Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A more biologically plausible learning rule for neural networks.

P Mazzoni1, R A Andersen, M I Jordan

  • 1Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge 02139.

Proceedings of the National Academy of Sciences of the United States of America
|May 15, 1991
PubMed
Summary

This study introduces a biologically plausible reinforcement learning rule for artificial neural networks, demonstrating it can model brain functions like visual spatial representation in monkeys without unbiological back-propagation.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Quantifying physical degradation alongside recording and stimulation performance of 980 intracortical microelectrodes chronically implanted in three humans for 956-2246 days.

medRxiv : the preprint server for health sciences·2024
Same author

Neural subspaces of imagined movements in parietal cortex remain stable over several years in humans.

Journal of neural engineering·2024
Same author

Neural subspaces of imagined movements in parietal cortex remain stable over several years in humans.

bioRxiv : the preprint server for biology·2023
Same author

Bis-cyclooctatetraenyl Thulium(II): Highly Reducing Lanthanide Sandwich Single Molecule Magnets.

Angewandte Chemie (International ed. in English)·2022
Same author

A shared neural substrate for action verbs and observed actions in human posterior parietal cortex.

Science advances·2020
Same author

Machine learning: Trends, perspectives, and prospects.

Science (New York, N.Y.)·2015

Area of Science:

  • Computational neuroscience
  • Artificial intelligence
  • Neurobiology

Background:

  • Artificial neural networks (ANNs) are increasingly used to model brain information processing.
  • Current ANN training methods, like back-propagation, lack biological plausibility.
  • Understanding neural computation requires biologically realistic models.

Purpose of the Study:

  • To develop and test a biologically plausible learning rule for ANNs.
  • To model the representation of visual space in head-centered coordinates in area 7a of the posterior parietal cortex.
  • To investigate if biologically plausible learning can yield meaningful neural network behaviors.

Main Methods:

  • Implemented a reinforcement learning rule as an alternative to back-propagation for training ANNs.

Related Experiment Videos

  • Applied the reinforcement learning network to model visual spatial representation in area 7a.
  • Compared the network's behavior to ANNs trained with back-propagation and to actual neural recordings from area 7a.

    • Main Results:

    • The reinforcement learning network successfully modeled visual spatial representation.
    • The network's behavior mirrored that of back-propagation trained networks.
    • The network's performance was comparable to recorded neural activity in area 7a.

    Conclusions:

    • Biologically plausible learning rules, such as reinforcement learning, can train ANNs effectively.
    • Artificial neural networks do not require back-propagation to exhibit biologically relevant properties.
    • This work offers a more biologically realistic approach to modeling neural information processing.