Parallel distributed processing

from Wikipedia, the free encyclopedia

Parallel Distributed Processing (short: PDP ) is a u. a. Theory applied in artificial neural networks in the further development of connectionism (see: Cognitive Psychology and Neuroinformatics ).

It is a theoretical approach to information processing in neural networks, which particularly considers the activation patterns that exist across the networked neural elements.

In the scientific discourse on research in perceptual psychology , James L. McClelland and David E. Rumelhart proposed a new type of model for the neurological explanation of visual pattern recognition (in this case: letter recognition) as early as 1981. To this end, they designed a network model based on the principle of combining stimulus and context information in visual processing. The behavior of this network depends heavily on activation propagation mechanisms, a distinction being made between excitatory and inhibitory processes.

Massaro (1989) criticizes the PDP model for reacting too insensitively to the effects of the stimulus information when these run counter to the context information.

See also


  • David E. Rumelhart, James L. McClelland: Parallel Distributed Processing. Explorations in the Microstructure of Cognition. 2 volumes. MIT Press, Cambridge MA et al. 1986, ISBN 0-262-18123-1 .
  • Helge Ritter , Thomas Martinetz , Klaus Schulten: Neural Networks. An introduction to the neuroinformatics of self-organizing networks. 2nd revised edition, 2nd unaltered reprint. Addison-Wesley, Bonn et al. 1994, ISBN 3-89319-131-3 .
  • John R. Anderson : Cognitive Psychology. 3rd revised and updated edition. Spectrum - Akademischer Verlag, Heidelberg et al. 2001, ISBN 3-8274-1024-X .
  • Robert L. Solso: Cognitive Psychology. Springer, Heidelberg 2005, ISBN 3-540-21270-1