This version has typo corrected after posting. Posted Sun Aug 11 23:22:12 BST 1996 Newsgroups: comp.ai.philosophy Distribution: world References: <4s9m1c$1304@useneta1.news.prodigy.com> <4sj1id$23go@useneta1.news.prodigy.com> <4sooae$125a@useneta1.news.prodigy.com> <4t2pr7$11bu@useneta1.news.prodigy.com> <4t5ke8$1kdk@useneta1.news.prodigy.com> <4tal5u$1pc0@useneta1.news.prodigy.com> <4tj9pm$1pic@useneta1.news.prodigy.com> <4tmhrq$h2a@useneta1.news.prodigy.com> <4tpkft$1teg@useneta1.news.prodigy.com> <4tqnmb$159g@useneta1.news.prodigy.com> <4trd98$1kcu@useneta1.news.prodigy.com> <4tvq06$mjo@useneta1.news.prodigy.com> Subject: Re: Free Will Lewis Jorgenson (LSVL06A@prodigy.com), whose style and wit often liven up this group, does not like some of what I wrote, and seems to find it threatening. > Date: 3 Aug 1996 15:07:50 GMT > > In the thread 'Note on design stance (Re: Sloman (Part 5))' > A.Sloman@cs.bham.ac.uk wrote in part: > >> This is usually based on the notion that understanding is an >> all-or-nothing matter, and since the machine and its subsystems do >> not have full human-like understanding of the internal structures, >> they therefore have no understanding, and no intentional states. > Jorgenson: > What internal structures are you talking about? I can't recall the context, but was probably thinking of information structures in a virtual machine. E.g. o information that a parser or compiler might build up about the structure of a sentence or program, o information in an operating system about user processes, o information in an internet network node about states of the network and routes to use for transmitting messages, o information in a planner about subgoals still not yet satisfied, o information in a vision system about the structure of an image or about possible interpretations of a part of an image and many more. [LJ] > ..Is a concept > or qualia to be thought of having location and shape like a > pigeon? Not necessarily: structures in a virtual machine need not have properties and relations like those of a physical object. E.g. in a list processing virtual machine, List A can be a part of list B while B is simultaneously a part of A. But if A and B are physical objects (e.g. pigeons) then it is not possible for each to be part of the other. (This is one of the reasons why the concept of a *physical* symbol system introduced by Newell and Simon is misplaced. The symbol systems required by AI (and software engineering) do not need to be physical, though they may be *implemented* in physical structures. So physical analogies can often lead us astray when we think about information systems. Likewise pigeon analogies.) [LJ] > Do you actually think of the brain as a pigeon roost > where some pigeons fly in and out and also raise their young? So > a concept or perception flies in and couples with another concept > or perception and a new concept or perception is born which can > fly out. I have no idea why you bring in pigeons, flying in and out, breeding etc. There are lots of examples of what I was talking about in text books on AI (including vision and robotics), compilers, operating systems, networks, etc. I don't see any need for something like your pigeon metaphor as there are already far more precise ways of characterising the design and behaviour of systems that acquire, manipulate and use information. There's also some work on how such things happen in animal brains, but my impression is that at present only a tiny subset is understood by scientists. [LJ] > Directed thinking is a breeding exercise where certain > pigeons force other certain pigeons to mate and then have a baby > pigeon that reflects certain attributes of the parents. > Understanding is when one or a specific group of pigeon holes have > pigeons living there, perhaps poking their little heads out to > look at the other pigeons. You may find such metaphors useful. I don't. I don't know why you use them to summarise what you think I wrote. If that's how you think about computation, then maybe you need a course on computer science, compiler design, operating systems, etc.? [LJ] > Then emotions are pigeons too. A fear pigeon and a loud noise > pigeon and a jerk pigeon are mated when a human infant is born > and so we have the startle reflex. > .... [etc. etc. etc.]..... I don't find that pigeon metaphor contributes anything useful to our understanding of emotions. There's a great deal written about emotions that we share with other animals (being startled, terrified, tensely expectant etc.) and which bits of the brain are involved (the limbic system?). My own work on emotions (including work done by my research students) has mainly been concerned with states and processes that we do not (as far as I know) share with other animals, e.g. being publicly humiliated, excited anticipation of an event happening next week, long term debilitating grief, etc. I am interested in questions about the sorts of information processing architectures that make different kinds of emotions possible. See the papers in ftp://ftp.cs.bham.ac.uk/pub/groups/cog_affect/ Your pigeon metaphors don't seem to have the required explanatory richness. [AS] >> The answer to this is to examine patiently all the different sorts >> of understanding that there are, in order to see what they have in >> common and how they differ, and how they fit into a larger >> framework of varieties of intentionality, e.g. some of which do >> not require a social system with public language, etc. > > Jorgenson: > I should think it would be easier to start with all the different > sorts of misunderstandings that there are. What's easier is not necessarily more relevant. But I agree that types of misunderstanding would need to be explained by a comprehensive theory: a system that is incapable of understanding anything is incapable of misunderstanding anything, and vice versa: misunderstanding is a type of understanding where some sort of mistake has been made or wrong conclusion drawn. [LJ] > ..For instance, why > must the concept of free will be jettisoned in order for AI > research to continue? Who said it has to be jettisoned? I certainly did not. (It may be that beginners who are not well versed in philosophy think that AI has that consequence.) I would argue that the only way we have to make sense of what it is to act freely is in terms of doing things because of what you want, believe, prefer, choose, etc. Currently I don't know of any good way to explain how that is possible EXCEPT in the framework of an information processing mechanism. I.e. AI, done properly, explains how humans are capable of being free in a way that is not possible for clocks, rocks, clouds, or the wordprocessor in your computer. (A huge amount has been written on freedom. For people who have not read much philosophy a good starting point is Dan Dennett's little book: Elbow Room.) If you want free will to be essentially mysterious inexplicable and unrelated to any explanatory mechanism then I can't help you. I don't see any use for such a notion of freedom. (Except perhaps in theological debates that attempt to absolve an omniscient and all powerful god of responsibility for horrible diseases, earthquakes, wars, torture, etc. by blaming them on some kind of free choices of devils or men. But I have no interest in trying to save anyone's god from blame.) [LJ] > Why do some choose to hand over their fate? I don't recognize this as relevant to the discussion of AI. To whom or what would you hand over your fate? People in some sense partly hand over their fate when they vote for politicians, when they join clubs or societies that dictate their behaviour, when they try parachute jumping, when they experiment with dangerous drugs, when they choose marriage partners, when they decide to have children, when they choose a university at which to study. All these are part of everyday life: total control of your own destiny is never possible, so why get worried about it? In any case, people don't necessarily do anything like "handing over their fate" when they produce a philosophical theory explaining in AI terms or any other terms how our minds work. (Especially if they do it with an open mind, being prepared to pay attention to counter evidence, for example.) In fact it is possible that such a theory helps us understand (a) how things can go wrong (e.g. how you can become obsessed, how your vision of what is possible can become constrained, how your concepts may not develop fully) and (b) how you can, in some cases, remedy such things. If so, it may thereby give you (or your counsellors, therapists, teachers and friends) the opportunity to extend your freedom. [AS] >> But there is enormously strong resistance to work which requires >> deep revision of some of our basic conceptual frameworks... > .. [LJ] > You are demanding that I accept your conditions based on your > knowledge and logic. I don't *demand* anything. I merely point out facts observed over many years reading and teaching philosophy and also found in myself. (For two or three years after Max Clowes started pointing out to me in 1969 that the best way for me to make progress with some of my philosophical problems was to study and do AI, I resisted: it's hard to abandon intellectual investments built up over many years.) The strong resistance to conceptual revision is probably something we can explain better when we have a better theory of how minds works. We may also then understand why some people are more flexible and open to new ideas than others. [LJ] > You know best? How or why do you know best? Who knows best? Do you ask these questions every time someone formulates an opinion you don't like? [LJ] > I do not care to join your religion. You have chosen logic over > justice and beauty and I do not like your religion. Frankly this sounds to me like the paranoid raving of someone who regards what I've written (and also AI?) as a kind of horrible threat, but who has not read it carefully. What's the threat? Where's the injustice. Who has ruled out beauty? It's hard to read something carefully if you START with the presumption that it's going to undermine everything you value most. (By the way I have no commitment to *logic* as the sole form of representation of knowledge. I've often argued that we need a broader framework. My first attack on logicist AI was published in the proceedings of the second international joint conference in AI, held in London in 1971. The article was also reprinted in the first volume of the journal Artificial Intelligence, that year. See also the 1995 MIT Press book on Diagrammatic Reasoning, edited by Glasgow, Narayanan, Chandrasekaran, for more on non-logicist AI.) [LJ] > ..I demand that ^^^^^^^^ > the existence of free will be acknowledged as a matter of > scientific fact and your contrived logic does not permit you to > honestly meet my demand. You are the one who is making DEMANDS of me and others(?) while going on about free will. How come? I make no demands of you. If you wish, you can join in attempts to learn more about how we work and what we are, and what makes it possible for us to choose some things freely (e.g. I can choose whether to post this reply to you or not) and what makes other things impossible (I cannot look at the words on this screen without seeing them as meaningful words of English). You can starve yourself to death, but you cannot, no matter how fed up with life you are, kill yourself by holding your breath. If you don't wish to find explanations for such contrasts then why are you bothering with a news group concerned with AI and Philosophy? [AS] >> The resistance is even stronger when doing the work requires >> breaking disciplinary barriers, e.g. between philosophy and >> science. > Jorgenson: > Why should scientists get to define selfhood? My point was that most of the barriers academic departments set up between themselves and others are unjustified. It's wrong to think there is a discipline that studies insects that can be separate from the discipline that studies flowering plants. It's equally wrong to think there can be a discipline that studies notions like thought, belief, desire, choice, that can be separate from disciplines that study and try to explain human learning, perception, cultural influences, and how the brain works. Of course some people will try to defend such distinctions by saying that philosophical questions are purely conceptual questions, to which empirical observations and explanatory theories of science are irrelevant. And they then go on to do shallow and narrow philosophy which ignores everything we know about evoluation and developmental psychology, about brain damage, about the mentality of other animals, about the capabilities of new forms of machines. They are free to do that if it makes them happy. Others are equally free to attempt a broader synthesis. [LJ] > Whoever attempts to > define selfhood for all can and should meet strong resistance. ^^^^^^^^ > Who are you to define selfhood for all? And what happens to > those that disagree with your contrived, quite possibly self > serving proclamations? This is extraordinary: it almost sounds paranoid. You write as if you think people are out to get you. Do you really think I want to DO anything to you if you disagree with me? I may wish to point out errors of fact, or invalid arguments, or limitations of theories. But you are equally free to do that to me. Both of us are free to respond to such criticism. Would you prefer it otherwise? Moreover, I have no interest in "defining selfhood". "Selfhood" is not a concept I ever recall showing any interest in at all. I am interested in mental states and processes: how they evolved, how they work, how they develop, how they differ between different people, or between different species or different sorts of machines, and what sorts of architectures can support them. In all this I don't recall finding any need or use for a state, process, or thing called selfhood. I do, however, find it useful to talk about personality, character, traits, attitudes, capabilities, moods, desires, preferences, likes, dislikes, intentions, hopes, expectations, delusions, fears, thrills, pains, and many more. As for this talk about [LJ] > ....quite possibly self > serving proclamations? I am not interested in making proclamations unless saying what I think is making proclamations. How saying what one thinks is self-serving in this context I don't know. You also can and do say what you think. What's the difference? I have the impression that there's something about AI that you really fear, possibly because you don't know what it's about? If you are afraid of it because you think it will undermine some religious beliefs about free will then I can't help you. By the way, I am entirely unmoved by religious people trying to assimilate the search for scientific explanations to a form of "religion". I can see the differences well enough and if you can't then I may try to explain them another time. (Of course, if you define "religion" in a broad enough way then every belief is a religion, and the word becomes useless.) If you are afraid of AI (or a science of mind) because you think your freedom has to be incompatible with any sort of mechanism, and because you don't believe that the only sort of freedom worth having is one in which your choices flow from what you are, then I suggest you read the book by Dennett. If you don't like it, then you could post a summary and criticism of the arguments that you think are flawed, and maybe we'll all learn something. An alternative viewpoint, closer to what you seem to be saying is to be found in Joseph Weizenbaum's book Computer Power and Human Reason: From Judgement to Calculation W.H.Freeman 1976. I found him extremely paranoid about AI: he got upset just because his secretary did not wish him to look at what she was typing to Eliza. Aaron ====