X-Message-Number: 7991 From: (Thomas Donaldson) Subject: to.Mr.Clark Date: Tue, 1 Apr 1997 22:47:33 -0800 (PST) Hi again! For Mr. Clark: Some comments on your answers. First, it's not an issue of biological versus "something else". It is a simple fact that neurons are far more complex than any chip yet built, and not only that, they have some abilities at growth and self repair which are essential to memory itself. To really emulate even a single neuron in a computer would take a computer of great power and (I suggest) a large and extensible memory. And that's just one neuron, and I mean here specifically those behaviors of the neuron which bear on our behavior and affect other neurons. If we replace our neurons all with another set of devices (not computers) which did match their behavior, then the chaos problem from a computer simulation would not arise, because we were not EMULATING neurons, we were actually providing substitute neurons, as working devices rather than programs. My problem with emulation is the same as the problem a meteorologist has with predicting the weather --- or a geologist with predicting earthquakes. An object simulated inside a computer will inevitably lack may inputs (and particularly inputs of which we are still unaware) than the actual object responding in the world. And in a previous message, I went on to point out that not only would this computer emulation fail to follow your behavior, but that it would quite drastically fail --- something which physical devices substituting for neurons would not do. As for the Tortoise, yes, if thinking is to be identified with computing then we can at best rarely think. Why didn't you answer my question about this? I did not interpret the Tortoise as saying nothing could reason, but as drawing a distinction between thinking and computation. You also (so far as you do) claim that intelligent behavior implies consciousness, thus begging a number of questions. Is the behavior of the computer in the Turing Test intelligent? Again, just what is to be intelligent behavior in the first place? Is the ability to play chess, for instance, a sign of intelligent behavior? If you identify the Turing Test with intelligent behavior, as a DEFINITION, what about chess or other activities which can be played in total silence? Doing mathematics? Even driving an automobile. None of these are signs of intelligent behavior? Surely you have in mind a more general definition than just the Turing test. As you should have worked out by now, I am doubting the Turing Test, not just as a test of consciousness but as a test of intelligent behavior. And I even explained a test I would take more seriously: the computer or person would have to not only converse with me but also interact with the world separately from me. And as you recall, I am not referring here to the Chinese room --- though the Chinese room story does raise questions. Searle used Chinese ideograms as a way of indicating, in the story, that the ability to play games with tokens does not automatically prove intelligence. As for "information", we may have run into different definitions. But for what I wished to say, call it what you wish, there is an important difference between the world and any symbolic version we may construct of it or part of it, and a difference (in terms of how we work) between feelings and statements of fact. Certainly someone else might describe our thoughts and feelings on film or verbally, and that description would certainly be information. But our thoughts and feelings are not the same as this person's description, and differ in how they affect us. No amount of verbal description can be identified with what it describes, no matter how accurate that description may be. IF we define "information" as a symbolic description of something, then no amount of information about an object can be identified with the object itself. (Sure, in cryonics terms it may allow us to recreate the object, but that re-creation is then different from the information which described it). Finally, I will point out that my test could conceivably be passed by something which was not human, even something which we had built ourselves. It would be something which did NOT just play with symbols but interacted with the fuzzy, sometimes even bewildering, universe, with no help from me. Whether or not it could also compute better than me is irrelevant here --- it could at least show intelligent behavior in many fields rather than just one very restricted field. Long long life, Thomas Donaldson Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=7991