IUBio

Rickert on embedded computation (was re: science of consciousness.)

Patrick Juola patrick at gryphon.psych.ox.ac.uk
Thu Apr 30 10:25:07 EST 1998


In article <6ia3l2$lea at ux.cs.niu.edu> rickert at cs.niu.edu (Neil Rickert) writes:
>patrick at gryphon.psych.ox.ac.uk (Patrick Juola) writes:
>>In article <6i88a8$j7g at ux.cs.niu.edu> rickert at cs.niu.edu (Neil Rickert) writes:
>>>andersw+ at pitt.edu (Anders N Weinstein) writes:
>
>>>>Formal models of computation treat *interaction* in terms of 
>>>>symbols getting written onto the tape memory. This seems a 
>>>>pretty good model of how the operating system code views a device. 
>
>>>Right.  But you generally have to model this as a Turing machine with
>>>oracle.  You can't model it as a plain Turing machine.  The oracle
>>>places the interactive symbols on the tape in a way that is not
>>>predictable from within the Turing machine model.
>
>>But Turing machines neither need nor are expected to predict their
>>inputs.
>
>Bah! Humbug.
>
>Turing machines have no input.  The data is all there on the tape
>before they begin execution.

But it needn't be.  Because the Turing machine can't control its
inputs, it might just as well all be on tape at the beginning.

That's the difference between a TM with input and a TM with an
oracle.  An oracle-TM can phrase a question *and get an answer*.
The inputs to a TM -- or to a real computer -- need have nothing
to do with the current machine state.  Even when your Mac apparently
locks up waiting for you to click the "confirm" button, what it's
really doing is reading successive bits from the input tape waiting
for one that it likes....

	-kitten




More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net