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.

COGNITIVE SCIENCE

Imprecise learning and uncertainty

Nature Human Behaviour volume 5, pages 7–8 (2021)Cite this article

Subjects

A study in Nature Human Behaviour proposes a biologically plausible algorithm producing near-optimal behaviour in uncertain and volatile environments through computational imprecision. A complementary study in the same issue shows that, depending on context, uncertainty itself guides different decisions and is differentially represented in the brain.

Computational imprecision is usually detrimental and may underlie suboptimal choices5. However, Findling and colleagues propose that a minimal model making first-order inferences on stimulus–action–outcome contingencies where computations are corrupted by imprecisions conforming to Weber’s law counterintuitively shows near-optimal performance without explicitly tracking volatility. They compared this Weber-imprecision model with models of optimal adaptive computations, their biologically plausible algorithmic approximations and non-inferential adaptive reinforcement learning models in various volatile environments. Indeed, the Weber-imprecision model reaches near-optimal performance, outperforms reinforcement learning and accounts for human performance better than all other models. The Weber-imprecision model is a computationally frugal heuristic. It quickly infers locally accurate but globally inaccurate simplified world models assuming stable contingencies. Computational imprecisions enable rapid adaptations that optimize short-term predictability. In contrast, higher-order inferential models infer more general, more accurate and more complex world models over a longer time, which reduces their short-term effectiveness. Many everyday situations are rather stable and require no higher-order inferences. Computational imprecisions enable humans to adapt to unstable environments in a near-optimal fashion without the need of computationally expensive higher-order inferences. In addition, imprecise neural computations may be more robust against pathology of the underlying substrate. For these reasons imprecise neural computations may have been preserved throughout evolution.

This is a preview of subscription content

Access options

Buy article

Get time limited or full article access on ReadCube.

$32.00

Buy

All prices are NET prices.

References

  1. Behrens, T. E., Woolrich, M. W., Walton, M. E. & Rushworth, M. F. Nat. Neurosci. 10, 1214–1221 (2007).

    CAS  Article  Google Scholar 

  2. Payzan-LeNestour, E., Dunne, S., Bossaerts, P. & O’Doherty, J. P. Neuron 79, 191–201 (2013).

    CAS  Article  Google Scholar 

  3. Bossaerts, P., Yadav, N. & Murawski, C. Philos. Trans. R. Soc. Lond. B Biol. Sci. 374, 20180138 (2019).

    Article  Google Scholar 

  4. Findling, C., Chopin, N. & Koechlin, E. Nat. Hum. Behav. https://doi.org/10.1038/s41562-020-00971-z (2020).

  5. Drugowitsch, J., Wyart, V., Devauchelle, A. D. & Koechlin, E. Neuron 92, 1398–1411 (2016).

    CAS  Article  Google Scholar 

  6. Deco, G., Scarano, L. & Soto-Faraco, S. J. Neurosci. 27, 11192–11200 (2007).

    CAS  Article  Google Scholar 

  7. Morales, J., Lau, H. & Fleming, S. M. J. Neurosci. 38, 3534–3546 (2018).

    CAS  Article  Google Scholar 

  8. Lebreton, M., Abitbol, R., Daunizeau, J. & Pessiglione, M. Nat. Neurosci. 18, 1159–1167 (2015).

    CAS  Article  Google Scholar 

  9. Trudel, N. et al. Nat. Hum. Behav. https://doi.org/10.1038/s41562-020-0929-3 (2020).

  10. Findling, C., Skvortsova, V., Dromnelle, R., Palminteri, S. & Wyart, V. Nat. Neurosci. 22, 2066–2077 (2019).

    CAS  Article  Google Scholar 

  11. Lebreton, M., Jorge, S., Michel, V., Thirion, B. & Pessiglione, M. Neuron 64, 431–439 (2009).

    CAS  Article  Google Scholar 

  12. Klein, T. A., Ullsperger, M. & Jocham, G. Nat. Commun. 8, 16033 (2017).

    CAS  Article  Google Scholar 

  13. Jocham, G., Neumann, J., Klein, T. A., Danielmeier, C. & Ullsperger, M. J. Neurosci. 29, 7489–7496 (2009).

    CAS  Article  Google Scholar 

  14. Fischer, A. G., Bourgeois-Gironde, S. & Ullsperger, M. Nat. Commun. 8, 1690 (2017).

    Article  Google Scholar 

  15. Boorman, E. D., Rushworth, M. F. & Behrens, T. E. J. Neurosci. 33, 2242–2253 (2013).

    CAS  Article  Google Scholar 

Download references

Author information

Affiliations

  1. Otto-von-Guericke University Magdeburg, Magdeburg, Germany

    Markus Ullsperger

  2. Center for Behavioral Brain Sciences, Magdeburg, Germany

    Markus Ullsperger

Authors
  1. Markus Ullsperger
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Markus Ullsperger.

Ethics declarations

Competing interests

The author declares no competing interests.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ullsperger, M. Imprecise learning and uncertainty. Nat Hum Behav 5, 7–8 (2021). https://doi.org/10.1038/s41562-020-00992-8

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41562-020-00992-8

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