Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/5/84; site uwmacc.UUCP Path: utzoo!linus!philabs!cmcl2!seismo!uwvax!uwmacc!edwards From: edwards@uwmacc.UUCP (mark edwards) Newsgroups: net.philosophy,net.math,net.physics Subject: Re: Mind as Turing Machine: a proof *and* a disproof! Message-ID: <1637@uwmacc.UUCP> Date: Mon, 4-Nov-85 09:15:44 EST Article-I.D.: uwmacc.1637 Posted: Mon Nov 4 09:15:44 1985 Date-Received: Tue, 5-Nov-85 07:20:09 EST References: <1996@umcp-cs.UUCP> <667@hwcs.UUCP> <2031@umcp-cs.UUCP> <509@klipper.UUCP> <1096@jhunix.UUCP> Reply-To: edwards@uwmacc.UUCP (mark edwards) Organization: UWisconsin-Madison Academic Comp Center Lines: 42 Keywords: minds, Turing machines Xref: linus net.philosophy:2760 net.math:2115 net.physics:3238 In article <1096@jhunix.UUCP> ins_apmj@jhunix.ARPA (Patrick M Juola) writes: >In article <509@klipper.UUCP> biep@klipper.UUCP (J. A. "Biep" Durieux) writes: >> Psycholinguistics has >> found that humans can search their memory in < log n time, n >> being the number of items. Turing machines clearly can not do >> better than order n time. Proof that humans are not Turing machines. > > I'm sure that a Turing machine can search its memory faster than order >n : all it would have to do is store the stuff in its memory in some sort of >order. I'm thinking specifically of the structure called a binary tree, where >everything in the right sub-tree is > the root and the left is < the root. >Program the machine to start at some designated root (call it position 1) on >the tape. If the item to be searched for is < position n, shift left (for >example) to position n*2. If the item is >, shift left to position n*2+1. >This, on the average, will find any item in memory in log(base 2)n comparisons, >and you've still got an infinite amount of tape to the right for storage of >other items. > Pat Juola While your premise is true for simple data forms, it breaks done as the complexity goes up. Namely semantic concepts vs a simple binary number search. This is not to say that I support the log n vs n search times. There are many other attributes on which to search. One such would be a picture (A picture is worth a thousand words). This is not to say that a turing machine can not do it, only it can not do it today.(tomorrow ?) The brain has a highly parallel interconnected architecture. Each neuron has 100 to 10000 other neurons connected to it. This means after 3 cycles using the low figure 1000000 neurons can be activated (in parallel). While the computer will have only 3 things done (what ever that means). Granted that the computer is much faster (100 to 1000 x ?) th brain still has it beat because of parallelism. Scientific America says that there are 10 to the 10 or 11 neurons in the brain. If only a tenth of these are used the brain still has a large magnitude more memory then a modern processor has. mark =============================================================== Flame away