Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Letter
  • Published:

Coordination of entorhinal–hippocampal ensemble activity during associative learning

Abstract

Accumulating evidence points to cortical oscillations as a mechanism for mediating interactions among functionally specialized neurons in distributed brain circuits1,2,3,4,5,6. A brain function that may use such interactions is declarative memory—that is, memory that can be consciously recalled, such as episodes and facts. Declarative memory is enabled by circuits in the entorhinal cortex that interface the hippocampus with the neocortex7,8. During encoding and retrieval of declarative memories, entorhinal and hippocampal circuits are thought to interact via theta and gamma oscillations4,6,8, which in awake rodents predominate frequency spectra in both regions9,10,11,12. In favour of this idea, theta–gamma coupling has been observed between entorhinal cortex and hippocampus under steady-state conditions in well-trained rats12; however, the relationship between interregional coupling and memory formation remains poorly understood. Here we show, by multisite recording at successive stages of associative learning, that the coherence of firing patterns in directly connected entorhinal–hippocampus circuits evolves as rats learn to use an odour cue to guide navigational behaviour, and that such coherence is invariably linked to the development of ensemble representations for unique trial outcomes in each area. Entorhinal–hippocampal coupling was observed specifically in the 20–40-hertz frequency band and specifically between the distal part of hippocampal area CA1 and the lateral part of entorhinal cortex, the subfields that receive the predominant olfactory input to the hippocampal region13. Collectively, the results identify 20–40-hertz oscillations as a mechanism for synchronizing evolving representations in dispersed neural circuits during encoding and retrieval of olfactory–spatial associative memory.

This is a preview of subscription content, access via your institution

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Prices may be subject to local taxes which are calculated during checkout

Figure 1: 20–40-Hz coupling between LEC and dCA1 during successful odour discrimination.
Figure 2: Evolution of 20–40-Hz coupling between LEC and dCA1 during odour–place learning.
Figure 3: 20–40-Hz coupling on non-cued trials and novel-odour trials.
Figure 4: Development of odour-specific representations in dCA1 and LEC cells.

Similar content being viewed by others

References

  1. Freeman, W. J. Spatial properties of an EEG event in the olfactory bulb and cortex. Electroencephalogr. Clin. Neurophysiol. 44, 586–605 (1978)

    Article  CAS  Google Scholar 

  2. Gray, C. M., Konig, P., Engel, A. K. & Singer, W. Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338, 334–337 (1989)

    Article  ADS  CAS  Google Scholar 

  3. Singer, W. Synchronization of cortical activity and its putative role in information processing and learning. Annu. Rev. Physiol. 55, 349–374 (1993)

    Article  CAS  Google Scholar 

  4. Buzsaki, G. Rhythms of the Brain (Oxford University Press, 2006)

    Book  Google Scholar 

  5. Fries, P. Neuronal gamma-band synchronization as a fundamental process in cortical computation. Annu. Rev. Neurosci. 32, 209–224 (2009)

    Article  CAS  Google Scholar 

  6. Colgin, L. L. Mechanisms and functions of theta rhythms. Annu. Rev. Neurosci. 36, 295–312 (2013)

    Article  CAS  Google Scholar 

  7. Squire, L. R. Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans. Psychol. Rev. 99, 195–231 (1992)

    Article  CAS  Google Scholar 

  8. Buzsáki, G. & Moser, E. I. Memory, navigation and theta rhythm in the hippocampal-entorhinal system. Nature Neurosci. 16, 130–138 (2013)

    Article  Google Scholar 

  9. Buzsáki, G., Leung, L. W. & Vanderwolf, C. H. Cellular bases of hippocampal EEG in the behaving rat. Brain Res. 287, 139–171 (1983)

    Article  Google Scholar 

  10. Bragin, A. et al. Gamma (40–100 Hz) oscillation in the hippocampus of the behaving rat. J. Neurosci. 15, 47–60 (1995)

    Article  CAS  Google Scholar 

  11. Chrobak, J. J. & Buzsaki, G. Gamma oscillations in the entorhinal cortex of the freely behaving rat. J. Neurosci. 18, 388–398 (1998)

    Article  CAS  Google Scholar 

  12. Colgin, L. L. et al. Frequency of gamma oscillations routes flow of information in the hippocampus. Nature 462, 353–357 (2009)

    Article  ADS  CAS  Google Scholar 

  13. Witter, M. P. & Amaral, D. G. in The Rat Nervous System (ed. Paxinos, G. ) 635–704 (Elsevier Academic, 2004)

    Book  Google Scholar 

  14. Day, M., Langston, R. & Morris, R. G. Glutamate-receptor-mediated encoding and retrieval of paired-associate learning. Nature 424, 205–209 (2003)

    Article  ADS  CAS  Google Scholar 

  15. Martin, C., Beshel, J. & Kay, L. M. An olfacto-hippocampal network is dynamically involved in odor-discrimination learning. J. Neurophysiol. 98, 2196–2205 (2007)

    Article  Google Scholar 

  16. Montgomery, S. M. & Buzsaki, G. Gamma oscillations dynamically couple hippocampal CA3 and CA1 regions during memory task performance. Proc. Natl Acad. Sci. USA 104, 14495–14500 (2007)

    Article  ADS  CAS  Google Scholar 

  17. Tort, A. B., Komorowski, R. W., Manns, J. R., Kopell, N. J. & Eichenbaum, H. Theta-gamma coupling increases during the learning of item-context associations. Proc. Natl Acad. Sci. USA 106, 20942–20947 (2009)

    Article  ADS  CAS  Google Scholar 

  18. Eichenbaum, H., Kuperstein, M., Fagan, A. & Nagode, J. Cue-sampling and goal-approach correlates of hippocampal unit activity in rats performing an odor-discrimination task. J. Neurosci. 7, 716–732 (1987)

    Article  CAS  Google Scholar 

  19. Ahmed, O. J. & Mehta, M. R. Running speed alters the frequency of hippocampal gamma oscillations. J. Neurosci. 32, 7373–7383 (2012)

    Article  CAS  Google Scholar 

  20. Fries, P., Reynolds, J. H., Rorie, A. E. & Desimone, R. Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291, 1560–1563 (2001)

    Article  ADS  CAS  Google Scholar 

  21. Buschman, T. J. & Miller, E. K. Top-down versus bottom-up control of attention in the prefrontal and posterior parietal cortices. Science 315, 1860–1862 (2007)

    Article  ADS  CAS  Google Scholar 

  22. Berke, J. D., Hetrick, V., Breck, J. & Greene, R. W. Transient 23–30 Hz oscillations in mouse hippocampus during exploration of novel environments. Hippocampus 18, 519–529 (2008)

    Article  Google Scholar 

  23. Bi, G. Q. & Poo, M. M. Synaptic modifications in cultured hippocampal neurons: dependence on spike timing, synaptic strength, and postsynaptic cell type. J. Neurosci. 18, 10464–10472 (1998)

    Article  CAS  Google Scholar 

  24. McNaughton, B. L., Douglas, R. M. & Goddard, G. V. Synaptic enhancement in fascia dentata: cooperativity among coactive afferents. Brain Res. 157, 277–293 (1978)

    Article  CAS  Google Scholar 

  25. Engel, A. K. & Fries, P. Beta-band oscillations — signalling the status quo? Curr. Opin. Neurobiol. 20, 156–165 (2010)

    Article  CAS  Google Scholar 

  26. Ravel, N. et al. Olfactory learning modifies the expression of odour-induced oscillatory responses in the gamma (60–90 Hz) and beta (15–40 Hz) bands in the rat olfactory bulb. Eur. J. Neurosci. 17, 350–358 (2003)

    Article  Google Scholar 

  27. Howe, M. W., Atallah, H. E., McCool, A., Gibson, D. J. & Graybiel, A. M. Habit learning is associated with major shifts in frequencies of oscillatory activity and synchronized spike firing in striatum. Proc. Natl Acad. Sci. USA 108, 16801–16806 (2011)

    Article  ADS  CAS  Google Scholar 

  28. Kopell, N., Ermentrout, G. B., Whittington, M. A. & Traub, R. D. Gamma rhythms and beta rhythms have different synchronization properties. Proc. Natl Acad. Sci. USA 97, 1867–1872 (2000)

    Article  ADS  CAS  Google Scholar 

  29. Henriksen, E. J. et al. Spatial representation along the proximodistal axis of CA1. Neuron 68, 127–137 (2010)

    Article  CAS  Google Scholar 

  30. Steward, O. Topographic organization of the projections from the entorhinal area to the hippocampal formation of the rat. J. Comp. Neurol. 167, 285–314 (1976)

    Article  CAS  Google Scholar 

  31. Mitra, P. P. & Pesaran, B. Analysis of dynamic brain imaging data. Biophys. J. 76, 691–708 (1999)

    Article  ADS  CAS  Google Scholar 

  32. Kepecs, A., Uchida, N. & Mainen, Z. F. Rapid and precise control of sniffing during olfactory discrimination in rats. J. Neurophysiol. 98, 205–213 (2007)

    Article  Google Scholar 

  33. Sargolini, F. et al. Conjunctive representation of position, direction, and velocity in entorhinal cortex. Science 312, 758–762 (2006)

    Article  ADS  CAS  Google Scholar 

  34. Solstad, T., Boccara, C. N., Kropff, E., Moser, M. B. & Moser, E. I. Representation of geometric borders in the entorhinal cortex. Science 322, 1865–1868 (2008)

    Article  ADS  CAS  Google Scholar 

  35. 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)

    Article  ADS  CAS  Google Scholar 

  36. Fyhn, M., Molden, S., Witter, M. P., Moser, E. I. & Moser, M. B. Spatial representation in the entorhinal cortex. Science 305, 1258–1264 (2004)

    Article  ADS  CAS  Google Scholar 

  37. Macmillan, N. A. & Creelman, C. D. Detection Theory: A User’s Guide (Cambridge Univ. Press, 1991)

    Google Scholar 

  38. Jezek, K., Henriksen, E. J., Treves, A., Moser, E. I. & Moser, M. B. Theta-paced flickering between place-cell maps in the hippocampus. Nature 478, 246–249 (2011)

    Article  ADS  CAS  Google Scholar 

  39. Naya, Y. & Suzuki, W. A. Integrating what and when across the primate medial temporal lobe. Science 333, 773–776 (2011)

    Article  ADS  CAS  Google Scholar 

  40. Skaggs, W. E., McNaughton, B. L., Gothard, K. M. & Markus, E. J. in Advances in Neural Processing Systems Vol. 5 (eds Hanson, S. J., Cowan, J. D. & Giles, C. L. ) 1030–1037 (Morgan Kaufmann, San Mateo, 1993)

    Google Scholar 

  41. Skaggs, W. E., McNaughton, B. L., Wilson, M. A. & Barnes, C. A. Theta phase precession in hippocampal neuronal populations and the compression of temporal sequences. Hippocampus 6, 149–172 (1996)

    Article  CAS  Google Scholar 

  42. Langston, R. F. et al. Development of the spatial representation system in the rat. Science 328, 1576–1580 (2010)

    Article  ADS  CAS  Google Scholar 

  43. 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)

    Article  ADS  CAS  Google Scholar 

  44. Igarashi, K. M. et al. Parallel mitral and tufted cell pathways route distinct odor information to different targets in the olfactory cortex. J. Neurosci. 32, 7970–7985 (2012)

    Article  CAS  Google Scholar 

  45. Fanselow, M. S. & Dong, H. W. Are the dorsal and ventral hippocampus functionally distinct structures? Neuron 65, 7–19 (2010)

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank A. M. Amundsgård, K. Haugen, K. Jenssen, E. Kråkvik, R. Skjerpeng and H. Waade for technical assistance, and M. Witter and members of the Moser laboratory for discussions. This work was supported by two Advanced Investigator grants from the European Research Council (‘CIRCUIT’, Grant Agreement no. 232608; ‘ENSEMBLE’, Grant Agreement no. 268598), the Kavli Foundation, the Centre of Excellence scheme of the Research Council of Norway (Centre for the Biology of Memory and Centre for Neural Computation), the Mishima Kaiun Memorial Foundation, and the Japan Society for the Promotion of Science.

Author information

Authors and Affiliations

Authors

Contributions

K.M.I., M.-B.M. and E.I.M. designed experiments and analyses; K.M.I. and L.L. performed the experiments; K.M.I. performed the majority of the analyses, with input from L.L.C, M.-B.M. and E.I.M.; K.M.I and E.I.M. wrote the paper with input from all authors.

Corresponding authors

Correspondence to Kei M. Igarashi or Edvard I. Moser.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 Brain sections showing recording positions in CA1, LEC and MEC.

a, Recording positions (red) in CA1 in coronal brain sections from representative animals (top; animal and tetrode numbers indicated; distal–proximal (DP) distances as in Extended Data Fig. 5) and in a projection of a 3D reconstruction of the CA1 pyramidal cell layer (bottom; see Extended Data Fig. 5 for the full 3D reconstruction). Abscissa shows the lateral–medial axis and ordinate the anterior–posterior axis. Dorsal–ventral is colour-coded. Each dot corresponds to one recording location in distal (red), intermediate (green) and proximal (blue) parts of CA1. Stippled line indicates septotemporal axis of the hippocampus (see Supplementary Methods). b, Tetrode positions (red) in coronal sections through LEC for all 5 animals with simultaneous recordings in LEC and dCA1. Animal numbers are indicated. c, Tetrode positions (red) in sagittal sections through MEC (all 4 animals with simultaneous MEC and dCA1 recordings). d, Tetrode positions (red) in coronal brain sections through LEC from all 3 animals with simultaneous recordings in pCA1 and LEC. Orientation: D, dorsal; L, lateral; A, anterior.

Extended Data Figure 2 Learning was not associated with changes in motor behaviours.

a, Left, example trajectories of one animal on sessions T1 to T5, on error trials (T5e) and on correct trials downsampled to the same number of trials as the error trials (T5d). Red parts of each trajectory indicate positions covered during the cue sampling interval, black indicates positions during runs from cue port to food cups, and grey shows positions from food cups back to the cue port. Right, polar plots showing distributions of head direction during the cue sampling interval (T1–T5, T5d and T5e, as indicated to the left). Values to the right indicate mean head angle relative to the centre of the cue port, designated as north (0°). b, Time course of instantaneous speed before, during and after cue sampling, averaged across 5 rats. Shading denotes s.e.m. T1 is shown at the top; T5, T5d and T5e at the bottom. For every 10-ms time bin during T2–T5, speed was compared with the corresponding bin at T1. Speeds were not significantly different from T1 at any time bin (q > 0.05; false discovery rate (FDR) corrected for multiple comparisons). Speed was not different at T5e and T5d. c, Mean head angle (top) and mean vector length for head direction (bottom) during the cue sampling interval (mean ± s.e.m. for all 5 animals). Neither head angle nor mean vector length changed significantly from T1 to T5 (repeated measures ANOVA: F(4, 16) = 0.44, P = 0.78 for mean angle; F(4, 16) = 0.13, P = 0.97 for mean vector length). There was also no change in these parameters on the error trials (T5e compared with T5d using two-tailed paired t-test: t(4) = 0.70, P = 0.52 for mean angle; t(4) = 0.18, P = 0.87 for mean vector length). d, Mean instantaneous speed (top), path length (middle) and run duration (bottom) for trajectories from cue port to food cups (left column) and from food cups to cue port (right column). None of these parameters changed significantly during learning (repeated measures ANOVA for time-points T1–T5; for cue port to food cups: F(4, 16) = 0.14, P = 0.97 for mean speed, F(4, 16) = 0.55, P = 0.70 for path length, F(4, 16) = 0.16, P = 0.95 for run duration; for food cups to cue port: F(4, 16) = 0.35, P = 0.84 for mean speed, F(4, 16) = 0.47, P = 0.76 for path length, F(4, 16) = 0.13, P = 0.97 for run duration). None of these parameters changed on error trials (T5e compared with T5d using two-tailed paired t-test; cue port to food cups: t(4) = 0.31, P = 0.77 for mean speed, t(4) = 1.87, P = 0.52 for path length, t(4) = 0.91, P = 0.42 for run duration; food cups to cue port: t(4) = 0.14, P = 0.89 for mean speed, t(4) = 0.50, P = 0.64 for path length, t(4) = 0.02, P = 0.99 for run duration). e, To check if changes in neural activity from T1 to T5 are associated with changes in body position, LEDs were attached to the back of 4 rats and LED positions were recorded from T1 to T5. Left, example trajectories of body position in one animal at different stages of learning. Red indicates body positions covered during the cue sampling interval, black indicates positions during runs from cue port to food cup, and grey show positions from food cup back to the cue port. Right, polar plots showing distribution of body direction (deflection from south) during the cue sampling interval (T1–T5). f, Cumulative body movement (top), mean body angle (middle) and mean vector length for body direction (bottom) during the cue sampling interval (mean ± s.e.m. of 4 animals). In the middle panel, body angle is shown as the absolute change in mean angle compared to T1, since different rats may turn in different directions. None of these parameters changed significantly from T1 to T5 (repeated measures ANOVA: F(4, 12) = 0.77, P = 0.57 for cumulative body movement; F(3, 9) = 1.06, P = 0.41 for mean angle; F(4, 12) = 1.11, P = 0.40 for mean vector length). g, Example recording showing changes in sniffing amplitude and frequency during cue sampling. A temperature sensor (thermocouple) was implanted in the right nostril to measure respiration (top). In each breathing cycle, inhalation is associated with decaying voltage output from the thermocouple, and exhalation with increasing voltage. Instantaneous sniff frequency was determined from the voltage change (bottom). Note that sniff frequency increased during cue sampling (between t = 0 s and t = 1 s) and when the animal approached the food cup. The food cup was reached at t = 3.8 s in the present example. h, Time course of sniff frequencyaveraged for 5 rats at time-points T1–T5 as well as on error trials (T5e) and correct trials down-sampled to the same number of trials (T5d). Shading denotes s.e.m. At time points T2–T5, sniff frequency was compared for every 10 ms bin with corresponding bins at T1. None of the comparisons were significantly different (q > 0.05; FDR corrected for multiple comparisons). Sniff frequencies between T5d and T5e were also not different. i, Mean sniff frequency during the cue sampling interval. Sniff frequency did not change during the course of learning (repeated measures ANOVA: F(4, 16) = 0.24, P = 0.91) or on error trials (T5e compared with T5d using two-tailed paired t-test, t(4) = 0.03, P = 0.97). j, Left, example traces showing position of the animal on non-cued control trials (T5n). Colours indicate positions during cue sampling (red), positions from cue port to food cups (black) and positions from food cups back to cue port (grey). Right, polar plot showing head direction distribution during the cue sampling period for the session shown on the left. Values on the right indicate mean head angle. The centre of the cue port is north, or 0°. k, Instantaneous speed on non-cued trials, averaged across 5 rats. Shading denotes s.e.m. Speed on non-cued trials (T5n) was compared at successive 10-ms bins with speed on corresponding bins of cued trials (T5). Differences were not significant at any time bin between 2 s before and 4 s after poke onset (q > 0.05, FDR corrected for multiple comparisons). l, Mean head angle (top) and mean vector length of head direction (bottom) during the cue-sampling interval. Neither head angle nor mean vector length was different from corresponding values on cued trials (two-tailed paired t-test, t(4) = 1.85, P = 0.13 for mean angle; t(4) = 1.99, P = 0.12 for mean vector length). m, Mean speed (top), path length (middle) and run duration (bottom) for trajectories from cue port to food cups (left column) and from food cups to cue port (right column). None of these parameters were different between cued and non-cued tasks (two-tailed paired t-test; cue port to food cup movements: t(4) = 0.16, P = 0.88 for mean speed, t(4) = 0.50, P = 0.64 for path length, t(4) = 1.03, P = 0.36 for run duration; food cup to cue port movements: t(4) = 0.21, P = 0.84 for mean speed, t(4) = 0.30, P = 0.78 for path length, t(4) = 0.44, P = 0.68 for run duration). n, Sniff frequency at the cue port on non-cued trials, averaged across 5 rats. Frequency on cued trials is shown as a reference. Shading denotes s.e.m. Sniff frequency was compared at successive 10-ms bins between T5 and T5n. Differences were not significant at any time between 2 s before and 4 s after poke onset (q > 0.05; FDR corrected for multiple comparisons). o, Mean sniff frequency during the cue sampling period was not different between T5 and T5n (two-tailed paired t-test, t(4) = 0.89, P = 0.43). p, Mean sniff frequency during the cue sampling interval in novel odour trials. Sniff frequency did not change over the course of trials with novel odours (repeated measures ANOVA: F(6, 18) = 0.77, P = 0.60).

Extended Data Figure 3 Power and coherence over a broader spectrum of frequencies in dCA1 and LEC at the end of training.

a–d, Time-resolved power spectra averaged across tetrodes in a, dCA1 (n = 9 tetrodes), b, pCA1 (n = 14 tetrodes), c, LEC (n = 10 tetrodes) and d, MEC (n = 8 tetrodes), as in Fig. 1e and g, but for a broader band of the frequency spectrum (0–140 Hz). Power is shown as percentage change from power during the pre-cue period. dCA1 and pCA1 data are from the same animals (rats with tetrodes along the entire proximodistal CA1 axis), whereas LEC and MEC data are from separate animals (5 and 4 rats, respectively). Right panels show mean power for frequencies up to 140 Hz during cue sampling (red) and during running from cue port to food cups (blue). Peak frequencies are indicated. In CA1, the mean power in the 20–40-Hz frequency band increased from (1.07 ± 0.04) × 10−3 before cue sampling to (1.36 ± 0.05) × 10−3 during cue sampling and then reverted to (0.89 ± 0.05) × 10−3 mV2 after cue sampling (repeated measures ANOVA: F(2, 78) = 118, P < 0.001). In LEC, the power was (1.05 ± 0.12) × 10−3, (1.62 ± 0.16) × 10−3 and (0.78 ± 0.08) × 10−3 mV2, respectively (F(2, 18) = 67.1, P < 0.001). e, f, Time-resolved coherence spectra averaged across tetrode pairs in LEC and dCA1 (e) and in MEC and dCA1 (f), as shown in Fig. 1h, but across a broader band of the frequency spectrum (same animals as in c and d; data are averaged across all EC-dCA1 tetrode pairs). Note that, during running, MEC shows fast gamma oscillations (60–100 Hz) that are coherent with dCA1 LFPs (arrows). Right panels show mean coherence spectra during cue sampling (red) and during running (blue), with peak frequencies indicated. LEC did not show fast gamma oscillations.

Extended Data Figure 4 Effect of running speed on oscillation frequency.

a, Power spectra of oscillations in dCA1, pCA1, LEC and MEC as a function of running speed at T5 during random foraging in the 1 × 1 m square box. Power is normalized at each frequency bin and shown as a z-score18. Data were averaged across 9 (dCA1), 14 (pCA1), 10 (LEC) and 8 (MEC) tetrodes from 5 dCA1 and pCA1, 5 LEC and 4 MEC rats, respectively. b, Peak oscillation frequency (mean ± s.e.m. for all tetrodes) as a function of speed bin in a. As in previous work18, a significant correlation was observed between speed and peak frequency of oscillations in the 20–140-Hz range (P < 0.001; r(58) = 0.71, 0.44, 0.60 and 0.59 for dCA1, pCA1, LEC and MEC, respectively). c, d, Speed–frequency relationship in the cue-place association task. Same tetrodes as in a. Data were plotted during the cue sampling period (c) and during subsequent running from the cue sampling port to the food cups (d). Note that during cue sampling, speed was minimal (0–1 cm s−1). Strong 20–40-Hz oscillations were observed in dCA1, pCA1 and LEC (arrows in c) but not at similar speeds during running (d). e, In all four brain regions (dCA1, pCA1, LEC and MEC), the power at very low speeds (0–1 cm s−1) was significantly stronger during cue sampling than during running (paired t-test, P < 0.01; t(8) = 3.9, t(13) = 13.7, t(9) = 13.4 and t(7) = 4.5 for dCA1, pCA1, LEC and MEC, respectively), suggesting 20–40-Hz oscillations do not reflect low speed as such. f, During running from cue port to food cups, a significant positive correlation between speed and frequency was observed only in pCA1(P < 0.001, r(58) = 0.48). g, Time spent in each speed bin during the running part of the cued task (means ± s.e.m. for all 14 rats). Note that all speed bins (including low speeds) were sampled for 2.5 s or more.

Extended Data Figure 5 20–40-Hz oscillations in dCA1 and pCA1 are functionally decoupled.

a, Time-resolved power spectra averaged across all tetrodes in the distal one-third of CA1 (dCA1, top) as well as the proximal one-third (pCA1, bottom) (9 and 14 recording sites, respectively). Data were pooled over 5 rats with hyperdrive implants spanning a wide transverse range of CA1. To control for impedance differences between tetrodes, LFP power during cue sampling was normalized, for each tetrode, to power during the pre-cue period when the animal was stationary at the odour port before the cue was delivered. Normalized power was averaged across tetrodes. Time t = 0 indicates cue onset. b, Positions of all tetrodes from which data were recorded in a, plotted in a 3D reconstruction of the CA1 cell layer. Distance from the anterior tip of subiculum is shown in micrometres. Position on the distal–proximal (DP) axis (0, distal; 1, proximal; see Supplementary Methods) was calculated in 3D space for each tetrode. CA1 tetrodes were grouped into distal (red point), intermediate (green point) and proximal (blue point) groups, each corresponding to one-third of the DP axis. c, Power spectra of LFP averaged over tetrodes in distal and proximal one-thirds of CA1 during cue sampling. Power is shown as percentage change from the pre-cue level. During cue sampling, power in the lower part of the 20–40-Hz band was stronger in dCA1 than pCA1 (11–25 Hz, green dots; q < 0.05; false discovery rate (FDR) corrected for 60 multiple comparisons at 1–60 Hz using 1 Hz bins, q < 0.05), whereas power in the higher part was lower in dCA1 (33–47 Hz, green dots; q < 0.05). d, To compare power in the lower and higher parts of the 20–40-Hz oscillation across the DP axis in CA1, the ratio of power in the slower part (20–25 Hz) and the faster part (33–40 Hz) was plotted as a function of DP distance. Each dot refers to one tetrode. Red, distal one-third; green, intermediate; blue, proximal one-third. Low/high power ratio was negatively correlated with DP distance (r(38) = −0.50, P < 0.001). e, Time-resolved coherence spectrum for dCA1 versus pCA1. Data were averaged over 28 recording pairs in dCA1 and pCA1 from animals with tetrodes in both regions. f, Mean coherence between dCA1 and pCA1. Note lack of change in 20–40-Hz coherence during cue sampling compared to pre-cue or run periods (repeated measures ANOVA, F(2, 54) = 0.75, P = 0.48).

Extended Data Figure 6 Comparison of power and coherence development during learning.

a, b, Power of 20–40-Hz oscillations in dCA1 (a, n = 10 tetrodes) and LEC (b, n = 10 tetrodes) from 5 rats at time points T1–T5, on error trials at T5 (T5e) and at T5 after down-sampling (T5d), shown as values normalized by power at time point T1. Power of 20–40-Hz oscillations did not change significantly during the course of learning (repeated measures ANOVA: F(4, 36) = 0.72, P = 0.58 for dCA1; F(4, 36) = 1.71, P = 0.17 for LEC). Power on error trials (T5e) was not significantly different from correct trials (T5d; two-tailed paired t-test, t(9) = 1.40, P = 0.20 for dCA1; t(9) = 0.06, P = 0.94 for LEC). c, d, Power of theta oscillations in dCA1 (c) and LEC (d). Theta power did not change during learning (repeated measures ANOVA: F(4, 36) = 0.90, P = 0.47 for dCA1; F(4, 36) = 0.63, P = 0.64 for LEC). Theta power on error trials was not different from correct trials (two-tailed paired t-test, t(9) = 0.24, P = 0.82 for dCA1; t(9) = 0.40, P = 0.70 for LEC). e, Power of 20–40-Hz oscillations in dCA1 (left, n = 8 tetrodes) and LEC (right, n = 8 tetrodes) at the end of training with the original odours (A/B; time point T5), and with odours C/D and E/F (mean ± s.e.m.) (4 rats). Power of the 20–40-Hz oscillations did not change significantly with the new odours, neither between T5 and the first day with C/D or E/F (t(7) < 0.5, P > 0.6 for CA1, t(7) < 0.9, P > 0.4 for LEC) nor during the course of learning with C/D and E/F (repeated measures ANOVA: F(6, 42) = 0.34, P = 0.91 for dCA1; F(6, 42) = 0.44, P = 0.85 for LEC). f, Power of theta oscillations in dCA1 (left) and LEC (right) (mean ± s.e.m.). There was no significant change in the power of theta (repeated measures ANOVA: F(6, 42) = 0.13, P = 0.99 for dCA1; F(6, 42) = 0.71, P = 0.64 for LEC). g, Mean of LEC–dCA1 coherence spectra during cue sampling at successive time points during learning (T1–T5) and on error trials (T5e) and down-sampled correct trials (T5d), plotted in the same way as the time-resolved coherence spectra in Fig. 2b (n = 20 recording pairs from 5 rats). Dots above spectra denote frequencies with significant difference (FDR corrected for 62 multiple comparisons at 0–60 Hz, q < 0.05). h, Coherence averaged across the theta-frequency band during cue sampling in novel odour trials. No change was observed for coherence in the theta frequency band (repeated measures ANOVA: F(2, 30) = 0.02, P = 0.98 for cues C/D and F(2, 30) = 0.25, P = 0.78 for cues E/F).

Extended Data Figure 7 Cross-frequency coupling of 20–40-Hz oscillations to local theta oscillations in LEC and pCA1.

a–d, Relationship of 20–40-Hz oscillations in LEC to phase of local theta oscillations, plotted as in Fig. 2d–g (n = 10 tetrodes from 5 rats). a, Theta phase distribution of 20–40-Hz oscillation maxima at T1, T3, T5 and on error trials at T5 (T5e). 0° was defined as the trough of the theta cycle. Note that theta oscillations exist in LEC and that 20–40-Hz oscillations are moderately phase-coupled with theta oscillations already at T1. b, Mean vector length calculated from theta phase distribution of 20–40-Hz oscillation maxima at T1–T5. The degree of cross-frequency coupling did not change significantly across the learning period (T1–T5, repeated measures ANOVA: F(4, 36) = 1.7, P = 0.17). No difference was observed on error trials (T5e, compared with T5d using a two-tailed paired t-test, t(9) = 0.78, P = 0.45). c, Representative cross-frequency coherence plot showing for LEC that power of 20–40 Hz oscillations (y-axis) is modulated by theta phase (x-axis) during cue sampling at T5. Coupling strength is colour-coded (dark blue, no coupling; red, maximal coupling). d, Top, time-resolved power spectrum averaged across all theta cycles with 20–40-Hz oscillations at T5 in LFP from 10 tetrodes in all 5 rats. t = 0 corresponds to the theta trough. Bottom, averaged unfiltered theta cycle. 20–40-Hz oscillations occurred at the falling phase of the theta wave. e–h, Similar plots for pCA1 (n = 14 tetrodes from 5 rats; e, theta phase distribution; f, mean vector length; g, cross-frequency coherence plot; h, time-resolved power spectrum). Animals with implants in pCA1 were recorded only after the completion of learning, that is, only at T5. In pCA1, oscillations at 30–50-Hz were phase-coupled with theta oscillations. The degree of coupling did not change on error trials (two-tailed paired t-test, t(13) = 1.29, P = 0.22). i, Same plots as in Fig. 2f, but with wider 60° bins. The diagram shows the relationship of 20–40-Hz oscillations in dCA1 to the phase of local theta oscillations (n = 10 tetrodes from 5 rats). The use of wider bins did not change the results. Significant cross-correlations between cell pairs in LEC and CA1 were not found, as expected due to the sparse connectivity between cell pairs in these areas13. j, Theta phase distribution of 20–40-Hz oscillation maxima for LFP from dCA1 as in Fig. 2f, but during the pre-cue period, at time points T1, T3, T5 (n = 10 tetrodes from 5 rats, means ± s.e.m.). k, Mean vector length calculated from theta phase distributions of 20–40-Hz oscillation maxima during the pre-cue period did not change during the course of training (repeated measures ANOVA: F(4, 36) = 0.26, P = 0.90; n = 10 tetrodes from 5 rats). Vector lengths during cue sampling was increased compared to the pre-cue period at T5 (paired t-test, P < 0.05; t(9) = 2.5) but not at T1–T4 (P > 0.05; t(9) < 1.9).

Extended Data Figure 8 A large fraction of CA1 and MEC cells with activity at the cue port were place cells and grid cells, respectively.

a, Spatial distribution of firing in example cells with cue-port activity recorded in dCA1, pCA1, LEC or MEC in either the odour–cue association task (top) or a random foraging task in a 1 m square box (bottom). Each column shows results for one representative cell. For each cell, spike position (red) is overlaid on the trajectory of the rat (grey) at the top and a colour-coded frequency map is shown at the bottom. Red is maximum firing rate, as indicated by the scale bar. Rat number, tetrode number (t) and cell number (c) are indicated above each path diagram. Peak frequencies are indicated at the top right of the colour map. Note that dCA1/pCA1 and MEC cells had clearly distinguishable place fields or grid fields in the foraging task, whereas the LEC cell showed no clear spatial modulation. b, Top, distribution of spatial information scores calculated from firing rate distributions in the random foraging task for cells with cue-port activity in dCA1, pCA1, LEC and MEC. Results in dCA1 are shown for both microdrive (MD) and hyperdrive (HD) implants. Bottom, distribution of shuffled data based on 100 permutations per cell. Red lines indicate 95th percentile value (chance level) for a distribution based on all permutations in each area. 95th percentile value is indicated in red. Percentage of cells that exceeded chance level is shown for each region. Note that 74% and 81% of the cue-port cells were spatially modulated in dCA1 and pCA1, respectively, that is, they were place cells45. c, As in b, but for the distribution of gridness scores. Note that 89% of cue port cells in MEC had gridness scores that exceeded chance level and so were defined as grid cells45.

Extended Data Figure 9 Phase-locking of individual neurons to the 20–40-Hz rhythm.

a, Raster plots showing cue-port activity of example dCA1 cell on successive trials at T5. Rows correspond to individual trials; ticks indicate spikes. Top, trials with food in left cup; middle, trials with food in right cup; bottom, peri-stimulus time histogram (PSTH); orange, left-predicting trials; black, right-predicting trials. Between T1 and T5, the number of dCA1 cells with mean rates above 1 Hz during the cue interval ranged from 63 to 75 (5 rats). b, Spike-time distribution for dCA1 principal cells with cue-port activity across phase of local 20–40-Hz oscillation at T1–T5 and T5e. c, d, Percentage of significantly phase-locked cells (c) and mean vector length of distribution in b (d), averaged across dCA1 cells with cue-port activity. T5d as in Fig. 2c. *#, as in Fig. 2c. eh, LEC cells, as in ad. Between T1 and T5, the number of LEC cells with mean rates above 1 Hz during the cue interval ranged from 76 to 82 (5 rats). The number of cue-port cells phase-locked to the local 20–40-Hz LFP increased significantly from T1 to T5 in both dCA1 and LEC (main text) and there was a significant increase in the phase locking of each cell (ANOVA for mean vector length: dCA1: F(4, 348) = 2.81, P = 0.026; LEC: F(4, 383) = 21.0, P < 0.001). On error trials, the mean vector for the spike-phase distribution decreased in both dCA1 and LEC (dCA1: two-tailed paired t-test, t(72) = 2.97, P = 0.004; LEC: t(81) = 4.05, P < 0.001). i, j, Interregional phase-locking of individual dCA1 and LEC cells. i, Phase-locking of dCA1 spikes as shown in c and d, but against 20?40-Hz oscillations in LEC. Left, percentage of significantly phase-locked cells. Right, mean vector length of the spike-phase distribution. The percentage of cells that was significantly phase-locked to the oscillations increased from 7.9 at T1 to 17.4 at T5 (P < 0.005; binomial test with Bonferroni correction). This increase was matched by an increase in the mean vector length of the spike-phase distribution (ANOVA for mean vector length: F(4, 348) = 2.8, P = 0.03). T5e and T5d indicate T5 error trials and down-sampled correct T5 trials, respectively. Mean vector length on T5e was significantly reduced compared to T5d (two-tailed paired t-test, t(72) = 2.60, P = 0.011). j, Phase-locking of LEC spikes as shown in g and h, but against 20–40-Hz oscillations in dCA1. The percentage of LEC cells significantly phase-locked to 20–40-Hz oscillations in dCA1 increased from 13.3 at T1 to 24.4 at T5 (P < 0.005). This was accompanied by a significant increase in the mean vector length of the spike-phase distribution (F(4, 383) = 2.7, P = 0.03). Mean vector length on error trials (T5e) decreased significantly compared to down-sampled correct trials (T5d; two-tailed paired t-test, t(81) = 2.42, P = 0.017). *P < 0.05, Bonferroni post-hoc test; #P < 0.05, paired t-test.

Extended Data Figure 10 ROC-based and firing rate-based analysis of odour-specific representations in dCA1 and LEC.

a, Schematic representation of odour-type spike representations based on receiver operating characteristic (ROC) analysis (left) and direct comparison of firing rates (right). Odour-specific representation was assessed using the metric Selectivity. Selectivity was computed by comparing firing rates during sampling periods for left-associated and right-associated cues in successive 100-ms bins: Selectivity was first expressed as (FR1 − FR2)/(FR1 + FR2), where FR1 and FR2 are mean firing rates for multiple trials on individual 100-ms bins during left- and right-associated cue samplings, respectively. To confirm the development of odour-specific representations, we subsequently computed Selectivity using ROC analysis, a method based on signal detection theory: Selectivity (ROC) was computed as 2 × (auROC − 0.5), where auROC is the area under the ROC curve computed from spike numbers for left and right trials on individual 100-ms bins. After scaling, for both Selectivity metrics, Selectivity = 1 indicates that the cell fired only on left-associated odour trials, whereas Selectivity = –1 denotes firing only on right-associated odour trials. b, Trial-type representations for all dCA1 cells with activity at the cue port shown as in Fig. 4a but using ROC analysis. Note development of selectivity also with this method. c, As in b, but for LEC cells.

Supplementary information

Odour-place association task

On each trial, the rat sampled odours in a cue port for 1 s and then, depending on odour identity, ran to either of two cups for food reward. In this video, the rat was trained above criteria of 85% (time point T5) and performed the task with odours A and B. Tone signals were delivered 1.0 s after initiation of cue sampling, after which food reward was made available in the associated food cup. If the rat withdrew its nose from the cue port before the tone signal, no food reward was delivered in any of the cups. (MOV 9124 kb)

PowerPoint slides

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Igarashi, K., Lu, L., Colgin, L. et al. Coordination of entorhinal–hippocampal ensemble activity during associative learning. Nature 510, 143–147 (2014). https://doi.org/10.1038/nature13162

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nature13162

This article is cited by

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing