Posted by R. John Howe on January 29, 1999 at 12:33:10:
In Reply to: Re: A comment posted by Yon Bard on January 25, 1999 at 13:53:39:
I would like to comment, albeit, belatedly on something Yon Bard has said here in response to Dr. Salingaros' claim that human aesthetic are at bottom a set of discoverable rules, perhaps hardwired in our brains/minds. Yon made a familiar objection. He said in part:
"Your comments are dotted with terms such as 'correct judgment,' Of course the trouble is that there is no such thing. What is correct for one person at one time or place is incorrect for another person, time, or place.
: While neural networks or other expert systems may, with sufficient effort, be trained to emulate certain aspects of human judgment, there is little evidence that this actually represents how the mind works (I have spent my career in computer science and have dabbled in expert systems and the like).
: I would like to ask one question: Does anybody seriously believes that the set of rules in our checklist, or anything at all like them, can capture the esthetic perceptions of the human mind?"
My own experience is similar to Yon's in some ways but in others quite different. I build custom job training for several groups of professional employees, including white collar investigators and some species of medical insurance claims examiners. The central move we make in this work (I have been doing it now for about 35 years) is to discover the "rules" followed by expert practitioners in these jobs and to build experiental learning exercises that make it possible for most of a much larger target population to perform at levels approaching that of the expert practitioners.
From time to time we hear that expert systems have been or are being developed that can take over the diagnosic process we go through with the expert practitioners to discover the rules they follow, to bring them to self-conscious and to formulate, with them, the best statements of these rules. I approach such expert systems with a mixture of fear and potential awe, and after examining them I always so far have left them with feelings of disappointment and relief. They do not it seems to me yet pose much opportunity or threat in the area of diagnosis. My particular criicism of them is that "intelligence" always still has to be insert from outside the expert system. The ones I have seen can use the rules they are given but have no real ability to recognize new rules reliably. And they appear to be utterly blind to the question of what is the best statement of a given rule. So that far I want to go with Yon. Our experiences are I think fairly congruent.
I would also agree with Yon that I think it more likely that the rules that govern human perception are relative, learned and variable rather than absolute and commonly held capabilities as it seems to me Mr. Alexander and Dr. Salingaros claim.
But I have become very impressed with the character of rules themselves and do think that it is like that most human thought processes can in principle not only reduced to them but that human acts like recognition of a beautiful rug are very likely comprised of nothing else. On this point I find myself with Dr. Salingaros. I would be interested to hear someone describe what they believe the components of human recognition of a beautiful rug are that move beyond instances of rule-following.
R. John Howe
Post a Followup