The society of mind: Marvin Minsky

✍ Scribed by Matthew Ginsberg

• Publisher: Elsevier Science
• Year: 1991
• Tongue: English
• Weight: 240 KB
• Volume: 48
• Category: Article
• ISSN: 0004-3702
• DOI: 10.1016/0004-3702(91)90033-g

✦ Synopsis

Artificial intelligence is fundamentally an engineering discipline; our collective goal of constructing an intelligent artifact is fundamentally an engineering one. Good engineering builds on good science. Good science, quite frequently, builds on good mathematics.

The Society of Mind does not pretend to be good engineering or good mathematics. My aim in this review is to discuss the question of whether or not it is good science.

I will do this by examining two separate but related issues. The first involves Minsky's abandonment of the usual methods of scientific inquiry and a discussion of the problems to which this leads; the second is an analysis of Minsky's criticisms of formal logic and its role in AI.

1. Falsifiable claims

Consider Newton's observation that the acceleration imparted to objects in a gravitational field is independent of their mass. If I can find two objects for which this rule fails, Newton is wrong and his theory needs to be revised. I do not need to seek him out and discuss the matter; he has no choice. The theory of gravitation allows us to make precise claims about observations to be made in the future, and we evaluate the theory by testing the validity of these claims.

In The Society of Mind, Minsky makes no such precise claims (about gravity or anything else). Instead, he builds up a loose framework in which the behavior of a single conscious entity is analyzed in terms of the collective * (Simon & Schuster, New York, 1986); 339 pages, $10.95 (paperback).