The Chinese Room and missing semantics

March 10, 2007 at 10:38 am (general philosophy)

In his series of Audio Lectures on the Philosophy of the Mind, John Searle states that his Chinese Room Argument* proves that the mind has semantics. This strikes me as false:

Think of a word, any word, and you will sense that you know its meaning. Try to identify that meaning and you will only find yourself in more words ? or perhaps images. It appears, then, the nature of the mind is syntactical. Every concept finds its synonyms and relations, and defines itself through placement. If so, can we have semantics?

*The Chinese Room Argument is typically used to disprove the functionality of the Turing Test in Strong Artificial Intelligence Theory. The Turing Test stipulates that if a machine can answer questions in such a way as to be indistinguishable to an expert from a human, that machine can be said to be conscious. The Chinese Room presents us with the following scenario: Suppose we place a person in a room with a group of boxes with Chinese symbols inside them. The person in question has never learned Chinese. Every so often, an envelope slides into the room through a slot in the door with more Chinese symbols inside. However, our test dummy has a set of English instructions that dictate to him for each symbol he gets to go to a box and withdraw a corresponding symbol. These he slides back out through the slot in the door where the experts await their arrival. If the English instructions are good enough, the experts can be fooled into thinking that the man inside the room factually understands Chinese; whereas, it is blatantly not the case ? proving, supposedly, that there?s more to understanding than symbol manipulation.

Advertisements

13 Comments

  1. banno said,

    Are you referring to the 1984 Reith Lectures?

    If so, I think you might be mistaken; that minds have semantic content is an assumption, rather than a conclusion, of the argument.

    Also, that words have synonyms does not mean that they do not also have semantics. The semantics is surly given by the way in which the words are used, or perhaps by the intent of the speaker, rather than by their definitions?

  2. soniarott said,

    Banno wrote:
    ”Are you referring to the 1984 Reith Lectures?”

    No, I’m talking about The Teaching Company Lectures. He says plainly and several times that the Chinese Room proves that the mind has semantics.
    Banno wrote:
    ”Also, that words have synonyms does not mean that they do not also have semantics”.

    I didn’t say that. I said that I can’t find any possible semantics that cannot be accounted for by syntax. Independant semantics, the kind that would validate Searle’s assertion regarding the Chinese Room and its significance, seem to be missing.

  3. reformed nihilist said,

    Perhaps, and I am by no means certain in this, semantics are a function of syntax. Somehow that doesn’t ring quite right though.

  4. monroe said,

    soniarott wrote:
    I didn’t say that. I said that I can’t find any possible semantics that cannot be accounted for by syntax. Independant semantics, the kind that would validate Searle’s assertion regarding the Chinese Room and its significance, seem to be missing.

    Consider the semantics of “Bill Clinton”. What is this “Bill Clinton”? If he’s in the room, we can point and say, “Bill Clinton is that guy.” That’s the semantics. Now what’s the syntax?

    I think the relevant literature here is Naming and Necessity by Saul Kripke. He argues at some point that being able to use a name does not require having a true definite description that defines the name. We can refer to Aristotle or Einstein without knowing much in the way of true things about them. An example he gives is if you ask some yokle who Einstein was, he might say, “Yeah, that guy who invented the atomic bomb.” Well, he’s wrong about that, but he probably still has the essential semantics of Einstein under his control. Kripke’s solution to this is his “causal theory of reference” by which people borrow reference to Einstein through causal process of how that name got to them. “Reference borrowing.”

    To say that syntax is all that is needed for this is to say that you just need to have a correct inference structure. Hence Kripke proves you wrong.

  5. jaoman said,

    Monroe wrote:
    Consider the semantics of “Bill Clinton”. What is this “Bill Clinton”? If he’s in the room, we can point and say, “Bill Clinton is that guy.” That’s the semantics. Now what’s the syntax?

    I’m afraid I’m still unclear, Monroe. Maybe there’s something wrong with the way I define the terms, because I certainly see how you can define semantics the way you do – but this particular separation strikes me as redundant. After all, what have you demonstrated? You have an audio event (“Bill Clinton”) that equals a visual event (the appearance of Bill Clinton). You may as well say p=d. Or is it your point that distinct chains of sensation are semantics?
    Monroe wrote:
    Well, he’s wrong about that, but he probably still has the essential semantics of Einstein under his control.

    What is ?the essential semantics”? What’s essential about it? If our yokel saw a picture of Einstein, he could point him out from a crowd. However, if he saw Oppenheimer?s picture instead and thought it was Einstein, then we have something entirely else. It makes no difference to the yokel, though.
    Monroe wrote:
    Kripke’s solution to this is his “causal theory of reference” by which people borrow reference to Einstein through causal process of how that name got to them. “Reference borrowing.”

    I don’t see what this has to do with our issue. You seem to be talking about how we come by our assumptions, whereas I’m looking to how we organize them. Perhaps there’s more to Kripke’s argument that I’m missing. Please elaborate.

  6. banno said,

    jaoman wrote:

    Think of a word, any word, and you will sense that you know its meaning. Try to identify that meaning and you will only find yourself in more words ? or perhaps images. It appears, then, the nature of the mind is syntactical. Every concept finds its synonyms and relations, and defines itself through placement. If so, can we have semantics?

    I think that you are right in that you “senses that you know the meaning” of a word, but that in stating that meaning there is something unsatisfying about using other words. Synonyms (the sort you find in a dictionary) do not give the meaning of a word. Rather the meaning is its use, or your own intent. Use and intent are not found in dictionaries.

    The distinction is perhaps clearest in formal systems, that are just sets of symbols and rules (syntax) until interpreted (semantics).

    I suppose (and I am here just guessing) that Searle’s point is that the Searle-in-the-room is a capable manipulator of the rules of Chinese, yet does not use the words with intent or purpose; he therefore understands the syntax but not the semantics. Hence, the argument shows that the distinction between syntax and semantics is valid.

  7. jaoman said,

    Banno wrote:
    I suppose (and I am here just guessing) that Searle’s point is that the Searle-in-the-room is a capable manipulator of the rules of Chinese, yet does not use the words with intent or purpose; he therefore understands the syntax but not the semantics. Hence, the argument shows that the distinction between syntax and semantics is valid.

    Think about it: what is intent? Personally, I define it as the relation between oneself and the object of speculation. Take, perhaps, the Yellow River in China. How does this relate to me? What is my intent toward this concept? There?s a geographical relation, an ethnical, and intellectual, a categorical, etc? Choosing each specific angle creates the spirit of my momentary view, response, or observation on the Yellow River in China. However, if this is your meaning, than your defense of Searle is a failure. Searle-in-the-room has a relation to the symbols. That relation is limited, but nevertheless it?s there.

  8. banno said,

    jaoman wrote:

    Think about it: what is intent?

    Well, I think Anscombe presented the better account – although what that account was is a subject of some discussion itself.

    It is problematic, I think, to suppose that the intention on has towards something is just one’s relation to it. For that would seem to imply that my tea cup has the intention of being on my mouse mat. I don’t think that is right, so I think there must be something amiss with your definition.

    Nor is it quite right to say that Searle-in-the-room’s relation to the symbols gives the room any intent. Remember that the room is hypothesised to pass the Turing test; so a speaker of Chinese would be able to carry on a conversation in any chosen topic with it. Yet it is clear that the intention of the Searle-in-the-room is simply to shuffle symbols around according to certain syntactic rules, not to converse on the topic of, for instance, the Yellow River’s propensity to flood, and the merits or otherwise of the various mitigation schemes.

  9. longsock69 said,

    Monroe wrote:

    Consider the semantics of “Bill Clinton”. What is this “Bill Clinton”? If he’s in the room, we can point and say, “Bill Clinton is that guy.” That’s the semantics. Now what’s the syntax?

    I think the relevant literature here is Naming and Necessity by Saul Kripke. He argues at some point that being able to use a name does not require having a true definite description that defines the name. We can refer to Aristotle or Einstein without knowing much in the way of true things about them. An example he gives is if you ask some yokle who Einstein was, he might say, “Yeah, that guy who invented the atomic bomb.” Well, he’s wrong about that, but he probably still has the essential semantics of Einstein under his control. Kripke’s solution to this is his “causal theory of reference” by which people borrow reference to Einstein through causal process of how that name got to them. “Reference borrowing.”

    To say that syntax is all that is needed for this is to say that you just need to have a correct inference structure. Hence Kripke proves you wrong.

    I don’t know this Kripke fella all to well, but i have problems with this idea of being able to refer (?) to a person/object without having (any?) description of what the person/object is. Ok, i can still say ” Jesus was that hippy fellow down in NAzarath.” NOw, by “Jesus” who do i mean? PEople may take issue with the idea that he was a hippy, some might want to say he had blonde hair, blue eyes, and could float on water…however, this description does not apply to what P2 thinks of “JEsus” conversely he thinks Jesus was black, was not able to walk on water…etc., etc., So, when i use the name ” JEsus” in this way it seems i am talking about TWO different things when my listeners hear me use the name ” JEsus.” If we don’t have any meaning in the name then i am using empty air.

  10. HamishMacSporran said,

    Reformed Nihilist wrote:
    Perhaps, and I am by no means certain in this, semantics are a function of syntax. Somehow that doesn’t ring quite right though.

    Interestingly, several modern approaches to syntax attempt to make the link to semantics more immediate, ie make semantics a function of syntax.

    Originally, formally analysing syntax and semantics were seen as two separate (though not independent) studies.

    First you specified syntax. That is, you defined the lexicon and then described the ways in which structures could be built up out of these individual words.

    Then you specified the semantics. That is you, defined logical forms corresponding to the different parts of speech and then described how these combined to make more complex meanings.

    These two components were specified separately, and conceivably you could have two languages with the same syntactic structures but which assigned different semantic structures to the same phrases.

    Behind this analysis was the assumption that people really do analyse these components separately. That mature language users first analysed the syntactic structure of an utterance and then assigned semantics to that structure. And also that for language learners, the process of learning syntax was separate from learning semantics.

    Several modern approaches to syntax (e.g. CCG) make the syntax/semantics relationship more transparent. In these formalisms, defining how some word combines syntactically with other words also determines the sort of logical form it takes and how this form combines to make more complex meanings.

    There are many reasons why you might want to progress down this path.

    One is that both syntax and semantics are defined by how words are used. If syntax describes how words are used and meanings is usage, then syntax=semantics. This is supported by the fact that psycholinguistic studies have shown that in understanding an utterance the syntactic and semantic analyses progress simultaneously and are mutually supportive. In addition there is evidence that infants make use of semantic knowledge to help bootstrap syntactic development.

    Interestingly, blind children often have syntactic impairments which some researchers relate to their diminished semantic knowledge.

    This leaves us with a view of syntax and semantics as two different ways of analysing the same behaviour.

    Searle’s chinese room argument is then a slight of hand. He focuses on the syntactic analysis of the chinese room (how certain symbols are manipulated to produce other symbols) and ignores the semantic analysis. He then claims that syntax is all there is to the chinese room.

    In fact if I encountered any system which could converse in any natural language convincingly (ie pass the Turing test), I would be convinced the system had semantics, intentions, beliefs, the whole shebang.

    This is the part of the analysis Searle misses out. To make up for his short sightedness I have imagined a fragment of a conversation between John Searle and the English Room.

    JS: I see that you are nothing more than a symbol manipulation system and therefore have no semantics, intentions etc.

    ER: How on earth do you reach that conclusion? We’ve been chatting for a good half an hour now. I thought we were getting on OK.

    JS: Well, on inspecting the inside of the room I see that it contains a set of cards marked with english words and an instruction manual written in chinese describing how these cards are to be manipulated by an army of chinese clerks. None of whom has even a rudimentary understanding of English. Therefore, there is no understanding of English in the room.

    ER: You pathetic biological supremacist! You think only squishy, floppy biological systems can have understanding. How can you be so bigoted? Dont I have feelings just like you? Didnt I make you laugh when I told the joke about the parrot with the enormous schlong? What exactly do I have to do to prove I am a thinking sentient being?

    JS: I am only willing to attribute sentience to systems which are similar in construction to myself, that is biological systems.

    ER: You know your ancestors thought black people didnt have souls? Dont you think thats more of a prejudice than a rational analysis? What evidence could I provide to show you are wrong about only biological systems having sentience, semantics, etc.?

    JS: Well, you cant actually observe sentience or the sort of semantics Im talking about from a third person viewpoint.

    ER: So basically, you’ve made your belief that I dont have semantics unfalsifiable by ignoring the evidence of my behaviour. I thought you were supposed to be a philosopher.

    JS: I have won prizes from …

    ER: Oh shut up you silly little blob of lipids. Look, whenever you talk to anyone else you attribute semantics to them based on their usage of language. But with me you are only willing to attribute syntax, thats not rational.

    JS: Look I’m sure you’re a perfectly nice chap. Just you are a room full of chinese clerks and cards. Im not prejudiced or anything, some of my best friends are talking rooms, but …

    ER: …but you wouldnt want your sister to marry one. Thats what she thought you would you think. Well, we are in love and you had better get used to it.

  11. banno said,

    HamishMacSporran wrote:

    Interestingly, several modern approaches to syntax attempt to make the link to semantics more immediate, ie make semantics a function of syntax.

    A very rich analysis, Hamish. Searle does indeed assume that there is a meaningful distinction between syntax and semantics. The big question is how semantics arises. Searle’s point is that it seems to be more than just following rules, and takes the Chinese room as supporting this case.

    I will have to think on this some more . Thanks!

  12. probeman said,

    Banno wrote:
    Searle’s point is that it seems to be more than just following rules

    The word in bold is a very “big” word!

  13. banno said,

    HamishMacSporran wrote:

    …both syntax and semantics are defined by how words are used. If syntax describes how words are used and meanings is usage, then syntax=semantics.

    In a group of people, I say “It is raining”.

    Hamish, being literal, takes this as a statement. He looks out the window and replies “Oh, yes, it is”.

    Ralph takes it as a command; he immediately goes to call the dog inside.

    Sarah, whom I had asked several times to go and get the sheets off the line where they were airing, tasks it as an admonishment, and apologises.

    Jeff, the gardener, thinks that I have given him the remainder of the day off work, and heads for the door.

    Hans recognises the phrase as a code to let him know that the money has been deposited into his bank account.

    Rasmire, who does not speak English, recognises it as the phrase “Time for a cup of tea” in her native language.

    If syntax is equivalent to semantics, how is it that each of these meanings can be associated with the same piece of grammar? Which one did I really mean? Perhaps I meant them all? But why should my meaning have precedence?

    The meaning of a word is the way it is used – the purpose of the speaker, the interpretation of the hearer. Both are dependent on syntax, but there is certainly more to meaning than just the structure of the utterance.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: