The knowledge argument | Hacker News Reader

The knowledge argument

en.wikipedia.org

31 pointsHooke54 comments

To everyone insisting that the argument is silly: while in this particular case I do disagree with the conclusion, I'd encourage you to pause for a moment before dismissing it. I often find that when I approach philosophical theories with too critical a mindset, I dismiss it too quickly, and, if I spend the time afterwrd, find the issue more complicated than it appeared at first. It's something like being posed a coding problem, going "Oh, that's simple!", and then discovering later that the problem, is, in fact, quite complex.

On the topic of philosophy, in general, though, Bertrand Russell says it better than I can: "Two things are to be remembered: that a man whose opinions and theories are worth studying may be presumed to have had some intelligence, but that no man is likely to have arrived at complete and final truth on any subject whatever. When an intelligent man expresses a view which seems to us obviously absurd, we should not attempt to prove that it is somehow true, but we should try to understand how it ever came to seem true. This exercise of historical and psychological imagination at once enlarges the scope of our thinking, and helps us to realize how foolish many of our own cherished prejudices will seem to an age which has a different temper of mind."

Except it's not a good argument - if you believe knowledge is equivalent to experience. In order to understand what ammonia smells like I need to understand how the mind as a whole interprets the presence of ammonia. Once you have understood how the mind interacts with ammonia then you know what it is like to experience the smell of ammonia. You might say well, I can imagine how it must feel to understand how the mind works and my internal model of how it must be to understand how the mind works indicates to me that this would not allow me to experience the smell of ammonia. But you would be wrong. Just as the machine in the Chinese room argument understands Chinese and understands that part of the world described within Chinese.

The use of outlandish thought experiments is fraught with problems, not the least of which is philosophers' predilection to accept intuition and common sense in supposing what the outcome would be. This is particularly clearly demonstrated in Searle's Chinese Room argument, which is really just an attempt to disguise a personal opinion as a rational conclusion.

The fundamental problem here is that, unlike in science, philosophical thought experiments are not constrained by reality, so they tend to ultimately become self-referential arguments over which axioms to accept, but with no objective basis for deciding between them.

The argument seems to assume that "knowledge" is some sort of indivisible base class. "Knowledge" is shorthand for memories, compressed sensory input videos (plus smell, touch, and taste) of finite length in the brain. You can remember facts (beliefs, really) about wavelengths, and also the experience (believed experience, really) of sight. Not sure what there is to write a book about.

These philosophers seem to think that humans matter a lot more than those who "subscribe" to physicalism (if you can call believing in the least extraordinary option "subscribing") do. I don't care about all of those "the book is red" koans; "red" is just a word, and words are invented, primarily for convenience.

The only thing I find a bit hard to get from physicalism is why one's locus of consciousness happens to be in the brain that it is in. If the time I come into the world as a baby is "random", what "makes" it random?

I think the problem overloads the term knowledge.

First, there's knowledge about the external world, the kind which enables you to predict the conseqeuences of your actions and to manipulate the world to achieve desired results. Concretely, Mary will be able to use a, say, hex editor on a BW terminal to "paint" a picutre which a "normal" will recognize as a, say, red poppy flower. She manipulated another person into having the experience of seeing a RED poppy flower.

Second, there's knowledge about your internal world, how you experience things. If Mary is shown the poppy flower picture on a color monitor, she will gain new, experiential, knowledge about how her own body reacts to that particular stimulus. It's a knowledge, but it's not a knowledge that gives her abilities to manipulate the external world in new ways.

The premise of the experiment doesn't account for the fact that each individual is a sum of their past experiences, encoded physically in their brains in some way. One person upon hearing "red" will think about tomatoes, another about blood, yet another about poppy flowers. (I grew up at a place where they were common and we as kids even ate the seeds :))

Thus the premise is false, the fallcy is in the formulation "what goes on when we see ripe tomatoes"; emphasis on we. There is no collective "we" in this case; she doesn't know all physical information shee needs to know in order to manipulate the external world (e.g., knowing whether to draw a poppy flower or a tomato in order to evoke a particular emotional response from a particular individual).

One friend of mine told me once that no matter how much you read about sex, you have to practice it in order to understand it. We, as men, will never actually understand female orgasms and women will never know how ours are felt, no matter how accurate one tries to describe it.

I don't know what the above means in the context of "the knowledge argument" but these are easier to understand examples than "not experimenting the red color" in the article

Furthermore, each individual will feel it differently http://en.wikipedia.org/wiki/Qualia

:( don't talk as if your whole audience is men.

I am not exactly sure in what way his discourse seems as though it were addressed to an audience of men only. My political correctness meter is tingling.

> We, as men, will never actually understand female orgasms and women will never know how ours are felt, no matter how accurate one tries to describe it.

My position is almost the same as btilly's. When he says "We, as men, will never actually understand female orgasms and women will never know how ours are felt... " anaolykarpov is merely including himself in the more general set of men, and thus communicating an opinion as an element representative of that set, regardless of the audience. The disjunction with the set of women is necessary to underline the fact that there might be a fundamental qualitative/quantitative difference in the way orgasm is felt by the two.

In fact, the statement wouldn't have been any less correct had the audience been only women.

The same applies if a woman were to say "We, as women, will never ..."

Invalid complaint. The statement was grammatically correct if anaolykarpov and the friend having the conversation were both male. In that case "we" refers to the two of them, and not to those of us in the audience to whom that conversation is being recounted.

This kind of silliness is why I ignore philosophers whenever I can.

It is clear that the subject of how we personally experience X is not, strictly speaking, part of the body of knowledge about X. But the reverse is also true. Learning more about how we personally experience X does not increase our understanding of X. It is a different topic.

There is an important asterisk on this. And that is that we actually learn about things through our subjective experience. And therefore understanding our subjective experience can be necessary to get clear understanding of that thing.

This is true in a very literal way. "How long did that take?" If you give three people a stopwatch and have them record a common event, you will get 3 answers. And the answers will be consistently different in similar ways. In astronomy this phenomena is well-known as your personal equation. See http://en.wikipedia.org/wiki/Personal_equation for verfication.

But now I return you to philosophers arguing that something obvious is something deep that has consequences that it really does not...

>This kind of silliness is why I ignore philosophers whenever I can.

And avoiding this type of hasty comments are why it pays to study philogophical problems.

>It is clear that the subject of how we personally experience X is not, strictly speaking, part of the body of knowledge about X. But the reverse is also true. Learning more about how we personally experience X does not increase our understanding of X. It is a different topic.

That "it's a different topic" is a definitive statement that is not derrived from anywhere.

If anything, it is itself of philosophical nature, albeit a dogmatic one.

Who said that the actual experience of X doesn't increase "our understanding of X"? Even if it might be so, it remains to be argued.

Whereas the opposite, that experience, as an extra informational element regarding X, and as the main component of X as it concerns us, adds to our understand of X, is the obvious (correct or not) conclusion.

>* This is true in a very literal way. "How long did that take?" If you give three people a stopwatch and have them record a common event, you will get 3 answers. And the answers will be consistently different in similar ways.*

No, you wont, except within the margin of acceptable error (e.g. time to press the stopwatch button for each, etc).

First on the personal equation, the actual measurement difference from person to person could range up to 0.7 seconds. See http://www.jstor.org/stable/1627157?seq=1#page_scan_tab_cont... for verification. This was far above acceptable error for measuring the positions of the stars, and is far longer than most could believe possible.

As for "it's a different topic", well, it is. The one topic is the objective reality of how something works. The other topic is the subjective reality of how we experience that same thing.

Take the example that was given. Suppose that I tell you about electromagnetism. I tell you about the absorption spectrum of various colors. I tell you about the fact that when we process color we subtract red from green and yellow from blue so that we can never perceive a mix of those. I tell you about various forms of color-blindness, and the rare forms of tetrachromacy. I can tell you everything there is to know about how the process works, and what colors will be perceived, when. And how that differs for different people. You still know nothing about what it is like to experience it for yourself.

And conversely, learning to experience it for yourself tells you very little about everything that I just told you. For example you've been seeing in color for your whole life and likely are not color-blind. If so, were you aware that it is impossible for you to see a mix of red and green as a mix of red and green? Probably not!

Which returns me to my point. Subjective experience is a necessary filter through which we perceive the objective world around us. That experience is neither necessary nor sufficient to gain objective knowledge. And conversely objective knowledge is neither necessary nor sufficient to gain subjective knowledge. They are different kinds of things.

While this link is pretty good (as far as wikipedia philosophy entries are concerned), as always you should consult the SEP for a more in-depth discussion.

http://plato.stanford.edu/entries/qualia-knowledge/

And if you prefer your philosophical thought experiments to be accompanied by a pop-rock sound track, you're in luck:

https://www.youtube.com/watch?v=8Y37cAqdq0g

Obviously in any system that does perception there will have to be different classes of knowledge. Talking about the different mechanics of acquiring such knowledge seems like an odd way to attempt to disprove something called "physicality".

The question seems to be a mere quibble about the meaning of the word "knowledge". It presumes that there is something special about one of the ways people learn things. Does a recording video camera gain "knowledge" about the colour red when it records a scene with a red object in it? What, if anything, does the camera learn about the shape of a cube it sees? It can learn colour but it can't learn shapes. Should we conclude from this that shape is some sort of special knowledge outside of mere physicality?

"Knowing everything" implies far more than our intuitions may initially tell us, and can lead to "a set of all sets" type of paradoxical problems (i.e. the premise of the argument is absurd). This and other philosophical thought experiments like the Chinese room abuse our intuitions about being human, learning things, performing tasks, etc.

In this particular experiment, if Mary's omniscience allows her to invoke the specific pattern of neural activation that is triggered by observing a blue sky, then she won't learn anything new, otherwise she will. It all depends how far we take the absurd claim of her omniscience.

I'm guessing someone smarter than me has already thought of this, but imagine another alien scientist observing Mary through this entire experiment, with the ability to inspect at the neuronal level the state of her brain at any instant.

When Mary steps out of the room, we (humans considering this problem) would intuitively expect her to say "Wow, this sensation is like nothing I could ever have predicted based on my previous knowledge. I understand this in a way I couldn't have before!" How would the alien interpreter analyze the situation?

The alien may say that these sensations activate a different part of Mary's brain and alter the structure to create memories. If she were put back into the room, she could recall what it was like to perceive color. These memories could be defined as knowledge, but the alien might not consider that knowledge important. It might just say that is an artifact of the operation of her primitive brain that these sorts of neuronal patterns can only be created by perception with her eyes. From it's perspective, the entire interaction is still entirely physical. The alien could have predicted her reaction just prior to her stepping out of the room by looking at the structure and state of her brain.

The alien may add prior to her stepping out "By the way, she's going to say that experiencing color is different than knowing of it (qualia). This is a curious artifact the way her brain works." I think it would be impossible for me (a human) to prove the alien wrong. From the aliens perspective, this wouldn't be fundamentally different than me writing a program 'printf("I feel hungry");'. It would be difficult to convince me that the computer is actually hungry. It is programmed to say that.

The reason I added the alien is because we tend to empathize with Mary when we imagine the experiment. But, if consciousness is an illusion that our brain is programmed to tell, it's difficult for us to consider this problem objectively.

If a mind is what brains do, then Mary cannot know every physical fact about even her own neural system - that is too much information for her neural system to hold. This argument, therefore, implicitly presupposes that there is something more to a mind, and so is begging the question.

Certain types of knowledge cannot be encoded effectively.

What I mean by an encoding that's effective is one that translates my subjective experience of the idea or the thing so that you can "imagine" or "visualize" the thing by the mere act of decoding my encoding.

Experiencing the thing directly does not count as an encoding in this case.

Defining the precise meaning of "encoding" is left as an exercise to the reader.

So if you've never experienced seeing red, you can't "imagine it" by just reading or hearing about it.

If you've never experienced reading a Chinese sentence and "understanding it" effortlessly, you won't really know what it feels like exactly; you can only approximate the feeling by comparing with your own native language.

I often summarize it in programming: a mock class, or a very deeply copied instance, can not have the original `this` pointer and never will.

I don't think that's a good analogy. In your example, we still end up with a perfect "copy" of all the data and behavior.

When it comes to subjective perception, there's simply no way of even "copying" the information by simply writing it down. * (caveat follows below)

So, in the Mary example, no matter how much she "reads" about the color red, she really would not "know all there is to know" about it.

* Now it might be possible to encode a neural state on paper, and if you have a machine that can "apply" that state to your brain, then you can read that paper, enter codes in the machine, and let the machine "fix" some neural state in your brain so you can experience the same "thing".

At least it seems like this should be possible in principle. Although it might not be practical .. depending on how perception is "implemented" in the brain.

    Nagel takes a slightly different approach.
    In an effort to make his argument more adaptable and relatable, he takes the stand of humans attempting to understand the sonar capabilities of bats.
    Even with the entire physical database at one's fingertips, humans would not be able to fully perceive or understand a bat's sonar system, namely what it is like to perceive something with a bat's sonar.

My car emits a sound while parking in reverse. As the car behind gets closer, the sound is emitted more frequently, and becomes constant under a certain distance. When I'm parking a different car, I'm unable to 'feel' the distance from the next car. It's like being blind.

See also human echolocation (e.g. https://en.wikipedia.org/wiki/Human_echolocation). The experience, however, will not be like a bat's, not only because of the latter's greater acuity but also because humans are conscious, self-aware and have language to reason about their experiences.

I don't understand why this is a valid argument against physicalism. Maybe I can't get it because the definition of "physicalism" seems muddy and unintuitive to me.

Or maybe because there's an actual difference between knowing how something will work (light -> retina -> neurons etc.) and experiencing it while it works. But all of this is physical: both the knowledge, as well as the memory of experience.

That's the problem with many philosophers: they create some concepts, define them in arbitral (not necessarily adequate, often misleading) way and then other philosophers discuss with these concepts. IMHO most of the time it doesn't bring useful conclusions and doesn't advance human knowledge.

To assign complete knowledge to someone in a thought experiment like this is to open a can of worms that leads to counter-intuitive conclusions. The subjective experience of color vision is based on physical truths that extend beyond theory. If Mary truely possessed the knowledge of the receptor layout in her eyes, her exact neural layout, the composition of her specific signal substance soup, how a given input by extension would affect her emotional state etc. etc., she would be able to accurately foresee the impact of stepping out of that room. So, as the problem is phrased, she would not learn anything new.

She would also be a god.

On one hand, reductionists say that qualia (qualitative propeties of experience) is reducible to physical states. On the other hand, some (anti-reductionists) say, qualia is irreducible. How to settle this dispute?

Only theories from the object level domain (say, neurophysiology, vision, human experience) can settle this. When theories from the object level domain do not exist or when we have not understood enough about the phenomena, people engage is all philosophical disputes. Which philsophical theory one should choose? Pick your thought experiment:)

The same thing you see in moral philosophy: tons of thought experiments.

This is slightly off topic but it reminds me of a question I've pondered for years. Does every person see red as the same color in their mind? Obviously it applies to all colors. Furthermore - do I see color as the same "concept" as everyone else. I.e. to some color is more of a feeling, while to others it's no more than a property of an object (like the surface being smooth or rough).

I've also often wondered if this is why some people like some colors better than others - and why some people are said to have good taste when putting some colors together.

I don't understand how the thought experiment isn't simply begging the question. It states in its premises that (A) Mary already has all physical knowledge and (B) she later gains new knowledge. If the conclusion (that not all knowledge is physical) wasn't true, the premises could not be consistent.

Am I missing something obvious? Why should we accept that the conclusion is true, rather than discarding one of the premises?

Philosophers typically assume certain "fixed points" in arguments that they then refuse to budge from under any circumstances. It is a mental dysfunction, a kind of inbuilt rigidity of mind, that is peculiar to the breed.

For example, there is an actual argument in the philosophy of science that is based on the supposition that there is a substance that is "identical to water in every respect" but "is not H2O but rather XYZ".

Philosophers take this "argument" quite seriously, and get rather testy when you point out that the premise is a contradiction: if something has "all the properties of water", then that must include properties under electrolysis, and under electrolysis water decomposes to 2H and O, not X, Y and Z (whatever those might be). If you say this to a philosopher they will reply, "Yes, but what if it doesn't?" as if that was a coherent statement.

So the answer to your question is, "Because if we allowed ourselves to discard one of the premises we'd have a much harder time motivating this whole ridiculous discussion, and since our livelihood depends on such ridiculous discussions, we're not about to let that happen!"

Without disputing your main point about how philosophers typically behave, I want to clarify the is-water-H2O argument. As I understand it, it's making a point about how we use language: people have been using the word "water" to refer to various puddles, lakes, oceans, beverages, etc for a long time before electrolysis was discovered. We know now that all those instances have being-H2O in common, but it's not a necessary property for the word "water" to apply: if some other chemical fulfilled the everyday requirements for water (being clear, wet, thirst-quenching, etc) without being H2O, it would be "water," and if a particular variation of H2O doesn't fulfill those requirements (^2H2O, heavy water, is poisonous) then it's not water. See "Water is Not H2O" (Michael Weisberg, 2003). http://www.phil.upenn.edu/~weisberg/papers/waterfinal.pdf

I have a suspicion that many philosophical arguments, probably including the knowledge argument, are ultimately about language use - in this case, words and phrases like 'knowledge', 'physical' and 'physical knowledge'. Is this what Wittgenstein was going on about?

Yes, philosophers have such arguements: Water on some alternative universe without H2O structure vs water on our earth with H2O structure.

Even here, the issues are related to metaphysics, not so much about chemistry:) There is a jump from empirical world (the world we are in) to possible worlds. Here is the article: Is water necessarily identical to H2O, by David Barnett http://spot.colorado.edu/~barnetdb/my_papers/Barnett_Water.p...

> "Because if we allowed ourselves to discard one of the premises we'd have a much harder time motivating this whole ridiculous discussion, and since our livelihood depends on such ridiculous discussions, we're not about to let that happen!"

So saith the Amalgamated Union of Philosophers, Sages, Luminaries, and Other Professional Thinking Persons, eh?

This seems to be more about a limitation of language than a problem with the logic. Using "knowledge" in both premises looks clumsy. But looking at it purely conceptually, where Mary would intellectually understand the complete theory in the B&W room, and then gain new experiential insight when leaving the room, it seems to at least prove that experiential knowledge is inherently different than theoretical knowledge.

That's more what confuses me about the structure of this argument, though - I would take the experiential insight to be more "physical". So I still don't understand what they mean by physicalism.

If Mary only knows about colours "intellectually" then she might be a SHRDLU for colour, knowing the calculations of blocks symbolically but not knowing blocks and gravity, having never experienced it. http://hci.stanford.edu/winograd/shrdlu/

That's exactly the point - she has all physical information. If you could show that some piece of information fits the definition of "physical information", then it's assumed that Mary has it, and someone defending the knowledge argument would have to agree.

The point is that even IF you give Mary all that information, she's still missing some information. Then, that information must not be physical information. More to the point, there's a difference between the points you presented: In (A), you say Mary has all _physical_ knowledge (Jackson uses the word information, but the difference is unimportant). In (B), you note that she gets new _knowledge_. It follows that this knowledge must be non-physical, since we assumed that she had all physical knowledge, and yet, she didn't have this new knowledge.

But this looks identical to the "immovable object vs. unstoppable object" paradox. It presupposes something that doesn't actually exist.

Experience is information. It doesn't make sense to imagine a situation in which she has all the information about a thing, except experience. Either she has all of the information about it, or she doesn't.

Let's take the claim "experience is information" to be true (I've yet to find a reasonable definition of "information", so I agree).

Jackson's point here is specifically to show that there are at least two kinds of information: physical information, and information that is not physical. Bearing that in mind, what do you mean by: > "Either she has all of the information about it, or she doesn't."

What I mean is that all information is physical information -- I'm rejecting outright his starting premise, because no amount of research on wavelengths or poetry will excite the exact same neurons in your brain that are excited by actually seeing a color.

I suppose that my objection falls under 4.2 in http://plato.stanford.edu/entries/qualia-knowledge/, specifically, metaphysical physicalism: "there are no non-physical individuals, properties or relations and no non-physical facts".

What would be a good example of non-physical information?

That's not what I'm asking about. This thought experiment is meant to argue that not all knowledge is physical, right? Set aside any notion of what those terms mean and consider it in the abstract. It sets out to argue that "Not all X has property Y", and then asks us to consider a hypothetical situation with the premises "Assume that Mary has all X with Y" and "Assume that later she finds more X". The thought experiment doesn't support the claim that not all knowledge is physical, it requires it - i.e. the definition of begging the question.

Perhaps the point here is that "Mary gains new knowledge" is supposed to be so manifestly obvious that it doesn't need to be argued? If so, this seems far from obvious to me.

If Mary is never allowed to actually see the colour of objects in her room, she is being deprived of a physical fact regarding colour vision, and premise 1 doesn't stand. Put another way, how is Mary's later epiphany to be induced in her but by physical means? She must physically experience (photons in eyeballs) colour.

Can someone explain why the whole thing isn't simply the abuse of the universal quantifier? It SOUNDS like there are two "alls" here - ALL[0] is a proper subset of ALL[1].

If set theory is an obscene tool for this subject, I promise to rend my garment and heap ashes on my head. But only in an aphysical manner...

I think Daniel Dennett nailed it.

https://www.youtube.com/watch?v=X8aWBcPVPMo

That's a joke. Not exactly a dismisal of the argumentation any more that a creationism making a joke is a dismissal of evolution.

Was OP by any chance at the recent San Francisco premiere of Ex Machina, a film which briefly mentions this thought experiment?

Odd -- contrary to other users' comments, I don't find this topic silly at all. As I see it, "science" is the ability to predict the future†. That's all. And thus a better prediction corresponds to better science. For instance, quantum electrodynamics has been used to calculate the gyromagnetic ratio of an isolated electron to 10 digits of experimental accuracy. That's incredible. In terms of predictive capability, the order goes: physics > chemistry > biology > psychology. That's not to say that any field is less valuable than any other; it's just that the higher level fields are more complex, and thus it is harder to formulate predictive models for them.

So if the value of science is its predictive capacity, then how does one verify that capacity? Through reproducibility. This property is essential to modern research; all scientific results must be able to be reproduced by anyone (assuming they have the appropriate equipment and the ability to adhere strictly to a specific procedure). For if reproducibility were not a requirement, then science would be of little value.

For some reason, a lot of scientists (or maybe just people on the internet who like the idea of science) pigeonhole themselves into the philosophies of positivism, physicalism, and materialism. I don't know why. I do research in quantum chemistry, and even as I learn more about the subject (and quantum field theory), no increase in knowledge has ever changed my philosophical stance. The only thoughts that run through my mind as I learn more are "Cool, now I can make better predictions" or "Hey, that's kind of elegant that nature works that way".

I find it odd, then, that so many people derive their entire life philosophy from sensory input corresponding to output from other humans. It's weird to me. If one were to plant a microchip that affected optical nerves into the brain of one of these people, and the microchip caused them to see an object that wasn't there, I would hardly be surprised if they denied seeing the object at all once they realized the object's visibility wasn't "reproducible".

Is it really the case that so many believe that what is real is not only what is personally experienced, but what is also supposedly personally experienced (and thus communicated) by others?

Playing devil's advocate, assume the simulation hypothesis (I don't actually believe it, but it's useful to illustrate my point). Furthermore, suppose that each individual human has been fed sensory input that only they are capable of experiencing, but the input is both personally predictable and reproducible. And lastly, the program running the universe does not allow any chain of interactions to exist that would permit the information describing these individual experiences to be transmitted to other humans.

What's wrong with this? Nothing. It's perfectly reasonable that every person in the simulation observes aspects of the universe that are unique to them, and them alone. It meets the "predictive" requirement for science at an individual level, and while it's personally reproducible, it's not consensually reproducible. Does that really make it any less real?

At least for me, I'm confident that physicalism is false. Why? Because if I were to rank everything I am certain about on a scale of most certain to least certain, at the top of the list would be "I am certain that I am experiencing my own existence". I cannot think of anything I am more sure about. Take away my senses (or manipulate them somehow), but as long as my brain is still functioning and I'm conscious, I'm still experiencing. The universe didn't have to be this way. It could have existed almost like it does now, except that all the people in it would have been philosophical zombies. (And perhaps all of you are, which would make me a solipsist, but at least I'm a solipsist that knows I'm not a p-zombie then.)

† My claim that "science is the ability to predict the future" and "better science corresponds to better predictions" is a semantic issue, and a whole debate could center around developing a consensus on good definitions. Wikipedia lists other attributes of the word "science" in addition to predictability. If you don't like my definition, make up a new word and substitute it for "science"; the argument holds regardless.

Just to summarize the article:

People doing a "non physical" experiment, while using physical proof methods.

This is scientific comedy at its finest, people.

Yeah, because professors of philosophy and epistemologists toil for years about the subject, but some 2-minute comment shows it for what it is...

You even got the facts wrong. They are not doing a "non physical experiment". They are proposing a thought experiment -- you know, the kind lots of scientists did, Einstein and Shroedigger among them.

The "non physical" part is just that it tries to establish that there is "non physical" knowledge (qualia). That is, nothing about the experiment as a process (as opposed to the output) is necessary to be "non physical"...

> They are proposing a thought experiment -- you know, the kind lots of scientists did, Einstein and Shroedigger among them.

Neither Einstein nor Schrodinger "did" thought experiments. They used thought experiments (which is a really terrible name for imaginary scenarios) to motivate arguments and guide thinking, not to demonstrate anything.

No one who has been paying attention to the history of knowledge uses thought experiments for anything other than illustrative, explanatory or motivational purposes. In particular, the conceit that imaginary scenarios can teach us anything about the way reality actually is long outdated, as no such imaginary scenario has ever done any such thing.

Relativity was motivated by thought experiments, but so too have been numerous false ideas. Therefore thought experiments can do nothing to distinguish false ideas from true ones, so anyone who uses them for that purpose is doing them wrong.

Mary's Room is a thought experiment that attempts to establish that there are non-physical properties and attainable knowledge that can be discovered only through conscious experience.

From the beginning the argument pre-supposes there is objective knowledge that can learnt. As we know from relativity and quantum theory, there is no objective truth. The location of every atom, cell or neuron is unique to every observer.

Reading the RGB encoding of a photo, one number at a time, compared to seeing the photo with your own eyes, are different experiences. Imagine if each of our senses are different dimensions of perception, different ways of learning knowledge (reading about color through a book vs experiencing color with eyes) are orthogonal to each other. You can only connect those experiences and imagine colours after you've experienced it with your own eyes - your brain can "fake" perceptions, but only after having experienced the same kind of perception previously.

This blind man regained vision at 68:

When he went to the corner store with his sister, he handed the cashier a $10 bill when he really should have handed a $5 bill...The idea of remembering things visually is hard as well, and he still resorts to touching and feeling the world around him to store those memories. Of course while blind he'd been able to differentiate between $5 and a $10, but once he's regained his vision the touch-knowledge isn't connected to his vision-knowledge immediately. http://www.dailymail.co.uk/news/article-2382619/Canadian-man...

Knowing how neurone in our brains work is different to having those neurons interacting with each other.

I argue "knowledge", if exists, can be interpreted only through the filter of qualia. No human knows anything without experience. Even reading a page of C code can only be done through our eyes and our mind and compared through previous experience in reading the same. Our minds are qualia, period.

As this was a philosophical argument I'll post a poem from a philosopher.

    Thus Something and Nothing produce each other; 
    The difficult and the easy complement each other; 
    The long and the short off-set each other; 
    The high and the low incline towards each other; 
    Note and sound harmonize with each other; 
    Before and after follow each other.

http://wengu.tartarie.com/wg/wengu.php?m=NOzh&l=Daodejing&no...

With your babel fish you listen to an alien recording "A tall being and a short being once stood here". You wonder, having never met or seen any sign of aliens besides this audio recording, how tall and how short were those beings?

What grounds are there for excluding "how the knowing subject that is an emergent property of the matter of the brain experiences colour" from "all knowledge about colour"?

fenomas correctly points out the argument is begging the question based on the I've raised here, but the argument is even more deeply flawed than that because it assumes that "knowledge" is a thing rather than an activity of a knowing subject.

If you take the epistemology of a knowing subject seriously then knowing is an activity of the subject, with the ontology of an action rather than a thing. Activities are not conserved in the way things are. It is easy to slip past what "know everything there is to know" means when we assume (incorrectly) that knowledge is thing-like: we can image books full of sentences that exhaustively list laws and properties. One might say that if Mary knows Maxwell's equations and a long list of information about matter she might "know everything", but considered as an action, to "know everything" means "to be able to have any thought that anyone might be able to have about the thing", or something similar.

From this, we can immediately draw an important inference: Mary is not a human being, so this is not an argument about human beings. Not even in the wildest philosopher's imagination could any human being be able to have every thought about light that could be had. There are no-doubt "single thoughts" of such enormous length and complexity that they could not be fit into a human lifetime. And because the actions of the mind are serial (mostly) rather than parallel, the absurdity of the proposition that we can "know everything there is to know" is revealed. Human life is simply too finite for it.

Any expert will attest to this: no matter how much we learn, with ongoing experience there is always more.

So like all imaginary arguments, the primary purpose of this one is to mislead us about the subject. It isn't about human beings or human knowledge. It is about some other kind of being, unrelated to humanity. Which I guess might be interesting if you're in to that kind of thing, but personally I'd rather spend my time focused on the world that exists, not the world of some philosopher's imagination.

If the argument were honestly stated it becomes a tautology:

"Imagine a knowing subject that is capable of thinking every thought about a given subject, which necessarily includes all the thoughts a knowing subject might have when it physically interacts with the material reality the subject describes (otherwise it wouldn't be able to think every thought about the subject). Is such a knowing subject capable of knowing the thoughts a knowing subject might have when it physically interacts with the material reality the subject describes?"

Reply on news.ycombinator.com