Recognition Automata Based on Selective Neural Networks

  • George N. ReekeJr.
  • Gerald M. Edelman


The critical requirement for learning and other higher mental functions is the prior ability to categorize objects and events based on sensory signals reaching the brain. The theory of neuronal group selection provides an explanation for this ability based on a kind of Darwinian selection operating in somatic time on groups of interconnected neurons. These groups are formed during development with varied and overlapping abilities to respond to patterns of environmental stimulation. They are connected in networks to form repertoires of recognizing elements, and these repertoires, many of which are arranged in maps, are further arranged in parallel hierarchies which communicate with each other to carry out categorization with generalization according to attributes of adaptive significance to the organism. A key feature of such systems required to maintain spatiotemporal continuity is reentry of output signals, both within the system and globally, via changes in sensory input resulting from interactions of the system with the environment. These forms of reentry provide a basis for context-dependent figure/ground discrimination and for perceptual invariance to object transformations such as those resulting from motion.

Another key element of the selective paradigm is a form of differential amplification of groups which contribute to responses of adaptive value. In selective neural networks, such groups are selectively modified to enhance the organism’s response to future instances of the same or similar stimuli. This modification consists of enduring changes in the efficacies of the synaptic connections between cells. Rules for these changes based on known biochemical and biophysical properties of neurons have been devised and their behavioral properties studied.

Working computer models of categorizing automata based on these principles have been constructed. Examples are presented to demonstrate their ability to carry out a variety of tasks involving recognition, categorization, generalization, and visual tracking. The computer models give insight into how biological pattern recognizing systems might operate and point the way toward construction of improved recognition automata.


Visual Tracking Selective Amplification Connection Strength Stimulus Object Repertoire Size 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    A.R. Anderson, “Minds and Machines,” Prentice-Hall, Englewood Cliffs, N.J. (1964).Google Scholar
  2. 2.
    D.E. Rumelhart and J.L. McClelland, “Parallel Distributed Processing. Vol. I: Foundations,” MIT Press, Cambridge, Mass. (1986).Google Scholar
  3. 3.
    J.L. McClelland and D.E. Rumelhart, “Parallel Distributed Processing. Vol. II: Psychological and Biological Models,” MIT Press, Cambridge, Mass. (1986).Google Scholar
  4. 4.
    D. Marr, “Vision,” W.H. Freeman & Co., San Francisco, Calif. (1982).Google Scholar
  5. 5.
    E.E. Smith and D.L. Medin, “Categories and Concepts,” Harvard University Press, Cambridge, Mass. (1981).Google Scholar
  6. 6.
    L. Wittgenstein, “Philosophical Investigations,” the English text of the third edition, Macmillan, New York (1958).Google Scholar
  7. 7.
    R. Herrnstein, D. Loveland and C. Cable, Natural Concepts in Pigeons, J. Exp. Psychol. Animal Behav. Processes, 2, 285 (1976).CrossRefGoogle Scholar
  8. 8.
    J. Cerella, Mechanisms of Concept Formation in the Pigeon, in “Analysis of Visual Behavior,” D.J. Ingle, M.A. Goodale and R.J.W. Mansfield, eds., MIT Press, Cambridge, Mass. (1982).Google Scholar
  9. 9.
    K. Lashley, In Search of the Engram, in “Physiological Mechanisms in Animal Behavior” (Society of Experimental Biology Symposium No. 4), Academic Press, New York, p. 454 (1950).Google Scholar
  10. 10.
    G.M. Edelman, Antibody Structure and Molecular Immunology, Science, 180, 830 (1973).PubMedCrossRefGoogle Scholar
  11. 11.
    G.M. Edelman, Group Selection and Phasic Reentrant Signalling: a Theory of Higher Brain Function, in “The Mindful Brain,” G.M. Edelman and V.B. Mountcastle, eds., The MIT Press, Cambridge, Mass. p. 51 (1978).Google Scholar
  12. 12.
    S. Kirkpatrick, C.D. Gelati, Jr., and M.P. Vecchi, Optimization by Simulated Annealing, Science, 220, 671 (1983).PubMedCrossRefGoogle Scholar
  13. 13.
    G.M. Edelman and L.H. Finkel, Neuronal Group Selection in the Cerebral Cortex, in “Dynamical Aspects of Neocortical Function,” G.M. Edelman, W.E. Gall and W.M. Cowan, eds., Wiley, New York, p. 653 (1984).Google Scholar
  14. 14.
    G.M. Edelman, Group Selection as the Basis for Higher Brain Function, in “Organization of the Cerebral Cortex,” F.O. Schmitt, F.G. Worden, G. Adelman and S.G. Dennis, eds., MIT Press, Cambridge, Mass., p. 51 (1981).Google Scholar
  15. 15.
    M. Minsky, A Framework for Representing Knowledge, in “The Psychology of Computer Vision,” P.H. Winston, ed., McGraw-Hill, New York, p. 211 (1975).Google Scholar
  16. 16.
    D.A. Norman and D.G. Bobrow, On the Role of Active Memory Processes in Perception and Cognition, in “The Structure of Human Memory,” C.N. Cofer, ed., Freeman, San Francisco, Calif., p. 114 (1976).Google Scholar
  17. 17.
    R.C. Schank, The Role of Memory in Language Processing, in “The Structure of Human Memory,” C.N. Cofer, ed., Freeman, San Francisco, Calif., p. 162 (1976).Google Scholar
  18. 18.
    G.M. Edelman and G.N. Reeke, Jr., Selective Networks Capable of Representative Transformations, Limited Generalizations, and Associative Memory, Proc. Natl. Acad. Sci. USA, 79, 2091 (1982).PubMedCrossRefGoogle Scholar
  19. 19.
    D.O. Hebb, “The Organization of Behavior,” Wiley, New York (1949).Google Scholar
  20. 20.
    L.H. Finkel and G.M. Edelman, Interaction of Synaptic Modification Rules within Populations of Neurons, Proc. Natl. Acad. Sci. USA, 82, 1291 (1985).PubMedCrossRefGoogle Scholar
  21. 21.
    G.N. Reeke, Jr. and G.M. Edelman, Selective Networks and Recognition Automata, Ann. N.Y. Acad. Sci., 426, 181 (1984).PubMedCrossRefGoogle Scholar
  22. 22.
    T.J. Kellman and E.S. Spelke, Perception of Partly Occluded Objects in Infancy, Cognitive Psych., 15, 483 (1983).CrossRefGoogle Scholar
  23. 23.
    O. Selfridge, Pandemonium, a Paradigm for Learning, in “Mechanisation of Thought Processes, I, Natl. Phys. Lab. Symp. No. 10,” H.M. Stationery Office, London (1959).Google Scholar

Copyright information

© Plenum Press, New York 1986

Authors and Affiliations

  • George N. ReekeJr.
    • 1
  • Gerald M. Edelman
    • 1
  1. 1.The Rockefeller UniversityNew YorkUSA

Personalised recommendations