The Chinese room experiment has a fundamental function of giving the system/person interacting with it the illusion that the room understands chinese, but it seems flaky to me what the term understanding really means. What is understanding in humans for contrast?

The way I see it, the human operator inside the chinese room is just like a traditional computer who consults the "rule-book" which happens to be written in English and here we should stop and ask what exactly does it mean when the human understands English?

Searle calls this rulebook as the program but I think that obfuscates the matter of understanding rather than telling us straight out what is meant by the word understanding. A computer understands machine instructions is like saying, a computer obeys the laws of physics and electronics. Of course it does that, it was built that way, and all a program does is change the way electricity flows in the components.

Similarly, the human operator who is reading and "understanding" the rule-book written in English, is just following the laws of physics, deriving "meaning" or rather the illusion of meaning by behaving in a way that is expected of & consistent with that of an English speaker. Of course he can read the rule book in English because he was raised in that society, where his brain could make relations with certain words and behaviors. What he is now, as a mature English speaker, then a complex electronic circuit that can respond in expected ways to how we expect English speakers to respond to English statements.

Where in this picture does understanding creep in? Should we just say that understanding English is an umbrella term for shaping an individual's brain circuitry in such a way that they can respond to English words?

Why does he insist then, that he "really" understands English? Nothing about him understands except his physiology/neurology is now shaped in a way consistent with English speaking.

If a non-native speaker is asked a question about his non-native language in English, he must first associate the best English words consistent with that experience before uttering those, giving the asker the impression of being understood and responded to in English, very similar to what the Chinese room achieves.

I also believe that in order to be intelligible enough to pass the Turing test, Searle's Chinese room should have the property of updating their rule book in addition to performing token manipulation on Chinese symbols. Thus, the rule book is not static and grows over time with each speaker asking specific questions on a topic. The rule book must keep track of this. Therefore, is this change in rule-book volume the component of understanding?

This issue is made worse when we take into account the conscious first person experience of the person. An English speaker's persistence is based upon the confidence he "feels" (another word I'm not sure about), in his consciousness. Seeing that we have just explained his confidence in terms of how good he gets at shuffling around English tokens that falls on his ears and spitting the right ones back, consistent with his experience of the world, there is no room left for the understanding bit of the story.

Summary: We are English-rooms, and our rule book is our neurology, so how are we any different from Chinese rooms in the "understanding" bit of the question?

  • 2
    If you start with the premise that a human is "a complex electronic circuit that can respond in expected ways" then you are missing the point of the Chinese room. It is not to define "understanding", but to give an example where it is lacking, intuitively, whatever it might mean. The idea is that intuition picks up what computer models of the mind do not have, and hence their premise is false.
    – Conifold
    Mar 7, 2020 at 9:39
  • @Conifold I do understand the design of the CR experiment, it was to refute the claims of the computational theory of mind that simulation is equivalent to actualization of a mind. My question is about the nature of intuition/understanding itself. What is it, that differentiates a human english speaker, and AI that is an English room? Why is not the Human too an English room? Just different hardware. I use the Chinese room as analogy to point this issue out.
    – Weezy
    Mar 7, 2020 at 9:44
  • 3
    Searle's strategy is to not answer what it is, but to claim that there is something, nonetheless, because we have intuition to that effect. So it is pointless to pose such questions to him. Instead, a response has to produce its own theory of "understanding" that explains why we "feel" that something is lacking. His reaction to the standard responses from CTM supporters has been that what they have to offer isn't credible.
    – Conifold
    Mar 7, 2020 at 9:55
  • The English room metaphor with a Chinese-speaking person works just as well. There's nothing special about English or Chinese in the argument. Your question about "what does it really mean to understand English?" is already addressed by the Chinese room argument itself, if you swap the roles of English and Chinese in the standard argument.
    – user253751
    Mar 9, 2020 at 12:48
  • @user253751 I don't see how that question was addressed in the original experiment. The point of CRA was to show that computational theory of mind cannot have a full understanding. The machine, at the end of the day, only "understands" machine language which is specific to its processor and that is a direct function of the structure how the CPU was architected. In other words, the functionalist philosophy doesn't address the problem of understanding at all. It is behaviorism in disguise.
    – Weezy
    Mar 9, 2020 at 13:15

1 Answer 1


I can offer an answer from a philosophical perspective that I think answers this question. But first -- I will try to outline what I think is the approach you are questioning. There are, to my understanding, two significantly different Identity Theories. the original one, which is what is usually how Identity Theory is described, is that Mind == neuro-structure (or neuro-dynamics, in a variation). The application of Multiple Relaizability to neurology brought this veriosn of Identity Theory into question, and a more robust proposal was that Mind == Function. Which is generalizable into Mind == Computation or Algorithm. I, and a fair number of other thinkers, see this as pretty significantly different from Mind == neuro-matter, and Jaegwon Kim among others considers the computational/algorithmic view of mind to be a form of dualism -- just not Spiritual/mental dualism.

Enter Searle, Jackson, and the arguments for qualia demonstrating the inadequacy of the computational view of mind. One can do computations, and have access to facts, without having "understanding". I will cite Kim once again, who asserts that the anti-reductionists have won this dispute over decades. He asserts that the consensus in Philosophy of Mind is that qualia indubitably exist, and they are not reducible to either algorithms or neurology (Jaegwon Kim -- Physicalism or Something Near Enough).

In epistemology, there are generally three ways knowledge is asserted to be gained: direct experience, rational deduction, and indirect inference. Direct experience is what is often asserted in Direct, or Naive Realism. Another term sometimes used for this is "intuition". Rationalism is the application of logic to a question, and has a challenge -- how can one know logic to be valid or true? The most plausible answer seems to be that we INTUITED it, which would make rationalism dependent on some prior direct experience(s). Indirect realism is the method science uses, to infer things like quark, and species. It requires observed data inputs, and the operation of logic processes, so is actually dependent on both prior forms of epistemology. This discussion puts all knowledge ultimately as dependent on direct experiences, or intuitions.

Qualia are direct experiences. But something of note is that Searle and Jackson refer to very different kinds of qualia. Jackson refers to the fundamental sensory experiences that we use to build up perceptions. Searle is referring to qualia of a very different kind -- our experience of an IDEA. Discussions of qualia often focus primarily on the base features of sensory experience, and leave out the second type. If you are not realizing there are two fundamental types of qualia, I can see how your question could come up quite naturally. Searle is asserting that we have direct experience of ideas.

Now, there is a philosophic POV that this discussion fits into perfectly and that is a foundationalist POV, in which the qualia of sensing, and the qualia of thinking, form the foundation for everything we later construct. In this POV, we have direct experience of sense data, and of very basic reasoning, and we have a natural pattern-forming proclivity, which leads us to build up our complete models of self, of the world, of other minds, and of reasoning. And the Chinese Room lacks these qualia, while the english speaking person in the room, does have those qualia, and therefore can build up the prerequisites for knowledge using the three forms of epistemology noted.

Neither the foundationalism I have outlined, nor the associated Representationalist view of perception it ties in with are popular in philosophic circles today, so you may not accept this answer completely. But I hope if not, that pieces of it will prove useful to you.

Aside -- one of the main objections to representationalism is that it basically requires a "homonculus" to receive the representation, and hence is intrinsically dualist in philosophy of mind. I think this objection is basically true, and I at least consider it to be an evidence in favor of dualism. Note, however, that a lot of the objection to dualism is anti-spiritualism, and most philosophic dualists are not spiritual dualists, they are emergent dualists, where they propose that mind emerges from either algorithms or matter. This could possibly make dualism more palatable for you.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .