Date: Thu 2 Jun 1988 00:48-EDT From: AIList Moderator Nick Papadakis Reply-To: AIList@AI.AI.MIT.EDU Us-Mail: MIT Mail Stop 38-390, Cambridge MA 02139 Phone: (617) 253-2737 Subject: AIList Digest V7 #12 To: AIList@AI.AI.MIT.EDU Status: R AIList Digest Thursday, 2 Jun 1988 Volume 7 : Issue 12 Today's Topics: Twin Studies Philosophy Symbolics stock ---------------------------------------------------------------------- Date: 16 May 88 03:59:48 GMT From: quintus!ok@sun.com (Richard A. O'Keefe) Subject: Re: AIList V6 #86 - Philosophy In article <523@wsccs.UUCP>, dharvey@wsccs.UUCP (David Harvey) writes: > lives. Even a casual perusal of the studies of identical twins > separated at birth will produce an uncanny amount of similarities, and > this also includes IQ levels, even when the social environments are > radically different. ONLY a casual perusal of the studies of separated twins will have this effect. There is a selection effect: only those twins are studied who are sufficiently far from separation to be located! A lot of these so-called "separated" twins have lived in the same towns, gone to the same schools, ... ------------------------------ Date: 20 May 88 09:00:08 GMT From: mcvax!ukc!strath-cs!glasgow!gilbert@uunet.uu.net (Gilbert Cockton) Subject: Twin Studies: Problems of Confounding Variables and Sample Populations In article <2865@cvl.umd.edu> harwood@cvl.UUCP (David Harwood) writes: > Can you substantially prove that there is not sound research >which shows comparatively significant psychological similarity of >identical twins, even when growing up apart? To start, yes there are similarities. The results weren't being questioned, it was the interpretation, and the original 'experimental' design. These are the two major weak links in psychological research, and for that matter in mathematical modelling (e.g. sociobiological applications of game theory). On experimental design, there is a problem in assuming that any population of separated identical twins share all the variation likely in the full population. The role of adoption agencies is particularly important, as they all have ideals of parenthood which many social groups will not be able to fulfil. Hence the separated identical twins will be less environmentally separated than the Bronx and the Berkshires. Another problem in experimental design is the very measure of 'radically different environments'. The twin studies cannot rely on assuming that ANY difference in environment could be relevant to development; the relationship has to be established by separate research. It largely because of the uncritical approach to social environments that any interpretations of 'results' will be invalid. If you don't control for confounding factors, your results aren't worth the paper they're printed on. I have no published work here either, but had to write on the topic as part of my Education degree. All the above is so obvious in psychology that it wouldn't be worth publishing, except within a more thorough review article. I've bothered to post this to a) defend Richard's argument b) improve some people's awareness of experimental design and thus hopefully encourage more constructive criticism and less credulity about BIG twin studies. Apologies to anyone who wants this sort of stuff out of comp.ai, but if you are interested in computer simulation of human behaviour, I don't see how you can justify the exclusion of anything to do with the study of humanity from this group. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs !ukc!glasgow!gilbert The proper object of the study of humanity is humans, not machines ------------------------------ Date: 20 May 88 21:35:25 GMT From: wlieberm@teknowledge-vaxc.arpa (William Lieberman) Subject: Re: Twin studies (was AIList V6 #86 - Philosophy) In article <8805201729.AA28919@decwrl.dec.com> cooper@pbsvax.dec.com writes: >> <> invalid because of a tendency for their environments to continue to >> be the same.>> > >I can't say that I am overly familiar with this area, but all the "separated >twin" studies which I have seen discussed as evidential (rather than merely >suggestive of further research) have seemingly controlled for this by >comparing the variance of the characteristic under study in identical >twins separated at birth (100% genetic similarity) against fraternal >twins separated at birth (50% genetic similarity). Is there a significant >body of studies which I am unfamiliar with, or is there some reason to >believe that the treatment of identical twins after separation is >substantially different from the treatment of fraternal twins after >separation? > > > Topher Cooper > >USENET: ...{allegra,decvax,ihnp4,ucbvax}!decwrl!pbsvax.dec.com!cooper >INTERNET: cooper%pbsvax.DEC@decwrl.dec.com > or cooper@pbsvax.dec.com Topher Cooper's remarks are well-thought out and relevant. His thoughts should be extended a little, though, I feel. At first, there would not seem to be any reason to believe the treatment of identical twins (after separation) should be substantially different from the treatment of fraternal twins (after similar separation). And there may, in fact, not exist substantive difference in treatment. But what is difficult conceptually (and therefore, in practice) to control for are phenotypically-based differences (factors, such as looks, which are observable) between, on the one hand, the set of identical twins (basically no obvious differences), and on the other hand, the set of fraternal differences (plenty of obvious, overt differences, such as in their looks - say handsome vs ugly). If the fraternal twins differ only in LOOKING different (to the adult adopting parents, etc), that fact ALONE MAY cause differential behavior TOWARD those children, chain-reacting a cause and effect cycle that winds up as being measured as "differences in intelligence (or behavior) "due to" genetic-based differences! Thus, while the difference observed within the set of fraternal twins is demonstrably due to the fact of fraternal vs identical origins, the thesis that the difference is DUE to neurologically-related differences in the nervous system is NOT thus demonstrated! All that will have been demonstrated, and I think most can agree has been demonstrated, is that the differences observed are due to SOMETHING related to genetics - but one must be very cautious, until a specific anatomical, biochemical, etc. analysis has been done on the complete developmental structure of the brain to show one way or another (which we are years away from being able to do) of drawing the conclusion that psychological factors, such as intelligence, are necessarily the sole initial CAUSE of later observed behavioral differences. In other words, until such time as it will be possible to specifically measure every aspect of TOTAL behavior, one may not conclude that a genetic difference is solely (or at all) linked to a conjectured fundamentally neurologically-based difference. Bill Lieberman ------------------------------ Date: 21 May 88 20:03:15 GMT From: mind!clarity!ghh@princeton.edu (Gilbert Harman) Subject: Re: Twin studies (was AIList V6 #86 - Philosophy) Could someone please post references to the twin studies being referred to? I am only familiar with older ones that have turned out to be based on fraudulent data. Gil Harman Princeton University Cognitive Science Laboratory Princeton, NJ 08542 ghh@princeton.edu HARMAN@PUCC.BITNET ------------------------------ Date: 27 May 88 19:46:50 GMT From: dan@ads.com (Dan Shapiro) Reply-to: dan@ads.com (Dan Shapiro) Subject: Re: [DanPrice@HIS-PHOENIX-MULTICS.ARPA: Sociology vs Science Debate] I'd like to take the suggestion that "bigger decisions are made less rationally" one step further... I propose that irrationality/bias/emotion (pick your term) are *necessary* corollaries of intelligence - that they arise because of the need to make decisions on partial information, to fill in "the gap", so to speak, between what an agent knows and the responses it might apply. The claim is that it is not in general possible to prove one's way from situation and goals to action, and that some force encoding bias is required. I.e., if you place a donkey between two equally attractive bales of hay, it doesn't starve. It chooses the left one because it *likes* the left barrel of hay. In a deeper form, this is an argument against deductive planning as an action selection technique. ------------------------------ Date: 27 May 88 22:31:07 GMT From: ejs@orawest.sri.com (e john sebes) Reply-to: ejs@orawest.uucp (e john sebes) Subject: Re: [mcvax!ukc!its63b!aiva!jeff@uunet.uu.net: Re: Sorry, no philosophy allowed here.] In article <19880527050233.8.NICK@MACH.AI.MIT.EDU> > >In article <1069@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk >(Gilbert Cockton) says: >> If you can't write it down, you cannot possibly program it. > >Not so. I can write programs that I could not write down on paper >because I can use other programs to so some of the work. So I might >write programs that are too long, or too complex, to write on paper. Yes so. You can write such programs (such as a YACC application) because someone else has written the other program (such as YACC). And that someone else couldn't have written that other program unless he could have written it down, or used some other other program..... If you want to be more precise, try this version of Gilbert Cockton's remark: If nobody can write it down, then nobody can possibly program it. I should have hoped that the essential point was obvious. If anyone really beleives that programming languages are some kind of priviledged formalism in which otherwise impossible things become possible, I'd like to hear their views. --John Sebes ------------------------------ Date: Fri, 27 May 88 23:04:18 -0200 From: Antti Ylikoski Subject: the human mind as a logical system It would seem that the human mind is very fault-tolerant with respect to locigal oddities. Example: a human being can be a queer reasoner in the sense of Smullyan. I recall that a queer reasoner believes a proposition p (Bp) and simultaneously believes he/she doesn's believe p (B - (Bp)), the minus sign denoting logical negation. Let John be a true believer of some obscure faith. Say the Tur religion by Edgar R. Burroughs in his Tarzan books. Let p be the proposition "Tur exists". Let John lament his lack of faith to a Tur priest. Then John believes in Tur (Bp) but believes he doesn't believe in Tur (B - (Bp)). Andy Ylikoski ------------------------------ Date: 28 May 88 11:20:19 GMT From: cae780!leadsv!esl!ssh@hplabs.hp.com (Sam) Subject: Re: Symbolics stock ->marsh@mbunix (Ralph Marshall) sez -> ->Summarized advice: ->1) Face squarely in directions of Symbolics shares. ->2) Turn 180 degrees. ->3) Run like hell; don't look back or you turn into a pillar of cons cells. -> ->Symbolics sells GREAT software; they just can't push boxes worth a damn. ->Their equipment is way too expensive for deliverable systems in almost any ->realistic situation, their maintenance costs even for research use is ->exhorbitant, and they don't seem to get the message from what customer ->base they have left. ...(More stuff deleted)... For this reason I've recommended any project I've been related with NOT be Symbolics-based for the last four years. Obviously, I'm not alone. I also regret that many mistakes killed the D-machines from Xerox, which were great to work in/on, but were doomed by brain-damaged sales / marketing strategists at Xerox. -- Sam ------------------------------ Date: 29 May 88 01:57:42 GMT From: glacier!jbn@labrea.stanford.edu (John B. Nagle) Subject: Re: Symbolics stock I don't know of anybody buying Symbolics boxes in quantity around here. Everybody seems to have a few, but no new ones are being acquired. Besides, they're no longer a status symbol; nowadays you have to have a Connection Machine to impress anybody. But the general machine for getting work done seems to be a Sun III. John Nagle ------------------------------ Date: 1 Jun 88 22:21:46 GMT From: bbn.com!pineapple.bbn.com!barr@bbn.com (Hunter Barr) Subject: Re: Symbolics stock In article <692@esl.UUCP> ssh@esl.UUCP (Sam) writes: >->marsh@mbunix (Ralph Marshall) sez -> I'm no investment expert, but it looks to me like you have Symbolics confused with LMI. LMI hung on at the edge of bankruptcy for a very long time, whereas Symbolics seems to gave plenty of cash to see them through this development cycle and into the next one. All the indications are that the coming batch of hardware and software is very solid. Symbolics is taking exactly the right steps to get out of the "box" business, by putting their effort into the Ivory chip and their software development. As someone who uses Symbolics Lisp Machines regularly (as well as VAXen, SUN workstations, and other machines), I can tell you that their latest release of software (Genera 7.2) shows that they are responsive to the demands of the market: It contains many popular improvements and enhancements. It was delivered on time. It marks the return of the "source included" policy, with a very reasonable price. It actually contains more of the source than 7.1 even without the fee! I don't have enough money to outfit my VAX or SUN like a Lispm; the memory, software, and OS source-code licenses are far too expensive. Moreover, it is obviously going to be a couple of years until the development tools on these machines catch up to where Lispms are now. (I am betting on Saber C, but maybe SUN's SPE will surprise us.) If I did have that much money, I would buy more Symbolics stock. I think the only way they are going out of business is if they are bought by Sony, or DEC, or a very big defense contractor. In almost every large project on which I've worked, there has been some component which was best implemented on Symbolics machines, usually for its development environment, but sometimes for the unique hardware itself. I will continue to recommend them where they are the best solution, which I expect to be often. ______ HUNTER ------------------------------ End of AIList Digest ********************