Chinese Room Argument

John Searle - Chinese Room ArgumentThe argument and thought-experiment now generally known as the Chinese Room Argument was first published in a paper in 1980 by American philosopher John Searle. It has become one of the best-known arguments in recent philosophy. Searle imagines himself alone in a room following a computer program for responding to Chinese characters slipped under the door. Searle understands nothing of Chinese, and yet, by following the program for manipulating symbols and numerals just as a computer does, he produces appropriate strings of Chinese characters that fool those outside into thinking there is a Chinese speaker in the room. The narrow conclusion of the argument is that programming a digital computer may make it appear to understand language but does not produce real understanding. Hence the “Turing Test” is inadequate. Searle argues that the thought experiment underscores the fact that computers merely use syntactic rules to manipulate symbol strings, but have no understanding of meaning or semantics. The broader conclusion of the argument is that the theory that human minds are computer-like computational or information processing systems is refuted. Instead minds must result from biological processes; computers can at best simulate these biological processes. Thus the argument has large implications for semantics, philosophy of language and mind, theories of consciousness, computer science, and cognitive science generally.

–David Cole
The Chinese Room Argument

4 thoughts on “Chinese Room Argument

  1. Yeah, it’s a famous argument, and it’s total bullshit.

    Let’s go back to the Chinese room setting, with one difference. The guy in the room speaks Chinese, Someone yells at him in Mandarin Chinese, “Hey its Sunday morning. You want to go out to eat?” And our chap contemplates this for a fraction of a second and calls back, “Yeah but not dim sum, I’m sick of eating stuff I could be eating in Peking. Let’s go to Olive Garden.”

    Of course, nothing is ever quite so simple. That first voice for instance wasn’t just Chinese. It was a sequence of signals, vibrations in the local atmosphere with frequencies of 40 to 60 Kilohertz, modulated by signals of 0 to 20,000 hertz, yadda yadda yadda. It’d take a fairly big document to describe that signal as accurately as we’d like. And this complex signal shoved its way into our man’s ears, shaking microscopic hairs this way and that, pushing on the thin layer of skin that formed his eardrum … generating electrical impulses which traveled via an elaborate pathway to specific neurons in the man’s brain, sending forth waves of chemicals through thousands of neurons to bring up related memories and impressions and wishes….

    And let’s realize we’re talking in very general high level terms. To be really accurate, we’ve to examine billions of brain cells permeated by blood and various glandular secretions and the ordinary liquid content of those cells, we’ve got to watch chemicals dancing though the cells, we’ve got DNA chains splitting open and furnishing templates for groups of amino acids formed out of the whirling chaos and thrown like escaping hostages on the bareback steeds of tRNA to be speedily delivered to waystations of the cellular circuit…..

    And we haven’t even talked yet about this guy talking back yet! It’s complicated, right? And maybe come the year 2200 or so, we’ll be able to record and playback all that complicated chemistry and biology. We’ll be
    to describe what goes on in books, and with computer databases. AND THAT DESCRIPTION OF THIS EVENT IS GOING TO LOOK EXACTLY LIKE SEARLE’S PILES OF BOOKS AND COMPLICATED INSTRUCTIONS. We’re going to look a hundred thousand neurons brimming over with chemicals in varying concentrations, we’ll meter the electrical current flowing through the dendrites …. we’ll think we really understand the process of shaking up memories … but I don’t think we’re ever going to look at the chemical concentrations and voltages and put them together as “breakfast in Peking in February 2017”

    Now, you want to tell me this looks so complicated, Searle’s kind of made his case — AI is hard. But that’s not the point Searle wanted to make. He’s saying that meaning and understanding don’t exist in books or complicated sets of directions or computer algorithms — instead, we know what we know because we are living intelligent organisms. We comprehend because we are conscious and alive — either because comprehension is a possible characteristic of all living things, or because God has so blessed us. (I don’t think he explicitly draws a connection to God; it’s just implicit in the argument. As a good agnostic, I find that a little hard to swallow.)

    • I salute you, sir, as an eloquent constructor of sentences. They’re really good.

      I’m not 100% sure what your ultimate point is; I am not a mental giant. Are you saying Searle is correct, and true AI is impossible to achieve? Or is Searle wrong, and within a few centuries, we’ll puzzle out this stuff?

      I have no strong opinion either way, I dispassionately find the debate interesting, I’m just not clear on your conclusion, since I am slow on the uptake. But you do write some wicked good sentences.

  2. Chuckle! Sentences not clearly understood cannot be very good sentences. And when the topic being discussed is the comprehension of sentences …. Oh dear, oh dear, oh dear!

    Searle is attempting to show that Artificial Intelligence — at least in the naïve sense that 6th graders might have, as a machine or computer which is self-conscious, aware of its surroundings, able to control its own activity, and capable of choosing paths of action based on intellectual, ideological, emotional or other non-physical grounds — is impossible.

    He does this by imagining a Black Box device. We provide inputs to the BB, by sliding messages to it under the door, and after some time, outputs emerge, via messages which slide back to us. These answers are sensible, we are told. How is this possible, given that — by definitions — we can’t imagine what goes on inside a Black Box? And Searle imagines an answer: some sequence of mechanical acts or other physical phenomena must transform the incoming messages to something which can be understood or at least deciphered as instructions within the BB, these instructions cause the performance of various other actions, and ultimately the BB responds to its inputs by producing an output message.

    We are to understand that these are “generalized” factors. The “input message” might be a sheet of paper bearing the question “What do you want for supper?” It might also be flashing lights and sirens and TV images of enemy ICBMs exploding over American cities. The “output” might be a paper on which someone has scrawled “Cheeseburgers and fries,” or a series of very precise instructions sent through multiple communication channels to a large number of military bases, manned bombers, drones, satellites, and other government facilities. In any event, the output will be appropriate to the input — we will not start World War 3 by asking about supper.

    Since the inputs and outputs are so freeform, Searle supposes that the hidden processing within the Black Box must be equally arbitrary, with the sole restriction that it must be physically realizable. He therefore visualizes a hidden prisoner — a Captive Processing Unit — who is totally unaware of what events are transpiring.. The CPU speaks English, we can imagine, and the incoming messages are all in Chinese. But the prisoner has reference to a number of books written in an unrecognizable language –Mandarin, or perhaps Fortran 98 — and on a convenient table is yet another set of books with the comforting series title “What To Do When Then They Slide Pieces Of Paper Under Your Door” — obviously an analog of a computer program. These latter volumes are completely comprehensive, and will guide our prisoner to producing a satisfactory output to ANY input, no matter what his state of hunger, ignorance of the outside world, etc. He’ll be a totally perfect UC Berkeley graduate,

    Searle’s point is that all this activity is “untouched by human mind” even though we recognize that much of it is valuable ONLY because it is shaped by human minds. “Hamburgers and a chocolate shake” the prisoner replies to a menu request, and it is what the prisoner actually wishes for when his supper comes, not the “Chicken noodle soup and liverwurst on crackers with a side of horseradish” was daydreaming of while preparing his message. “Is there a God, and does He love us? Explain your answer.” we ask of the prisoner and for weeks to come he discusses the sources of his atheism, never imagining in the slightest that he has bared his (nonexistent) soul to anyone.

    This sounds absurd, and Searle concludes it IS absurd, and that his depiction of a black box with an idealized prisoner performing actions which mimic intelligence must be absurd. Rote behaviors cannot produce deliberate, conscious, intelligent outcomes. Artificial intelligence is impossible.


    I point out, FIRST, that it is too much to expect perfection of AI. I have neighbors, for example, who eat and drink and breath and drive to work in the mornings and who will be irritable if poked with sticks, They are all presumably alive and conscious and reasonably intelligent, although some of them may have voted for Hillary Clinton in the last election and some for Donald Trump. Surely it’s just this variety of responses that shows us we are dealing with consciousness, rather than the uniformity? (Let’s concede, I live in the Bay Area, so there weren’t that many varied responses …)

    The fact that there is no “give and take” in Searle’s account of the prisoner’s existence shows the failure of his model. The nice long “program” that dictates the actions of the CPU gives him no leeway to produce an individualized response; it no more mimics human understanding than a falling rock, and ALL of Searle’s tale, from beginning to end, is thus preposterous.

    SECONDLY, my neighbors are not all fully human. They include dogs and cats and squirrels and some young children, all without any interest at all in presidential elections. and seldom displaying great individuality of behavior when presented with pieces of fish or candy or dog biscuits. Are we to deny that they live, or have some sort of free will, or even souls, because their actions — viewed from our superior perspectives — seem so constrained in comparison to our freedom? Surely if we can conceive of a mechanism with AI, we can conceive of one with a squirrel’s degree of intelligence? Or a young child’s?

    Shouldn’t Searle have considered the possibility of flawed or immature AIs?


    And then, THIRDLY, we might conceive of sweeping aside this Searlean veil of ignorance and suppose we had complete and detailed records of the physiological and biological and mechanical phenomena which go on within a subject’s brain as he perceives and deciphers sense impressions, forms judgments, and expresses his thoughts. Surely at each instant, we would perceive some physical state — since brains and bodies, after all, are creatures of the physical universe. Surely the record of these states would correspond to the books and instructions that guided the existence of Searle’s prisoner. And surely our scrutiny of this physical record would be equally unsatisfactory in disclosing the actuality of free will, consciousness, deliberation, intelligence!

    This succession of records, these frozen physical moments, somehow fail to exhibit life! Just as Zeno’s arrow, shot into the air, could always be visualized as drawing nearer to a target but never actually reaching it.

    Yet the arrow does go to the target, the physical record did correspond to the mental activity of a living intelligent being, etc. So we must recognize our models are flawed, Searle’s imagined books and prisoners are not complete descriptions of some automaton responding to arbitrary signals. Our record of physical states does not realize the full complexity of dynamic behavior within organisms.

    Perhaps we are not as intelligent and as possessed of free will as we believe. Perhaps what we view as consciousness is not so independent from our physical being as we think, but merely something that rides about on the surface of our mental activity, visible much as scum is on the surface of a polluted bay, registering the existence of tides but not causing them. Perhaps it is the dance of dust motes along my brow and itching garments which shape my thoughts, rather than my memory of Cicero’s orations, or a pastor’s sermons. Perhaps we have fewer great moments of originality or wisdom than we’d like to think and get along on auto-pilot much of the time. Do we insist, after all, that statesmen consider the rumbles of their stomachs as louder and more pressing than the complaints of the plebes? Perhaps we overrate consciousness and the values of intelligence, imputing much more power to them then is deserved, and our very notions of what AI might produce are flawed. Which is not a happy thought, but I must form it for you.


    And FOURTH and finally, we might hold open the possibility that brighter folks than us, or those with more developed sciences, might visualize more of the universe’s possibilities than we, and behold lovely powerful visions where we can only find confusion and darkness.


    The TL;DR version: Searle didn’t think matters through as deeply as he thought.

    • Fascinating! And I wasn’t being snarky when I praised your sentences. “Chemicals dancing through the cells” is quite good. IMO.

      It sounds like a subject you’ve really given some thought to. I’ll admit, my life is too mundane for these kinds of questions. I believe they’re important questions, though, and I’m glad you and others are asking them. It’s like finding comets. I can’t do it. It’s terrific that others can.

      How much free will do we really have? How are we that much different from a squirrel that runs up a tree when it hears a dog bark? I suspect, not so different as we like to think. But I am definitely out of my league when discussing such matters.

Leave a Reply

Your email address will not be published. Required fields are marked *