On 5/15/2015 4:40 PM, Bruce Kellett wrote:
meekerdb wrote:
On 5/14/2015 7:24 PM, Bruce Kellett wrote:
LizR wrote:
On 15 May 2015 at 06:34, meekerdb <[email protected]
<mailto:[email protected]>> wrote:
I'm trying to understand what "counterfactual correctness" means in
the physical thought experiments.
You and me both.
Yes. When you think about it, 'counterfactual' means that the antecedent is false. So
Bruno's referring to the branching 'if A then B else C' construction of a program is
not really a counterfactual at all, since to be a counterfactual A *must* be false. So
the counterfactual construction is 'A then C', where A happens to be false.
The role of this in consciousness escapes me too.
It comes in at the very beginning of his argument, but it's never made explicit. In
the beginning when one is asked to accept a digital prosthesis for a brain part, Bruno
says almost everyone agrees that consciousness is realized by a certain class of
computations. The alternative, as suggested by Searle for example, that consciousness
depends not only of the activity of the brain but also what the physical material is,
seems like invoking magic. So we agree that consciousness depends on the program
that's running, not the hardware it's running on. And implicit in this is that this
program implements intelligence, the ability to respond differently to different
externals signals/environment. Bruno says that's what is meant by "computation", but
whether that's entailed by the word or not seems like a semantic quibble. Whatever you
call it, it's implicit in the idea of digital brain prosthesis and in the idea of
strong AI that the program instantiating consciousness must be able to respond
differently to different inputs.
But it doesn't have respond differently to every different input or to all logically
possible inputs. It only needs to be able to respond to inputs within some range as
might occur in its environment - whether that environment is a whole world or just the
other parts of the brain. So the digital prosthesis needs to do this with that same
functionality over the same domain as the brain parts it replaced. In which case it is
"counterfactually correct". Right? It's a concept relative to a limited domain.
That is probably right. But that just means that the prosthesis is functionally
equivalent over the required domain. To call this 'counterfactual correctness' seems to
me to be just confused.
What makes the consciousness, in Bruno's view, is that it's the right kind of program
being run - which seems fairly uncontroversial. And part of being the right kind is that
it is "counterfactually correct" = "functionally equivalent at the software level". Of
course this also means it correctly interfaces physically with the rest of the world of
which it is conscious. But Bruno minimizes this by two moves. First, he considers the
brain as dreaming so it is not interacting via perceptions. I objected to this as missing
the essential fact that the processes in the brain refer to perceptions and other concepts
learned in its waking state and this is what gives them meaning. Second, Bruno notes that
one can just expand the digital prosthesis to include a digital artificial world,
including even a simulation of a whole universe. To which my attitude is that this makes
the concept of "prosthesis" and "artificial" moot.
I don't think you would consider just *any* piece of software running to be conscious and
I do think you would consider some, sufficiently intelligent behaving software, plus
perhaps certain I/O, to be conscious. So what would be the crucial difference between
these two software packages? I'd say having the ability to produce intelligent looking
responses to a large range of inputs would be a minimum.
Brent
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.
