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

Evolving neural models of path integration.

R J Vickerstaff1, E A Di Paolo

  • 1Centre for Computational Neuroscience and Robotics, School of Life Sciences, University of Sussex, Brighton, BN1 9QG, UK. robertvi@sussex.ac.uk

The Journal of Experimental Biology
|August 20, 2005
PubMed
Summary

Researchers evolved neural networks to model ant navigation, successfully replicating desert ant (Cataglyphis fortis) homing behavior and search patterns using a genetic algorithm.

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

Environmental regulation can arise under minimal assumptions.

Journal of theoretical biology·2008
Same author

Rhythmic and non-rhythmic attractors in asynchronous random Boolean networks.

Bio Systems·2001
Same author

Artificial life: discipline or method? Report on a debate held at ECAL '99.

Artificial life·2000
Same author

Ecological symmetry breaking can favour the evolution of altruism in an action-response game.

Journal of theoretical biology·2000

Area of Science:

  • Computational neuroscience
  • Animal behavior
  • Artificial intelligence

Background:

  • Path integration is crucial for animal navigation, enabling animals to track their position relative to a starting point.
  • The desert ant Cataglyphis fortis exhibits remarkable homing abilities, making it a model organism for studying navigation.

Purpose of the Study:

  • To evolve neural network models of path integration without pre-defined internal representations.
  • To reproduce the homing and searching behaviors of Cataglyphis fortis ants.

Main Methods:

  • Utilized a genetic algorithm to evolve neural network models.
  • Integrated models within a complete system representing animal movement and environment.
  • Analyzed evolved networks for resemblance to existing navigation models.

Main Results:

  • Successfully evolved a neural network capable of path integration and homing behavior reproduction.
  • The best evolved network showed similarities to the bicomponent model and incorporated leaky integration.
  • The model naturally replicated systematic navigation errors and searching behavior observed in desert ants.

Conclusions:

  • Neural networks can learn path integration without explicit internal vector representations.
  • Leaky integration and cosine-shaped compass responses may play key roles in ant navigation.
  • The evolved model provides insights into the mechanisms underlying insect navigation and homing.

Related Experiment Videos