I shall interject, albeit briefly, here: --- In Wittrs@xxxxxxxxxxxxxxx, Gordon Swobe <wittrsamr@...> wrote: > > No "non-causation claim" exists in the third premise in the first place. > The CRA is about a non-causation claim and the third premise is intended to support the CRA as in providing a basis for its conclusion. The only way it can do that is to say something about causation in the case at hand. > A3: "Syntax by itself is neither constitutive of nor sufficient for > semantics." > > A3 means that no agent can *derive* or *ascertain* or *know* the meaning of a > given symbol X from knowledge only of the form "X" or from rules based on > that form. > You persistently confuse this idea of "agent", Gordon. As Josh once pointed out to me, an agent needn't be a subjective entity like ourselves. An agent can be anything that has a causal implication for something else (i.e., brings it about). However, when you talk here about the man in the room you persistently make this a matter of that entity, a clearly subjective agent, being unable to find the meaning in the symbols from syntactic rules concerning the symbols alone. This is a confusion because the man in the room isn't a "man" at all in his CR capacity. He's a man playing a CPU. That still makes him an agent in Josh's broader sense but, in being so, his ability to guess meanings from syntactical rules is no longer the issue because that implies that understanding requires the presence of a meaning recognizing homunculus which I'm sure Searle would not claim is necessary to have understanding in the CR. So the fact that we subjective agents cannot recognize the meanings in symbols from information about their syntax alone isn't relevant. The question, rather, is whether such rote symbol manipulation meets the standard we apply to instances of our own mental behavior which we recognize as understanding. And, of course, the answer is it doesn't because understanding things, for us, implies a whole complex of connections between the pictures or representations we carry of the world. If a computationally system could do that it is certainly arguable that it would have what we call understanding in ourselves. (I guess I wasn't as brief as I planned to be, huh?) > Syntax does not equal, nor does it SUFFICE FOR semantics, i.e., we cannot use > syntax IN PLACE OF semantics. That is true in the context we are speaking of here (though it is at least arguable that part of understanding language is being able to use the syntax of the word symbols correctly. So in that other sense we could certainly say that knowing the syntax is understanding. > And this is a general truth about syntax and semantics independent of any > considerations about the CRA or AI. > That syntax is not semantics (not the same as) is not in dispute. What is in dispute is what does it take to "cause" semantics as brains do. The system reply holds that it requires the right system and that such a system would have to be a more robustly specked "room" than the CR though it is not at all inconceivable that it could be achieved using the same constituent elements as make up the CR. > To put A3 yet another way: form does not equal substance, where substance > equals meaning. > To put my respons another way, no one says it does nor does the system reply depend on it doing that. > And per A1, programs are formal. > No, per Searle they are formal though we aren't disputing that for the purposes of this argument -- though there is some dispute as to what it means to be "formal", after all, Searle does slide into the strange position that programs, being formal, lack the capacity to make anything happen in the real world but, if so, that's to take no account of the role of the physical platforms on which the computers run, namely computers. Budd thinks Searle finally just means that computers, being mere hardware, are simply irrelevant to the computers they run on (on the grounds of multiple realizability). But THAT is to confuse the idea of multiple realizability, as in any platform with adequate capacity to run the more robustly specked system will do, with the notion that certain non-computational (and entirely unspecified) machine features must be added to the mix in which case it's no longer what Searle calls "Strong AI" which he is opposing. But that's just silly because capacity always matters in any programmed system and because NO ONE in the field of AI thinks we're talking about programs in some idealized isolation from the physical platforms on which they run. > When we say that programs are formal we mean that programs operate on and > according to the forms of data, not on or according to the meanings of data. > In other words they operate according to syntactic rules. > No one disputes that. What are disputed are the confused notions that to demand a platform with certain capacity is to depart from the claim of what Searle calls "Strong AI" and also that understanding is not conceivable as a system level function rather than one that occurs at some basic level, co-existent with one of more of the basic constituent processes that the CR system runs on. If understanding is found at the system level then the only reason the CR doesn't achieve it is it is an inadequate system. > Because of the nature of their architecture, S/H systems will never know what > they're talking about. That is the mistake, i.e., the one that depends on a predetermined notion of understanding that ASSUMES it is irreducible, hence the dualist implication! > By design, they operate on and according to the forms of symbols, not on or > according to the meanings of symbols, and form does not equal or suffice for > meaning. > > -gts > And the proponents of so-called "Strong AI" are saying that by design they can be built to operate according to meanings. That form is not the same as meaning doesn't say anything about the potential for a system to have understanding if understanding is the ability to make the right kinds of connections. I once asked you to explicate what you think understanding is if you think my account is mistaken. To date you have never answered that request. But perhaps, if you would try to do so and have a discussion about this, comparing the notion I have been offering with whatever counter notion you think pertinent, we could actually get beyond this yes-it-is/no-it-isn't level of engagement. If you would even attempt to formulate what you think understanding is, we could look at it in light of the explanation I've already given and see whose account seems more sensible. If mine is more sensible, then there is no reason to insist on an account of understanding as a kind of irreducible. If yours is, then your insistence will look much stronger. Why not take a flyer on this and offer an opinion of what it is YOU think understanding is (besides simply telling us what you are certain it isn't)? SWM ========================================= Need Something? Check here: http://ludwig.squarespace.com/wittrslinks/