Design of a Feeling-Thinking Machine
Abstract
Machine intelligence for autonomous systems must be capable of learning and especially thinking, if we are to go beyond the 'islands of autonomy' presently envisioned for teleoperated and remotely piloted vehicles. One approach is to use the mammalian brain as a model and investigate the possibility of duplicating its functions in electronic circuitry. This extension of neural network design is called non-living intelligence (NLI). The brain consists of approximately 10**12 neurons intricately interconnected. Only a small part of this circuitry has been unravelled. The NLI effort at Benet Laboratories does not describe how the brain works, but involves electronic and computational experiments that provide insight into how the brain might work. We pursue NLI through the design and construction of feeling-thinking machines. Feeling is essential because without motivation, there is nothing. The machine must want to do things. In doing things, it will learn; and having learned, it will think. This report does not describe machines that exhibit intelligent behavior; but rather machines that feel, want, and think. The distant goal is to create a machine that thinks and acts like a man. This report discusses the first of a series of feeling-thinking machine designs.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1988
- Accession Number
- ADA194107
Entities
People
- Mark A. Johnson
- Raymond Scanlon
Organizations
- United States Army Armament Research, Development and Engineering Center