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

Multiagent cooperation and competition with deep reinforcement learning.

Ardi Tampuu1, Tambet Matiisen1, Dorian Kodelja1

  • 1Computational Neuroscience Lab, Institute of Computer Science, University of Tartu, Tartu, Estonia.

Plos One
|April 6, 2017
PubMed
Summary
This summary is machine-generated.

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

An effort recalibration framework for digital media use and cognition.

Nature human behaviour·2026
Same author

Electrophysiological mechanisms of psychedelic drugs: A systematic review.

Neuroscience and biobehavioral reviews·2026
Same author

Effects of LSD, DMT and psilocybin on cognitive and psychological functions: A systematic review of the literature.

Journal of psychopharmacology (Oxford, England)·2026
Same author

Oxidative stress-driven transcriptomic remodeling in human astrocytes reveals network signatures associated with neurodegenerative and cardiovascular processes.

Computational and structural biotechnology journal·2026
Same author

On biological and artificial consciousness: A case for biological computationalism.

Neuroscience and biobehavioral reviews·2025
Same author

Comparing a computational model of visual problem solving with human vision on a difficult vision task.

PLoS computational biology·2025
Same journal

Analysis of strength degradation of coal and rock masses and stability of mined areas under long term immersion environment.

PloS one·2026
Same journal

Biogenic Silver-Selenium nanocomposite with anticancer activity and potent efficacy against vancomycin-resistant Staphylococcus aureus.

PloS one·2026
Same journal

Preparation and physicochemical characterization of a biodegradable chitosan/carboxymethyl cellulose hydrogel synthesized in NaOH/urea medium.

PloS one·2026
Same journal

Action-guilt, survivor-guilt, and depression in combat-related PTSD.

PloS one·2026
Same journal

Explainable machine learning for predicting activities of daily living at discharge in stroke patients: A retrospective study using SHAP interpretability.

PloS one·2026
Same journal

Deep learning based two-way feature depiction model for brain tumor detection.

PloS one·2026
See all related articles

This study shows how artificial agents using Deep Q-Networks learn to cooperate or compete in the game Pong. Increasing rewards for cooperation shifts behavior from competitive to collaborative, creating robust strategies.

Area of Science:

  • Artificial Intelligence
  • Game Theory
  • Multi-agent Systems

Background:

  • Cooperation and competition are key evolutionary dynamics in shared environments.
  • Autonomous agents learning via reinforcement present opportunities to study emergent behaviors.

Purpose of the Study:

  • To investigate the emergence of cooperation and competition in multi-agent systems.
  • To explore how reinforcement learning agents adapt their strategies based on reward structures.

Main Methods:

  • Utilized the Deep Q-Learning framework extended to multi-agent settings.
  • Trained two autonomous agents using raw visual input in the game Pong.
  • Manipulated the reward scheme to influence agent behavior.

Main Results:

Related Experiment Videos

  • Demonstrated the emergence of both competitive and collaborative behaviors by altering Pong's reward system.
  • Observed a transition from competitive to collaborative strategies as cooperation incentives increased.
  • Showcased that learning against adaptive agents yields more robust strategies than against fixed algorithms.

Conclusions:

  • Deep Q-Networks are effective tools for studying decentralized learning in high-dimensional multi-agent systems.
  • Reward manipulation can guide the evolution of cooperation and competition in artificial agents.
  • Adaptive opponents foster more resilient and sophisticated agent strategies.