Computational Templates, Neural Network Dynamics, and Symbolic Logic

— The relationship between subsymbolic neural networks and symbolic logical systems is discussed from the point of view of the account of computational science by Paul Humphreys [Humphreys, Paul (2004), Extending Ourselves. Oxford: Oxford University Press.] This philosophical account of the organization of scientific knowledge considers the units of analysis constituting scientific knowledge to be computational models, and computational templates. ional templates are more abstract computational schemata underlying specific models. This view is contrasted with the received views where concepts, theories or laws constitute the units of analysis. Humphreys’examples are mainly from physics, biology and statistics, but neural networks can also be considered computational models/templates in Humphreys’ sense. Implications regarding psychoneural reduction vs. emergence, and the symbolic-subsymbolic debate are also discussed.

• Type: Conference
• Year: 2007
• Where: IJCNN
• Authors: Otto Lappi
• Keywords: Abstract Computational Schemata, IJCNN 2007, Neural Networks, Subsymbolic Neural Networks