Neil Rickert wrote:
>>modlin at concentric.net writes:
>> >In <6htajh$rq2 at ux.cs.niu.edu>, rickert at cs.niu.edu (Neil Rickert) writes:
> > [snip]
>> >>You seem to have entirely missed a point I made. Namely, there might
> >>be a completely different way of describing the internal operations
> >>of a computer, such that under this different internal description the
> >>computer is executing a completely different algorithm. If it is
> >>the abstract computation that matters, then I am suggesting that the
> >>abstract computation being performed is not determined by what happens
> >>in the machine, in the sense that there are completely different
> >>ways of assigning algorithmic descriptions to what happens physically.
>> >Let's focus on this point very closely.
>> >Consider a very simple computer. It's called an "OR gate". It has
> >two external inputs, and one external output.
>> Is this an abstract Turing OR gate (if there is such a thing).
>> Let's instead consider a real OR gate. To keep it simple, lets suppose it
> uses CMOS with a +5v voltage supply. What it does, is it produces a voltage
> output depending on two voltage inputs.
>> >The function it computes is "logical OR". If either of the inputs is
> >active, the output is active. If both inputs are inactive, the output
> >is inactive.
>> You are quite mistaken. Of course it computes AND. We assign TRUE
> to voltages near 0, and FALSE to voltages near +5v. Using this
> assignment of TRUE and FALSE to generate a truth table, you can
> easily see that it computes AND. But, of course, we are
> oversimplifying. We are ignoring one of its inputs, the input
> connected directly to the power supply. If we take that input as
> part of the data being processed, then it is computing a rather more
> complex function.
>> >I say that the function it computes is determined by whatever is inside
> >the black box of the computer, and will remain the same no matter how
> >you choose to describe it.
>> And I say that what it computes depends on how we describe it. I
> have just demonstrated this by describing it such that it computes
> AND. Nevertheless the causal operations remain unaffected by the
> choice of description.
As Chalmers has pointed out, an implementation can
compute more than one function, but that doesn't mean that it computes
*every* function and that we can simply arbitrarily pick one as
*the* function being computed or that, just because the molecules
were in a particular state at a particular moment, that the system can
be said to be implementing an computation that goes through that state.
An implementation only computes a finite number of functions, and
through our interest we select one or a small number of them as *the*
function(s) being computed. Discovering that other functions are also
being computed doesn't alter this. This is a common problem with false
dichotomies in philosophical discussions -- just because one branch of
the dichotomy is shown to be false (e.g., "a machine objectively
computes a single function"), that does not mean that the other branch
must be true (e.g., "what a machine computes is arbitrarily
subjective").
--
<J Q B>