• A Raybould
    86


    ↪A Raybould
     What (if anything) Mary learns from seeing colors is either discursively-learnable or it is not. We can consider these two cases separately.

    OK, so this argument puts off the question of accepting whether or not Mary learns anything [my emphasis.]
    jkg20


    I am puzzled by this. The claim that Mary learns something from seeing colors is a premise of Jackson's argument, and yours too, if I am not mistaken. We only reach the two cases here if that premise holds up; if it does not, the argument has already failed.


    I guess against this argument, the skeptic would have to push the investigation of the supposed distinction between knowledge being discursively learnable and knowledge being physically encodeable.jkg20

    It is not clear (at least, to me) to which argument you are referring, or what 'push the investigation' means. The meaning of 'discursively learnable' and 'physically encodable' seem to be straightforward, at least within the context of Jackson's argument:

    Discursively-learnable: Knowledge that can be learned by discourse, which is to say linguistic communication [1].

    Physically-encodable: Claude Shannon developed the canonical theory of information, in which information is present whenever things are arranged one way when they might be another, and any specific piece of information is encoded in such differences. Physically-encoded information is information that is actually encoded in physical states that might be otherwise - a coin that is heads-up when it might be tails-up, for example - and physically-encodable information is that which could be so encoded. I suppose that any information can be physically-encodable, precisely because the medium of encoding is immaterial (pun intended) to the information content.


    So, a premise of the Churchland/Crane/whoever argument is that there is a distinction between knowledge that is discursively learnable and knowledge, the content of which is entirely encodeable in a physical medium.jkg20

    To be clear, the distinction is not one of mutual exclusion; in fact, if all information is physically-encodable, as I suggested above, then discursively-learnable knowledge is a subset of knowledge having physically-encodable information content. The distinction is that they are two different aspects of knowledge.

    It seems odd to call this distinction a premise of the argument, as it is just a matter of defining two terms, and it is simply a fact that they have different definitions; furthermore, denying it would not seem to help Jackson's case.


    Content is propositional. Even if one accepts the idea of non conceptual content, that is not to accept the idea that there is content that cannot be expressed in terms of propositions, it is just to accept that the person having the mental state with that content need not have the resources to state those propositions.jkg20

    This is a very big assertion, and the justification you offer here does not even begin to do the job. It is not just a possibility, but a fact, that some people have a pathology that prevents them from stating something they know; nevertheless, you cannot deduce that content is propositional from that - it is simply an invalid argument, one for possibility where one for necessity is necessary. As we shall see, it is also moot in this case.


    So if all content is propositional, then not only is all discursively learnable knowledge physically encodeable, all physically encodeable knowledge is expressable as propositions and so is discursively learnable[my emphasis]jkg20

    The italicised conclusion does not follow, even if the conditional here were to be true, as, to be discursively-learnable, knowledge must be communicable. For one thing, note that, in what you said to justify saying that all content is propositional, you made the case that people might have propositional knowledge that they cannot state, which would seem to be an insurmountable impediment to it being communicated!

    Furthermore, if this argument were valid, then it would apply to all knowledge, not just all physically-encodable knowledge (if there is a difference), and so all knowledge would be discursively-learnable, which is the other horn of the dilemma: In that case, what would have prevented Mary from learning it prior to her release?


    the Crane/Churchland line seems to force the skeptic to a position in which they will have to defend the following modus tollens:

    1 If some item of knowledge is physically encodeable, then it is discursively learnable

    2 The item of knowledge Mary gains is not discursively learnable

    _________________________

    The item of knowledge Mary gains is not physically encodeable
    jkg20

    It is not clear to me that anyone has to defend this argument, but, as we have seen above, your current argument for premise 1 does not succeed, even on its own terms. 


    [1] On writing this, it occurred to me that it might be disputed whether anything Mary deduces is discursively-learnable. I think it is, because the deduction process itself could be expressed in words and communicated. The issue is moot as far as Jackson's argument is concerned, however, as completed physics already includes everything that can be deduced from other facts - in fact, Jackson stresses this point in his papers, as his argument depends on it.
  • jkg20
    405
    The claim that Mary learns something from seeing colors is a premise of Jackson's argument, and yours too, if I am not mistaken.
    You are mistaken. My argument that you are referring to contained the conditional premise that if one accepts that Mary gains knowledge, then there is an obligation to say something substantive about what she gains. Jackson's argument states that Mary gains knowledge. But your misunderstanding on that point is a peripheral issue anyway, since we have now moved on to a different, and more fundamental, disagreement. Let's try a new argument:
    1: All information content that is encodeable is expressable as a finite set of propositions.
    2: Any finite set of propositions is discursively learnable.
    3: From 1 and 2; all information content that is encodeable is discursively learnable.
    4: Anything that is physically encodeable is information content that is encodeable.
    5: From 3 and 4 anything that is physically encodeable is discursively learnable.
    6: At time t, Mary knows everything that is discursively learnable.
    7: Mary gains new knowledge when at t1 she sees a red thing for the first time.
    8: From 5, if what Mary learns at t1 is physically encodeable, then it is discursiveley learnable.
    9: From 8 and 7, if what Mary learns at t1 is discurvively learnable, then she did not know at t everything that is discursively learnable.
    10: From 8 and 9, if what Mary learns at t1 is physically encodeable, then Mary did not know at t everything that is discursively learnable.
    12: Since the consequent of 10 contradicts 6, by modus tollens what Mary learns is not physically encodeable.

    The premises here are 1, 2, 4, 6 and 7.
    Nothing you have said would lead me to think you would reject 2.
    Your remarks about the notion of physical encodeability suggest to me you would also accept 4.
    I'm not sure whether you accept 7 or not. You have not stated explicitly that you do, although many of your remarks seem geared up to defend physicalism even if 7 were true. Nevertheless, without explicitly accepting it, I am not going to force it upon you. As I have said a few times, the motivation for it is just an intuition pump.
    I am also not clear whether you allow the possibility of premise 6 being true or not. It may be incoherent, although I do not see any obvious contradiction.
    So, this leaves 1. Why accept 1? Well it is totally in accord with the information theory that Shannon developed. True, Shannon's original paper talks about the transmission of messages rather than content, but it is not a stretch to identify the term "information content" as used in the argument above with "message" as used by Shannon, and a message is very definitely something expressable in a finite set of propositions. Having said that, Shannon takes the notion of a message pretty much for granted.
    Are there counterexamples to 1? Note that it is not a counterexample to point out that there might be people with contentful states that are pathologically unable to express that content. That some content cannot actually be expressed by a some specific individual in a set of propositions does not entail that the content is not expressible by such a set. In fact, if you tried to come up with a detailed counterexample of this type, it would probably be self stultifying, since you would have to identify, through propositions, what the content of the state was that the person was unable to express.
    Might there be unecodeable information? Sounds like that might be a contradiction in terms, and in any case, it is not really a line a physicalist would want to push. Might there be information content not encodeable in a finite set of propositions? Well, transfinite information theory does not exist as far as I am aware, but that may be ignorance on my part.
  • A Raybould
    86

    The claim that Mary learns something from seeing colors is a premise of Jackson's argument, and yours too, if I am not mistaken.
    You are mistaken. My argument that you are referring to contained the conditional premise that if one accepts that Mary gains knowledge, then there is an obligation to say something substantive about what she gains. Jackson's argument states that Mary gains knowledge.
    jkg20

    You say I am mistaken in saying that it is a premise of yours that Mary learns something from seeing colors, but your premise 7 is that "Mary gains new knowledge when at t1 she sees a red thing for the first time." Do you make a distinction between "learns something" and "gains new knowledge"? If so, what is it? Alternatively (or additionally), do you suppose that her new knowledge is not a consequence of seeing red - that these two events are just coincidental? I do not suppose you do, as this latter supposition would invalidate the argument, but I am at a loss for why you think I was mistaken.

    Also, note that "new knowledge" is ambiguous here. It is new to Mary, but as Jackson goes to some length to point out, it is not new as in having just become true at t1, as, for his argument to succeed, it must have been a fact that Mary could have learned prior to t1, and one cannot learn non-true facts, as there are no such things. This is the point behind Jackson's insistence that what she learns is a fact that was already true about other people.

    As for what you thought I was referring to, your "conditional premise that if one accepts that Mary gains knowledge, then there is an obligation to say something substantive about what she gains": I have said something substantive: either it is discursively-learnable, or it is not. As Jackson's argument fails in either case, a skeptic is under no logical obligation to choose one over the other.

    If you think this is insufficient to meet the obligation, then it should be possible for you to show that there is a logical obligation to say more, by showing that the counter-argument is invalid without it. This you have not done.


    Now, to your latest argument. There are quite a few claims that I could quibble over, but I will focus on the ones that are decisive.


    1: All information content that is encodeable is expressable as a finite set of propositions.

    To avoid equivocation later on, we have to be careful here over what it is about the information that is expressible as propositions (especially given the distinction you made between 'expressible' and 'can be stated' in your previous post.) The physical state encoding any specific case of physically-encoded information can be described (in the language of completed physics) but, to be clear, the description is not the state (the map is not the territory), and even though the description may itself be physically-encoded, the physical state in which that description is encoded is not the physical state which is described by that description - for example, the physical state being described can change without the physical state encoding the description changing. Therefore, I do not accept this premise, though I will accept this:

    1': The physical states encoding all physically-encoded information can be described in a set of propositions. 


    2: Any finite set of propositions is discursively learnable.

    I will accept that; as this is a thought experiment, we can put aside the objection that Mary's ability to simply remember propositions is limited.


    3: From 1 and 2; all information content that is encodeable is discursively learnable.

    Once we substitute 1' for 1, it is clear that this does not follow; the most that one can deduce from 1' and 2 is this:

    3': Any description (in the language of physics), of any physically-encoded information content (of someone's knowledge), is discursively-learnable.


    4: Anything that is physically encodeable is information content that is encodeable. 

    I cannot accept this, either, as only a small amount of all physically-encodable information is the information content of knowledge. For example, the angular momentum of some undiscovered exoplanet encodes some sort of information, but it is not the information content of anyone's knowledge. For a bit of physically-encoded information to be part of the information content of someone's knowledge, it must be encoded by a physical state within that person's brain. If Mary could have knowledge which had its information content encoded in someone else's brain, she would have telepathy, and there would be no justification for the belief that, before seeing colors herself, she could not know what it was like for other people to see colors (this is another manifestation of the dilemma faced by Jackson.)

    From here on out (if not before), the deductions fall like a house of cards, and I can no longer even say "here's a valid deduction that is similar in spirit to yours."

    In summary, this argument fails because it equivocates between two different things: on the one hand, knowledge of the physical state encoding the information content of some other item of knowledge, and on the other hand, that other item of knowledge itself.
  • jkg20
    405
    Therefore, I do not accept this premise, though I will accept this:
    You make it sound as though what preceded this sentence was an argument against premise one of my argument, but it was not. All you do is point out a pretty obvious, iterative distinction between a state and its description. That is not a counterexample to premise 1, nor any kind of reason to reject it. If you want to reject premise 1 you need to do better. Premise 1 seems perfectly in harmony with what you yourself referred to as canonical information theory.

    Against premise 4 you state:
    I cannot accept this, either, as only a small amount of all physically-encodable information is the information content of knowledge.
    But that does not speak to premise 4 at all, within which there is no mention of knowledge, it is a metaphysical, not epistemological premise. The only epistemological premises of the argument are 6 and 7. Up to premise 4 all the issues are metaphysical. Note that the background assumption of the argument is a realism about information content, of course: i.e. there can be information content that is not the content of any actual mental state. But why would any physicalist object to such a realist assumption?

    In summary, this argument fails because it equivocates between two different things: on the one hand, knowledge of the physical state encoding the information content of some other item of knowledge, and on the other hand, that other item of knowledge itself.

    No it does not. You have misunderstood that premises 1 and 4 are metaphysical not epistemological in nature. Of course, if you want to discuss the possible collapse of that distinction, go at it, but it is probably not something a physicalist should attempt to do.
  • jkg20
    405
    (especially given the distinction you made between 'expressible' and 'can be stated' in your previous post.)
    That is not the distinction I made. Please reread the post.
  • A Raybould
    86


    ↪A Raybould
    (especially given the distinction you made between 'expressible' and 'can be stated' in your previous post.)
    That is not the distinction I made. Please reread the post.
    jkg20

    I did, and I see that it can be read in the light of these being synonyms. I assume, therefore, that you have no objection to 'can be stated' being substituted for 'expressible' everywhere the latter is used in your latest argument?
  • jkg20
    405
    At the risk of falling into a trap: no objections. :wink:
  • jkg20
    405
    Actually, "could be stated" would be better as a synonym for "expressible". Sorry to split hairs, but there are nuances of difference between "can" and "could" that might make a difference in these kinds of debates.
  • A Raybould
    86

    Therefore, I do not accept this premise, though I will accept this:
    You make it sound as though what preceded this sentence was an argument against premise one of my argument, but it was not. All you do is point out a pretty obvious, iterative distinction between a state and its description.
    jkg20

    I am glad that this, at least, is obvious, but there is an implicit question here: suppose we have some system having information on account of being in physical state ψ. if the corresponding propositions you have in mind are not a description of that state, then what are they? It is far from obvious that such propositions must exist, whether as a consequence of Shannon's information theory or of anything else. As things stand, proposition 1 lacks any justification.

    Against premise 4 you state:
    I cannot accept this, either, as only a small amount of all physically-encodable information is the information content of knowledge.
    But that does not speak to premise 4 at all, within which there is no mention of knowledge, it is a metaphysical, not epistemological premise.
    jkg20

    Premise 4 is "Anything that is physically encodeable is information content that is encodeable." Given the context, I took "information content" to mean the information content of knowledge. If it is not that, then what is it the content of?
  • A Raybould
    86
     
    The SEP is an excellent resource, and in section 2 we have a discussion showing that it is not straightforward to find an inverted spectrum scenario that is clearly behaviorally-invariant (and as we will see, my argument is not defeated by the existence of some scenarios that are.)
    — A Raybould
    You're confusing your opinion with your argument. The stuff in section 2 is a different argument than what you've presented...
    InPitzotl
     
    Indeed it is, and on the contrary, you seem to have overlooked (or, as you would put it, "are confused about") the relevance of this issue. It goes back to your attempt to claim that the possibility of behaviorally-invariant inverted spectra somehow invalidated my hypothesis, and I pointed out that mere possibility would not do; at the very least you would have to show actual prevalence. This section shows that, as a matter of fact, undetectable inverted spectra cases are sparse among all possibilities, and an explanation of why they are does not, of course, dispute the fact that they are. 

    Furthermore...
    some of the arguments in that section are in fact decent and relevant, but they do also presume things about color processing for which we really need more detail... Since these are L1 like properties, inverting the mappings from L1 to L2 colors may still carry these propertiesInPitzotl
     
    I see that you have no hesitation over needing more details when you wish to speculate on how things might be - but if you want to argue against my hypothesis with a speculation that is consistent with it, but which we would have no justification to believe if the mappings routinely differed between people as a first-order effect, then go ahead!
     
     
    "Too vague" seems to have become your default response, but by itself, it is too... vague?
    — A Raybould
    It's not exactly a default response so much as it is prompted:
    InPitzotl
     
    On the contrary, as I pointed out in several cases, you have attempted to manufacture that impression by quoting out of context. And so it is here, also; you did not quote the rest of my response here, but only part of it, and elsewhere. For the record, here it is:
     
    Are there any non-vague metrics of vagueness, and more-or-less objective thresholds? The falsifiability criterion of science provides one (despite not being directed specifically at vagueness alone), and it is an appropriate one here: Vague claims (such as of vagueness exceeding some unspecified threshold) are not falsifiable.
     
    What I am saying here could be falsified; a (not-yet-achieved) causal model of how the human brain produces a mind would reveal whether, to a first-order approximation, most humans are undergoing the same physical functions when they experience color. That's good enough for a hypothesis.
    A Raybould
     
     
    A more parsimonious expectation, however, is that this multi-generational training has produced brains that function alike, to a first approximation. If they did not, how likely is it that they would have, so to speak, 'learned the lessons' imparted by evolution?
    — A Raybould
    ...in relation to the topic at hand, this is indeed too vague. Your expectation is more parsimonious than what exactly? Function alike in what ways? What "learned lessons"? I'm perfectly happy to say that human brains evolved in human like ways, but that does not really imply same-experience unless you can connect the similarity of human evolution to the similarity of color experience, which I've yet to see. Other than that, it's yet another nature versus nurture debate. Truth is, both nature and nurture make brains, especially human brains.
    InPitzotl

    Raising questions about specific points, as you have done here, is much more to-the-point than vague claims of vagueness, and I have no problem with answering them. The "learned lessons" are evolutionary adaptations. My expectation is more parsimonious than the hypothesis that first-order differences, between individuals, in their high-level functional responses to color stimuli, are pervasive. It is more parsimonious because the latter has a problem with ensuring that, of all the possible ways that a child might differ from its parents, only the ways that are behaviorally-close seem to arise. As for nature vs. nurture, nurture is not going to 'correct' a child into thinking that her tongue is red if her experiential correlates to how she sees her tongue maps close to those of grass and cucumbers, and distant from those of tomatoes, pimentos, and the setting sun. 


    My point is based on empirical fact: if variation is routinely producing children that have markedly different functional responses to color stimuli than their parents, then how come we only very rarely see those variations that do not result in observable differences?
    — A Raybould
    Because, for example, predators aren't examining your brain with fMRI to see if you represent redness on this spot or that spot or using this average frequency of pulses or that frequency? The important thing from a fitness perspective is that you run away, hide, or fight the predator appropriately.
    InPitzotl
     
    This is a non-sequitur. There will, in general, be behaviorally-observable consequences if human brains routinely have first-order differences in their functional responses to color stimuli, unless, mysteriously, this variation can only occur in the relatively few ways that are behaviorally inconsequential.
     
     
    The alternative is so lacking in plausibility that I will not bother to reply until specific arguments for it are presented.
    — A Raybould
    That's your choice, but in effect, given that you're the one claiming to be supporting a proposal that the experiential correlates have a fitness advantage, not answering the question absolutely equates to not addressing the very thing your proposal is supposed to be about. If your proposal is about a tie between fitness and particular experiential correlates of color, it is backed if and only if you can demonstrate what it is about... i.e., tie fitness to experiential correlates of color. This is what I mean by relevance [my emphasis.]
    InPitzotl
     
    Again, some relevant context is missing:
    I do not think it is very speculative to say that our mental abilities are strong determinants of fitness (unless, of course, one thinks our experiences are epiphenomenal.)
    — A Raybould
     
    In what way?
    — InPitzotl
     
    The alternative is so lacking in plausibility that I will not bother to reply until specific arguments for it are presented. Extraordinary claims require extraordinary evidence.
    A Raybould
      
    The premise that, in general, brain function is adaptive, is just one premise in my hypothesis. If you want to hang your case on refuting that particular premise, I rest my case!
      
    Oh, you're not taking a position on the issue? Well, then I am not putting anything at risk by not responding until it is more than just creationists who are making that claim.
      
    As for the italicised statement, this is an inaccurate paraphrase of my position, and therefore not germane. The evolutionary issue is just one point in an argument that involves ontogeny, the mechanism of inheritance, and the scope of variation: it is that if variation routinely produced first-order differences between offspring and their parents, adaptation would not occur.
     
     
    For physicalists, where there is no physical difference, there is no difference simpliciter.
    — A Raybould
    Sure. So let's focus on the correlates, since that's where the difference would be.
    And what are those correlates?
    — A Raybould
    I don't know, but I think that's the key question. In terms of L1 colors, there are reasonable explanations of development that are works in progress but involve self organization; these generally produce opponent color processes. Example:
    https://www.frontiersin.org/articles/10.3389/fncir.2014.00016/full
    This itself requires a bit more study. In terms of color perception in the brain, as I said before, we know there are multiple places of analysis. But there's also the possibility of potential signal space (such as these toy models of potential Poisson rates)... and since the signals potentially start this way there may be transference from these signal rates to positions, or not. But "qualia" may be more complex than many philosophers tend to treat it as well; e.g., there's such a thing as pain that doesn't hurt, suggesting that qualia aren't necessarily singular in the first place. This might be explained as pain in practice actually having separable pieces that typically are co-associated but not essentially so, and the same could possibly be true of color. What we're looking at developmentally for different-experience would be any sort of theory that starts with this brain in a semi-random developmental state and just "settles into" the "nearest corners".  The entire question here (that has not yet actually been addressed) is how many "corners" are there, why are there that many, and how do the L1 colors map into them? Assuming opponency-sized "corners", a same-experience theory would postulate four to six of them, and some sort of specific developmental pathway whereby particular opponent signals latch into the specific four (or six) that they're supposed to map to.[my emphasis.]
    InPitzotl
     
    I will just note that this is no less vague than what I have been saying, but only to make a point about your mode of argumentation: what you have said here is perfectly reasonable speculation about how things might be, and it also satisfies the standard of falsifiability.
     
    But more substantively, I admit that I am puzzled by why you would think the emphasized sentence would present a problem for my hypothesis - or if it does not, why you think it is relevant. Also, how might the developmental pathway distinguish between the latchings that are supposed to happen and those that are not? Are you  proposing that there could be pervasive first-order differences in how this occurs in most individuals? If not, at what point could these differences manifest themselves?
     
     
    What I am saying here could be falsified
    — A Raybould
    ...I'm fine with that, but "adequately justified" is more akin to what you're saying being verified.
    InPitzotl
     
    Talk about vagueness! I am not convinced that this makes it to the status of being a complete sentence, to the point where I cannot even guess what you mean here.
     
     
    Argument and counter-argument are the principal methods of philosophy
    — A Raybould
    Yes, but counterarguments should not have to require an opposition taking a side to provide the counterarguments, and most certainly should not require the opposition to hold the countering view as an opinion...
    InPitzotl

    You are not required to take a side, but if anyone floats a putative problem for my hypothesis, then I am, of course, going to probe it regardless of who, if anyone, might hold it. It is not all about you.
     
     
    Also, two people discussing a thing need not necessarily each pick a corner and box; it's entirely possible, and may even be more productive, for the two to simply walk hand in hand from corner to corner.InPitzotl
     
    Right at the start of this thread, and subsequently, I have said that we could simply agree to differ. Or we could consider the issue dispassionately, agreeing to differ on how plausible each argument is.
     
     
    So, let's get back to tetrachromacy. Let's suppose we introduce a new gene in the human gene pool, call it OPN1MW3. OPN1MW3 expresses in people who have it by producing an M cone with spectral sensitivity shifted towards blue by the same amount (measured in frequency) that M shifts L spectral sensitivity towards blue; let's call this a N cone. This gene is an allele for the M on the X chromosome. So suppose we have: (a) Adam, who has L, M, and S cones; (b) Bill, who has L, N, and S cones; (c) Cindy, who has L, M, N, and S cones. So here are some questions. (a1) Is Adam likely to be a trichromat? (b1) Is Bill? (c1) Is Cindy more likely to be a tetrachromat or a trichromat?
     
    (c1) is the interesting question... but regardless of its answer we still have followups. If the developmental process is such that Cindy's likely to become a tetrachromat, then there are questions about what exactly causes the L2 colors that Bill sees; your hypothesis suggests it's some built-in gene, but if self organization suffices to establish L2 colors this is questionable. But if Cindy is only a tetrachromat if she has some other specific gene, some BN (brain-gene-N analog; here, this is just a generic referent... it could be more than one gene), that pre-structures her brain, then we must also ask whether Bill can have that gene as well, or under what conditions precisely Bill sees what Cindy sees that Adam doesn't see.
     
    Your hypothesis seems to require a "BM" gene of sorts. Okay, we have the human genome mapped out... so which gene is BM?
    InPitzotl
     
    Let me just point out here that if this name-the-gene standard of whatever-it-is-that-my-hypothesis-does-not-meet is applied consistently, then a good deal of evolutionary and general biological theory, including the fundamental one that life as we see it today is a result of natural selection, must be thrown out.

    More substantively, it is completely unclear to me why my hypothesis requires a BM gene of any sort.
     
     
    The point isn't to simply maintain some position with unreasonable standards though. The point is to require relevance. The thing being talked about here is the actual stuff happening between our ears in our soft pink squishy warm brains, that has to do with our subjective conscious experience of colors. Some discussion of and/or constraints on how that subjective experience's correlates develop is necessary to provide a theory of how much the subjective experience's correlates can vary. Without having that discussion or addressing what those constraints are, you're just plain not having the required conversation [my emphasis.]InPitzotl
     
    What I am proposing is precisely that there is a plausible constraint, from empirical facts about biological inheritance and ontogeny, on the degree to which the high-level functional correlates of visual experience can vary between individuals.
     
     
    I am guessing that your emphasis on 'established' indicates that you were aware of this, but I had in mind cerebral and congenital achromatopsia and dyschromatopsia, color agnosia
    — A Raybould
    Achromatopsia and dyschromatopsia are the same modes of L1 level color deficiencies previously discussed (though there are acquired forms). Color agnosia as far as I'm aware is a defect of the ventral stream, which is particularly interesting for awareness of L2 colors at all (if not L2 modes of color at all). I would be interested in an L2 specific defect.
    InPitzotl
     
    Not that I know of.

    I had regarded your focus on L2 as just a case of unnecessary specificity, but when I looked for something justifying this specificity, I did not find any use of L1/L2 terminology with regard to color perception and experience - perhaps you can provide some references, and explain the relevance of your specificity?

    While your stated goal, apparently, is to show that the notion that our experiences are similar to a first approximation does not achieve hypothesis-hood
    — A Raybould
    I've no objection to same-experience as a hypothesis. My objection is claiming that the hypothesis is adequately justified prematurely.
    InPitzotl
     
    You have always been rather vague about what your objection is. We are in agreement, apparently, that it makes the grade as a falsifiable hypothesis. We are also in agreement, apparently, that it is at least several facts short of a theory. We disagree, apparently, on how plausible it is. On the key issue, however, of whether there is a plausible mechanism, for maintaining the observed first-order similarity between child and parent behavioral similarities, if it is the case that there is little commonality in the functional correlates of their visual experiences, neither of us are aware of one. You also still seem to be arguing against mistaken paraphrases of my argument. I feel that you have not yet made the case that a third party should summarily dismiss this hypothesis.
     
                                                                                                                                                 
     
    Nevertheless, while I disagree that what I have presented here is too vague to be considered, it could be less vague, especially as my responses to misunderstandings are scattered over several posts. I intend to set it out more precisely, addressing the misunderstandings that have surfaced in this discussion. In doing so, I might come to the conclusion that it is not very plausible, and I will say so if I do.
  • A Raybould
    86

    There is alternative way of looking at your argument, by introducing premise 0: Everything that anyone could know about color vision is physically-encodable.

    If we also assume that the argument through point 6 is valid (I am not saying I do!), we can conclude that, contrary to our intuitions, by time t, Mary knows everything that anyone could know about color vision, and therefore premise 7 cannot be added to this argument without invalidating it.

    Given that the conclusion being sought here, by yourself and Jackson, is the antithesis of premise 0, it would be begging the question to deny that premise.
  • InPitzotl
    880
    It goes back to your attempt to claim that the possibility of behaviorally-invariant inverted spectra somehow invalidated my hypothesis,A Raybould
    Stop right here. No. The possibility does not invalidate your hypothesis; that's not the point of it. Your hypothesis is inadequately justified. The possibilities simply highlight ways in which you can be wrong that you haven't actually addressed.

    (FYI, your confusion of my rejection that your argument is adequate justification with invalidation of your hypothesis suggests the possibility that you're potentially susceptible to argument from fallacy, so I'd like to just point that out quickly).
    We disagree, apparently, on how plausible it is.A Raybould
    We disagree over how well your justification supports your hypothesis.

    Let me suggest a device... call it the surprise test. Your hypothesis suggests we see the same colors. Your support is that evolution basically just rolls that way. Let's just suppose someone proves your hypothesis wrong. Based on your presented hypothesis, and the support you've given for it, how surprised should I be to find it wrong?
    I feel that you have not yet made the case that a third party should summarily dismiss this hypothesis.A Raybould
    Wrong criteria. "Adequately justified" is the phrase I've used, in multiple posts. It's not about dismissal, it's about the criteria for acceptance. So apply the surprise test. I feel you have not yet made a case sufficient for a third party to find it surprising, based on the case you've made, to find the hypothesis wrong.
    This section shows that, as a matter of fact, undetectable inverted spectra cases are sparse among all possibilities, and an explanation of why they are does not, of course, dispute the fact that they are.A Raybould
    Just to clarify, section 2 is talking about behaviorally undetectable inverted spectra, not undetectable inverted spectra. They're talking about different things that the guy with inverted spectra might do. If you can analyze his brain and find different representations of color, that's not what they mean.

    Also, opponent color theory is mainstream (at least among color theorists). The analysis in section 2 of the SEP article didn't take into account the effects of color opponency on the modes of asymmetries they were discussing. Opponent color analysis is presumed to occur in the ganglia cells in the retina (cones connect to bipolar cells which connect to ganglia).
    I had regarded your focus on L2 as just a case of unnecessary specificity, but when I looked for something justifying this specificity, I did not find any use of L1/L2 terminology with regard to color perception and experience - perhaps you can provide some references, and explain the relevance of your specificity?A Raybould
    I hope you didn't search too hard, because the terms L0, L1, and L2 are simply defined by me in this thread for your specific benefit, for the purpose of this discussion, in this post.

    Refer to this picture again:
    fauxlotto-cf.png
    L0 is spectral. This is the physical form of colors; objects reflect different frequencies of light by different amounts.

    L1 is colorimetric:
    It is similar to spectrophotometry, but is distinguished by its interest in reducing spectra to the physical correlates of color perception, most often the CIE 1931 XYZ color space tristimulus values and related quantities.Colorimetry
    Colorimetry is based roughly on Grassmann's laws. L1 colors in short then are produced by our three cones under photopic conditions, and result from our trichromaticity. This collapses the full visible spectrum (L0) into three distinct cone channels. There is also the opponent color process, which is triggered primarily by processes in the eye (namely, ganglia cells that connect the signals from the cones, lying behind bipolar cells); so I'm lumping this into L1 (from a colorimetric perspective, opponent processes don't change anything; a spectral metamer is still that metamer and therefore still the same point in a colorimetric space). L1 colors are critical for experiences, because everything gates through them... if a person's eye cannot distinguish one metamer from another, then neither can his brain, and neither can he. But it's pretty out there to speculate that eyeballs experience color.

    L2 is experiential. Experience is not produced in the eyeball; it's produced in the brain. Various places in your brain (visual cortex) are responsible for color analysis; here's an outline:
    https://en.wikipedia.org/wiki/Colour_centre
    ...which I post as a raw link for emphasis (you really should read at least the first paragraph). Per a physicalist account, at some point, something in your brain generates consciousness, and your personhood arises. Within your mental scape, you experience colors, such as the colors in the color illusion displayed in this diagram. Since mind states are brain states, your experience should have a correlate. Those are L2 colors. Qualia are experiences:
    Those new to the topic should first read the entry on qualia. ... Qualia (singular ‘quale’), in a common modern usage, are properties of experiences that type them in phenomenological respects.Inverted Qualia
    ...and as such, inverted qualia are experiences. So qualia, as in the things being inverted in qualia inversion, are at L2, not at L1.

    More substantively, it is completely unclear to me why my hypothesis requires a BM gene of any sort.A Raybould
    Oh that's easy... that falls out pretty much directly from this:
    The "learned lessons" are evolutionary adaptations.A Raybould
    Let me just point out here that if this name-the-gene standard of whatever-it-is-that-my-hypothesis-does-not-meet is applied consistently, then a good deal of evolutionary and general biological theory, including the fundamental one that life as we see it today is a result of natural selection, must be thrown out.A Raybould
    Well, then I am not putting anything at risk by not responding until it is more than just creationists who are making that claim.A Raybould
    What are you talking about? These genes are just portions of a molecule, and they encode photopsins. Photopsins photoisomerize (fold in response to light), and this leads to the chain reaction that causes the cones to send signals. Change the shape of the photopsin and you change the absorption characteristics. For example, L and M's sensitivity curves are close, because erythrolabe and cyanolabe are chemically close, because the encodings for them on the associated genes encode for those particular close molecules.

    Evolution doesn't hang in the balance, and if I suddenly managed to become a creationist, someone forgot to inform me. I've no idea where this nonsense came from... did you perhaps think those genes coded for color percepts?

    ETA: Ah, it occurred to me that this nonsense may be due to you not understanding the question I asked you. You're proposing that same-experience in particular is selected due to it being an evolutionary adaptation. I'm asking you, why you think same-experience is subject to evolutionary pressures, which in my mind equates to the core of your hypothesis. But I'm guessing you're interpreting that as questioning evolution. Evolutionary adaptation is a process where selective pressures (what you're adapting to) influence the frequency of genes (in "competition" for genomic representation; i.e., alleles) in a gene pool (the change in allele frequency being evolution) to the degree that adaptive gene variants become predominant in a population. Your hypothesis presumes that same experience is adaptive, which further presumes that different experience would be maladaptive, which in turn means that there's a distinction between same experience and different experiences sufficient to warrant a distinction between an adaptive and a maladaptive behavior, in the sense that having same experiences of some particular sort somehow winds up increasing the probability that genes remain in the gene pool (e.g., more women with this trait survive to give birth; more men survive to mate; women give birth to more offspring; women find mates more easily, and so on). All of these are presumptions that basically equate to your core hypothesis, but you seem to argue for none of them, because you seem to think simply invoking the word "evolution" is enough to support your hypothesis. That's not true. Evolution only supports your hypothesis if it supports your hypothesis. For same experience to even be adaptive in the first place, it has to influence something that selective pressures can work on.

    You have supplied no adequate reasoning as for why this would be the case; effectively, you just presumed it. (Mind you, you did attempt it, by pointing out someone else's argument in section 2 of that SEP article regarding asymmetries of perception, but since that can be explained in terms of L1 properties it's not quite adequate; note that I actually granted relevance to that argument, but have specific technical reasons to require more details). That is precisely what I'm calling guessing, and what I'm demanding you need to do before you can call your hypothesis adequately justified. If you cannot justify why same experience would be selected, you ipso facto have not justified that it is. So long as there's a giant presumptive gap in your argument, it does not rise to the level of adequacy.
  • jkg20
    405
    There is alternative way of looking at your argument, by introducing premise 0: Everything that anyone could know about color vision is physically-encodable.
    Let me deal with this first, I'll speak to the more significant issues you raise in your previous reply to me, concerning the notion of information content, in a later post.
    First, by introducing a premise to an argument you do not look at an argument in an alternative way, you change the argument. The premise you wish to introduce, in this context, is effectively the claim that physicalism is true for colour vision. Why would anyone presenting an argument with the aim of proving precisely the opposite accept that their argument can be "looked at in a different way" by adding a claim they oppose?
    and therefore premise 7 cannot be added to this argument without invalidating it.
    As I have pointed out on a number of occasions throughout our discussion, the physicalist can object to the claim, expressed in premise 7 of my latest argument, that Mary learns anything new when she sees that red tomato. As far as I can see, all Jackson has on his side for motivating acceptance of it is intuition, and not in the Cartesian sense of that word, but the colloquial sense in which it is akin to something like inspired guess work. Is intuition in that sense a sound basis for support? Maybe in the end all we have to go on is intuition, but that is a metaphilosophical question. However, if a physicalist does want to reject premise 7, note that they are rejecting the argument for reasons other than those first put forward by Churchland and later expanded on by Crane etc. Those objections were based on the claim that Jackson was guilty of more or less subtle forms of equivocation, which may well be true. My final argument, however, has removed those equivocations. Right from the beginning, you may remember, my aim was to circumvent Churchlandish responses to Jackson's argument, whilst leaving something behind that at least retained the spirit of Jackson's argument. If, now, after amendment in the form of my latest argument, the physicalist is actually being pushed into the corner of rejecting the claim that Mary learns anything new, then that particular aim has been accomplished. My adapted argument also focuses the issues where, to my mind, they need to be focussed, which is on the theory of content that lies in the shared representationalist background of both the physicalist and dualist positions....which takes us to the points you make in your other reply, which I will reflect upon a little more before responding.
  • A Raybould
    86

    There is nothing wrong with introducing a new premise into a discussion. If doing so were wrong, then we should reject your introduction, in your latest argument, of several premises that you had not raised in previous posts. Furthermore, in this case, premise 0 (the only additional premise; all the rest are yours) is not just any premise, it is the antithesis of your conclusion.

    Of course, you would not want to accept premise 0 in any argument, but that is beside the point, as the purpose of debate in the search for knowledge is to persuade third parties of certain things (and even if you were to claim that you are arguing only for yourself, it would then be inconsistent with that claim to object to anyone else adopting any premise; furthermore, you would be in great jeopardy of fooling yourself if you did not examine your argument as a skeptic would.) The point is that you cannot simply reject the skeptic's assumption of the antithesis of your argument without begging the question.

    As for your objection that this goes beyond your initial limited claims against Churchland's reply to Jackson, it would be inconsistent for you to do what you are doing - trying repeatedly to find an argument that holds up - while insisting that I must stick with only my first presentation of Churchland's argument (if the latter rule were applied consistently, there would be no debate about anything!) It would also be a contradiction to say that your latest argument is on-topic, while attempts to refute it are off-topic.

    Your final argument has not removed the Churchland-Crane charge of equivocation, at least until you deal with the argument in my other reply above.

    You seem to be under the impression that I am implying that premise 7 must be false, but the point here is, instead, that if your argument, for the proposition that all knowledge having physically-encodable information content is discursively-learnable, holds, then what non- question-begging reply do you have to the claim that Mary could have learned everything beforehand? One does not have to deny 7 to see that this is a problem for your argument.

    Saying that premise 7 is just an intuition is not a get-out-of-jail card, logically speaking, as it does nothing to fix this problem. Also (somewhat unintuitively, perhaps), the reasons why one might believe any premise are not germane to validity. 

    Far from the physicalist being pushed into the corner of of rejecting the claim that Mary learns anything new, you are in the corner of having not yet given a non- question-begging reply to the problem of why Mary was unable to learn it before t1. That, of course, was in my original presentation of the Churchland/Crane reply to Jackson, so all the stuff about not addressing your original claim is beside the point, even if it were valid.

    For reference, here is the counter-argument (points 1-7 copied from your original):
    [0]: Everything that anyone could know about color vision is physically-encodable.  
    [1]: All information content that is encodeable is expressable as a finite set of propositions.
    [2]: Any finite set of propositions is discursively learnable.
    [3]: From 1 and 2; all information content that is encodeable is discursively learnable.
    [4]: Anything that is physically encodeable is information content that is encodeable.
    [5]: From 3 and 4 anything that is physically encodeable is discursively learnable.
    [5.1]: From 0 and 5, everything that anyone could know about color vision is discursively learnable.
    [6]: At time t, Mary knows everything that is discursively learnable.
    [7.1]: From 5.1 and 6, at time t1, Mary already knows everything that could otherwise be learned from seeing a red thing for the first time.

    Which contradicts 
    [7]: Mary gains new knowledge when at t1 she sees a red thing for the first time.

    As noted above, this does not necessarily imply that 7 is false, just that not all of the above can hold together.
  • jkg20
    405
    There is nothing wrong with introducing a new premise into a discussion.
    Certainly there is not. However, the only acceptable premise to add to an argument that is presented as a complete argument is a premise that is hidden. The premise 0 that you are offering is certainly NOT a hidden premise of my argument.
    If I gave an argument such as:
    1. If there is a pattern to the universe then a non finite sentient being exists.
    2. God is the only non finite sentient being
    3. There is a pattern to the universe.
    Therefore God exists
    and then you come along and add premise 0,
    0: All sentient entities are non divine.
    and then you point out that by adding in premise 0 to the argument, we can draw out a contradiction with 2, all you are doing is rejecting the conclusion of the argument but without pinpointing the premises of it that you actually reject.

    Of course, you would not want to accept premise 0 in any argument, but that is beside the point,

    It is entirely to the point, since the argument is presented as complete and with a conclusion that physicalism is false for colour vision. Your premise is effectively the claim that physicalism is true for colour vision. It is for you to show that the argument has hidden premises that can be objected to OR that there are explicit premises that can be objected to. It is no use just coming along with another premise that is effectively just a rejection of the conclusion.
    You seem to be under the impression that I am implying that premise 7 must be false, but the point here is, instead, that if your argument, for the proposition that all knowledge having physically-encodable information is discursively-learnable, holds, then what non- question-begging reply do you have to the claim that Mary could have learned everything beforehand?

    The argument requires not that Mary could have learned everything beforehand but that Mary could have learned everything physically encodeable because she can learn everything discursively learnable. The argument then adds a premise, 7, that she learns something new. If that premise is accepted, then it just follows from the other premises that what she learnsis not physically encodeable. There is no question begging going on.
  • jkg20
    405
    Premise 4 is "Anything that is physically encodeable is information content that is encodeable." Given the context, I took "information content" to mean the information content of knowledge. If it is not that, then what is it the content of?
    Information content can of course be the content of knowledge, but since realism about information content is being assumed, it need not be the content of any actual knowledge. A simple example, if we assume realism quite generally, there could well be a fallen tree in a forest somewhere that no one has seen, the trunk of which has 70 rings. That physical state, which we can describe in a finite set of propositions at whatever level of detail we like, has the information content that that particular tree had survived 70 seasons of growth. Note the the information content is here identified with a single proposition. There may be more information content than just this encoded in the phyiscal state of the tree, but there is at least this. One fine day, someone walks through the forest and comes across the fallen tree and counts the rings and forms the belief that that particular tree had survived 70 seasons of growth. Let's assume a version of physicalism and that that person is in a physical state that we can identify with her believing that that tree had survived 70 seasons of growth. That physical state is describable by a finite set of propositions, this just follows from the finitude of human beings. The information content of that state is the same as the information content physically encoded in the tree itself, i.e. the information content identifiable by the proposition that that tree had survived 70 seasons of growth. Is it now clear to you what I mean by information content and that it need not necessarily be the content of anyone's actual knowledge, and so objections to premise 4 need to be metaphysical in nature and not epistemological, unless you wish to challenge realism.

    suppose we have some system having information on account of being in physical state ψ. if the corresponding propositions you have in mind are not a description of that state, then what are they? It is far from obvious that such propositions must exist, whether as a consequence of Shannon's information theory or of anything else. As things stand, proposition 1 lacks any justification.

    I presume that the above talk about tree trunks makes it clear what the difference is between, on the one hand, the propositions that describe a phyiscal state which encodes some information content and, on the other, the propositions that describe the information content itself. As to the apparent issues you have with premise 1, it is in fact pretty much axiomatic for all theories of information content that I am aware of, that information content is identifiable by sets of propositions. It is certainly an assumption, not a consequence, of Shannon's theory of information transference, it is an assumption embedded in his free and undefined use of the term "message". It also runs through all of the philosophers who defend representational theories of mind, including physicalist representational theories of mind. It might be false, of course, but since it seems to be widely held amongst physicalists themselves, who are the target of the argument after all, I would like to see a counterexample or be referred to a relevant text where a physicalist non propositional account of information content is expounded. It might even be possible to come up with an actual argument that you could not have a coherent theory of informational content that allowed for content that was not identifable with finite sets of propositions, but on that particular point premise 1 is playing a card dealt to it by physicalists anyway, so I am content for the moment to sit on my laurels. However, if that is a point you want to pursue in detail, have at it.
  • jkg20
    405
    The point is that you cannot simply reject the skeptic's assumption of the antithesis of your argument without begging the question.
    But I am not simply rejecting it, I have presented an argument against it. Assuming the argument is deductively valid, which on the surface it is, the skeptic who wants to contest its soundness, and thus reject its conclusion, has a number of choices, whilst remaining rational. These include:
    1: Show, by counterexample or some other means, that one or more of the explicit premises is false.
    2: Show that there are hidden premises in the argument that are or might be false.
    3: Linked to 2, show that there is an ambiguity in one or more of the premises and that when the ambiguity is clarified, the premises become false or at least possibly false.
    4: Argue that we have more reason to reject the conclusion of the argument than we do to accept any of the premises.

    4 would seem difficult to go for, after all physicalism and dualism are at least as much up for grabs as any of the premises of my argument. This leaves strategies 1 through to 3. Notice that I did not add.

    5: Create a new argument by adding an entirely new premise to the existing ones that directly or implictly contradicts the arguments conclusion, thus turning the argument into a deductively invalid one. This is certainly a novel approach, but hardly a rational one.
  • jkg20
    405
    you are in the corner of having not yet given a non- question-begging reply to the problem of why Mary was unable to learn it before t1.

    Mary was unable to learn it before, because what she learns is not discursively learnable, and since everything that is physically encodeable is discursively learnable, according to the argument, what she learns is not physically encodeable either. So what does she learn? She learns what red things look like, and learning what red things look like is not discursively learnable. Of course; to be consistent I also have to say that what she learns is not identifiable by any finite set of propositions, so have I refuted myself by saying she learns what red things look like? Here I suppose we have to go Wittgensteinian and say that we do not identify what Mary learns with the expression "Mary learns what red things look like", but rather show or gesture at what she learns by doing so. The main point, in any case, is that whatever it is, it is not physically encodeable nor identifiable with any finite set of propositions.
  • jkg20
    405
    As for your objection that this goes beyond your initial limited claims against Churchland's reply to Jackson, it would be inconsistent for you to do what you are doing - trying repeatedly to find an argument that holds up - while insisting that I must stick with only my first presentation of Churchland's argument (if the latter rule were applied consistently, there would be no debate about anything!) It would also be a contradiction to say that your latest argument is on-topic, while attempts to refute it are off-topic.

    I did not make any such objection. I pointed out simply that the original aim was to provide an argument that circumvented Churchland's objections about equivocation of epistemological notions. The argument I present may not be sound, but it does not equivocate in the way that Jackson's original argument does. Thus Churchland's objections are, I claim, circumvented. This is done by introducing specific metaphysical claims about informational content that are absent from Jackson's argument, so it should not be a real surprise that Churchland's epistemological objections become otiose. I am also not insisting that you stick to your guns. Within the bounds of rationality, you are free to find fault points in my argument. There may even be equivocation going on somewhere within it, and perhaps even epistemological equivocation, but thus far you have not identified that equivocation.
  • A Raybould
    86
     
    firstly. let me say that I have come to agree with you that I should say something about what physicalists assume about color vision, though it will not - and need not - be a complete explanation of how it works. I will say more on that below, in replying to your second post of this series.

    Your claim that "the only acceptable premise to add to an argument that is presented as a complete argument is a premise that is hidden" only applies if it is being claimed that the resulting argument is effectively the same as the original. Any modification to an argument can be seen as creating a new one, and one can use any premises in a new one. In this case, I am creating a new argument, for the conclusion that at time t1, Mary already knows everything that could otherwise be learned from seeing a red thing for the first time.

    This argument has several features of interest. Firstly, it contradicts your premise 7, and as such, it stands as a counter-argument: This argument and yours cannot both be valid and sound. So how could you show that this new one is either invalid or unsound?

    This is another of its interesting points: except for one premise and two deductions, it is taken directly from your argument. If you could show that those parts are invalid or unsound, that would apply equally to your argument.

    That leaves you with a few options. You could show it to be invalid by showing that lemma 5.1 (everything that anyone could know about color vision is discursively learnable) or conclusion 7.1 (at time t1, Mary already knows everything that could otherwise be learned from seeing a red thing for the first time) do not follow from the premises, but so far, you have not.

    Your only other option is to show that the new premise (everything that anyone could know about color vision is physically-encodable) is unsound. You cannot simply say that your argument shows this to be so, as then your argument would be circular: "your argument against my argument fails, because my argument shows that your premise does not hold!" You could simply deny, without offering any additional argument, that my premise is false, but now we get to the third interesting point: this premise is the antithesis of your conclusion[1]. Therefore, you would be begging the question if you just did that. What else is there? That is the question I have been asking.

    This is unlike your proof-of-God example, where the reply technically is a counter-argument, but not a useful one, as it does nothing more than deny the conclusion. While the only new premise I have introduced does that, I use several additional premises to show that your argument is not valid as it stands. 


     
    Let's follow Douglas Hofstadter's advice, and 'twiddle the knobs' on your tree-ring example.

    When your subject discovers the tree trunk, no knowledge about its age can be learned discursively. It only becomes so once your subject has processed it, together with additional information that cannot be derived from the trunk. Note that a person who did not know the origin of tree rings would not form the proposition that the tree had survived 70 seasons of growth, and would not be able to communicate this fact (unless it was learned another way, which would be beside the point.) 

    Before any of that happens, the rings have to be perceived, which requires a modulated reflection of the light falling on it, stimulation of the viewer's receptor cells, and neural processing of the resulting signals.

    Some of the physical changes in the neural system will be persistent, at least if the viewer has any memory at all of seeing the trunk. Let's call the physical state resulting from these changes due to perceiving the tree rings P (note that the physical state P is unique to each individual, as they have physically different brains simply as a result of not being made from the same atoms.). A physicalist can reasonably propose that P is essential to the subject knowing what it is like to perceive the rings.

    Getting to the discursively-learnable proposition that "this tree has survived 70 seasons of growth" requires additional processing, which, for the first time, requires linguistic abilities, which are known to be located in a different region of the brain than visual processing. This will result in some additional state changes within the viewer's brain, and we can call the physical state resulting from these changes L. Note that the person who does not know the cause of tree rings does not form the proposition, and does not enter her version of state L, yet presumably does know what it is like to see the rings. Consequently, we can conclude that knowing this discursively-learnable knowledge is not necessary for knowing what it is like to see the rings.

    (As it happens, one of Churchland's arguments for non-propositional qualia is that many animals lacking linguistic abilities can be shown to perceive, remember and compare colors (including to memories of colors), all without contemplating any propositions whatsoever.) 

    Regardless of how many ways this discursively-learnable information is the same as that encoded in the trunk's state, it is not identical to the latter information in at least three ways: 1) it is encoded as a physical state in the observer's brain, not a physical state of the trunk; 2) unlike the information in the trunk, it is the information content of someone's knowledge; 3) the trunk can be destroyed, while the observer continues to know something about its age when it fell.

    Now let's introduce Mary, who knows all of completed physics. She has not seen the trunk, but has been given a complete physical description of it (there is no doubt that the physical description is communicable, and, in fact, it would be part of her complete knowledge of physics, which she has learned discursively.) From this description, she could deduce that the trunk has 70 rings, and, with the additional knowledge about the origin of tree rings (which is also part of her complete physical knowledge), she could achieve physical state L by deducing the discursively-learnable proposition that the tree has survived 70 seasons of growth.

    Let's assume she can also deduce what her state P would be if she had seen the tree trunk, and give a description of it. In synthesising that description, she undergoes a third set of state changes, to state D, encoding (among other things) the propositions comprising the description of the state P. That state is not P itself.

    To get to state P from this description of state P, Mary would need to target the specific neurons and synapses involved in P, and produce in them specific state changes. This would require a degree of connectedness and influence that certainly is not necessarily so, and has never been demonstrated to science (can you do it?) We cannot even just get the state of specific synapses, let alone change them.

    And if we think that her total knowledge would give her this ability, then, per Dennett, she could get from D to P and would know what seeing the trunk is like - this is one of the horns of Jackson's dilemma again. In fact, all she would have to do would be to just modify the state of the first layer of neurons that the optic nerve connects to, replicating the effect of seeing the trunk, and so come to know what it is like to see it without having done so.

    I don't generally like computer analogies, but there is a very appropriate one here: if I have a secure operating system, it will have plenty of internal-use-only state, and you will be unable to write a program that an unprivileged user could use to modify that state, even if you know everything there is to know about that operating system.

    If you have read Curchland's Reply to Jackson, you will recognize this as being very similar in spirit to what he wrote there, and as he said, physicalists do not have to provide a complete and completely accurate account of vision in order to show that, quite plausibly, there is an entirely physical explanation for Mary not learning everything from her studies, even though she has full knowledge of physics (which is not the same as having all physically-encoded knowledge.) So long as this goes unanswered, Jackson's argument fails to make its case.

    If it is the case that many philosophers, including many physicalist ones, hold that all information content is identifiable by sets of propositions (what, precisely, does 'identifiable by' mean?) then that might explain the popularity of 'old knowledge seen in a new way' replies to Jackson, which seem to me to be wrestling with a problem that has a much simpler answer. So long as philosophers like the Curchlands and Dennett are unconvinced by this view, I am not convinced that I am overlooking something obvious - and if I am, it should be easy to show that I am doing so, without appealing to authority.



    This is all covered above - in particular, see the bit about circularity.



    Again, see the above. It seems to me that your metaphysical claim is merely avoiding pertinent issues raised by Churchland and others. Metaphysical arguments have to make their case, just like any other.

    ---------------------------------------------------

    [1] I did not actually write "it is not the case that what Mary learns is not physically encodeable", but your  conclusion does not actually refute either physicalism, or Churchland's reply, if she doesn't learn anything about color vision at t1. The conclusion you need here is something equivalent to what Jackson said, such as "at t1, Mary learns something that is not physically encodable." Of course, in your argument, you can get there directly by using premise 7, but I can also get to its antithesis directly from my conclusion that Mary does not learn anything at time t1.
  • jkg20
    405
    Your claim that "the only acceptable premise to add to an argument that is presented as a complete argument is a premise that is hidden" only applies if it is being claimed that the resulting argument is effectively the same as the original.
    I thought I had been clear about this point, and here is a quotation from an earlier post of mine, emphasis added:
    "One can circumvent objections by changing the argument, which is effectively what I did: it is certainly not Jackson's original argument." I have never claimed that my arguments are the same as the original argument, if by "original argument" you mean Jackson's. Thus my claim about what is acceptable to add to my last argument stands. However, on reflection there is at least one "hidden" premises in that argument, so please see a new version of it below and treat it on its own merits. The thing it shares with Jackson's argument is the intuition pump that Mary does learn something new about colour vision when she sees that tomato for the first time.


    Regardless of how many ways this discursively-learnable information is the same as that encoded in the trunk's state, it is not identical to the latter information in at least three ways: 1) it is encoded as a physical state in the observer's brain, not a physical state of the trunk; 2) unlike the information in the trunk, it is the information content of someone's knowledge; 3) the trunk can be destroyed, while the observer continues to know something about its age when it fell.

    None of this is relevant if we are assuming realism about information content, and most of what you talk about in your last post concerns how sentient beings might gain access to information content, but does not concern the what that content is in itself. Assuming realism about content allows the same content to be encoded in different ways. The points in the metaphysical premises of my argument concern identity of content conceived realistically, not identity of vehicle for that content. Nearly all of the points you make in your post concern differences in the vehicle for content and how different types of vehicle might come to encode the same content.

    So, with my one concealed premise now exposed, here is the new argument against physicalism.

    1 If physicalism is true, then everything anyone can know, in any sense of the term "know", about colour vision is physically encodeable.
    2: All information content that is encodeable is identifiable by a finite set of propositions.
    3: Any finite set of propositions is discursively learnable.
    4: From 2 and 3; all information content that is encodeable is discursively learnable.
    5: Anything that is physically encodeable is information content that is encodeable.
    6: From 4 and 5 anything that is physically encodeable is discursively learnable.
    7: At time t, Mary knows everything that is discursively learnable.
    8: Mary gains new knowledge of some kind about colour vision when at t1 she sees a red thing for the first time.
    9: From 6, if the new knowledge Mary gains at t1 is physically encodeable, then it is discursiveley learnable.
    10: From 9 and 8, if the new knowledge Mary gains at t1 is discurvively learnable, then she did not know at t everything that is discursively learnable.
    11: From 10 and 6, if the new Knowledge Mary gains at t1 is physically encodeable, then Mary did not know at t everything that is discursively learnable.
    12: Since the consequent of 10 contradicts 7, by modus tollens what Mary learns is not physically encodeable.
    13: From 1 and 12, physicalism is not true.

    Note that if you now add your premise 0 to this argument, you get a deductively invalid argument. So, if you insist of creating an new argument by adding your premise 0 to the one above, your argument can be immediately rejected on the grounds that it is not deductively valid.


    From what I gather, your issues with this argument focus on 2 and perhaps on 5. As indicated, everything you have thus far said against 5 conflates epistemological issues about how sentient creatures get access to information content, and metaphysical issues about information content itself. Do not misunderstand me, I am aware that realism about information content is a fraught issue. However, physicalism entails realism about information content, so if antirealism about information content is true, then physicalism is false anyway. So, for the purposes of arguing against physicalism it is perfectly acceptable to assume realism about content.

    As a quick argument for premise 5, the very notion of encoding is that it is the conversion of data from one form to another, and data just is information content.

    As concerns 2, I am not really appealing to authority. I am saying that it is a premise accepted or assumed by all physicalists and generally realist scientists and philosophers that deal with the notions of information, its encoding and its transference. Dennett and Churchland amongst them. However, there has been some recent work in philosophy that has attempted to make viable the notion of non propositional intentionality. If all intentional phenomena have content consisting entirely of information content, and not all intentional phenomena have content that is identifiable by propositions, then there will indeed be information content that is not identifiable by propositions, whatever "identifiable" means, and so premise 2 fails. However, if the objection to my argument is that there is non propositional intentionality, we are entering into very complex territory, and it is terrain that is very far from Churchland's fair equivocation points against Jackson's argument. If I have some spare time, I may well read a collection such as this Non Propositional Intentionality, but I feel I've said all I can or want to say about this for the moment.
  • A Raybould
    86

    I see that you still have no clear idea what my argument is, largely, it seems, as a result of your attempts to paraphrase it into something that you can dispute. I will reply with a lot more detail, but as I am off on a trip to the backwoods, that probably will not be until sometime next week.
  • A Raybould
    86

    I'm leaving for a week in the backwoods, so it will be more like next week before I reply, as there are many things here to correct.

    One thing I did notice immediately, however, is that you have not yet given a clear and coherent explanation of what you mean when you use 'identifiable' / 'identified by', even though you use the term in a new premise. If you could do that, it would be helpful - for one thing, no-one should accept any premise containing the term, without there being a clear understanding of what it means.
  • InPitzotl
    880
    I see that you still have no clear idea what my argument isA Raybould
    Perhaps, but, 25 days ago. you responded specifically to this:
    "... but to simply conclude that the color experiences are the same because we're all human sounds to me more like guesswork. — InPitzotl"A Raybould
    ...with this:
    It doesn't approach being a proof, but, IMHO, it is a plausible hypothesis. Per my earlier post, I would not propose that everyone's experience is the same in detail, but that for most people, there is a broad degree of functional equivalence.A Raybould
    largely, it seems, as a result of your attempts to paraphrase it into something that you can dispute.A Raybould
    So while you're speculating, is that because you're inexplicably the most important person in the world to me, or can you explain why my post count isn't a lot higher than it is given I'm just out to dispute people at random? (E.g., note that all this time, I haven't intervened on either side with your discussions with jkg20... isn't that curious?) Also, why do you feel the need to share your ridiculous speculations about my motivations with me, who one would think would be in a much better place to know said motivations?

    Allow me to speculate about my own motivations to you... it's... perhaps... because 25 days ago you responded to the aforementioned quote with the aforementioned response, to me, and that you seem to be under the mistaken impression that this somehow should change my view about the thing I said that you quoted. Also, if you haven't guessed, color per se is a personal passion of mine. (Now doesn't that sound a lot more viable than he's-out-to-get-me [<-being literally some random guy on the internet]?)

    So my issue with you is your alleged defense of this being "plausible" above the level of being "guesswork".
    I will reply with a lot more detailA Raybould
    I'm sure you will.
  • jkg20
    405
    A premise of my argument is that information content is propositional. This means, amongst perhaps other things, that if someone were to ask: "what is the information content encoded in such and such a physical state?", the answer to that question will require giving expression to a specific set of propositions. Specifying thusly that set of propositions is at least one wayto identify the information content of the physical state concerned. Another way of putting it is that the physical state has information content that is identifiable by specifying that set of propositions. This invites, but does not beg, a number of questions of course, including "What is a proposition?". An acceptably realist answer to that would perhaps be that propositions are sets of possible worlds. In any case, the point stands, if the only way my argument against physicalism can be rationally challenged is by delving into the complexities of such metaphysical questions as "What is content?", "What are propsitions?" etc, then it certainly avoids the equivocation objections that Churchland raised against Jackson's argument.
  • Julianne Carter
    10


    I’d like to critique two objections that you pose to Jackson’s argument: the first regarding Jackson’s assertions about Mary’s knowledge, and second, your claim that Jackson’s argument contains a begging the question fallacy.

    P1 of the version of this argument that you posted states, “Mary knew everything physical about the color red but has never seen the color red.” This refers to the fact that Mary’s knowledge of the color red encompasses everything except the actual experience of seeing it. However, you claim that this “implies Jackson is of the view that physicalism entails that knowing what brain-state corresponds to actually seeing red should evoke the brain-state of actually seeing red,” and that because of this, the argument “fails to achieve its intended objective of refuting physicalism.” I don’t see how this implication follows from P1. Even if Mary knew the exact brain-state that occurs when a person sees red (perhaps through brain imaging, descriptions, or information about the operation of the senses), she would not have any reason to experience the brain-state of actually seeing red, and she would still be gaining knowledge after seeing the color for herself. I acknowledge your assertion, based on this premise, that the knowledge of red and the experience of seeing red are both brain-states, but I think it fails to note one of Jackson’s points: Mary still gained new knowledge through the experience of seeing red, in a way that could not be taught to her. Further, even after acquiring the new brain-state of seeing red, she experiences yet another: if she goes back into the black-and-white bunker, she has the memory of seeing red, and is in the new state of having knowledge of experiencing red. Based on these points, I have a difficult time accepting your claim that the argument fails to refute physicalism.

    You further claim that Jackson’s argument contains a begging the question fallacy, which you label as Premise 1a: “Complete physical knowledge can't account for actually seeing red.” This fallacy is present in the version of the argument that you posted, but I’m curious as to where it came from: is it your interpretation of the argument, or another’s? The versions of the argument that I’ve seen don’t suggest this fallacy. I’ll focus on the first half of the argument, because that’s the part that you’re taking issue with. Most versions that I’ve seen go something like:

    P1: In the bunker, Mary knows every physical fact about the color red.
    P2: Mary learns something about the color red after her release.
    P3: Therefore, Mary didn’t have complete knowledge of the color red in the bunker.

    This first part of the argument does show that there was some knowledge that Mary lacked, but I don't see a fallacy there, when it's organized and worded this way. It just illustrates Mary’s example, before the remainder of the argument states that physicalism is false. Therefore, I think that the idea of the fallacy that you mention doesn’t harm Jackson’s argument.
  • creativesoul
    11.4k
    Jackson's use of the word "knew" needs to be clarified. When Jackson claims Mary "knew everything physical about the color red" he refers to everything except the direct experience of rednessTheMadFool

    That experience consists - in part at least - of physical interactions. Thus, Mary did not know everything physical about the color red.
  • Pfhorrest
    4.6k
    Mary knew everything that could be known in the third person about color vision.

    She did not know what it was like to experience color vision in the first person.

    This does illustrate that first-person experience is not knowable from the third person.

    It does not show that first-person experience is a non-physical phenomenon.

    It’s entirely possible that there is a first-person experience to any physical phenomenon. In fact, it gets trickier if you try to assert that somehow humans have it but other things don’t, because then you have to posit something metaphysically different between humans and other physical things. Far more elegant to just grant that humans are like anything else — and that since we each know we have a first-person experience, we should assume by default that everything else does too.
  • Isaac
    10.3k
    Mary knew everything that could be known in the third person about color vision.

    She did not know what it was like to experience color vision in the first person.

    This does illustration that first-person experience is not knowable from the third person.
    Pfhorrest

    No it doesn't. It simply declares it by the definition of the imaginary world in which the experiment takes place. We don't have an actual Mary who was actually told all there is to know about Red and who then actually reported a new sensation on leaving the room. We imagine such circumstances. All it tells us about is what we imagine to be the case. It tells us absolutely nothing at all about what actually is the case.
  • TheMadFool
    13.8k
    That experience consists - in part at least - of physical interactions. Thus, Mary did not know everything physical about the color red.creativesoul

    Reminds me of a comment made about emotions - actual feelings can't be put into words in such a way that they actually evoke those emotions. I guess Jackson's thoughts are in the same vein. There's something about immediate/direct/first-person experience that escapes description. Is it going to stay this way or are we just in need of the writer/speaker/interlocutor who has, let's just say, a way with words, having such powers of expression that faers words are capable of giving us the first-person experience I'm talking about here?
bold
italic
underline
strike
code
quote
ulist
image
url
mention
reveal
youtube
tweet
Add a Comment

Welcome to The Philosophy Forum!

Get involved in philosophical discussions about knowledge, truth, language, consciousness, science, politics, religion, logic and mathematics, art, history, and lots more. No ads, no clutter, and very little agreement — just fascinating conversations.