One-shot learning and behavioral eligibility traces in sequential decision making

  1. Marco P Lehmann  Is a corresponding author
  2. He A Xu
  3. Vasiliki Liakoni
  4. Michael H Herzog
  5. Wulfram Gerstner
  6. Kerstin Preuschoff
  1. École Polytechnique Fédérale de Lausanne, Switzerland
  2. University of Geneva, Switzerland

Abstract

In many daily tasks we make multiple decisions before reaching a goal. In order to learn such sequences of decisions, a mechanism to link earlier actions to later reward is necessary. Reinforcement learning theory suggests two classes of algorithms solving this credit assignment problem: In classic temporal-difference learning, earlier actions receive reward information only after multiple repetitions of the task, whereas models with eligibility traces reinforce entire sequences of actions from a single experience (one-shot). Here we show one-shot learning of sequences. We developed a novel paradigm to directly observe which actions and states along a multi-step sequence are reinforced after a single reward. By focusing our analysis on those states for which RL with and without eligibility trace make qualitatively distinct predictions, we find direct behavioral (choice probability) and physiological (pupil dilation) signatures of reinforcement learning with eligibility trace across multiple sensory modalities.

Data availability

The datasets generated during the current study are available on Dryad, at the following address https://dx.doi.org/10.5061/dryad.j7h6f69

The following data sets were generated

Article and author information

Author details

  1. Marco P Lehmann

    School of Life Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
    For correspondence
    marco.lehmann@alumni.epfl.ch
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5274-144X
  2. He A Xu

    School of Life Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
    Competing interests
    The authors declare that no competing interests exist.
  3. Vasiliki Liakoni

    School of Life Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
    Competing interests
    The authors declare that no competing interests exist.
  4. Michael H Herzog

    School of Life Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
    Competing interests
    The authors declare that no competing interests exist.
  5. Wulfram Gerstner

    School of Life Sciences, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
    Competing interests
    The authors declare that no competing interests exist.
  6. Kerstin Preuschoff

    Swiss Center for Affective Sciences, University of Geneva, Genève, Switzerland
    Competing interests
    The authors declare that no competing interests exist.

Funding

Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (CRSII2 147636 (Sinergia))

  • Marco P Lehmann
  • He A Xu
  • Vasiliki Liakoni
  • Michael H Herzog
  • Wulfram Gerstner
  • Kerstin Preuschoff

Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung (CRSII2 200020 165538)

  • Marco P Lehmann
  • Vasiliki Liakoni
  • Wulfram Gerstner

Horizon 2020 Framework Programme (Human Brain Project (SGA2) 785907)

  • Michael H Herzog
  • Wulfram Gerstner

H2020 European Research Council (268 689 MultiRules)

  • Wulfram Gerstner

Horizon 2020 Framework Programme (Human Brain Project (SGA1) 720270)

  • Michael H Herzog
  • Wulfram Gerstner

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Ethics

Human subjects: Experiments were conducted in accordance with the Helsinki declaration and approved by the ethics commission of the Canton de Vaud (164/14 Titre: Aspects fondamentaux de la reconnaissance des objets : protocole général). All participants were informed about the general purpose of the experiment and provided written, informed consent. They were told that they could quit the experiment at any time they wish.

Reviewing Editor

  1. Thorsten Kahnt, Northwestern University, United States

Publication history

  1. Received: April 5, 2019
  2. Accepted: November 1, 2019
  3. Accepted Manuscript published: November 11, 2019 (version 1)
  4. Version of Record published: December 6, 2019 (version 2)

Copyright

© 2019, Lehmann et al.

This article is distributed under the terms of the Creative Commons Attribution License permitting unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,210
    Page views
  • 337
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Marco P Lehmann
  2. He A Xu
  3. Vasiliki Liakoni
  4. Michael H Herzog
  5. Wulfram Gerstner
  6. Kerstin Preuschoff
(2019)
One-shot learning and behavioral eligibility traces in sequential decision making
eLife 8:e47463.
https://doi.org/10.7554/eLife.47463

Further reading

    1. Neuroscience
    Stefanie Engert et al.
    Research Article

    Gustatory sensory neurons detect caloric and harmful compounds in potential food and convey this information to the brain to inform feeding decisions. To examine the signals that gustatory neurons transmit and receive, we reconstructed gustatory axons and their synaptic sites in the adult Drosophila melanogaster brain, utilizing a whole-brain electron microscopy volume. We reconstructed 87 gustatory projections from the proboscis labellum in the right hemisphere and 57 from the left, representing the majority of labellar gustatory axons. Gustatory neurons contain a nearly equal number of interspersed pre-and post-synaptic sites, with extensive synaptic connectivity among gustatory axons. Morphology- and connectivity-based clustering revealed six distinct groups, likely representing neurons recognizing different taste modalities. The vast majority of synaptic connections are between neurons of the same group. This study resolves the anatomy of labellar gustatory projections, reveals that gustatory projections are segregated based on taste modality, and uncovers synaptic connections that may alter the transmission of gustatory signals.

    1. Neuroscience
    Vladislav Ayzenberg, Stella Lourenco
    Research Article

    Categorization of everyday objects requires that humans form representations of shape that are tolerant to variations among exemplars. Yet, how such invariant shape representations develop remains poorly understood. By comparing human infants (6–12 months; N=82) to computational models of vision using comparable procedures, we shed light on the origins and mechanisms underlying object perception. Following habituation to a never-before-seen object, infants classified other novel objects across variations in their component parts. Comparisons to several computational models of vision, including models of high-level and low-level vision, revealed that infants’ performance was best described by a model of shape based on the skeletal structure. Interestingly, infants outperformed a range of artificial neural network models, selected for their massive object experience and biological plausibility, under the same conditions. Altogether, these findings suggest that robust representations of shape can be formed with little language or object experience by relying on the perceptually invariant skeletal structure.