1

The description of the strong AI position in the wikipedia article on the Chinese room describes the blurring of the difference between a simulation and the real thing with respect to the mind:

The definition hinges on the distinction between simulating a mind and actually having a mind. Searle writes that "according to Strong AI, the correct simulation really is a mind. According to Weak AI, the correct simulation is a model of the mind."

If we acknowledge the existence of the real thing, it is often possible to produce an emulation of it. Often, this emulation is better than the real thing for all practical purposes. The purpose of an emulation is to reproduce all relevant features of the external measurable behavior of the real thing. The advantage (and disadvantage) of the notion of emulation is that its intention and meaning is more restricted than the intention and meaning of the notion of simulation.

A simulation related to some real thing can have many different purposes. The perfect reproduction of the external measurable behavior of some real thing is normally not among these purposes. In fact, simulation is often used in situations where there is considerable uncertainty about the actual behavior of the real thing. Simulation is sometimes used to enable observation of internal states which would otherwise be unobservable.

In the context of Searle's Chinese room, we have a computer program related to the understanding of Chinese. The Chinese room is assumed to emulate the operation of this computer program. Now we ask ourself the question whether the Chinese room understands Chinese.

Questions

  • Because the Chinese room emulates the operation of the computer program, the question arises whether the statement that the computer program understands Chinese is equivalent to the statement that the Chinese room understands Chinese.

  • Is there some sense in which we can claim that the computer program simulates the understanding of Chinese? And I don't mean the claim that it is feigning the understanding of Chinese, because this just uses a different interpretation of the word "simulation" than I would expect here. Or is this really the intended interpretation of the word "simulation" in the quoted passage?

Improve this question
3
  • 1
    There's something that bugs me about talking about a computer program understanding. Shouldn't we really refer to the whole system the computer program runs on - everthing between and including input and output. Otherwise the computer program is the equivalent of "John" in the Chinese Room or the brain of the chinese speaker. We don't ask if a brain understands chinese rather we ask if a person understands.
    – leancz
    Dec 19, 2012 at 8:51
  • @leancz Perhaps I should have written "operation of the computer program" instead of "computer program". So we have the three statements: (1) "The correct simulation of understanding is equivalent to real understanding." (2) "The operation of the computer program simulates the understanding of Chinese." (3) "The Chinese room emulates the operation of the computer program." John Searle argues against (1), but I and many others get most confused about (3).
    – Thomas Klimpel
    Dec 19, 2012 at 9:36
  • @leancz An algorithm is an abstract construct, but something performing an algorithm is a physical thing... I agree in general with your point here, except I don't think the difference between brain and person is where to draw the line, as it is pretty clear that the brain is doing the heavy lifting when a person understands something.
    – A Raybould
    Jul 19 at 14:33

2 Answers 2

Reset to default
2

Because the Chinese room emulates the operation of the computer program, the question arises whether the statement that the computer program understands Chinese is equivalent to the statement that the Chinese room understands Chinese.

I think that's precisely what Searle's argument is getting at. Our intuition tells us that the room does not understand Chinese; therefore, an algorithm performing the same functions as the room does not understand Chinese.

Is there some sense in which we can claim that the computer program simulates the understanding of Chinese?

Yes; to the outside observer, it might appear that the Chinese room (or the computer equivalent) understands Chinese. If an investigator were given two teletypes, one of which goes to a trained Chinese translator, and one of which goes to the Chinese room (or its computer equivalent), there is no reliable way the investigator can tell the difference. In other words, they are functionally equivalent to the observer. However: as shown above, our intuition tells us that they are actually very different, and that understanding Chinese is very different than the mechanical simulation of understanding Chinese.

Improve this answer
1
  • The intuition that Searle is explicitly exploiting here is that the person in the room does not understand Chinese. It takes some question-begging to get to the intuition that there is no possible rule book whereby the room understands Chinese.
    – A Raybould
    Jul 19 at 14:24
-1

Brains are made of neurons, consciousness is an algorithm run on brains. Brains are the most complicated computers known to man, consciousness is the most complicated algorithm.

There is no such thing as "the real thing" wrt. minds. Consider in the future that we create a brain scanner so powerful it can see the configurations of synapses and enzyme activity in a human brain. I let some scientists scan me, and some neuro-scientists develop a way to simulate neurons and neuro-transmitter densities so precisely that my scan data can now be made a computer-simulated brain.

With a few VR systems attached to the various nerves governing sensory input/motor output, the scanned me is now a functional human in a simulated world. The simulated me is just as much a person as flesh me is, even though we run on differing computing substrates.

Consider that the flesh-and-blood me experiences waking up in a scanning facility in the real world, while the simulated me experiences waking up in a VR bed or something. We share memories but our experiences diverge. Flesh!me and Data!me are no longer the same person.

Flesh!I have hopes, dreams, desires and inner life, so does Data!I.

If you destroy the computer running Data!me, you can bet your free will Flesh!me will press charges for manslaughter.

Because personhood is an algorithm, not a computing substrate.

Improve this answer
2
  • 1
    Perhaps I should have left a comment @MichaelDorfman's answer that he really answered my question and helped me to better understand Searle's argument. I always wondered why Searle's argument generated so much discussion. The reference to "the real thing" in my question is just for explaining my understanding of "emulation" and "simulation". Your answer focuses on "the real thing" and details a strong AI position, but fails to answer my actual question. OK, the question was already answered before, so perhaps you just want to have a discussion. (I didn't downvote.)
    – Thomas Klimpel
    Jan 27, 2013 at 20:35
  • 2
    Knowledge of the basic units can sometimes be a bit misleading. Today, we know that the world around us is composed of atoms, that biological inheritance works via DNA, and that any feasible computation can also be done by a universal Turing machine. However, if you dig deeper in any field, you soon realize that the really useful knowledge is in the details. Understanding and manipulation of prokaryotes turns out to be much simpler than manipulation of eukaryotes. Everything you claim is as true (and untrue) as the atomic theory, but it doesn't explain consciousness.
    – Thomas Klimpel
    Jan 27, 2013 at 21:23

You must log in to answer this question.

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