Understanding biological computation: reliable learning and recognition.
- 1 November 1984
- journal article
- research article
- Published by Proceedings of the National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 81 (21) , 6871-6875
- https://doi.org/10.1073/pnas.81.21.6871
Abstract
We experimentally examine the consequences of the hypothesis that the brain operates reliably, even though individual components may intermittently fail, by computing with dynamical attractors. Specifically, such a mechanism exploits dynamic collective behavior of a system with attractive fixed points in its phase space. In contrast to the usual methods of reliable computation involving a large number of redundant elements, this technique of self-repair only requires collective computation with a few units, and it is amenable to quantitative investigation. Experiments on parallel computing arrays show that this mechanism leads naturally to rapid self-repair, adaptation to the environment, recognition and discrimination of fuzzy inputs, and conditional learning, properties that are commonly associated with biological computation.This publication has 2 references indexed in Scilit:
- Pigeon Perception of Letters of the AlphabetScience, 1982
- Neural networks and physical systems with emergent collective computational abilities.Proceedings of the National Academy of Sciences, 1982