1

1. One variety of functionalists holds the functional analysis of the

brain state to give the meaning-determining description of the mental

state term. If this is right, and the description is correct, then

anything that fills it would have to count as the mental state. (So

called analytic functionalists.) One who wanted to insist on their

description might consider the CRA just question-begging, and we would

dead end in a clash of intuitions.

Reply: The functionalist has offered an analysis of mental state terms and Searle has offered an intuitive counterexample to that analysis. The functionalist who pounds the table saying that the CRA is not really a counterexample is behaving badly, not arguing for her position.

2. But perhaps the CRA is supposed to draw our attention to times when

we would and would not apply mental state terms like 'thinking' and

'understanding'. In this case, the CRA is puzzling, because it's far

from clear what part of our ordinary account of mentality is not

satisfied by the case, given the various normal modifications.

Reply: It seems to me that the whole of our ordinary account of mentality fails to be satisfied by the case. The CRA doesn’t even approach understanding. The “normal modifications” either concede that symbol manipulation doesn’t suffice, or else they result in the same ‘nope, no understanding’ intuition.

3. The case is infuriating partly because it seems intentionally

confusing. Everyone admits that Searle himself doesn't understand

Chinese, but he repeatedly makes the point that 'it's obvious that he

doesn't understand Chinese' as if anyone denies it. It also seems to

me that if we start with a different beginning, one where we interact

only with the system from the outside, we get an exactly opposite

conclusion ...

More on this in a moment....

Reply: No one denies that Searle in the room understands Chinese because it’s blindingly clear that he doesn’t. But given that Searle in the room is instantiating an understanding-Chinese-stories program, the fact that Searle doesn’t understand Chinese stories seems to say something rather significant about the prospects for giving a machine intelligence by giving it a program. Of course you get a different intuition if you imagine interacting with the system from the outside. The system has been cleverly programmed to pass the Turing Test. So much the worse for the Turing Test.

4. How can we explain away the Searlian intuition? The strategy is to

show the intuition is not really about understanding or thinking, but

is caused by something else.

Reply: How did we get to the point where it’s appropriate to look for alternative explanations for the intuition? Seeking these alternatives presupposes that the intuition is false. But what are the reasons for thinking it’s false in the first place?

Here are some arguments:

(a) The original case exploits intuitions about externalism, and only

later do we imagine the figure interacting with the world.

Reply: Hard to say. We’d have to retell the story starting with the CR in the head of the robot. My guess: No difference in force of intuition.

(b) The case may rest on our lack of ability to imagine a fully

intelligent machine table. Such a table is held to be a mathematical

possibility, but not an actual one. So, it's not surprising that we

have trouble clearly imagining a case that is so far removed from

ordinary experience.

Reply: There’s no bar to imaginatively adding whatever complexities the functionalist claims are required.

we do not follow a system of rules in producing our behavior, but such

intuitions are not to be trusted. There are lots of rules that we

follow in, say, understanding language or visually interpreting the

world, that are not introspectively accessible. (In support of this,

Searle elsewhere criticizes the idea of unconscious rule following.)

Reply: The case doesn’t implicitly appeal to this. For one thing, it’s clear, introspectively, that some of our behavior is brought about by following rules. When I put together my futon couch, I followed some rules. The question raised by the CRA is not: Is intelligent behavior ever produced by following rules? For the answer to that question is “Yes, obviously.” The question is this more specific one: Does our intelligent behavior consist entirely in following rules for manipulating formal symbols?

(d) The case may exploit intuitions about phenomenal consciousness. I

admit the CRA analog for zombies does work. Perhaps Searle is just

reiterating the zombie argument. (In support of this, Searle elsewhere

supports the connection principle.)

Reply: There’s no mention of consciousness or the connection principle in the set-up of the CRA. Someone can quite easily have the anti-functionalist CRA intuition without endorsing the connection principle. I’m one of these someones.

(e) Apropos of (d), we can redescribe lots of genuine thinkers in ways

(e.g. causal terms) that lead us to doubt our intuitions that they are

intelligent persons.

5. Remember the intuitions about multiple realizability. Those

intuitions hold that if we encountered another:

(a) Artificially intelligent robots

(b) alien beings

(d) humans with alternative physiologies

(e) =91dualistic humans=92- i.e. humans of whom dualism is true

then we would be consider them intelligent. And the reason, according

to functionalism, is that they have functional intelligence. Searle's

argument denies this. I take it, again, that this results only from

the way he tells his story.

Reply: Searle doesn’t deny multiple realizability. He does deny the claim that functionally duplicating a human will duplicate its intelligence, but that’s different. Besides, there are cases of multiple realizability that seem troubling for the functionalist…

6. Importantly, I take it that it is an empirical possibility that the

way humans achieve the semblance of intelligent behavior is by symbol

processing. Suppose we accept this. (Even if it is not an empirical

possibility, surely it is a conceptual possibility). Suppose we accept

this and Searle's argument to the effect that symbol processing is

insufficient for genuine intelligence, thinking, understanding, etc.

Then, it turns out its an empirical (or at least conceptual)

possibility that we are not real intellects, thinking, understanding,

etc.But that's ridiculous, of course we are real intellects. And the

question of how we achieve our intellectual ends is an empirical one.

Reply: I know a priori that I think. And I know a priori, via the CRA, that formal symbol manipulation doesn’t suffice for thinking. So it turns out that the proposition that thinking is just formal symbol manipulation isn’t a real empirical possibility after all.

7. On another note: perhaps we should take terms like 'thinking' to=20

pick out natural kinds, the essence of which is symbol processing (e.g.=20

exactly analagous to water).