Meta and Meta

Go Fly A Kite

David Papineau reviews John Searle for the TLS. I mention this only as an excuse to point to a "common sense"  ('cause Searle is big on common sense) refutation of Searle's Chinese Room thought experiment (well, actually, a refutation of Searle's reply to the systems reply to Searle's Chinese Room thought experiment; still, by extension...). You might call it the "Chinese Kite" thought experiment.

(Via Arts & Letters Daily.)

Radio Head

Every once in a while you see the idea expressed that the mind might be pure consciousness, and that it's only "physicalized" when it interacts (in some unspecified way) with the organ of the brain -- sort of like how a radio receives radio waves from some transmitting source.

Call this model TRM (for "The Radio wave Model"). TRM actually has a lot going for it, at least, to the extent you've got a yen to save dualism: (1) It acknowledges the crucial role of the physical brain in realizing conscious behavior; yet (2) it preserves the intuitively appealing autonomy of the self (free will!) over and above physical causation; and (perhaps best of all) (3) it plausibly explains the physical (neural) correlates of consciousness in a way that still resists physicalist reduction.

Still, as a philosophical or scientific sketch, TRM is a nonstarter. The main problem, of course, is that while a radio receiver can modulate transmissions and interpret them in myriad ways, turning the radio off doesn't make the radio waves disappear. But apparently that's just what happens when we turn the "radio" off. It's not hard to think of examples: Certain types of anesthesia; sleep (in some stages); a well-aimed blow to the head; and so forth. What these all have in common is that they're result is counter to what TRM predicts: Immutable, disembodied consciousness.

(Via Butterflies & Wheels.)

The "Hard Problem" for Dualism

MIT's Alex Byrne surveys the philosophical back-and-forth regarding the so-called "hard problem" of consciousness. He ends up sketching a few reasons why the hard problem "may...not [be] so hard after all." I'm with Byrne, but want to add that even if the hard problem proves intractable, it certainly can't stand as an argument in favor of dualism.

The move from the hard problem to dualism relies on the notion that if physicalism is false, we need to appeal to nonphysical, or "intrinsic," properties to account for consciousness. But unless we construe intrinsic properties as essentially conscious-making, zombies are every bit as "clearly and distinctly" conceivable even on the dualist's picture.

Consider Al (no relation to Alex Byrne). On the assumptions of dualism, Al comprises or instantiates extrinsic properties Φ and certain intrinsic properties Ψ. The dualist postulates that Φ and Ψ are necessary and sufficient conditions for phenomen-Al.* Yet we can just as "clearly and distinctly" as in the original case conceive of zombie-Al, who is a duplicate of phenomen-Al in respect of both Φ and Ψ, but who is nonetheless phenomenally void.

This means that the zombie argument defeats dualism in just the same way it defeats physicalism. In fact, it would defeat any ontology that isn't stipulated a priori as containing ingredients essential to consciousness.

*SIDE NOTE: The argument is the same for substance dualism, except that  Ψ alone (construed as a class of nonphysical substance) is postulated as necessary and sufficient.

Replying to Searle's Reply to the Systems Reply

I was just reading about Searle's reply to the Systems Reply to his notorious Chinese Room thought experiment, and thought I'd pass along what I think is a knock-down refutation. First, though, a thumbnail sketch of the Chinese Room dialectic.

The Chinese Room ("CR"). A native English-speaking male who (as it happens) speaks no Chinese sits in a room. A printer prints out strings of Chinese characters. When he receives the printouts, the man consults a rule book (in English) that tells him how to correlate the strings of characters with other Chinese characters in his inventory, stringing these together in a rule-based way to provide appropriate "responses" to the input. Like a computer outfitted with the appropriate program, then, the man here has passed the Turing Test but clearly doesn't understand Chinese. Thus, the Turing Test fails.

The Systems Reply ("SR"). While it's true the man doesn't understand the sentences, he is nonetheless a part (the "CPU," as it were) of a system--memory, instructions, operations--that does.

Searle's reply to SR (SRSR). But the man could (in principle) internalize the system (i.e., memorize the rule book and do the operations in his head),  go out in a field somewhere and still pass the Turing Test without understanding a word of Chinese.

Now the refutation. Despite Searle's treating understanding as an all-or-nothing matter, it's pretty clear that Searle's hypothetical man in the field is demonstrating some kind of understanding of written Chinese. Granted, it's not the standard sort of understanding a competent speaker of Chinese would have. But then that's mostly¹ because Searle's man in the field hasn't been given the appropriate rule book--the book that provides instructions about what to do in the world when confronted with strings of Chinese symbols having such-and-such syntactical properties. A man who could perform the corresponding, rule-based tasks would "understand" written Chinese in the relevant way.² But in principle (as Searle would agree), nothing prevents our programming a robot to be able to perform these corresponding, rule-based tasks as well. And if a man so performing these tasks understands the Chinese, then so would the similarly competent robot.

NOTES

1. It's also because Searle analogizes at the wrong level of description. (See John Haugelean's remark in the Stanford Chinese Room article.)

2. For example, a man in a field has memorized my book of instructions. Upon reading the string of Chinese symbols that translates as "Go fly this kite and I will give you 1000 yuan," he takes the kite, flies it for a time, then returns with an outstretched hand.

The intuitive assessment here would seem to be that the man "understands" the Chinese offer. And yet to bite the bullet Searle would have to say, "Well, he still doesn't understand the Chinese sentence." To which our response might well be: Go fly a kite!

This Post Knows Something

Reflecting on the implication of Richard Dawkins' scientific work, Steven Pinker argues that "we can sensibly apply mentalistic terms to biology without shudder quotes." If that's so, he continues,

we would have a deep explanation of our own minds, in which parochial activities like our own thinking and wanting would be seen as manifestations of more general and abstract phenomena.

By "mentalistic" Pinker means dispositions like 'want,' 'try,' 'know, 'want,' or 'think,' properties like selfishness or intelligence, and so forth. Thus, for example, we might seek

a generic characterisation of “knowing” (in terms of the storage of usable information) that would embrace both the way in which people know things (in their case, in the patterns of synaptic connectivity in brain tissue) and the ways in which the genes know things (presumably in the sequence of bases in their DNA).

Pinker recognizes that such extended uses of folk terms are fraught with error (as with the "tendency to confuse the various entities to which a given mentalistic explanation may be applied"), but notes (by way of example) that such confusions are corrigible.

It might butress Pinker's argument to say that we use mentalistic terms with respect to the actions of human agents even when arguably they have no more basis in consciousness than the corresponding actions of a bot. E.g., while Al is asleep, a fly alights on his nose; Al intermittently shoos the fly away with his hand, the fly shortly returning to his erstwhile perch; and the cycle repeats.

If awoken, Al would report having no knowledge of his action; nonetheless, few would quibble with the statement that "Al was trying to get the fly off his nose." (If there's a challenge to such useage, it's a challenge to folk useage--quod erat disputandum in any case.)

More With Less

Some bird brains are more capacious than your brain might have thought.

Consciousness, "Thinking" and the Turing Test

MIT's Mark Halpern has an article up at the New Atlantis on the Turing Test. It's quite interesting.

Many AI opponents presuppose that to do bona fide thinking, the "thinking" entity has to be aware that it is thinking. But that metaphysic is something Turing seems quite explicitly to want to abstract away. (As should be clear from the structure of his test, Turing was concerned with the operational aspects of thinking.)

As such, it's at least arguable that while AI is nowhere near realizing human forms of thought in computation, modern computation nonetheless may be realizing some form of thought. And whether such thought is a conscious form of thought should be (but seemingly rarely is) approached as a completely different question.

(Via Arts & Letters Daily.)

Mind Hacks' "Essential Reading"

Good stuff for cogsci/philosophy of mind geeks here.

(Via Neuroethics & Law Blog.)

There's Nothing About Mary

Brian Weatherson posted the other day on the puzzlingly puzzling case of Mary the Color Scientist. (Brian analyzes the problem from the standpoint of defeasible knowledge. As always, he says some interesting things, but I'm not posting here to assess his argument.) 

I say puzzlingly puzzling because in my mind I can't imagine why Frank Jackson's hoary hypothetical is still considered a puzzle in the philosophy of mind.

The key, and very simple-to-understand, observation is this: Ambiguous talk of "knowledge" aside, there is no reason to suppose that the process of looking at color stimuli will induce the same brain state as looking at sentences about the process of looking at color stimuli. If anything, intuitions should run the other way.

Consider the case of MARy (which stands for Machine-based Algorithmic Robiomimicry) the Computer. MARy is endowed with sophisticated visual processing software that includes a discrete, heuristic color processing component and a discrete OCR component [/suitably technical-sounding gobbledygook], the upshot of which is that MARy's language-processing capabilities allow her lab mates to enter natural language queries about her visual "experiences" and receive appropriately detailed (and reasonably accurate) responses.

Now look at two scenarios, each of which for MARy is an alternative "first encounter" with a red-colored stimulus:

1. MARy is shown a red card.
2. MARy is shown the text of an appropriately detailed physical description of what would happen within MARy were MARy shown a red card.

Obviously, Jackson's argument can only be applied to MARy's "knowledge" about the color red if one assumes the procedures described in (1) and (2) elicit--or rather, are supposed to elicit--identical (in the relevant respects) physical states within MARy.

But it is plain that such an assumption is totally unmotivated by physicalism--or any other ontology, for that matter. Indeed, that assumption is almost certainly counterintuitive to anyone with even a smattering of knowledge about computers. (If you ever meet someone who thinks you can make iTunes play "Baba O'Riley" by typing source code into Word, let me know.)

An analogous argument holds with respect to neural states in the brain.

In sum, then, ambiguous talk of "knowledge" aside, there is no reason to suppose that the process of looking at color stimuli will induce the same brain state as looking at *sentences about* the process of looking at color stimuli.

(Note that this is argument is independent of, say, Ryle's distinction between knowing-how and knowing-that, bypassing ambiguous talk about "knowledge" altogether in favor of an analysis in terms of informational enrichment.)