Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ukma!rutgers!elbereth.rutgers.edu!harnad From: harnad@elbereth.rutgers.edu (Stevan Harnad) Newsgroups: comp.ai Subject: Re: Question on Chinese Room Argument Summary: Enter the neural-modelling delegation... Message-ID: Date: 25 Feb 89 06:21:47 GMT References: <4298@pt.cs.cmu.edu> <17923@iuvax.cs.indiana.edu> Organization: Rutgers Univ., New Brunswick, N.J. Lines: 102 dave@cogsci.indiana.edu (David Chalmers) of Concepts and Cognition, Indiana University writes: " ["Symbol"] is used [by Searle] to mean two different things: " (1)... a formal object which corresponds to some HIGH-LEVEL, semantic " concept in the real world... [e.g.,] a _word_ [and] (2)... any formal " object... manipulated by a computer program... low-level or " high-level [or meaningless] [e.g., a neuron] So far, so good, though I don't find this distinction particularly useful, because it just concerns how you INTERPRET the meaningless symbols you're manipulating -- here as words, there as neurons. (In principle, even the very same program could be interpreted either way.) But let's go on and see where this leads: " with sense (1)... I reject the PREMISE of Searle's argument; a formal " symbol-manipulator could never even display what _looked_ like " competent Chinese-speaking behaviour Well, this certainly gives away the store, and I'm inclined to agree. But I have reasons. Do YOU have better reasons than that you like neurons better than words? " [But] Contrary to what Harnad implies, Searle is not only arguing " against high-level symbol manipulators in the Newell/Simon/Fodor mould. " He wants to say that NO computer program could... have true " (subjective) understanding, not even an incredibly complex and subtle " program (such as a program that simulated a neural network the size of " the brain.) Actually, I don't imply otherwise: This is exactly what Searle would say, because for him it is immaterial how the symbols are interpreted by the programmer, as words or as neurons: To him they're all just meaningless symbols. And so are the inputs and outputs (Chinese symbols, remember? not Chinese neurons). Nor is Searle impressed with hand-waving about "incredible complexity and subtlety": Symbol manipulation is just symbol manipulation, no matter how complex the symbols or the interpretations. " [Searle] uses the word "symbol" in the low-level sense (2), while " appealing to our intuitions about symbol-manipulators which manipulate " high-level symbols of sense (1)!... But AHA - here we have him. These " low-level (sense 2) symbols... correspond to micro-structural entities " (such as neurons), which taken alone are devoid of semantics. Semantics " only emerges when we put enough of these neurons together to form an " incredibly complex SYSTEM. Despite the fact that the symbols taken " alone are meaningless, put enough of them together in the right way and " meaning will be an EMERGENT property of the system, just as it is with " the human brain. What we have here is exactly what it sounds like: Not an argument, but a statement of faith in the "emergent" properties of "incredibly complex" systems. I feel the same way about clouds sometimes. The human brain's another story. (The following is almost a paraphrase of some arguments from my paper, "Minds, Machines and Searle.") Of course we know the brain "has" semantics. But a symbolic simulation of a brain is not a brain, any more than a symbolic simulation of a plane is a plane. Hence there's no reason to believe that a brain simulation can think any more than a plane simulation can fly. On the other hand, there is every reason to believe that a correct brain simulation, like a correct plane simulation, could model symbolically all of the relevant causal principles we would need to know about thinking and flying in order to implement their mechanisms as a brain and a plane, respectively. The implemented brain and plane could then think and fly, respectively. But they wouldn't be just symbols anymore either. For one thing, they'd have to have the causal wherewithal for interacting with the outside world the way brains and planes do -- and that's not just symbols-in and symbols-out. They would have to include transducers and effectors (which, as I said before, are immune to Searle's Argument), and, if the other arguments I've been making have any validity, it would have to include a lot more nonsymbolic (analog, A/D, feature-detecting, categorical, D/A) processes in between the input and the output too. As long as the system's of the right type, you need make no special appeal to "incredible" complexity and "emergent" properties (though it'll no doubt be complex enough). Where you need inordinate amounts of complexity and equal amounts of credulousness is with a system of the wrong type, such as a purely symbolic one (or perhaps a purely gaseous one). " It is a very mysterious question indeed how real understanding, " subjective experience and so on could ever emerge from a nice physical " system like the human brain... nevertheless we know that it does, " although we don't know how. Similarly, it is a mysterious question how " subjective experience could arise from a massively complex system of " paper and rules. But the point is, it is the SAME question, and when we " answer one we'll probably answer the other. The first case is certainly a mystery that is thrust upon us by the facts. The second is only a mystery if we forget that there are no facts whatsoever to support it, just the massively fanciful overinterpretation of meaningless symbols. -- Stevan Harnad INTERNET: harnad@confidence.princeton.edu harnad@princeton.edu srh@flash.bellcore.com harnad@elbereth.rutgers.edu harnad@princeton.uucp BITNET: harnad@pucc.bitnet CSNET: harnad%princeton.edu@relay.cs.net (609)-921-7771