Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!munnari.oz.au!cs.mu.oz.au!ok From: ok@cs.mu.oz.au (Richard O'Keefe) Newsgroups: comp.ai Subject: Re: Discovering What Nature Wants Summary: Let's be clear about things Keywords: Consciousness and Intentionality. Message-ID: <2450@munnari.oz.au> Date: 17 Oct 89 11:25:35 GMT References: <357@massey.ac.nz> <2376@munnari.oz.au> <2394@uceng.UC.EDU> <1087@oravax.UUCP> Sender: news@cs.mu.oz.au Lines: 194 In article <1087@oravax.UUCP>, daryl@oravax.UUCP (Steven Daryl McCullough) writes: > In article <2433@munnari.oz.au>, ok@cs.mu.oz.au (Richard O'Keefe) writes: > > X has property P > > X is part of Y > > ---------------- > > Y has property P > > Let me call this principle "Kort's rule". > > So, far, this is a nice analysis of Barry Kort's argument. Your point, which > is well taken, is that this is not a valid rule of inference, in general. > Then you go on to give a million and one examples, enough to make it obvious > to someone of the meanest of intelligence that it is not a valid rule of > inference. Was this necessary? Yes it was, for the simple reason that Kort stated explitly that he didn't follow my reasoning. > Obviously Barry believes that "having > intentions" is a property such that Kort's rule applies. He is not claiming > that it always applies. His argument was not valid unless it was _assumed_ that Kort's rule applies in _this_ case. He presented no argument to show that it _does_ apply in this case. Many of my examples were directled at showing that the rule fails in cases which are obviously similar. > I think that "ability to form intentions" is a property such that, if a > part of an object has it, then the whole object has it. You may believe as you wish. If you want _me_ to believe the same thing (and I concede that it _may_ be true) you must present good reasons. > Common usage is > consistent with this: people often say "this committee intends to ...", > or "Congress intends to ...". It is not necessarily the case that *every* > intention held by the part is held by the whole, but it is surely possible > that *some* intentions are held by the whole are inherited from the parts. I'm afraid we can't appeal to common usage. Common usage "the sun rose" is pre-Copernican with respect to physics. A survey referred to in Discover found that 37% of the people interviewed believed that milk contanimated with radio-active material could be made safe by boiling it; no doubt their language would reflect this. Another point is that the usage is not common: in English (as opposed to American) one says "this committee INTEND to ..." or "Parliament INTEND to ...": the noun being taken as plural, just as one says "the University ARE going to cut funding for the Percy Grainger Museum by 20%". This is not a ``my language is better than your language'' point. The point is that common usage is an uncertain guide because usage _varies_. Since a committee collectively form a language-using spatio-temporally bounded individual, it would not challenge my position even if I accepted this example at face value. > More generally, it is often the case with abilities that, if the part has > a certain ability then the whole has. For example, I conclude that > "Richard O'Keefe can think" from the fact that "Richard O'Keefe's brain > can think". I do believe that I can think, but I do not believe that my brain can think. I don't believe that my brain CAN'T think, but saying that a brain can think sounds a lot like saying that a finger can feel. A simpler example: "being able to cook" is not a property of an element, but a property of the whole stove. It is necessary that the element be held level, that it be connected to electrical power by cables, &c &c. Thinking requires more than a brain; it requires a history. > Or I conclude "Elton John can play the piano" from the fact > that "Elton John's brain and fingers, working together, can play the piano". Elton John's brain and fingers can't play one note without his arms, shoulders, back muscles, spinal cord, skin, ears, ... (and I'm just listing parts which are directly involved in the act of playing). Piano-playing is not a property of individual parts. We can tell whether Elton John can play the piano by sitting him down in front of it and asking nicely. How do we test his brain-and-fingers on their own? Does it make sense to imagine his fingers waving in the air without any other part of Elton John around, playing a piano? No? Then sit his brain on the piano stool, underneath those waving fingers; can that ensemble play the piano? Does it really make sense to say that "Elton John's brain and fingers can play the piano"? No, it takes a more-or-less intact human being (prosthesis being included in "more or less intact"). In terms of what really happens, it is simply FALSE that anyone has yet concluded "that can think" from the fact that "'s brain can think" or "that can play the piano" from the fact that "'s brain and fingers can play the piano". In both cases, the inference goes the other way. We see people behaving in a way that we interpret as thinking, and believing that thinking takes place in the brain, infer that their brains are thinking. We see people playing the piano, and infer that their brains know how to play the piano. (Are you sure that none of the learning required for piano playing involves the spinal cord?) Basically, the ``Nature''-can-form-intentions-because-we-can viewpoint is just reductionism run backwards. > It is a perfectly legitimate question to ask "what does it *mean* for > nature to form intentions". Yes, and I wish someone would try to answer it. > I think that the question is just as relevent > (and its answer is no clearer) to human beings as to nature. Of course the question is relevant to human beings. But I'm not asking for any mysterious essence of things; I'm asking for a crude rule of thumb which will help me detect actual instances of human beings intending something and nature intending something. In the case of humans, we have a pretty good guide: if someone says they have an intention or a desire, that will usually do. I'm not asking for the Ultimate Truth about this, just something which works as well as that. > Usually > people are satisfied with something like "Entity E has intentions if it > shows signs of behaving in a goal-directed manner". That would give a thermostat intentions. It would elevate every equilibrium principle in physics and chemistry to the status of "proof of intentions". > However, something like this is the reason we ascribe intentions to > human beings. Back to front again. We ascribe intentions to human beings because we ARE human beings and have a particular kind of experience and project that experience onto other human beings in the belief that they are similar to us. People often worry about precisely the intentions of other people which the others have given no evidence of yet. > Given any definition you like of "goal-directed", it seems that nature > acting in a "goal-directed fashion" follows from human beings acting in > such a fashion. Nature accomplishes something using a part of itself, > namely human beings, as tools, in the same way a human being might use > parts of himself: his brain, his hands, his tongue. With any definition of "goal-directed" that does not take into account the notions of the agent having a model of an actual state of affairs, a model of a hypothetical state of affairs, some way of distinguishing between the agent's ATTITUDES and the agent's ACTIONS (so that we can talk about intentions an agent forms but does not carry out and actions an agent performs without having intended them), with any such limited definition of "goal-directed" we can talk about stones intending to fall down. But to say that 'nature acting in a "goal-directed fashion" follows from human beings act in such a fashion' is merely to reASSERT Kort's rule. If human beings act in a goal-directed fashion, then all that follows is that some PARTS of ``Nature'' act in a goal-directed fashion. We cannot conclude that ``Nature'' herself acts in a goal-directed fashion unless P="acts in a goal-directed fashion" obeys Kort's rule. Merely asserting that it does gets us not one step further. I have no difficulty with the idea that things happen in such a way that some measure is reduced (i.e. that ``Nature'' may be said in some sufficiently irrelevant way to ``act'' in a ``goal-directed'' way). But that doesn't get us one step further towards ``Nature'' having INTENTIONS. It is not enough that ``Nature'' should be heading towards the Heat Death of The Universe, just because that's happening doesn't mean ``Nature'' WANTS (intends) it to happen. One more example: Algol 68. That was a production of the Algol 68 committee. So if we are willing to accept committees as systems possessing intentions, the Algol 68 committee intended the language to be the way it was. But several of the members of the committee were so dissatisfied with it that they issued a Minority Report. Here we are a clear actual case where some of the parts (the members who signed the Minority Report) had an intention (that the language be other than it was) which the whole did NOT have. This is a proof by existence that even if a whole and its parts both have intentions, the intentions that the whole has may not be the same as the intentions of its parts. This tells us, for example, that simply because biological individuals behave as if they wished to live and wished their own species to continue in existence, it does not follow that ``Nature'' wishes either of these things. > I won't argue that this point of view is correct, or even that I believe > it, but I will argue that it isn't stupid. I am not arguing that it is stupid, but that it needs to be supported by argument, not by assertion. > Daryl McCullough > (I have a Master's Degree...in Science!) So have I. There's a lot of it about.