampal pattern separation mechanisms. eLife 4, e10499 (2015).
Steemers, B. et al. Hippocampal attractor dynamics predict memory-based decision making. Curr. Biol. 26, 1750–1757 (2016).
Stokes, J., Kyle, C. & Ekstrom, A. D. Complementary roles of human hippocampal subfields in differentiation and integration of spatial context. J. Cogn. Neurosci. 27, 546–559 (2015).
Witter, M. P. & Amaral, D. G. The hippocampal region. in The Rat Nervous System (ed. Paxinos, G.) 637–703 (Elsevier, 2004).
Diehl, G. W., Hon, O. J., Leutgeb, S. & Leutgeb, J. K. Grid and nongrid cells in medial entorhinal cortex represent spatial location and environmental features with complementary coding schemes. Neuron 94, 83–92 (2017).
Fyhn, M., Hafting, T., Treves, A., Moser, M.-B. & Moser, E. I. Hippocampal remapping and grid realignment in entorhinal cortex. Nature 446, 190–194 (2007).
Hafting, T., Fyhn, M., Molden, S., Moser, M.-B. & Moser, E. I. Microstructure of a spatial map in the entorhinal cortex. Nature 436, 801–806 (2005).
Doeller, C. F., Barry, C. & Burgess, N. Evidence for grid cells in a human memory network. Nature 463, 657–661 (2010).
Nadasdy, Z. et al. Context-dependent spatially periodic activity in the human entorhinal cortex. Proc. Natl Acad. Sci. USA 114, E3516–E3525 (2017).
Ainge, J. A., Dudchenko, P. A. & Wood, E. R. Context-dependent firing of hippocampal place cells: does it underlie memory? in Hippocampal Place Fields: Relevance to Learning and Memory (ed. Mizumori, S. J. Y.) 44–58 (Oxford Univ. Press, 2008).
He, Q. & Brown, T. I. Environmental barriers disrupt grid-like representations in humans during navigation. Curr. Biol. 29, 2718–2722 (2019).
Julian, J. B., Keinath, A. T., Frazzetta, G. & Epstein, R. A. Human entorhinal cortex represents visual space using a boundary-anchored grid. Nat. Neurosci. 21, 191–194 (2018).
Krupic, J., Bauza, M., Burton, S., Barry, C. & O’Keefe, J. Grid cell symmetry is shaped by environmental geometry. Nature 518, 232–235 (2015).
Stensola, T., Stensola, H., Moser, M.-B. & Moser, E. I. Shearing-induced asymmetry in entorhinal grid cells. Nature 518, 207–212 (2015).
Davies, G. M. & Thomson, D. M. Memory in Context: Context in Memory (John Wiley & Sons, 1988).
Nadel, L. & Willner, J. Context and conditioning: a place for space. Physiol. Psychol. 8, 218–228 (1980).
Schimanski, L. A., Lipa, P. & Barnes, C. A. Tracking the course of hippocampal representations during learning: when Is the map required? J. Neurosci. 33, 3094–3106 (2013).
Kennedy, P. J. & Shapiro, M. L. Motivational states activate distinct hippocampal representations to guide goal-directed behaviors. Proc. Natl Acad. Sci. USA 106, 10805–10810 (2009).
Jeffery, K. J., Gilbert, A., Burton, S. & Strudwick, A. Preserved performance in a hippocampal-dependent spatial task despite complete place cell remapping. Hippocampus 13, 175–189 (2003).
Bower, M. R., Euston, D. R. & McNaughton, B. L. Sequential-context-dependent hippocampal activity is not necessary to learn sequences with repeated elements. J. Neurosci. 25, 1313–1323 (2005).
Miller, J. F. et al. Neural activity in human hippocampal formation reveals the spatial context of retrieved memories. Science 342, 1111–1114 (2013).
Nau, M., Julian, J. B. & Doeller, C. F. How the brain’s navigation system shapes our visual experience. Trends Cogn. Sci. 22, 810–825 (2018).
Gulli, R. A. et al. Context-dependent representations of objects and space in the primate hippocampus during virtual navigation. Nat. Neurosci. 23, 103–112 (2020).
Nau, M., Schröder, T. N., Frey, M. & Doeller, C. F. Behavior-dependent directional tuning in the human visual-navigation network. Nat. Commun. 11, 1–13 (2020).
Turk-Browne, N. B. The hippocampus as a visual area organized by space and time: a spatiotemporal similarity hypothesis. Vis. Res. 165, 123–130 (2019).
Yassa, M. A. & Stark, C. E. Pattern separation in the hippocampus. Trends Neurosci. 34, 515–525 (2011).
Moser, M.-B., Rowland, D. C. & Moser, E. I. Place cells, grid cells, and memory. Cold Spring Harb. Perspect. Biol. 7, a021808 (2015).
McNaughton, B. L., Battaglia, F. P., Jensen, O., Moser, E. I. & Moser, M.-B. Path integration and the neural basis of the ‘cognitive map’. Nat. Rev. Neurosci. 7, 663–678 (2006).
Marozzi, E., Ginzberg, L. L., Alenda, A. & Jeffery, K. J. Purely translational realignment in grid cell firing patterns following nonmetric context change. Cereb. Cortex 25, 4619–4627 (2015).
Mathis, A., Herz, A. V. & Stemmler, M. Optimal population codes for space: grid cells outperform place cells. Neural Comput. 24, 2280–2317 (2012).
Sreenivasan, S. & Fiete, I. Grid cells generate an analog error-correcting code for singularly precise neural computation. Nat. Neurosci. 14, 1330 (2011).
Latuske, P., Kornienko, O., Kohler, L. & Allen, K. Hippocampal remapping and its entorhinal origin. Front. Behav. Neurosci. 11, 253 (2018).
Monaco, J. D. & Abbott, L. F. Modular realignment of entorhinal grid cell activity as a basis for hippocampal remapping. J. Neurosci. 31, 9414–9425 (2011).
Jun, H. et al. Disrupted place cell remapping and impaired grid cells in a knockin model of Alzheimer’s disease. Neuron 107, 1095–1112 (2020).
Schlesiger, M. I., Boublil, B. L., Hales, J. B., Leutgeb, J. K. & Leutgeb, S. Hippocampal global remapping can occur without input from the medial entorhinal cortex. Cell Rep. 22, 3152–3159 (2018).
Xu, Y., Regier, T. & Newcombe, N. S. An adaptive cue combination model of human spatial reorientation. Cognition 163, 56–66 (2017).
Alt, H. & Godau, M. Computing the Fréchet distance between two polygonal curves. Int. J. Comput. Geom. Appl. 5, 75–91 (1995).
Desikan, R. S. et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 31, 968–980 (2006).
Nielson, D. M., Smith, T. A., Sreekumar, V., Dennis, S. & Sederberg, P. B. Human hippocampus represents space and time during retrieval of real-world memories. Proc. Natl Acad. Sci. USA 112, 11078–11083 (2015).
Nolan, C. R., Vromen, J. M., Cheung, A. & Baumann, O. Evidence against the detectability of a hippocampal place code using functional magnetic resonance imaging. eNeuro 5, ENEURO.0177-18.2018 (2018).
Bellmund, J. L., Deuker, L., Schröder, T. N. & Doeller, C. F. Grid-cell representations in mental simulation. eLife 5, e17089 (2016).
We thank L. Sommer and L. Sandøy for help with data collection and M. Nau for discussions. This study was supported by the European Research Council (ERC-CoG GEOCOG 724836). We acknowledge further support to C.F.D. from the Max Planck Society; the Netherlands Organisation for Scientific Research (NWO-Vidi 452-12-009); the Kavli Foundation; the Centre of Excellence scheme of the Research Council of Norway – Centre for Neural Computation (223262/F50); the Egil and Pauline Braathen and Fred Kavli Centre for Cortical Microcircuits; the National Infrastructure scheme of the Research Council of Norway – NORBRAIN (197467/F50); and the Norwegian University of Science and Technology.
The authors declare no competing interests.
Peer review Information Nature Neuroscience thanks the anonymous reviewers for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Performance was measured as the mean distance error in object replacement in virtual meters (vm), separately for the Square (Sq) and Circle (Ci). Performance improved across training and did not differ between the Sq and Ci at the end of training (Student’s paired t-test, t23 = 1.18, p = 0.250) or during testing (t23 = 0.95, p = 0.352). Note that because the Sq had a larger walkable ground plane than the Ci, performance error expected by random chance was lower in the Ci (11.25 vm) than the Sq (13.78 vm). However, there was still no significant difference between Sq and Ci performance by training block 4 (t23 = 1.72, p = 0.098) or during testing (t23 = 1.92, p = 0.068) after normalizing distance error by the context-specific chance baseline. Error bars indicate ±1 SEM (n = 24). All t-tests two-tailed.
Extended Data Fig. 2 No systematic behavioral differences between Square- and Circle-consistent Squircle trials.
a–h, Comparison of behavioral measures between Square-consistent (Sq-con) and Circle-consistent (Ci-con) Squircle trials: a, response latency (Sq-con vs Ci-con: t23 = 0.30, p = 0.767), b, path tortuosity (t23 = 0.87, p = 0.393), c, distance of replaced objects from the context-consistent location (t23 = 0.76, p = 0.456), d, mean absolute angular difference in allocentric VR-facing direction between successive fMRI volumes, which reflects tendency to rotate VR-heading (t23 = 1.01, p = 0.325), e, percentage (%) of time spent in locations proximal to Sq (vs. Ci) boundary segments (t23 = 0.697, p = 0.394), f, % time viewing distal cues present in the Sq (vs. Ci) (t23 = 0.77, p = 0.448), g, % time spent viewing square (vs. circular) boundary segments (t23 = 1.58, p = 0.127), and h, % time spent VR-walking (vs. VR-stationary) (t23 = 0.06, p = 0.956). i, Comparison between the shapes of Squircle VR-walking path trajectories across trials. For each pair of trajectories (T1 and T2), (i) T2 was interpolated to have the same number of samples as T1 using a spline fit, (ii) a Procrustes transformation maximally aligned T1 to T2, and (iii) the distance between T1 and T2 was computed using a Fréchet distance metric (see Methods). j, The shape of VR-walking path trajectories did not reliably differ between Sq- and Ci-con Squircle trials: the average Frèchet distance between trajectories from trials sharing the same contextual memory (Same) did not differ from those with different contextual memories (Diff) (t23 = 0.39, p = 0.704). k, The Frèchet distance was computed between each Squircle trajectory and trajectories in the Sq and Ci. The difference between Squircle-Sq minus Squircle-Ci Frèchet distances were then computed, separately for Sq- and Ci-con Squircle trials. Positive (negative) values thus indicate that Squircle trajectories were more like those in the Sq (Ci). Squircle trajectories were more similar in shape to Sq trajectories, during both Sq-con (t23 = 2.56, p = 0.018) and Ci-con (t23 = 2.24, p = 0.035) trials. Critically, trajectory shapes did not differ in their similarity to the Sq or Ci across Sq- and Ci-con Squircle trials (t23 = 0.12, p = 0.902). l, Percentage of time occupying different locations in the Squircle during Sq-con (left) and Ci-con (right) trials, averaged across participants. Only 2% (1/49) of local Squircle regions showed differential sampling between Sq- and Ci-con trials (t-test, p < 0.05, uncorrected for multiple comparisons). Throughout the figure, error bars indicate ±1 SEM. All t-test two-tailed. Dots denote individual participants (n = 24). nsp > 0.05, *p < 0.05.
a, We examined five different possible determinants of trial-by-trial contextual memory retrieval in the Squircle. For each possible determinant, we tested whether that determinant predicted contextual memory retrieval on each Squircle trial (that is, a Square- or Circle-consistent replace location) and computed classification accuracy across all trials for each participant (% correct, compared to 50% theoretical chance baseline dotted line). The five determinants included: (i) The modal Square (Sq) or Circle (Ci) context during the N odd testing trials prior to each individual Squircle trial. Although trial order was balanced across all Squircle trials, for each individual trial, participants were tested more often in either the Sq or Ci every odd number of prior Sq and Ci trials, which may bias contextual memory. For example, for a given Squircle trial, if the three preceding trials were two Sq and one Ci, a participant may be more likely to retrieve a Sq-consistent memory. We thus tested whether Squircle memory retrieval on each individual Squircle trial was predicted by the modal context of the N = 1, 3, 5 prior Sq and Ci trials. However, replace locations in the Squircle were not predicted by the previous trials’ modal context (t23 = 0.98, 1.04, 1.75, pFWE = 0.508, 0.462, 0.140 for lags 1,3, and 5, respectively). (ii) Distance error during training. Although distance error during training was matched on average between the Sq and Ci, for each target object each participant had numerically lower distance error for that object in one of the two training contexts. For each participant, we tested whether Squircle contextual memory was determined by whichever training context had the lowest distance error for each object. Classification accuracy was not significantly higher than expected by chance (t23 = 0.985, p = 0.335). (iii) Squircle trial start location. Each Squircle trial started a random arena location, which may have biased contextual memory. For each Squircle trial, we tested whether contextual memory retrieval was predicted by whichever context-consistent location was nearest to that trial’s starting location. Prediction accuracy was not significantly higher than expected by chance (t23 = 1.41, p = 0.175). iv) Contextual memory stability (Mem Stability) across Squircle trials within (WI) the same participant. For each target object, did each participant retrieve the same contextual memory for that object across all Squircle trials? For each Squircle trial, we tested whether memory was predicted by the modal contextual memory retrieved on all other Squircle trials for the same target object. We repeated this separately for each participant. For each target object, participants tended to retrieve the same contextual memory for that object across Squircle trials (t23 = 5.28, p = 2.34 × 10−5). v) Mem Stability across Squircle trials between (BW) participants. For each target object, did each participant retrieve the same contextual memory for that object as the other participants? For each Squircle trial, we tested whether memory was predicted by the modal contextual memory retrieved for that trial’s target object by a different participant across all Squircle trials, jackknifed across all participants. For each target object, participants tended to retrieve the same contextual memory as other participants (t23 = 9.58, 1.70 × 10−9). However, choice stability was lower across participants than within the same participant (t23 = 4.73, p = 9.22 × 10−5), indicating that although there was some choice stability between participants, each participant tended to have a unique contextual preference. b, Participants replaced target objects in the Squircle closer to a context-consistent location than to the location intermediate between the two context-consistent locations (t = 10.86, p = 1.57 × 10−10), and this was the case in 23/24 (96%) individual participants, providing further evidence that participants retrieved either the Square or Circle-consistent contextual memories in the Squircle. Throughout the figure, error bars indicate ±1 SEM; All t-test two-tailed; Dots denote individual participants (n = 24). nsp > 0.05, ***p < 0.001.
A single estimate of the average bilateral hippocampal (Hipp) activation pattern for the Square (Sq) and Circle (Ci) was computed for each scan run. Across-run HIPP context classification accuracy for the Sq and Ci was significantly higher than in a permutation test (Perm). The dotted line indicates theoretical chance baseline (50%). Error bars indicate ±1 SEM. Dots denote individual participants (n = 24). ***p < 0.001.
a, To unambiguously compute CS Scores for each nonmnemonic behavioral factor given the fMRI signal’s hemodynamic lag, each factor must be temporally autocorrelated during Squircle trials within and across fMRI volumes. To determine the temporal autocorrelation within fMRI volume (TR), we computed the percentage of timepoints during which nonmnemonic factors were the same between timepoint t and t + N, where N corresponds to 19 lags from 51–969 msec from t (that is, at the behavioral sampling rate—20 Hz), within a single TR during Squircle trials. Nonmnemonic factors include VR-locomotion (VR-walking or -stationary), location (most proximal boundary segment, either Square or Circle), boundary view (toward Square or Circle boundary segment), distal cue view (toward Square or Circle distal cue). To control for potential nonuniform sampling within each nonmnemonic factor, percentages were converted to z-scores by randomly shuffling each nonmnemonic factor across timepoints separately for each participant (k = 500). Each nonmnemonic factor was more similar across successive timepoints within TRs at all lags than expected by chance (color bars denote p-values from one-sample t-tests compared to 0 baseline; all t23 > 9.82, p < 10−8). Light gray lines denote individual participants (n = 24), black line denotes group mean. b, Same as in (a) but between successive TRs (lags 1–10 TRs) during Squircle trials. Red dotted lines denote the lag N at which a given nonmnemonic factor was no more likely to be the same at time t than t + N (t-test, p > 0.05, uncorrected for multiple comparisons), if at all. Participants tended to occupy the same behavioral state for multiple TRs. c, Percentage of time during Squircle trials that participants VR-turned (mean angular velocity >0 deg/sec) between successive TRs during VR-walking and -stationary epochs. Participants VR-turned more often during VR-stationary than -walking epochs (t23 = 17.28, p = 1.13 × 10−14). This result also implies that both boundary view and distal cue view were more similar across successive TRs during VR-walking than VR-stationary epochs. Error bars indicate ±1 SEM. Dots denote individual participants (n = 24). Throughout the figure, all t-tests two-tailed. ***p < 0.001.
a, Leave-one-run-out cross-validated trial-by-trial context classification accuracy from the EVC was higher than chance (57.1% versus 49.9% in the permutation test, t23 = 2.31, p = 0.015; t23 = 2.24, p = 0.018 compared to 50% theoretical chance baseline). Though fewer participants had numerically greater context classification accuracy than expected in the permutation test in EVC than the hippocampus (n = 24, McNemar Χ2 test with Yates’ correction, Χ2 = 4.03, p = 0.045, two-tailed), this result nonetheless indicates that there tended to be different EVC activation patterns elicited in the Square and Circle, as expected given that the two contexts were visually distinct. b, Contextual similarity of EVC activity over time during Sq (red) and Ci (blue) trials. c, EVC CS Scores over time during Sq-con and Ci-con Squircle trials (Sq-con vs Ci-con replace phase: t23 = −1.63, p = 0.942, t-test, one-tailed). d, EVC Sq-con and Ci-con Squircle CS Scores binned by VR-locomotory behavior (contextual memory x VR-locomotion rmANOVA: VR-locomotion main effect, F1,23 = 1.69, p = 0.207; interaction, F1,23 = 0.25, p = 0.624). Throughout the figure, error bars or error bands indicate ±1 SEM. Dots denote individual participants (n = 24). Post-hoc t-tests were 1-tailed. nsp > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.
Trial-by-trial hippocampal (Hipp) Squircle contextual memory classification accuracy from the target-object-invariant context classifier, using data from VR-walking epochs, was significantly higher than expected by chance (61.6% versus 58.0% in the permutation test, t23 = 1.98, p = 0.030; t23 = 6.34, p = 9.0 × 10−7 compared to 50% theoretical chance baseline). Thus, hippocampal context representations during VR-walking epochs were tolerant to changes in target object. Error bars indicate ±1 SEM; Dots denote individual participants (n = 24). *p < 0.05.
a, The strength of reliable periodic modulation (beta weight) as a function of VR-walking direction in the entorhinal (EC) subregion of interest from Fig. 4a for 90° (4-fold) and 45° (8-fold) periodicities. These control periodicities were tested in particular because a recent study using human intracranial recordings found a large proportion of EC neurons had spatially periodic firing fields with 90° or 45° rotational symmetry23. Neither 90° nor 45° showed significant reliable periodic modulation aligned to their respective orientations in either the Square (Sq) or Circle (Ci) (90°: Sq, t-test t23 = 0.28, p = 0.779, sign-test p = 0.541; Ci, t test, t23 = 0.94, p = 0.359, sign-test p = 0.839; 45°: Sq, t-test t23 = 1.26, p = 0.110, sign-test p = 0.308; Ci, t23 = −0.419, p = 0.661). Note that these null effects were not specific to the entorhinal subregion based on the 60° periodicity analysis, as we saw no effect for 90° or 45° in any voxels in the entire entorhinal cortex at pSVC < 0.05. b, The strength of grid-like modulation in the early visual cortex (EVC) region of interest. There was no significant reliable grid-like modulation in EVC in either the Sq (t23 = 0.94, p = 0.180) or Ci (t23 = 0.498, p = 0.312). These null effects were not due to averaging over all voxels in EVC, as there was also no grid-like modulation observed in the entire EVC at pSVC < 0.05. c, The strength of grid-like modulation in the hippocampus (Hipp) region of interest. There was no significant reliable grid-like modulation in Hipp in either the Sq (t23 = 0.12, p = 0.454) or Ci (t23 = 1.01, p = 0.161). These null effects were not due to averaging over all voxels in Hipp, as there was also no grid-like modulation observed in the entire Hipp ROI at pSVC < 0.05. d, Difference in the percentage of VR-walking directions modulo 90°, 60°, and 45° across the Sq and Ci. To align VR-walking directions across the Sq and Ci, for each periodicity (90°, 60°, and 45°), 0° corresponds to the orientation that maximized entorhinal periodic modulation for that periodicity, separately for the Sq and Ci. Black lines denote means across all participants, and grey lines denote individual participants. There were no differences in periodic VR-walking direction bias (t-test against zero, α=0.05, FWE-corrected) between the Sq and Ci that would be confounded with the presence of context-specific 60° periodic fMRI signal dependent on VR-walking direction Throughout the figure, error bars indicate ±1 SEM. Dots denote individual participants (n = 24). nsp > 0.05.
Extended Data Fig. 9 Local environmental anchoring of entorhinal grid-like orientations in the Square and Circle.
a, Grid orientations (φ) can only be compared across contexts relative to some common reference frame. Here, we used the Squircle as the common reference frame; however, this reference frame choice may have been arbitrary, particularly because participants were not exposed to the Squircle until after having already learned the location-context associations in the Square and Circle. To provide further evidence for grid realignment, we therefore tested whether φs were differentially aligned to local orientational cues in the Square and Circle. In square environments, φ has been found to be offset 7.5° from the boundaries26,27. Since φ ranges between 0°–60°, we tested whether Sq φ clustered around four possible angles, each 7.5° from one of the two cardinal axes of the square boundaries. To do so, we folded the grid orientations of all participants by φ mod 15°, which would align all hypothesis-consistent alignments to 7.5° in a circular 0° to 15° space. b, We also tested whether φ was reliably aligned to the Sq or Ci distal cues, either directly (0° offset from a distal cue) or maximally offset between distal cues (45° offset from a distal cue, since distal cues were offset from each other by 90° in each context). In addition to 0° offset, a 45° offset was tested because a prior report suggests that in sparse circular VR-environments with only distal orienting cues, φ is maximally offset from the cues (rather than aligned) (Schroeder, T. N. et al. (2018) Entorhinal cortex minimises uncertainty for optimal behaviour. BioRxiv 166306). Since φ ranges between 0°–60°, we thus tested whether φ clustered around four possible angles in each context. To do so, we folded the grid orientations of all participants by φ mod 15°, which would align all hypothesis-consistent alignments to 12° in the Sq and 3° in the Circle, in a circular 0° to 15° space. c, Boundary and distal cue φ alignment. Middle: average φ in each participant (n = 24) in the Sq (red squares, top) and Ci (blue circles, bottom). Left and right: histograms of average φ across participants, modulo 15°. In the Sq (left), φ clustered around 7.5° (V-test, V = 5.96, p = 0.043), consistent with alignment to the square boundaries, but not around 12° (V = −4.077, p = 0.880) which would have been consistent with alignment to the distal cues. In the Ci (right), φ clustered around 3° (V = 8.63, p = 0.006), consistent with reliable distal cue alignment. Since grid orientations aligned to different local cues in each context, this provides further support for the existence of grid-like realignment between the two training contexts.
Extended Data Fig. 10 Entorhinal grid-like modulation in the ambiguous Squircle context control analyses.
a, Grid-like modulation (beta weight) in entorhinal cortex (EC), averaged over all EC voxels, aligned to a single Squircle-specific grid orientation (φ) across all Squircle trials independent of contextual memory. To compute the Squircle-specific φ we performed the same split-half analysis as for the Square (Sq) and Circle (Ci) contexts (for example, Fig. 6), but limited the analysis only to Squircle trials. No significant grid-like modulation (beta weight) was observed when we assumed there was a single Squircle-specific φ (t23 = 1.06, p = 0.149; sign-test p = 0.271, one-tailed). b, Left: Grid-like modulation in EC, averaged over all EC voxels, aligned to either the Sq φ or the Ci φ across all Squircle trials. No significant grid-like modulation was observed (Sq, t-test t23 = 0.50, p = 0.312; Ci, t test, t23 = 0.39, p = 0.350). Right: An alternative possibility is that EC grid representations have a single fixed φ (either Sq φ or Ci φ) across all Squircle trials, but whether the Squircle φ was the Sq or Ci φ differed across participants. To address this alternative, for each participant, we computed the maximum grid-like modulation observed by a single fixed φ (either the Sq φ or Ci φ, whichever yielded stronger grid-like modulation). We then compared the strength of grid-like modulation aligned to this fixed φ to the maximum grid-like modulation observed if φ realigned across trials (Realign; Fig. 6b), independent of contextual memory. Stronger grid-like modulation was observed in the Squircle in the majority of participants (75%; 19/24) if φ realigned on a trial-by-trial basis than if we assumed that there was a fixed φ across all Squircle trials (sign-test, p = 0.005, one-tailed; t-test t23 = 1.92, p = 0.034). c, Difference in the percentage of VR-walking directions modulo 60° between Sq- and Ci-consistent Squircle trials. There was no difference in periodic VR-walking direction bias between Sq- and Ci-consistent trials that would be confounded with the presence of contextual-memory-specific 60° periodic fMRI signals dependent on VR-walking direction. Throughout the figure, error bars indicate ±1 SEM. Dots denote individual participants (n = 24). **p < 0.01, nsp > 0.05.
About this article
Cite this article
Julian, J.B., Doeller, C.F. Remapping and realignment in the human hippocampal formation predict context-dependent spatial behavior. Nat Neurosci 24, 863–872 (2021). https://doi.org/10.1038/s41593-021-00835-3
Nature Reviews Neuroscience (2021)