Artificial Ethology and Computational Neuroethology: A Scientific Discipline and Its Subset by Sharpening and Extending the Definition of Artificial Intelligence

ARTIFICIAL ETHOLOGY AND COMPUTATIONAL NEUROETHOLOGY: A SCIENTIFIC DISCIPLINE AND ITS SUBSET BY SHARPENING AND EXTENDING THE DEFINITION OF ARTIFICIAL INTELLIGENCE THEODORE B. ACHACOSO and WILLIAM S. YAMAMOTO* As a physician observes the convergence of activities and interests of those concerned with neural network computation and the several facets of artificial intelligence, he may conclude that renaming the specific activities would provide organization and better boundaries, give greater coherence, and embed the activities in traditional and familiar domains. Etymology is a useful road map. Intelligence Almost no arguments arise if "intelligence" is defined in a contextual or "situation-specific" [1] manner. Consider that Bill, a child who hasjust learned to recognize the English alphabet visually, is presented with three playing blocks labeled C, B, and A laid left to right in a row on a table. Bill takes block B and places it immediately to the left of block C, and then takes block A and places it immediately to the left ofblock B, so that the row now reads A, B, C sequentially from left to right. The teacher witnessing this activity concludes that Bill is intelligent. Ted, another child in the same class who is presented with this situation , piles one block on top of the other so that the column of blocks reads as A, B, C from top to bottom. On the other hand, Mary, also in the: same class, upon seeing the blocks, points to each block and correctly recites the verbal equivalent of the label on the block. Certainly, in the context of Bill's performance, Ted and Mary can also be labeled as intelligent. *Department of Computer Medicine, The George Washington University School of Medicine, 2300 K Street, N.W., Washington, D.C. 20037.© 1990 by The University of Chicago. All rights reserved. 003 1-5982/90/3303-0684$01.00 Perspectives in Biology and Medicine, 33, 3 ¦ Spring 1990 \ 379 Although Stoddard in 1941 defined intelligence formally [2, 3], it seems that the formalization is more ofa description ofobservable activities , like the performances of Bill, Ted, and Mary above, rather than a definition of intelligence itself. Other definitions of intelligence include "a quickness of response, a scanning of possible solutions, and the capacity to perceive new relationships between aspects of a problem," and "versatility ofjudgment" [4]. Even the definitions provided by those directly involved in its study are, at best, vague or imprecise and emphasize the ability to reason [I]. More often than not, intelligence is equated with understanding [5]. Although some would disagree, doctors of medicine are generally labeled as an intelligent class of people. The facet ofthis so-called intelligence that is commonly manifested by a physician is his diagnostic logic, that is, the reasoning process by which he arrives at a correct or a défendable diagnosis of a patient's disease. The underlying reasoning process triggers a series of activities like getting a detailed medical history, performing a thorough physical examination, requesting laboratory examinations , requesting specialty consultations and clearances, reviewing pertinent literature, and so on. As these activities are observable to witnesses in the outside world, the witnesses may then conclude that the physician has enough intelligence to be able, through these series of activities, to correctly diagnose the illness ofa patient. Here again, intelligence is defined by observing or inferring from behavior, which leads one to conclude that divorced from the context of behavior, intelligence may never be truly definable. Rather than define intelligence, Chauvin asserts that historically, the concept of intelligence has already been laid aside as inadequate and arbitrary. Intelligence is just a simple category established by Greek philosophers for their own convenience, and one could very well study problem solving, rote learning, manipulatory activity, and so forth without employing this controversial term [6]. Artificial Intelligence If a camera acting as the sensor was hooked to a computer running a program that recognized the initial state of the playing blocks (C, B, and A in a row), and if a robot arm was effecting the movement of the blocks via electronic signals from the same computer with a preprogrammed goal of arranging the blocks in the order of A, B, and C either horizontally or vertically...