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.

  • Article
  • Published:

A spatiotemporal coding mechanism for background-invariant odor recognition

Nature Neuroscience volume 16, pages 1830–1839 (2013)Cite this article

Subjects

Abstract

Sensory stimuli evoke neural activity that evolves over time. What features of these spatiotemporal responses allow the robust encoding of stimulus identity in a multistimulus environment? Here we examined this issue in the locust (Schistocerca americana) olfactory system. We found that sensory responses evoked by an odorant (foreground) varied when presented atop or after an ongoing stimulus (background). These inconsistent sensory inputs triggered dynamic reorganization of ensemble activity in the downstream antennal lobe. As a result, partial pattern matches between neural representations encoding the same foreground stimulus across conditions were achieved. The degree and segments of response overlaps varied; however, any overlap observed was sufficient to drive background-independent responses in the downstream neural population. Notably, recognition performance of locusts in behavioral assays correlated well with our physiological findings. Hence, our results reveal how background-independent recognition of odors can be achieved using spatiotemporal patterns of neural activity.

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

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

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

Figure 1: Dynamic states of odor-evoked neural activity.
Figure 2: Ongoing background odor-evoked activity interferes with the response of individual sensory neurons to the freshly introduced stimulus.
Figure 3: Responses of individual projection neurons to the foreground odor can vary unpredictably depending on stimulus history.
Figure 4: Ensemble projection neuron activity makes dynamic transitions to create response overlaps between neural representations.
Figure 5: Projection neuron population response classification in a piecewise manner can allow background-independent recognition of odors.
Figure 6: Kenyon cells are sensitive to partial pattern matches in antennal lobe activity.
Figure 7: Behavioral ability of locusts to recognize odorants independently of background correlates with the classification analysis results.
Figure 8: Behavioral results validate the predictions from our physiology data.

Similar content being viewed by others

References

  1. Laurent, G., Wehr, M. & Davidowitz, H. Temporal representations of odors in an olfactory network. J. Neurosci. 16, 3837–3847 (1996).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Meredith, M. & Moulton, D.G. Patterned response to odor in single neurons of goldfish olfactory bulb: influence of odor quality and other stimulus parameters. J. Gen. Physiol. 71, 615–643 (1978).

    Article  CAS  PubMed  Google Scholar 

  3. Di Lorenzo, P.M., Chen, J.Y. & Victor, J.D. Quality time: representation of a multidimensional sensory domain through temporal coding. J. Neurosci. 29, 9227–9238 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Jones, L.M., Fontanini, A., Sadacca, B.F., Miller, P. & Katz, D.B. Natural stimuli evoke dynamic sequence of states in sensory cortical ensembles. Proc. Natl. Acad. Sci. USA 104, 18772–18777 (2007).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Seifritz, E. et al. Spatiotemporal pattern of neural processing in the human auditory cortex. Science 297, 1706–1708 (2002).

    Article  CAS  PubMed  Google Scholar 

  6. Smear, M., Shusterman, R., O'Conor, R., Bozza, T. & Rinberg, D. Perception of sniff phase in mouse olfaction. Nature 479, 397–400 (2011).

    Article  CAS  PubMed  Google Scholar 

  7. Griffiths, T.D., Uppenkamp, S., Johnsrude, I., Josephs, O. & Patterson, R.D. Encoding of the temporal regularity of sound in the human brainstem. Nat. Neurosci. 4, 633–637 (2001).

    Article  CAS  PubMed  Google Scholar 

  8. Ahissar, E., Haidarliu, S. & Zacksenhouse, M. Decoding temporally encoded sensory input by cortical oscillations and thalamic phase comparators. Proc. Natl. Acad. Sci. USA 94, 11633–11638 (1997).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Carlson, B.A. Temporal-pattern recognition by single neurons in a sensory pathway devoted to social communication behavior. J. Neurosci. 29, 9417–9428 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Machens, C.K. et al. Representation of acoustic communication signals by insect auditory receptor neurons. J. Neurosci. 21, 3215–3227 (2001).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Schnitzer, M.J. & Meister, M. Multineuronal firing patterns in the signal from eye to brain. Neuron 37, 499–511 (2003).

    Article  CAS  PubMed  Google Scholar 

  12. Friedrich, R.W. & Laurent, G. Dynamic optimization of odor representations by slow temporal patterning of mitral cell activity. Science 291, 889–894 (2001).

    Article  CAS  PubMed  Google Scholar 

  13. Dayan, P. & Abbott, L.F. Theoretical Neuroscience: Computational and Mathematical Modeling of Neural Systems (The MIT Press, 2001).

  14. Stopfer, M., Bhagavan, S., Smith, B.H. & Laurent, G. Impaired odour discrimination on desynchronization of odour-encoding neural assemblies. Nature 390, 70–74 (1997).

    Article  CAS  PubMed  Google Scholar 

  15. Vickers, N.J., Christensen, T.A., Baker, T.C. & Hildebrand, J.G. Odour-plume dynamics influence the brain′s olfactory code. Nature 410, 466–470 (2001).

    Article  CAS  PubMed  Google Scholar 

  16. Ito, I., Ong, R.C., Raman, B. & Stopfer, M. Sparse odor representation and olfactory learning. Nat. Neurosci. 11, 1177–1184 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Simões, P., Ott, S.R. & Niven, J.E. Associative olfactory learning in the desert locust, Schistocerca gregaria. J. Exp. Biol. 214, 2495–2503 (2011).

    Article  PubMed  Google Scholar 

  18. Kreher, S.A., Matthew, D., Kim, J. & Carlson, J.R. Translation of sensory input into behavioral output via an olfactory system. Neuron 59, 110–124 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Galán, R.F., Sachse, S., Galizia, C.G. & Herz, A.V.M. Odor-driven attractor dynamics in the antennal lobe allow for simple and rapid olfactory pattern classification. Neural Comput. 16, 999–1012 (2004).

    Article  Google Scholar 

  20. Mazor, O. & Laurent, G. Transient dynamics versus fixed points in odor representations by locust antennal lobe projection neurons. Neuron 48, 661–673 (2005).

    Article  CAS  PubMed  Google Scholar 

  21. Raman, B., Joseph, J., Tang, J. & Stopfer, M. Temporally diverse firing patterns in olfactory receptor neurons underlie spatiotemporal neural codes for odors. J. Neurosci. 30, 1994–2006 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Stopfer, M., Jayaraman, V. & Laurent, G. Odor identity vs. intensity coding in an olfactory system. Neuron 39, 991–1004 (2003).

    Article  CAS  PubMed  Google Scholar 

  23. Bathellier, B., Buhl, D.L., Accolla, R. & Carleton, A. Dynamic ensemble coding in the mamallian olfactory bulb: sensory information at different timesales. Neuron 57, 586–598 (2008).

    Article  CAS  PubMed  Google Scholar 

  24. Brown, S.L., Joseph, J. & Stopfer, M. Encoding a temporally structured stimulus with a temporally structured neural representation. Nat. Neurosci. 8, 1568–1576 (2005).

    Article  CAS  PubMed  Google Scholar 

  25. Rinberg, D., Koulakov, A. & Gelperin, A. Speed-accuracy tradeoff in olfaction. Neuron 51, 351–358 (2006).

    Article  CAS  PubMed  Google Scholar 

  26. Uchida, N. & Mainen, Z. Speed and accuracy of olfactory discrimination in the rat. Nat. Neurosci. 6, 1224–1229 (2003).

    Article  CAS  PubMed  Google Scholar 

  27. Abraham, N.M. et al. Maintaining accuracy at the expense of speed: stimulus similarity defines odor discrimination time in mice. Neuron 44, 865–876 (2004).

    CAS  PubMed  Google Scholar 

  28. Ditzen, M., Evers, J. & Galizia, C.G. Odor similarity does not influence the time needed for odor processing. Chem. Senses 28, 781–789 (2003).

    Article  PubMed  Google Scholar 

  29. Broome, B.M., Jayaraman, V. & Laurent, G. Encoding and decoding of overlapping odor sequences. Neuron 51, 467–482 (2006).

    Article  CAS  PubMed  Google Scholar 

  30. Spors, H., Wachowiak, M., Cohen, L.B. & Friedrich, R.W. Temporal dynamics and latency patterns of receptor neuron input to the olfactory bulb. J. Neurosci. 26, 1247–1259 (2006).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Spors, H. & Grisvald, A. Spatio-temporal dynamics of odor representations in the mammalian olfactory bulb. Neuron 34, 301–315 (2002).

    Article  CAS  PubMed  Google Scholar 

  32. Ito, I., Bazhenov, M., Ong, R.C., Raman, B. & Stopfer, M. Frequency transitions in odor-evoked neural oscillations. Neuron 64, 692–706 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Roweis, S.T. & Saul, L.K. Nonlinear dimensionality reduction by locally linear embedding. Science 290, 2323–2326 (2000).

    Article  CAS  PubMed  Google Scholar 

  34. Perez-Orive, J. et al. Oscillations and sparsening of odor representations in the mushroom body. Science 297, 359–365 (2002).

    Article  CAS  PubMed  Google Scholar 

  35. De Palo, G. et al. Common dynamical features of sensory adaptation in photoreceptors and olfactory sensory neurons. Sci. Rep. 3, 1251 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Nagel, K.I. & Wilson, R.I. Biophysical mechanisms underlying olfactory receptor neuron dynamics. Nat. Neurosci. 14, 208–216 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Baker, T.C. & Haynes, K.F. Field and laboratory electroantennographic measurements of pheromone plume structure correlated with oriental fruit moth behaviour. Physiol. Entomol. 14, 1–12 (1989).

    Article  Google Scholar 

  38. Laing, D.G., Eddy, A., Francis, G.W. & Stephens, L. Evidence for the temporal processing of odor mixtures in humans. Brain Res. 651, 317–324 (1994).

    Article  CAS  PubMed  Google Scholar 

  39. Martelli, C., Carlson, J.R. & Emonet, T. Intensity invariant dynamics and odor-specific latencies in olfactory receptor neuron response. J. Neurosci. 33, 6285–6297 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Pouzat, C., Mazor, O. & Laurent, G. Using noise signature to optimize spike-sorting and to assess neuronal classification quality. J. Neurosci. Methods 122, 43–57 (2002).

    Article  PubMed  Google Scholar 

  41. Saha, D., Leong, K., Katta, N. & Raman, B. Multi-unit recording methods to characterize neural activity in the locust (Schistocerca americana) olfactory circuits. J. Vis. Exp. 71, pii: e50139 (2013).

Download references

Acknowledgements

We thank N. Katta for performing double-blind behavior evaluation studies and D. Yang for help with behavioral experiments. We thank F. Gabbiani (Baylor College of Medicine), M. Stopfer and S. Reiter (Eunice Kennedy Shriver National Institute of Child Health and Human Development, US National Institutes of Health), D. Barbour, R. Wessel and members of Raman Lab (Washington University, St. Louis) for comments on previous versions of the manuscript. This research was supported by a McDonnell Center for Systems Neuroscience grant, an Office of Naval Research grant (N00014-12-1-0089) and startup funds from the Department of Biomedical Engineering at Washington University, St. Louis to B.R.

Author information

Author notes

  1. Debajit Saha, Kevin Leong and Chao Li: These authors contributed equally to this work.

Authors and Affiliations

  1. Department of Biomedical Engineering, Washington University, St. Louis, Missouri, USA

    Debajit Saha, Kevin Leong, Chao Li, Steven Peterson, Gregory Siegel & Baranidharan Raman

Authors
  1. Debajit Saha
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Kevin Leong
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Chao Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Steven Peterson
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Gregory Siegel
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Baranidharan Raman
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

B.R. conceived the study and designed the experiments. D.S., K.L. and G.S. performed the electrophysiological recordings. C.L. and S.P. did the behavioral experiments. C.L., K.L. and D.S. analyzed the data. B.R. wrote the paper, and D.S., C.L. and K.L. provided feedback on the manuscript.

Corresponding author

Correspondence to Baranidharan Raman.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Mean ORN, PN and KC responses for all odor pairs used in this study

PSTHs are shown for: (a) 2oct–hex, (b) chex–2hep, (c) bzald–iaa, (d) hxa–hex, (e) ger–cit, (f) mint–apple. The 4 s odor stimulation period is shown using a gray bar along the x-axis. For all odors used, three dynamical states can be clearly identified: an on-transient response following odor onset, an off-transient response after stimulus termination, and a steady-state between the two transient activity phases. n represents the number of neurons recorded. (g) The diverse set of background–foreground odor combinations chosen for the study is listed here. Diversity assessed based on functional groups, electroantennogram responses, vapor pressures, and complexity of the odorants (mono-molecular versus complex blend).

Supplementary Figure 2 Electroantennogram (EAG) responses to the chosen odor pairs

EAG recordings obtained from four locusts are shown for all six background–foreground odor pairs. Raw EAG signals (mean ± s.d.) obtained from one locust are shown on the left panels. Right panels reveal the distribution of peak EAG response amplitude to the background and foreground odors across different locusts to illustrate response consistency. Three groups of odor pairs can be easily identified based on their relative EAG response profiles: (a) odor pairs where the foreground odor has stronger EAG response: 2oct–hex and hxa–hex (b) odor pairs where the background odor has stronger EAG response: bzald–iaa and ger–cit, and (c) odor pairs with comparable EAG responses: chex–2hep and mint–apple. (*P < 0.05; paired t-test, n = four trials).

Supplementary Figure 3 Visualization of ensemble PN responses using linear principal component analysis (PCA)

PCA trajectories to all six odor pairs are shown. Same convention as that used in Fig. 4. The same sets of PNs used in Fig. 4 were used for generating the PCA plots.

Supplementary Figure 4 LLE plots showing PN ensemble response trajectories for 3 additional overlapping conditions

LLE plots showing PN ensemble response trajectories for 3 additional overlapping conditions. The three new presentation conditions include: background–500 ms latency–foreground, background–1000 ms latency–foreground, and background offset–500 ms latency–foreground. n = number of PNs recorded for each odor pair.

Supplementary Figure 5 Significance of PN classification results

(a) Percentage of time bins during pre-stimulus periods that were classified is less than 3% for all odor pairs. (b) Histograms revealing the distribution of angular distances between individual test patterns and their closest reference templates are shown. Same coloring scheme as used in panel a. Only vectors exceeding standard deviation test were included in this analysis. Black bars represent angular distance greater than 85°. The mean angular distance was between 68.65°–71.79°. (c) Classification of random vectors using reference vector templates obtained for each odor pair. Less than 5% of the random vectors were within the tolerance limit by chance. All other vectors exceeded the tolerance threshold and were not classified into any odor category.

Supplementary Figure 6 Average KC PSTHs for all odor pairs and for different overlapping conditions

Average KC PSTHs for all odor pairs and for different overlapping conditions. For each condition, firing rates were calculated over 100 ms non-overlapping time bins, averaged over ten trials and smoothed using a 3-point running average. For each odor pair, the plot follows the stimulus protocol scheme shown in Fig. 1b. n denotes the total number of KCs recorded for each odor pair. Max indicates the maximum firing rate observed.

Supplementary Figure 7 Locust retention tests and T-maze assay

(a) A schematic of locust palp opening response (POR). Dotted red line indicates the POR threshold used to determine a response. One or both maxillary palps have to cross this detection threshold at least once during the odor presentation period to be counted as positive response. (b) PORs for four consecutive blocks of unrewarded test trials are shown. Each block consisted of two test trials: presentations of the CST (iaa) and an untrained odor (bzald). Test trials started 10 min after the last training trial. A 10 min delay was maintained between test trials in a single block, and a 30 min delay was observed between consecutive blocks of test trials. Conditioned locusts had a significantly higher POR to the trained odor (iaa) during all four test trials (**P = 1.22×10−4, 6.10×10−5, 7.63×10−5, 3.05×10−5; McNemar's exact test, n = 28 locusts). The frequency of POR observed for trained and untrained odor remained consistent across the four consecutive test blocks (Cochran's Q test; for CST: Q = 0.67, df = 3, P = 0.87; for untrained odor: Q = 2.2, df = 3, P = 0.53). (c) The bar graph summarizes responses of locusts to a trained odor (CST – cit) and an untrained odor (ger). The POR to citral was low indicating that effective association between CST and unconditioned stimulus was not achieved (P = 0.50; McNemar's exact test, n = 26 locusts). n denotes the number of locusts used in the training set. (d,e) Bar graphs summarizing conditional POR probability in those locusts that responded only to the CST are shown (*P = 0.0351; NS indicates not significant, P > 0.05; McNemar's exact test with Bonferroni correction for multiple comparisons, n = 25 locusts for 2oct–hex, n = 27 locusts for bzald–iaa). (f) A schematic of the T-maze assay is shown. Locusts were restrained in a custom-designed holder and released just before the odor delivery. A test odor and the control (mineral oil) were simultaneously presented at the two odor delivery ports. An exhaust fan at the center of the maze ensured that there was a stable airflow inside the maze (flow patterns were confirmed with titanium tetrachloride). Each locust was given 4 min to make a decision: i.e. select a T-maze arm, reach and touch the sidewall at the end of the selected arm with its leg or antenna.

Supplementary Figure 8 Qualitatively similar results obtained from analysis of ensemble neural activity recorded from single locusts

Classification analysis using PN responses obtained from a single locust is shown. Results for all six different odor pairs are arranged as in Fig. 5b–g. n denotes the number of PNs recorded from both antennal lobes of the locust. Note that for each odor pair a different locust was used.

Supplementary Figure 9 Proposed attractor model of olfactory coding

(a) Each odor is encoded by a 'sub-space' or an 'attractor' in the state-space that encodes for its identity. Blue and red curves represent response trajectories evoked by pure odor presentations. The entire neural response dynamics during odor presentation (i.e. on-transient and steady-state activities) strive to keep the response within the attractor for the particular odor. Note that the steady-state activities are closer to the baseline response but are still odor-specific and aligned with the on-transient response. When a foreground odor is presented following a preceding stimulus, a strong excitatory input is required to overcome the inhibition offered by the on-going activity, and switch the system's response to the foreground odor attractor as shown by the dynamic trajectory in black. B denotes the baseline response where the neural activity eventually returns. (b) Percentage of PNs with excitatory or inhibitory responses to each background and foreground odor is shown (see Online Methods for details regarding the criteria used for this categorization). The height of the bar plot above and below the horizontal axis ('zero percentage') indicates the fraction of excitatory and inhibitory PNs respectively. As can be noted amongst all foreground odors (red bars), citral evoked the least excitatory and most inhibitory response. n represents the total number of PNs in each set (same as in Fig. 4).

Supplementary Figure 10 Examples of ORN, PN and KC spike-sorting

(a) An example of ORN recording and spike-sorting. (Left panel) Raw extracellular trace showing response of a single ORN. (Middle panel) Individual ORN spike events (black) and their mean (red). (Right panel) Inter-spike interval distribution for the identified ORN. (b) An example of PN spike-sorting. (Left panel) Extracellular waveforms from four independent channels of a tetrode are shown for all spiking events corresponding to two simultaneously recorded PNs. Individual events (black), mean (red), and s.d. (blue) are shown for both cells. (Right panel – top) Histograms obtained by projecting high-dimensional PN event representations (180 dimensional vector obtained by concatenating signals from all electrodes) onto the line connecting their means. To be considered a well-isolated unit, as in this case, a bimodal distribution with cluster centers separated by at least five times the noise s.d. is expected for every pair of simultaneously recorded cells. (Right panel – bottom) Distributions of inter-spike intervals are shown for these two PNs. (c) Similar plot showing an example for KC spike-sorting.

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–10 (PDF 4058 kb)

Source data

Rights and permissions

Reprints and permissions

About this article

Cite this article

Saha, D., Leong, K., Li, C. et al. A spatiotemporal coding mechanism for background-invariant odor recognition. Nat Neurosci 16, 1830–1839 (2013). https://doi.org/10.1038/nn.3570

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nn.3570

This article is cited by

Search

Advanced 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