[Wittrs] Re: What the Man in the Room Knows (and when does he know it?)

  • From: "SWM" <SWMirsky@xxxxxxx>
  • To: wittrsamr@xxxxxxxxxxxxx
  • Date: Tue, 23 Mar 2010 14:30:14 -0000

--- In Wittrs@xxxxxxxxxxxxxxx, Gordon Swobe <wittrsamr@...> wrote:

> --- On Tue, 3/23/10, SWM <wittrsamr@...> wrote:
> > You claim, on the other hand, that the man's
> > understanding is key because, as you have put it, try as he
> > might he doesn't understand Chinese (so, presumably, if he
> > weren't trying, then the thought experiment wouldn't
> > work?).
> >
> > I, on the other hand, have said that his trying to
> > understand Chinese is not the point of the CR.
> I must say this amounts to the most ridiculous line of argument against the 
> CRA that I've yet seen.

It's not an argument against the CRA. It's a description of it.

My argument against the CRA is that the CR on which it is based misses the 
point since it is wedded to a conception of consciousness as an irreducible 
and, hence, the CR model assumes that understanding can only be a process 
feature when, in fact, there is very good reason for thinking it's a system 
feature. And if it's the latter, then the system that is the CR is 
insufficiently specked to do what it does and so cannot be used as a basis for 
concluding anything about any other system made up of the same kinds of 
constituents (which is the point of the CRA).

> The entire idea of the thought experiment is for you, dear reader, to imagine 
> yourself implementing a formal program for understanding Chinese. You must 
> imagine yourself in that situation, and then look within yourself and ask 
> yourself honestly: "Self, do I understand Chinese by virtue of manipulating 
> these symbols according to these syntactic rules?"

That is beside the point. Of course, it's what it means to look inside the room 
and see if there is any understanding there (which is how I would put this). 
But we already agree that there isn't. The point, the thing that's wrong with 
the CR and its CRA, is that the CR is inadequately specked to do the job if 
consciousness isn't ontologically basic. So the argument only succeeds if you 
start out by thinking that that is all consciousness can be understood to be. 
But Dennett shows why that isn't the case, why there is another, perfectly 
reasonable option.

> Do you have a desire to understand the symbols? Of course you do! Presumably 
> you want to prove strong AI = true!

Our desire to understand the symbols is irrelevant to whether there is 
understanding happening in the CR and, of course, we have agreed that there 
isn't. The question before us is why there isn't!

Searle says there's no understanding to be found in the room because "nothing 
in the room understands Chinese and the Chinese Room doesn't either." That is, 
he says, look, there's no evidence of understanding anywhere so the appearance 
of understanding, through the successful passing of a Turing Test is manifestly 
not sufficient for understanding to occur. Therefore, something else must be 
needed for understanding to be present, something the room doesn't have.

But there's no evidence of understanding in the room because:

1) Understanding is not a process property (so there is no reason to expect 
anything in the room to understand what the room, itself, under the right 
conditions might understand); and

2) The room itself doesn't understand because it is insufficiently specked (the 
wrong conditions) if understanding is a system property rather than a process 

So the whole point of Searle's demonstration hinges on the belief that 
understanding (the proxy for consciousness in this example) is, and can only 
be, a process property.

It entirely misses the point that those processes might very well do the trick 
if consciousness is actually a system property!

> If I take your argument seriously, Stuart (that the man does not try to 
> understand the symbols) then I must suppose that your argument against the 
> CRA amounts to saying that the man cannot understand the symbols simply 
> because that lazy bum fell asleep at the wheel during the thought experiment.
> -gts

No, because THAT is not my argument. It is merely part of the description, 
i.e., that the man's efforts to understand aren't relevant. We already 
stipulate to the fact that he doesn't whether he is trying to understand or 
not! (That's the point of a thought experiment, after all, i.e., it doesn't 
depend on contingent occurrences.)

My argument against the CRA hinges on the inadequacy of the CR to demonstrate 
what the CRA purports to claim it demonstrates for all the reasons I've given 


Need Something? Check here: http://ludwig.squarespace.com/wittrslinks/

Other related posts: