A hierarchy of intrinsic timescales across primate cortex

Journal name:
Nature Neuroscience
Volume:
17,
Pages:
1661–1663
Year published:
DOI:
doi:10.1038/nn.3862
Received
Accepted
Published online

Specialization and hierarchy are organizing principles for primate cortex, yet there is little direct evidence for how cortical areas are specialized in the temporal domain. We measured timescales of intrinsic fluctuations in spiking activity across areas and found a hierarchical ordering, with sensory and prefrontal areas exhibiting shorter and longer timescales, respectively. On the basis of our findings, we suggest that intrinsic timescales reflect areal specialization for task-relevant computations over multiple temporal ranges.

At a glance

Figures

  1. Spike-count autocorrelation reveals a hierarchical ordering of intrinsic timescales.
    Figure 1: Spike-count autocorrelation reveals a hierarchical ordering of intrinsic timescales.

    (a) Data sets span seven cortical areas in the macaque monkey: MT, LIP, LPFC, OFC, ACC, S1 and S2. (b) Anatomical hierarchy of the areas, based on long-range projection patterns9, 10. (c) Spike-count autocorrelation was computed for neuronal spiking activity during the foreperiod of cognitive tasks. Each panel shows the data set for one of six research groups. The decay of autocorrelation was fit by an exponential decay with an offset. Some areas in data sets show refractory adaptation at short time lags, which were excluded from the fit (Online Methods). Solid lines show the exponential fit. Intrinsic timescale extracted from the fit is shown for each area. Autocorrelation was computed with 50-ms time bins. Error bars indicate s.e.m. (d) Intrinsic timescales across the visual-prefrontal hierarchy in five data sets (left) and the somatosensory hierarchy (right). Error bars indicate standard error of fit parameters.

  2. Links between intrinsic timescales and longer functional timescales.
    Figure 2: Links between intrinsic timescales and longer functional timescales.

    (a) The fit function for autocorrelation is defined by a timescale (τ), amplitude (A) and offset (B). Autocorrelation offset (B) reflects the strength of contributions with long timescales, which do not decay substantially within the fixation epoch. (b) Autocorrelation offset increases with intrinsic timescale. Colored lines show trends for individual data sets. The arrow shows the slope of dependence from a regression analysis (slope m = 0.8 ± 0.2 kHz). Error bars indicate s.e.m. (c) In the Lee data set, we previously measured timescales characterizing the decay of modulation of single-neuron firing rates by reward events while monkeys performed a competitive decision-making task11 for areas LIP (n = 160), LPFC (n = 243) and ACC (n = 134). The ordering of areas by reward timescale aligns with the ordering by intrinsic timescale. Error bars indicate standard error for fit parameter and median.

  3. Spike-count autocorrelations in time.
    Supplementary Fig. 1: Spike-count autocorrelations in time.

    Normalized autocorrelation matrices are shown for each area in a dataset. The matrix shows the mean correlation of the spike count in each time bin with the spike count in every other time bin, averaged across neurons. These show that the autocorrelation is roughly stationary across time during the foreperiod.

  4. Single neurons exhibit heterogeneous autocorrelations.
    Supplementary Fig. 2: Single neurons exhibit heterogeneous autocorrelations.

    Light grey traces show the spike-count autocorrelation as function of time lag for single neurons, averaged across time points. Circles mark the population mean at each time lag, and the curve shows the exponential fit to the population data. The observation of single-neuron heterogeneity reinforces the interpretation of intrinsic timescale as a characteristic at the population level rather than at the single-neuron level.

  5. Differences in mean firing rates across areas do not account for hierarchy of intrinsic timescales.
    Supplementary Fig. 3: Differences in mean firing rates across areas do not account for hierarchy of intrinsic timescales.

    Mean firing rates varied substantially across datasets and across areas within datasets. There was no significant dependence of intrinsic timescale on mean firing rate (P = 0.51, t(9) = −0.69, two-tailed t-test, regression slope m = −5.5 ± 7.9 ms/Hz; P = 0.16, rs = −0.34, Spearman’s rank correlation, two-tailed). Error bars mark s.e.

  6. Autocorrelation offset reflects trial-to-trial correlation.
    Supplementary Fig. 4: Autocorrelation offset reflects trial-to-trial correlation.

    Trial-to-trial correlation was calculated as the Pearson correlation coefficient between the foreperiod spike count in each trial and the spike count in the next trial. We hypothesized that autocorrelation offset would positively correlate with trial-to-trial correlation, and found a significant positive correlation between them. This indicates that the autocorrelation offset includes contributions from variability at timescales are comparable to or longer than the trial duration. Colored lines show trends for individual datasets. The arrow shows the slope of dependence from a regression analysis (slope m = 1.3 ± 0.3). Error bars mark s.e.

  7. Hierarchical ordering of areas by timescale of reward memory.
    Supplementary Fig. 5: Hierarchical ordering of areas by timescale of reward memory.

    In the Lee dataset, we previously measured timescales of the decay of memory traces for past rewards in single-neuron firing rates, while monkeys performed a competitive decision-making task. (a) The cumulative distribution of reward timescales in LIP (n = 160), LPFC (n = 243), and ACC (n = 134). For neurons fit with the sum of two reward timescales, we used the harmonic mean of the two timescales. (b) Median reward timescale for the three areas. Error bars mark s.e.

References

  1. Lennie, P. Perception 27, 889935 (1998).
  2. Badre, D. & D'Esposito, M. Nat. Rev. Neurosci. 10, 659669 (2009).
  3. Hasson, U., Yang, E., Vallines, I., Heeger, D.J. & Rubin, N. J. Neurosci. 28, 25392550 (2008).
  4. Honey, C.J. et al. Neuron 76, 423434 (2012).
  5. Churchland, M.M. et al. Nat. Neurosci. 13, 369378 (2010).
  6. Goris, R.L.T., Movshon, J.A. & Simoncelli, E.P. Nat. Neurosci. 17, 858865 (2014).
  7. Maimon, G. & Assad, J.A. Neuron 62, 426440 (2009).
  8. Churchland, A.K. et al. Neuron 69, 818831 (2011).
  9. Felleman, D.J. & Van Essen, D.C. Cereb. Cortex 1, 147 (1991).
  10. Barbas, H. & Rempel-Clower, N. Cereb. Cortex 7, 635646 (1997).
  11. Bernacchia, A., Seo, H., Lee, D. & Wang, X.-J. Nat. Neurosci. 14, 366372 (2011).
  12. Goldman, M.S., Compte, A. & Wang, X.-J. in Encyclopedia of Neuroscience (ed. Squire, L.R.) 165178 (Academic Press, Oxford, 2008).
  13. Buracas, G.T., Zador, A.M., DeWeese, M.R. & Albright, T.D. Neuron 20, 959969 (1998).
  14. Salinas, E., Hernandez, A., Zainos, A. & Romo, R. J. Neurosci. 20, 55035515 (2000).
  15. Wang, X.-J. Neuron 36, 955968 (2002).
  16. Wang, H., Stradtman, G.G., Wang, X.-J. & Gao, W.-J. Proc. Natl. Acad. Sci. USA 105, 1679116796 (2008).
  17. Wang, Y. et al. Nat. Neurosci. 9, 534542 (2006).
  18. Fuster, J. The Prefrontal Cortex (Academic Press, New York, 2008).
  19. Amatrudo, J.M. et al. J. Neurosci. 32, 1364413660 (2012).
  20. Elston, G.N. Cereb. Cortex 13, 11241138 (2003).
  21. Bisley, J.W., Zaksas, D., Droll, J.A. & Pasternak, T. J. Neurophysiol. 91, 286300 (2004).
  22. Zaksas, D. & Pasternak, T. J. Neurophysiol. 94, 41564167 (2005).
  23. Zaksas, D. & Pasternak, T. J. Neurosci. 26, 1172611742 (2006).
  24. Hussar, C.R. & Pasternak, T. Neuron 64, 730743 (2009).
  25. Hussar, C.R. & Pasternak, T. J. Neurosci. 32, 27472761 (2012).
  26. Freedman, D.J. & Assad, J.A. Nature 443, 8588 (2006).
  27. Swaminathan, S.K. & Freedman, D.J. Nat. Neurosci. 15, 315320 (2012).
  28. Seo, H., Barraclough, D.J. & Lee, D. J. Neurosci. 29, 72787289 (2009).
  29. Seo, H., Barraclough, D.J. & Lee, D. Cereb. Cortex 17 (suppl. 1), i110i117 (2007).
  30. Seo, H. & Lee, D. J. Neurosci. 27, 83668377 (2007).
  31. Kennerley, S.W., Dahmubed, A.F., Lara, A.H. & Wallis, J.D. J. Cogn. Neurosci. 21, 11621178 (2009).
  32. Kennerley, S.W. & Wallis, J.D. J. Neurosci. 29, 32593270 (2009).
  33. Hosokawa, T., Kennerley, S.W., Sloan, J. & Wallis, J.D. J. Neurosci. 33, 1738517397 (2013).
  34. Padoa-Schioppa, C. & Assad, J.A. Nature 441, 223226 (2006).
  35. Padoa-Schioppa, C. & Assad, J.A. Nat. Neurosci. 11, 95102 (2008).
  36. Cai, X. & Padoa-Schioppa, C. J. Neurosci. 32, 37913808 (2012).
  37. Cai, X. & Padoa-Schioppa, C. Neuron 81, 11401151 (2014).
  38. Ponce-Alvarez, A., Nácher, V., Luna, R., Riehle, A. & Romo, R. J. Neurosci. 32, 1195611969 (2012).
  39. Ogawa, T. & Komatsu, H. J. Neurophysiol. 103, 24332445 (2010).
  40. Nishida, S. et al. Cereb. Cortex 24, 16711685 (2014).

Download references

Author information

Affiliations

  1. Center for Neural Science, New York University, New York, New York, USA.

    • John D Murray &
    • Xiao-Jing Wang
  2. Department of Neurobiology, Yale University School of Medicine, New Haven, Connecticut, USA.

    • John D Murray,
    • Alberto Bernacchia,
    • Hyojung Seo,
    • Daeyeol Lee &
    • Xiao-Jing Wang
  3. School of Engineering and Science, Jacobs University, Bremen, Germany.

    • Alberto Bernacchia
  4. Department of Neurobiology, The University of Chicago, Chicago, Illinois, USA.

    • David J Freedman
  5. Instituto de Fisiología Celular, Universidad Nacional Autónoma de México, México D.F., Mexico.

    • Ranulfo Romo
  6. El Colegio Nacional, México D.F., Mexico.

    • Ranulfo Romo
  7. Helen Wills Neuroscience Institute, University of California, Berkeley, Berkeley, California, USA.

    • Jonathan D Wallis
  8. Department of Psychology, University of California, Berkeley, Berkeley, California, USA.

    • Jonathan D Wallis
  9. NYU-ECNU Institute of Brain and Cognitive Science, NYU Shanghai, Shanghai, China.

    • Xinying Cai &
    • Xiao-Jing Wang
  10. Department of Anatomy and Neurobiology, Washington University in St. Louis, St. Louis, Missouri, USA.

    • Xinying Cai &
    • Camillo Padoa-Schioppa
  11. Department of Neurobiology and Anatomy, University of Rochester, Rochester, New York, USA.

    • Tatiana Pasternak
  12. Center for Visual Science, University of Rochester, Rochester, New York, USA.

    • Tatiana Pasternak

Contributions

J.D.M., A.B. and X.-J.W. designed the research and wrote the manuscript. J.D.M. analyzed the data and prepared the figures. D.J.F., R.R., J.D.W., X.C., C.P.-S., T.P., H.S. and D.L. contributed the electrophysiological data. All authors contributed to editing and revising the manuscript.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Supplementary information

Supplementary Figures

  1. Supplementary Figure 1: Spike-count autocorrelations in time. (368 KB)

    Normalized autocorrelation matrices are shown for each area in a dataset. The matrix shows the mean correlation of the spike count in each time bin with the spike count in every other time bin, averaged across neurons. These show that the autocorrelation is roughly stationary across time during the foreperiod.

  2. Supplementary Figure 2: Single neurons exhibit heterogeneous autocorrelations. (199 KB)

    Light grey traces show the spike-count autocorrelation as function of time lag for single neurons, averaged across time points. Circles mark the population mean at each time lag, and the curve shows the exponential fit to the population data. The observation of single-neuron heterogeneity reinforces the interpretation of intrinsic timescale as a characteristic at the population level rather than at the single-neuron level.

  3. Supplementary Figure 3: Differences in mean firing rates across areas do not account for hierarchy of intrinsic timescales. (135 KB)

    Mean firing rates varied substantially across datasets and across areas within datasets. There was no significant dependence of intrinsic timescale on mean firing rate (P = 0.51, t(9) = −0.69, two-tailed t-test, regression slope m = −5.5 ± 7.9 ms/Hz; P = 0.16, rs = −0.34, Spearman’s rank correlation, two-tailed). Error bars mark s.e.

  4. Supplementary Figure 4: Autocorrelation offset reflects trial-to-trial correlation. (120 KB)

    Trial-to-trial correlation was calculated as the Pearson correlation coefficient between the foreperiod spike count in each trial and the spike count in the next trial. We hypothesized that autocorrelation offset would positively correlate with trial-to-trial correlation, and found a significant positive correlation between them. This indicates that the autocorrelation offset includes contributions from variability at timescales are comparable to or longer than the trial duration. Colored lines show trends for individual datasets. The arrow shows the slope of dependence from a regression analysis (slope m = 1.3 ± 0.3). Error bars mark s.e.

  5. Supplementary Figure 5: Hierarchical ordering of areas by timescale of reward memory. (120 KB)

    In the Lee dataset, we previously measured timescales of the decay of memory traces for past rewards in single-neuron firing rates, while monkeys performed a competitive decision-making task. (a) The cumulative distribution of reward timescales in LIP (n = 160), LPFC (n = 243), and ACC (n = 134). For neurons fit with the sum of two reward timescales, we used the harmonic mean of the two timescales. (b) Median reward timescale for the three areas. Error bars mark s.e.

PDF files

  1. Supplementary Text and Figures (7.27 MB)

    Supplementary Figures 1–5 and Supplementary Note

  2. Supplementary Methods Checklist (382 KB)

Additional data