Posted 25 Jul 1996 Newsgroups: comp.ai.philosophy References: <4rcir5$ioh@usenet.srv.cis.pitt.edu> <4t3e0k$fc6@usenet.srv.cis.pitt.edu> <4t5o16$mod@usenet.srv.cis.pitt.edu> Subject: Re: Note on design stance (Re: Sloman (Part 5)) Concerning moods and the design stance. andersw+@pitt.edu (Anders N Weinstein) writes: > Date: 24 Jul 1996 17:54:46 GMT > Organization: University of Pittsburgh > > In article , Jim Balter wrote: > >In article <4t3e0k$fc6@usenet.srv.cis.pitt.edu>, > > > >>[AW] Um, I think we already make pretty good sense of mood without reference to > >>underlying mechanisms. Perhaps I don't know what you mean by "make sense > >>of". [JB] > >I hate to be Longleyan, but I suggest you study the psychological literature. [AW] > This doesn't help me. I meant: we already understand what it means for > someone to be in a good or bad mood. True we don't always understand > what to do about it. > > So I still don't understand what you mean by understanding underlying > mechanisms being required to make sense of mood. Do you mean such > things as the effect of prozac and serotonin levels on mood, or what? The answer is quite complicated. I suspect people normally think of a mood (happy, sad, depressed, optimistic, irritable) as a unitary kind of thing that you have or have not got. By investigating a range of possible designs for more or less human-like systems, and also thinking about ways of introducing malfunctions, distortions, bugs, misinformation, or whatever into very human-like systems, we can come to see that what looked like one global state that is present or absent,is actually a large mixture of rather subtle dispositions, capabilities, etc. which are thought of as a unity because they never come apart in normal human life, but which can come apart revealing a space of conceptual possibilities that we would not previously have dreamt of, i.e. possibilities in which all sorts of combinations of the sub-capacities, sub-dispositions, etc. occur. Studies of various types of brain damage have already done something like this for notions like "seeing", including "seeing a face", "seeing my house". I can't say that this has been done for moods (though psychotherapists and psychiatrists might be able to provide examples I don't know about.). Those studies can be extended by thinking about a range of possible designs for visual systems (all used to sample the optic array to gain information about the environment) with different combinations of capabilities. This sort of exploration of design space is how I see AI, though it is still at a very early stage. Now when you grasp this larger realm of possibilities and see that the concepts you currently use form only a selection from that larger realm, suddenly your old concepts (be they socially determined , or not) look very different: they can be seen to form a relatively "arbitrary" subset in a space of possible concepts. (Not totally arbitrary: that subset happens to be most useful in normal human life.) Moreover you can start asking: are there subtle differences in human moods that we had not previously noticed and which we can now start taking into account (including differences whose effects may not be very evident except in highly unusual situations). (Compare starting to think about different isotopes of carbon.) You can also start asking questions like: what is it about our design, or interactions between our design and our environment, that ensures that normally we have only this combination of states and not those, and what goes wrong that can produce those? I.e. you not only relate pre-theoretic concepts to a richer space that can extend your concepts for non-scientific purposes, but you can also ask new scientific questions posed in terms of those concepts and look for explanations linking some of the previously known phenomena to subtle features of the information processing architecture and others to subtle features of the underlying physical and chemical architecture. E.g. are there interesting differences between pre-menstrual irritability and the irritability of a man who is worried about losing his job? Perhaps we'll need to enrich our ordinary vocabulary for talking about irritability when we get a deeper understanding of the range of possibilities. Or maybe we won't because the phenomena of irritability are all due to a particular high level feature that can certainly be affected by different sorts of causes deeper in the system but whose states do not reflect the subtle differences in the causes (like a TV set that has certain states that do not depend on whether they are switched by pushing buttons on its control panel or by using a remote control unit). Note that which it is is an empirical, not a conceptual question. I am aware that this is a shallow, programmatic, and partial answer. It may suggest ways in which a deeper answer might elaborate. Aaron ===