A thought on the Chinese room argument
The Chinese room argument says that -
Now consider the room to be our brain and the person is replaced by a chain of neurons. The visual symbols of the Chinese alphabets are converted into a series of action potentials which are transmitted by a chain of interconnected neurons. This gives rise to the conscious understanding in our brain. But no individual neuron have the understanding of Chinese or have any idea what these symbols mean, it is just opening/closing the ion channels in response to a neurotransmitter and passing on the action potential to the next neuron. The same thing occurs in even in the whole network of neurons. This is analogous to the person following a set of instruction.
Hence it appears to me that consciousness is the property of whole systems on not of its isolated part, this has already been posited out as systems reply to Searle.
Imagine a native English speaker who knows no Chinese locked in a room full of boxes of Chinese symbols (a data base) together with a book of instructions for manipulating the symbols (the program). Imagine that people outside the room send in other Chinese symbols which, unknown to the person in the room, are questions in Chinese (the input). And imagine that by following the instructions in the program the man in the room is able to pass out Chinese symbols which are correct answers to the questions (the output). The program enables the person in the room to pass the Turing Test for understanding Chinese but he does not understand a word of Chinese.
Now consider the room to be our brain and the person is replaced by a chain of neurons. The visual symbols of the Chinese alphabets are converted into a series of action potentials which are transmitted by a chain of interconnected neurons. This gives rise to the conscious understanding in our brain. But no individual neuron have the understanding of Chinese or have any idea what these symbols mean, it is just opening/closing the ion channels in response to a neurotransmitter and passing on the action potential to the next neuron. The same thing occurs in even in the whole network of neurons. This is analogous to the person following a set of instruction.
Hence it appears to me that consciousness is the property of whole systems on not of its isolated part, this has already been posited out as systems reply to Searle.
Comments (85)
A person who has a mind, and speaks Chinese and reads and writes in Chinese too, has an AI program. It processes not only simple input-output instructions, but makes decisions on avaliable other data.
For instance: You pass in the Chinese question "How is the weather today?" the instruction set may direct the dumb actor to say "nice" or "awful", but it will never say "12 Kilograms." When it's a choice of "nice" and "a\wful", the program alone can't decide. It needs a third input into the instruction set, "check the weather and answer accordingly."
If it's a simple translation set of instructions, the machine will be stuck with not knowing whehter to say "nce" or "awful". A human who has access to the third piece of information can pick the proper symbol.
I don't know what this proves or unproves, because, frankly, I don't follow the logic that brings you to the conclusion that the consciousness is the whole thing, not one of its part. That conclusion absolutely escapes my understanding.
Hence, I reject the concept that the Chinese room purports to prove according to the example you brought up.
This conceptualization of "nice", "awful" etc is due to the activity of one or more neuronal networks in our brain. But any single neuron within the neuronal network is unaware of what it is conceptualizing.
You don't just jump from "a single neuron" to "full human consciousness" like that.
Yes, I agree. I am trying to draw an analogy in which a neuron is the man in the chinese room and our whole brain is the room itself. Both the man and the neuron have no understanding of chinese yet the brain will understand chinese, hence the room should too.
It seems to me that to solve this riddle, we need a concise definition of "understanding".
Then the Turing Test isn't very good at determining some system's understanding of some symbol-system.
Ironically, the instructions themselves is a symbol-system and the man in the room understands the instructions, but not Chinese, so the instructions are not for understanding Chinese, but what do when you see a certain scribble.
So the man in the room does understand something, but not Chinese. This leads one to posit that understanding is possessing instructions for interpreting some symbol.
Searle is arguing much the same thing with the Chinese room.
How would the Chinese room deal with nonsense? How would it translate A Spaniard in the works?
Am I reading this right..? Are you using the Chinese room argument to suggest that individual neurons aren't conscious?
Am I reading this right..? Are you suggesting that we have billions of conscious entities within one brain? I wonder, which neuron in my brain is my consciousness?
Is the problem that the sentence is actually nonsense, or that there are no instructions for translating such an arrangement of scribbles? What does it mean for some string of scribbles to be nonsense?
Oliver Sacks in his book The River of Consciousness wrote about patients with Parkinsonism with bradykinesia who had altered perception of time. Time flowed more slowly for these patients in their Parkinsonian state. They were able to recognize this change in their temporal perception only when they were relieved of this Parkinsonian state by medication or deep brain stimulation. The region in the brain responsible for this change in temporal perception has been grossly identified to be the basal ganglia and substantia nigra. So a directed electrical stimulation in the brain can change the perception of time.
I think Searle's thought experiment was rather a reaction to reductive takes on consciousness, particularly computational, functionalist ones:
Quoting debd
Quoting debd
There are other variants of the thought experiment that are an even better fit for this, such as Ned Block's Chinese Nation thought experiment, where a large group of people performs a neural network computation simply by calling a list of phone numbers. The counterintuitive result here is that a functionalist would have to say that the entire system thinks, understands language, feels pain, etc. - whatever it is that it is functionally simulating - even though it is very hard to conceive of e.g. the Chinese nation as a single conscious entity.
But I think this people-as-computer-parts gimmick is a red herring. Of course a part of a system is not equivalent to the entire system - that was never in contention. A wheel spoke is not a bicycle either. The real contention here is whether something that is not a person - a computer, for example - can have a functional equivalent of consciousness.
Another issue is that the contents of a computer's mind (if it has one) are immune from discovery using scientific methods. The only access to knowledge of computer mental states would be through first-person computer accounts, the reliability of which would be impossible to verify. Whether machines are conscious will forever be a mystery. This suggests that consciousness is unlike all other physical properties.
Your take on this also implies that, if we flip it around, that consciousness maybe an illusion; after all, if one is under the impression that a Chinese Room is incapable of understanding then, we too must be incapable of doing so.
How is this issue different from not having a first-person experience of another person's consciousness? Unless your real issue is that it's a computer rather than a person - but that is the same issue that Chinese Room-type thought experiments try to capitalize on (confusingly, in my opinion).
Yes, that's why I used the analogy of the brain and its constituent neurons which we consider to be conscious.
Quoting SophistiCat
Consider a neuronal network that is responsible for the perception of time as in the case of Parkinson's patients I described earlier. Now consider a biological neuron in this network is replaced by an artificial one. This is already being done, although not at the level of the neurons but more crudely with deep brain stimulation and responsive neuro stimulation. Hopefully, technology will advance sufficiently to let us do this at the neuronal level. Now, these patients don't consider any otherness in their perception except that it normalizes from the diseased state, even when they know that implants are present within their brain. And now we keep replacing biological neurons with electronic ones - sort of like the ship of Theseus. I would argue that as this part biological part electronic construct retains its time perception, if we ultimately replace the whole network, it will retain the same perception. We can extend this to consciousness itself although the network will be much more complicated.
I second this. If we consider the statement of a supposedly conscious computer to be unreliable then the same should apply to any other person also and then everyone other than ourselves may be a zombie.
:up:
It's perfect as you wrote it. All I did was gild the lily.
So are we just gonna ignore the fact that the person in the room passed the program instruction, and not the understanding of the Chinese language?
Sometimes I feel that thought scenarios like this is more like a sleight of hand in logical argument -- as long as the reader keeps losing track of what is being said, the argument keeps its force.
What makes a neuronal network conscious but not a silicon network? Sounds like biological bias to me.
Also, this seems to be 3rd person view of understanding. What is the 1st person view of understanding or consciousness or perception. I know I'm conscious, understanding and perceiving by different means than you would know I'm conscious, understanding and perceiving. Why?
Quoting RogueAI
Quoting RogueAI
Only because of thinking of mind and body in conflicting dualistic terms creates the problem in the first place.
Well, yes. That's what an algorithm is.
Minds, Brains, And Programs
This version has replies to critics.
The problem is more that "A nice derangement of epitaphs" could not be translated into Chinese without losing the joke. Hence, there are aspects of language that are not captured by such an algorithmic translation process.
No, just that the person passing the instruction is replaced by a neuron or a network of neurons passing the instructions without having any intrinsic understanding of chinese even though the brain as a whole does.
If he internalizes all the rules in his head(brain) then effectively he is learning and understanding chinese. This is what we do when we learn a new language. Memorizing all the rules does not allow me to answer questions like "How do you feel today?", "What are you grateful for today?". This has been posited as a reply to Searle by D Cole.
I'm not saying a silicon network with a similar complexity as that of a neuronal network will not be conscious. Instead I think it will have and I said asmuch upthread with an analogy to the ship of Theseus.
I confess that I have no answer to the second part.
We can look inside our brains and see. Consider my brain and yours. We undergo fMRI and EEG scans when we are awake and find both of us have similar fMRI and EEG patterns. Now for a given brain state as represented by fMRI and EEG patterns, if I consider myself to be conscious, why shouldn't I consider the same for you when you too have a similar fMRI and EEG pattern as me?
And yet...
...so he has not understood Chinese
This doesn't strike you as problematic?
So yes, it is not the whole thing. No where near it.
You didnt answer my question. What makes some string of scribbles nonsense? What makes some string of scribbles a joke? I understand English but didn't find that string of scribbles funny.
Use. What you do with that string of scribbles. Think I've said that before.
Because consciousness had been in the domain of religion as the soul for so long. Science seems to want to dismiss it as an illusion, but then consciousness is what is used to observe the world and theorize about what is observed. If consciousness were an illusion then so are all scientific theories as they are based on what is observed via consciousness.
There have been many things considered outside the realm of science, but have eventually come under the fold of science. When science takes it seriously we should be able to have better theories. It will take a change in our view, just like other great discoveries like Newton discovering gravity, Einstein discovering the relationship between gravity and space and Darwins theory.
Putting idealism aside, yes, we have indirect evidence other people are conscious because they have brains like our own, but there's no way to know for sure if they're conscious. How do I know that there's not something unique to my brain, some little unnoticeable difference, that makes me (and me alone) conscious? How would I begin to even test such a theory?
With advances in technology we will have a far more detailed picture for comparing. Hence the gaps in which this uniqueness can hide will become smaller. For example, magnetoencephalography provides a much more detailed spatial map of our brain than EEG. But sure you can always argue that there is "some little unnoticeable difference". Similarly I can say that consciousness is due to a teapot orbiting the sun somewhere with no way to disprove it but that's not a very helpful way to go about it.
"consciousness is due to a three pound hunk of meat orbiting the sun somewhere". Does this make more sense than the teapot?
My non-pithy response: The fact that materialism can't disprove "consciousness is due to a teapot orbiting the sun" is a problem, don't you think? Shouldn't it be able to show the absurdity of such a thing? After all, if I said that "the earth's rotation is due to a teapot orbiting the sun" or "the sun's energy comes from a teapot orbiting the sun", I could easily be disproven.
Before you can claim consciousness depends on matter, you have to clearly define matter and conciousness. Science has not defined either adequately. It doesn't know what we are looking at, nor what is looking.
This is plainly obvious for all to see.
As far as we can tell, its impossible to know anything about objective reality. Science has not revealed a single objective truth as of yet, only subjective observations (repeated observations sure. But no amount of subjective observation will change the fact that its subjective)
The issues that are causing such consternation here are borderline.
To give an idea of the complexity, there are about 100 trillion connections within our brain and the present technology allows us to put leads with 4-8 channels in a lead and and usually only two leads are put in for long term use. It is like doing nanotechnology with hammers.
Nice. I shall use this in future.
None of the progress has been made on the causal explanation: How do brains produce consciousness? Also: Why are we conscious? There's an Explanatory Gap. Science hasn't filled it with anything except speculation. Integrated Information Theory and Panpsychism are all the rage, but they're just guesses.
Now, will this gap eventually be filled? I doubt it. If it were, we'd have seen some progress by now. The Hard Problem was ignored for a long time, but the lack of an explanation is starting to bother people.
So you are saying that the Chinese room is a brain with one neuron... as the man is analogous to the neuron. Yet you proved it that one neuron does not a brain or consciousness make.
And there is one man (1 man) in the Chinese room.
How do you explain a room / brain with one neuron to be conscious?
Like i said, it will require a change in the way we think about reality - like abandoning dualism, materialism and idealism. Everything is relationships, or information.
Quoting debd
But then you have to explain how neurons cause consciousness, or changes in consciousness. Is it a temporal or spatial change? How does something physical cause a change in something non-physical?
What language are the instructions in the Chinese room written in? Seems like the man in the room has to understand some language in order to know what to do in the room. If the man were one neuron then one neuron possesses an understanding of the language of neurons.
And we use scribbles to communicate. Think I've said that before.
You can also replace the man in the room with a single neuronal network that does the processing instead of a single neuron but the argument against the chinese room will still stand. If the superorganism that is the chinese room does not understand chinese, our brain will also not understand chinese.
At this point I'd like you to consider the nature of consciousness, specifically the sense of awareness, particularly self-awareness. The consciousness we're all familiar with comes with the awareness of the self, recognition of one's own being and existence, which unfortunately can't be put into words as far as I'm concerned. It's quite clear that the Chinese Room is, from the way it operates, aware, albeit in a very limited sense, of its external environment in that it's speaking Chinese fluently but is it self-aware?
I've already asked them to do that as well as define understanding, but they only seem willing to keep asserting their unfound notions.
They also ignore the fact that the man in the room still understands the language the instructions are written in and how the man learned THAT language, and then they're failure to define understanding and consciousness, this thread is just a bunch of smoke and mirrors. Interesting how you can learn another language using your language, hmmm?
I think the person in the Chinese Room, his knowledge of language, any language for that matter, isn't important. If I recall correctly, he doesn't know Chinese at all. All that this person represents is some mechanical computer-like symbol manipulation system that spits out a response in Chinese to a Chinese interlocutor and that's done so well that it appears the Chinese Room understands Chinese.
Perhaps this isn't the the right moment to bring this up but the issue of Leibniz's identity of indiscernibles seems germane. The Chinese Room is indistinguishable from a Chinese person - they're indiscernible - but does that mean they're identical in that the Chinese Room is ontologically a Chinese person? The issue of Nagel's and others' idea of an inner life as part of consciousness crops up.
First you say that knowledge of any language isn't important, then go on to explain how some entity knows Chinese or not.
Seems like we need to know how the "mechanical computer-like symbol manipulation system that spits out a response in Chinese" learned how to do just that.
Quoting TheMadFool
The difference is that the instructions in the room are not the same instructions that a Chinese person used to learn Chinese. People are confusing the instructions in the room with instructions on how to use Chinese. Since the man in the room already knows a language - the one the instructions are written in, he would need something that shows the Chinese symbol and then the equivalent in his language - you know, like how you use Google translate.
The memory of what to do when a chinese symbol enters the room is on the paper with the instructions. It retains the information of what those symbols mean, which is write this symbol when you see that symbol, which is not the same instruction set in a Chinese person's memory for interpreting these symbols. This is because symbol-use is arbitrary as you can use any symbol to point to anything. Limitations do arise, however, when you want to use those symbols to communicate. You have to not only remember how you are using the symbols, but how others use the same symbols.
I'm no linguist but I believe there are syntactic rules that govern all languages - these are computable i.e. can be reduced to an algorithm.
Semantics is, forgive my ignorance here, of two types: 1. Concrete and 2. Abstract. By concrete meanings I refer to ostensive definitions which is basically an exercise in matching words with objects. Abstract meanings are extracted patterns from, among other things, concrete meanings. Computers are fully capable of both assigning names to objects and pattern recognition.
All in all, computers are capable of both syntactical and semantic aspects of language. What this means is that language can be reduced to computation. If one wants to make the case that consciousness is something special then you can't do it using language.
:up:
Let us consider our sense of awareness as I'm most familiar with that. there are multiple levels of self-awareness in humans, the lowest being not in a coma or vegetative state and the highest being that of metacognition. How would you know if I am self-aware or not? You can only do that by looking to a comparator, yourself. At the crudest level we can do that by comparing behaviors - you compare my behavior with yourself and as it is fairly similar you assume with a certain degree of probability that I must be self-aware too.
With advancing technology, you can examine more closely and reduce the degree of uncertainty in your assumption. Along with behaviour, our EEG patterns are also similar, along with behavior and EEG, our fMRI are similar, along with behavior, EEG and fMRI our MEG patterns are also the same. Hence the uncertainty goes down. Ofcourse you can never be completely sure but with the increase in resolution/dimensions with which we can look into our brain, the uncertainty decreases. There are multiple neurological disorders in which there are varying degrees of loss of awareness, they are diagnosed in a similar way.
What is the underlying mechanism of self-awareness? I concede it is not yet known. Areas within the brain and their interconnections whose lesions lead to loss of self-awareness have been identified but we still lack their functional description as to how they do so. However, this does not mean we will never be able to. Our is brain is much more complex than it was assumed previously and only now the human connectomics project had been able to map the anatomical connections between different areas of the brain. And that is just the anatomical description. A functional description will be more difficult because we don't have a non-invasive way of doing it.
There are multiple large scale networks within our brain, a special one being the default mode network. Anatomically this has been identified with self-awareness. Actually what is going on inside the network is difficult to know but we will get there eventually.
My problem is this: the brain is a chinese room for just as the person who doesn't understand chinese inside the chinese room, neurons too don't understand chinese. The Turing test employed, we'd have to conclude that the chinese room is a chinese person. However, is the chinese room conscious in the sense we are in that direct, immediate, non-inferential, self-evident sense? Is the chinese room a p-zombie in that it lacks that inner life philosophers of consciousness talk about?
Meaning being the most important one.
Too bad. Thanks!
Seems to me that I have to first know that I am self-aware. What does that mean? What is it like to be self-aware? Is self-awareness a behaviour, feeling, information...?
Gamma frequency oscillations in paralimbic network and the default mode network in our brain has been identified to cause self-awareness in us.
Quoting Banno
Yes, this is a good way of putting it. That's why early on in the thread I tried to make a distinction between saying that the person "passed the program instruction" and that same person understanding the Chinese language. There is a big difference and the way the scenario is worded is a gloss over this distinction.
Your missing an important component - the instructions. The instructions are in the room, along with the man, but are two separate entities inside the room. What "physical" role does the instructions play inside the brain if the human is the entire neural network? And isn't the entire neural network really the brain anyway? So you haven't coherently explained all the parts and their relationship with each other.
Neural networks weren't born knowing Chinese, English or any other language. The neural network had to learn those instructions, which means that the instructions were initially external to the neural network. How does a neural network acquire instructions for learning a language, and where do the instructions go when they are learned, understood, or known?
How did a neural network learn to do what it does? It doesn't perform the same function as other cells in the body. What allowed it to do what it does and not some other job that some other type of cell does?
Cellular differentiation is a result of evolution. There are multiple different cells each performing specialized functions within our body.
Quoting Harry Hindu
For a sufficiently complex neural network, the basic underlying physiology and anatomy remains the same for learning a language as it is for estimating a trajectory and throwing a ball. Take the example of the C. elegans neural network. It has been shown to learn to balance a pole. No separate instruction set was provided, only the reward was specified - in a the natural environment this reward will ultimately be the survival of the organism.
Sure, but can this technology scale up many orders of magnitude to simulate human brain? And just as importantly, is such a neural net simulation fully adequate? It may reproduce some behavior, modulo time scaling factor, but not so as to make the simulation indistinguishable from the real thing - both from outside and from inside (of course, the latter would be difficult if not impossible to check).
I am not committed to this view though - just staking out a possibility.
The start of your second sentence contains a contingency which contradicts the first.