de.mpg.escidoc.pubman.appbase.FacesBean
English
 
Help Guide Disclaimer Contact us Login
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Conference Paper

Multisensory Perceptual Discrimination in Evolved Networks and Agents

MPS-Authors
http://pubman.mpdl.mpg.de/cone/persons/resource/persons84175

Rohde,  M
Research Group Multisensory Perception and Action, Max Planck Institute for Biological Cybernetics, Max Planck Society;

Locator
There are no locators available
Fulltext (public)
There are no public fulltexts available
Supplementary Material (public)
There is no public supplementary material available
Citation

Rohde, M. (2010). Multisensory Perceptual Discrimination in Evolved Networks and Agents. Artificial Life XII: proceedings of the Twelfth International Conference on the Simulation and Synthesis of Living Systems, 607-614.


Cite as: http://hdl.handle.net/11858/00-001M-0000-0013-BEC2-D
Abstract
The fact that humans and animals have several sensory modalities and use them together to make sense of the world imbues their behaviour with an immense richness and robustness. In this study, recurrent neural networks and minimal agents with active vision are evolved for a perceptual discrimination task (unimodal and bimodal). The purpose of this study is mainly exploratory: to test which of the characteristics of human perceptual discrimination evolve easily (with a focus on statistically optimal integration), how they are realised and what active perception does in this process. Whilst some of the systems evolved to perform perceptual discrimination well, they did not conform to the predictions from statistical optimality. Analyses of the systems point towards a number of relevant issues, noticeably towards the lack of a good account of ‘unimodality’ in existing models of multisensory perception.