A Mathematical and Physical Base for "A Standard Model of the Mind"

Abstract

This paper describes a mathematical and physical base for 'A Standard Model of the Mind'. It is a super-Turing model constrained by physical reality of the brain's construction. The constraints are noise and quantized charge transfer. The model has computing power beyond the Universal Turing Machine (UTM), which Turing himself claimed to be insufficient to model the brain. The super-Turing model meets Turing's desire for a more complex model. Allen Newell expressed difficulties in modelling the brain with the UTM not being as computationally complex as his functional analysis indicated. We will describe the model and note neuron operations compatible with it. Because of mathematical idealities in both Turing and super-Turing models, physical devices cannot directly implement either model. However, both Turing machines and super-Turing models can point the direction to the design and operation of physical devices. Brain modeling should use the more powerful super-Turing model to describe its operation. Our research seeks to develop artificial neural networks based on this model. The super-Turing model is guiding analog- and digital-hybrid hardware development of these neural networks. We will describe the progress on an optical implementation with encouraging, chaos-mimicking results and on designs for an electronic implementation. We describe a possible spectrum of super-Turing inspired devices. We will call on the community to help further the devices and their use in 'A Standard Model of the Mind'.

Department(s)

Physics, Astronomy, and Materials Science

Document Type

Conference Proceeding

Additional Information

Reprinted in Common Model of Cognition Bulletin 1, no. 2 (2020) pages 431-436, as part of a collection AAAI symposium papers that resulted in the concept of the Common Model of Cognition. Also referenced as AAAI Technical Report FS-17-05.

Keywords

standard model; super-Turing, analog, mixed-signal, physical, neural network

Publication Date

1-1-2017

Journal Title

AAAI Fall Symposium - Technical Report

Citation-only

Share

COinS