Contrasting action and posture coding with hierarchical deep neural network models of proprioception

  1. Kai J Sandbrink
  2. Pranav Mamidanna
  3. Claudio Michaelis
  4. Matthias Bethge
  5. Mackenzie W Mathis  Is a corresponding author
  6. Alexander Mathis  Is a corresponding author
  1. École Polytechnique Fédérale de Lausanne, Switzerland
  2. Eberhard Karls Universität Tübingen, Germany

Abstract

Biological motor control is versatile, efficient, and depends on proprioceptive feedback. Muscles are flexible and undergo continuous changes, requiring distributed adaptive control mechanisms that continuously account for the body's state. The canonical role of proprioception is representing the body state. We hypothesize that the proprioceptive system could also be critical for high-level tasks such as action recognition. To test this theory, we pursued a task-driven modeling approach, which allowed us to isolate the study of proprioception. We generated a large synthetic dataset of human arm trajectories tracing characters of the Latin alphabet in 3D space, together with muscle activities obtained from a musculoskeletal model and model-based muscle spindle activity. Next, we compared two classes of tasks: trajectory decoding and action recognition, which allowed us to train hierarchical models to decode either the position and velocity of the end-effector of one's posture or the character (action) identity from the spindle firing patterns. We found that artificial neural networks could robustly solve both tasks, and the networks'units show tuning properties similar to neurons in the primate somatosensory cortex and the brainstem. Remarkably, we found uniformly distributed directional selective units only with the action-recognition-trained models and not the trajectory-decoding-trained models. This suggests that proprioceptive encoding is additionally associated with higher-level functions such as action recognition and therefore provides new, experimentally testable hypotheses of how proprioception aids in adaptive motor control.

Data availability

The computational dataset and code to create it is available at https://github.com/amathislab/DeepDraw

Article and author information

Author details

  1. Kai J Sandbrink

    Brain Mind Institute, École Polytechnique Fédérale de Lausanne, Genève, Switzerland
    Competing interests
    No competing interests declared.
  2. Pranav Mamidanna

    Tübingen AI Center, Eberhard Karls Universität Tübingen, Tübingen, Germany
    Competing interests
    No competing interests declared.
  3. Claudio Michaelis

    Tübingen AI Center, Eberhard Karls Universität Tübingen, Tübingen, Germany
    Competing interests
    No competing interests declared.
  4. Matthias Bethge

    Tübingen AI Center, Eberhard Karls Universität Tübingen, Tübingen, Germany
    Competing interests
    No competing interests declared.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6417-7812
  5. Mackenzie W Mathis

    Brain Mind Institute, École Polytechnique Fédérale de Lausanne, Genève, Switzerland
    For correspondence
    mackenzie@post.harvard.edu
    Competing interests
    Mackenzie W Mathis, Reviewing editor, eLife.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7368-4456
  6. Alexander Mathis

    Brain Mind Institute, École Polytechnique Fédérale de Lausanne, Genève, Switzerland
    For correspondence
    alexander.mathis@epfl.ch
    Competing interests
    No competing interests declared.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3777-2202

Funding

Swiss National Science Foundation (310030_201057)

  • Mackenzie W Mathis

Swiss National Science Foundation (310030_212516)

  • Alexander Mathis

Rowland Institute at Harvard

  • Kai J Sandbrink
  • Mackenzie W Mathis
  • Alexander Mathis

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

Copyright

© 2023, Sandbrink 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

  • 2,104
    views
  • 360
    downloads
  • 9
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

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. Kai J Sandbrink
  2. Pranav Mamidanna
  3. Claudio Michaelis
  4. Matthias Bethge
  5. Mackenzie W Mathis
  6. Alexander Mathis
(2023)
Contrasting action and posture coding with hierarchical deep neural network models of proprioception
eLife 12:e81499.
https://doi.org/10.7554/eLife.81499

Share this article

https://doi.org/10.7554/eLife.81499

Further reading

    1. Biochemistry and Chemical Biology
    2. Computational and Systems Biology
    Shinichi Kawaguchi, Xin Xu ... Toshie Kai
    Research Article

    Protein–protein interactions are fundamental to understanding the molecular functions and regulation of proteins. Despite the availability of extensive databases, many interactions remain uncharacterized due to the labor-intensive nature of experimental validation. In this study, we utilized the AlphaFold2 program to predict interactions among proteins localized in the nuage, a germline-specific non-membrane organelle essential for piRNA biogenesis in Drosophila. We screened 20 nuage proteins for 1:1 interactions and predicted dimer structures. Among these, five represented novel interaction candidates. Three pairs, including Spn-E_Squ, were verified by co-immunoprecipitation. Disruption of the salt bridges at the Spn-E_Squ interface confirmed their functional importance, underscoring the predictive model’s accuracy. We extended our analysis to include interactions between three representative nuage components—Vas, Squ, and Tej—and approximately 430 oogenesis-related proteins. Co-immunoprecipitation verified interactions for three pairs: Mei-W68_Squ, CSN3_Squ, and Pka-C1_Tej. Furthermore, we screened the majority of Drosophila proteins (~12,000) for potential interaction with the Piwi protein, a central player in the piRNA pathway, identifying 164 pairs as potential binding partners. This in silico approach not only efficiently identifies potential interaction partners but also significantly bridges the gap by facilitating the integration of bioinformatics and experimental biology.

    1. Computational and Systems Biology
    2. Neuroscience
    Brian DePasquale, Carlos D Brody, Jonathan W Pillow
    Research Article Updated

    Accumulating evidence to make decisions is a core cognitive function. Previous studies have tended to estimate accumulation using either neural or behavioral data alone. Here, we develop a unified framework for modeling stimulus-driven behavior and multi-neuron activity simultaneously. We applied our method to choices and neural recordings from three rat brain regions—the posterior parietal cortex (PPC), the frontal orienting fields (FOF), and the anterior-dorsal striatum (ADS)—while subjects performed a pulse-based accumulation task. Each region was best described by a distinct accumulation model, which all differed from the model that best described the animal’s choices. FOF activity was consistent with an accumulator where early evidence was favored while the ADS reflected near perfect accumulation. Neural responses within an accumulation framework unveiled a distinct association between each brain region and choice. Choices were better predicted from all regions using a comprehensive, accumulation-based framework and different brain regions were found to differentially reflect choice-related accumulation signals: FOF and ADS both reflected choice but ADS showed more instances of decision vacillation. Previous studies relating neural data to behaviorally inferred accumulation dynamics have implicitly assumed that individual brain regions reflect the whole-animal level accumulator. Our results suggest that different brain regions represent accumulated evidence in dramatically different ways and that accumulation at the whole-animal level may be constructed from a variety of neural-level accumulators.