Hi Jason, and thanks for the comment.

I don't think you've got me quite right however; I'll try to make clear where we differ on my view, which will involve explaining how I am not begging the question against AH, which seems to be the danger you identify for my paper.

When you say 'Mr. Coleman suggests that phenomenal knowledge is required before Mary becomes able to identify colors.' it is revealing. I don't claim any such thing, at least not in a temporal sense of 'before'. The thought is rather that knowing what redness is like is the ground of Mary's ability to recognise, remember etc. the experience (by the way we disagree here on what the abilities have as their objects - you think it is objects of perception - but I won't pursue this further as it doesn't appear crucial to your objection). I agree that gaining the phenomenal knowledge and gaining the abilities are simultaneous events. Now this may be because gaining the abilities is gaining the phenomenal knowledge. That I leave open. My point is that gaining the abilities may (or may not) involve new factual knowledge on Mary's part. That is why reducing Mary's knowledge to abilities is a dialectically useless move.

Reductions do have 'orders of explanatory priority'. When it is said that 'mind = matter' by physicalists, we know what is being reduced to what. Just so when hypothesists say that 'knowing what it's like = having abilities'. But I argue that abilities of the relevant kind depend on knowing what sensations are like. This latter component (which I later on in the paper do call 'phenomenal knowledge') is unreduced by the claim that knowing what it is like is ability. That is why the reduction to abilities is neither here nor there for the issue of factuality.

Hi Sam, and thank you for your response.

I've taken another look at your paper with your recent comments in mind.  What follows is a detailed defense and elaboration of my criticism.

In your response, you said:  "When you ... (expand) say 'Mr. Coleman suggests that phenomenal knowledge is required before Mary becomes able to identify colors.' it is revealing. I don't claim any such thing, at least not in a temporal sense of 'before'."

I will point to the statement which suggests a temporal order, though I should point out that the weight of my objection does not depend on whether or not we are talking about temporal priority.  My objection extends to the notion of temporal priority, however, so I will follow through with this point.

You write:  "It seems that abilities essentially connected with a phenomenal state can only come into operation once the character of the phenomenal state involved is grasped by the subject/agent" (13).

The term "once" (emphasized in the original) indicates a temporal order between the ability to "come into operation" and the grasping of the phenomenal state.  Perhaps you did not intend to be read in this way, though in that case I think "once" was not the right word to use here.

Even if you had used the conjunction "if," your statement would still suggest a temporal order.  For you would have then been saying that the grasping of the phenomenal state is a logical condition of the coming into operation of the abilities, and not a logical condition of the abilities themselves, which suggests that the grasping occured before the abilities were operational. 

I do not wish to nit-pick here.  If you did not mean to imply a temporal order, I will not belabor the point.  Though I would urge you to reconsider your phrasing.

We can leave the issue of temporal priority aside. As I said, my criticism is not limited to the issue of temporal priority, and your recent post indicates that you do not wish to adhere to any notion of temporal priority. 

This leaves us with the purely logical distinction you draw between phenomenal knowledge and the relevant abilities.  In my understanding of the ability hypothesis, this is the ability to identify objects of experience, be they colors, sounds, or the feeling of being balanced on a bicycle.

You support this distinction by illicitly shifting from the phrase "phenomenal experience" to "phenomenal knowledge."  It occurs here, just as your paper moves from page 12 to page 13:

"So the bike rider must wait on some phenomenal experience, which underwrites the ability he’s in the market to gain. Just like Mary.

On this account, it turns out that the elusiveness of abilities from academic study is derived from the elusiveness of phenomenal knowledge: it is because acquiring the relevant abilities involves knowing what certain experiences are like that the abilities elude the classroom—these inherit their elusiveness from the elusiveness of the phenomenal knowledge they depend upon."

The shift from talking about phenomenal experience to talking about phenomenal knowledge is made without any explanation.  Yet experience and knowledge are not identical.  Your example of the bike rider illustrates why the experience of riding a bike "underwrites" the relevant abilities, but it does not follow that the relationship between phenomenal experience and abilities also holds between phenomenal knowledge and abilities.  Your conclusion is based on the dubious assumption that the same relationship holds.

The distinction between phenomenal experience and phenomenal knowledge is crucial.  Phenomenal experience does not automatically confer knowledge.  You suggest as much yourself when you write: "it makes sense that Mary cannot have the abilities with respect to a phenomenal state until she is acquainted with it; that is, until she experiences what it is like and cognises it" (12).

She must "cognise" the experience, which--in my view--is another way of saying she must learn how to identify the experience.

(By the way, your use of the phrase "to be acquainted" recalls that close, close cousin of the ability hypothesis, the "acquaintance hypothesis," which I am tempted to classify as one face of the ability hypotheses; Martine Nida-Rümelin [see the link] groups the ability and acquaintance hypotheses together as the "No Propositional Knowledge" defense of physicalism.)

So, I would say that the experience of phenomenal redness is temporally and even functionally prior to having the relevant abilities.  But that does not mean that knowledge of that experience is in any way prior to having those abilities.  According to the ability hypothesis, as I understand it, phenomenal knowledge is identical to having the relevant abilities, and so could not be in any way prior.

In your response, you expressed an openness to the possibility of this identity.  Yet, the argument in your paper depends on it not being possible.  For, if the relevant ability is logically identical to phenomenal knowledge, then there is no epistemic priority.  Phenomenal knowledge is neither a "necessary ground" nor a "component" of the abilities.  Your current position seems to embrace the possibility of identity whilst also embracing its impossibility.  I do not see how such a position could be maintained.

To make this as clear as possible, I will draw your attention to what is written on page 15: "I conclude, phenomenal knowledge is basic to ability-knowledge, and not vice-versa." 

The "not vice-versa" is unwarranted, and it precludes the possibility of phenomenal knowledge being identical to the relevant abilities.  Without a justification for this conclusion, we are left with an unmotivated distinction between phenomenal knowledge and abilities.

Incidentally, in my view, all knowledge is a matter of abilities, and even what we call "propositional knowledge" is a matter of certain linguistic abilities.  I only point this out as a matter of tangential interest, though it might help you see why I sympathize with your concern (which I am about to address) that the mere fact that we are talking about abilities does not preclude the possibility of us talking about propositional knowledge.  Ability hypothesists must specify which abilities they are talking about and how they differ from those we associate with propositional knowledge.

Now, I grant that none of what I have written so far contradicts your main point, which is that the abilities Mary gains when leaving her black-and-white room may involve new factual knowledge.  And, of course, if the abilities necessarily involve new factual/propositional knowledge, then the ability hypothesis fails.  But do they? 

The ability hypothesis, as traditionally formulated, says that the abilities to imagine, remember, and recognize phenomenal experiences can be gained without acquiring new factual information, and that this kind of knowledge is all Mary gains upon leaving her black-and-white room.  This possibility stands, as far as I can tell.  Your analysis does not seem to dig any deeper than that attained by the Hypothesists on this matter.

Perhaps the ability hypothesis hasn't been proven satisfactorily, in which case it casts only a shadow of doubt over the knowledge argument, preventing us from deciding on its soundness until the relevant abilities are better understood.  This conclusion--which we might call agnosticism about the knowledge argument--seems much weaker than the conclusion you wish to draw, which is that the ability hypothesis has nothing to offer.  And this agnostic conclusion does not involve any reversal of epistemic priorities.  It does not require us to regard phenomenal knowledge as being anything other than abilities.

Agnosticism may be the position you wish to maintain, though it does not appear to be the conclusion of your paper.  In any case, it is not a conclusion your paper convincingly supports.  If the full extent of the ability hypothesis were as you say, and phenomenal knowledge were merely pronounced to be an ability without explanation, then I would agree that the ability hypothesis was exceedingly weak.  But you have not shown that the explanations are insufficient.

Rather than explicitly take up and defend other people's accounts of the ability hypothesis, I will draw your attention to my own.  While I haven't been published in any journals, I formulated this argument in an email to Torin Alter not so long ago, and it is posted on my blog [link].  I explain why Mary's new phenomenal knowledge amounts to the ability to identify new objects of experience, and why this involves using language in new ways without acquiring new factual information.  If you have the time to take a look at it, I would appreciate your thoughts.  I think it clearly demonstrates why Mary's new abilities do not imply new factual information, and so successfully undermines the knowledge argument.

Of course, my criticism of your paper does not rest on the success of my own arguments for the ability hypothesis.  I think I have shown where your paper errs and why it undermines the conclusions you wish to draw.  If I am mistaken in any of my analysis, I hope you will take the time to correct me.

Regards,

Jason

Stimulating reply Jason, thanks. I'll try to address most of your points.

The 'once' is indeed not meant temporally, but thanks for the warning regarding its connotations. It may be that I could have picked a better word.

On the line I'm pushing, Lewis et al may be right that knowing what it is like is identical to ability. However, these kinds of abilities are founded on cognising the quality of phenomenal states, and and retaining this information. The bike rider needs to retain his cognition of the feeling of balance in order to have the ability to ride. So, in this sense, what I later call (perhaps this is loose of me) 'phenomenal knowledge' is basic to the abilities. But 'basic to' 'founded on' and such don't imply that the abilities are distinct from the phenomenal knowledge. I claim that the phenomenal knowledge partially composes the abilities. Any difference between them is just that between parts and wholes. That's why the priority I argue for is not temporal, but yes logical, or perhaps ontological is better.

I don't see that this is embracing the possibility and impossibility of the identity at the same time. This might be a poor analogy, but if so another will do: Sartre says that existence is prior to essence, for man. But that doesn't - it seems to me - imply that essence is separate from existence for us. Rather these are aspects, and this is a logical or ontological issue. Or what about this: triangulairty is prior to having one's angles add up to 180 degrees. Or: the substance of an object is ontologically prior to its propertiedness. And so on. Now it may be that I should have used some such analogies to make my position clearer. Fair enough.

As for your charge that this view is no advance on what Lewis et al anyway propose, I think this is mistaken. I do not concede in the paper that the abilities to recognise etc can be gained without new factual knowledge. What I say is that the appeal to abilities, even if granted, doesn't touch the question of factuality. Lewis et al take the appeal to settle this question. There we differ.

I will, when I get a moment, have a look at your argument, thanks for the link. In the meantime let me know if this answer is any more satisfactory.

Sam

Hi again, Sam.

I appreciate the attempt to resolve this issue, but I don't see any progress in our discussion yet.  It would help if you tried to justify, or at least clarify, the apparently illicit shift from the phrase "phenomenal experience" to "phenomenal knowledge" that occurs on pages 12 and 13.

I agree with you when you say, "these kinds of abilities are founded on cognising the quality of phenomenal states, and retaining this information. The bike rider needs to retain his cognition of the feeling of balance in order to have the ability to ride."

Which is another way of saying that the ability to identify/recognise involves cognition and memory, a point already made by the ability hypothesis.  This is why I do not think your paper has advanced our understanding.

I do not see how your next sentence follows:  "So, in this sense, what I later call (perhaps this is loose of me) 'phenomenal knowledge' is basic to the abilities."

When combined with your insistence that this relationship is only one-way (you deny that the relevant abilities could be "basic" to phenomenal knowledge), the word "basic" in your formulation implies an asymmetrical relationship that precludes the possibility of identity.  Yet, you say: "But 'basic to' 'founded on' and such don't imply that the abilities are distinct from the phenomenal knowledge."

I do not know how I am supposed to understand this.  It seems blatantly self-contradictory.

You also say, "I claim that the phenomenal knowledge partially composes the abilities. Any difference between them is just that between parts and wholes."

Again, this seems to clearly indicate that you do not claim a possible relationship of identity, but instead claim there is a relationship of part-to-whole.  So, again, this would contradict your claim to openness regarding the possibility of identity here.  And, unfortunately, the examples/analogies you offer do not resolve this problem for me.  They only seem to indicate that you are not allowing for the possibility of identity here, even though you say you are.

Finally, you say:  "What I say is that the appeal to abilities, even if granted, doesn't touch the question of factuality. Lewis et al take the appeal to settle this question. There we differ."

I think I addressed this already, but I will clarify my point.  Your charge against the ability hypothesis is that it gets the explanatory order backwards, and thus does not further our understanding.  Yet, your argument for this conclusion depends on an illicit use of the phrase "phenomenal knowledge," or so I maintain.  Thus, you have not made a solid case against the relevance of the ability hypothesis.  To make that case, you would have to analyze the ability hypothesis in greater detail.  You would have to demonstrate that the ability hypothesis is incoherent (perhaps on the grounds that it fails to clearly define a relationship between the abilities and the types of knowledge involved) or false (by demonstrating that acquiring the relevant abilities necessarily involves acquiring new propositional knowledge).

On some understandings of the part/whole relationship, in fact, wholes are identical to their parts.

But lets leave aside identity, which seems to be causing some trouble, and talk instead about reduction.

Hypothesists want to reduce knowing what it is like to abilities. Because they think gaining abilities doesn't involve new factual knowledge, this reduction is meant to render Mary's new knowledge non-factual (and so unproblematic for physicalism).

But part of having the ability to ride a bike is knowing how it feels to balance. Part of being able to remember redness is knowing what redness is like: this follows if that ability is meant to be structurally similar to being able to ride a bike, wield choptsticks etc. If these aren't meant to be of some salient type structurally, then it's hard to even know what the Hypothesists mean by 'ability' and their claim becomes even more dubious.

This is phenomenal knowledge, and it appears to be necessary for having abilities of the type Hypothesists invoke.

But if this is so, reducing knowing what it is like to abilities doesn't answer the question about factuality. For the phenomenal knowledge part of the ability is untouched as regards its factuality (or not) by the observation that it forms part of an ability.

One wants to say that a certain cake is nut-free. One could argue that the making of a cake is the making of a date-cake, that date-cakes don't involve nuts, and so that the cake is nut-free. But if part of making a date cake might be putting nuts in, then reducing the making of the cake to the making of a date cake won't answer the question whether the cake contains nuts. That will depend on whether making a date cake is a nut-free operation. And this is left, so far, an open question. So the claim that what is happening is the making of a date-cake is just irrelevant to (doesn't advance) the original claim of nut-freeness.

Analogously, the Hypothesist should simply stake their claim that knowing what it is like is non-factual knowledge. Adding the stuff about abilities is no advance on that claim, which may or may not be true in its own right.

Hi Sam,

I'm going to be a bit more formal in my argument to increase the chances of us coming to agreement.

You say, "But part of having the ability to ride a bike is knowing how it feels to balance. Part of being able to remember redness is knowing what redness is like: this follows if that ability is meant to be structurally similar to being able to ride a bike, wield choptsticks etc."

I don't think that follows.  Your argument, as stated, focuses on two sentences:

        A:  Part of having the ability to ride a bike is knowing how it feels to balance.

        B:  Part of being able to remember redness is knowing what redness is like.

We agree that knowing what redness is like is phenomenal knowledge, and that this is what the ability hypothesis attempts to explain in terms of non-factual (non-propositional) knowledge.  (You call this a reduction, though I am weary of that term, as it suggests an optional procedure, as though phenomenal knowledge could be explained in some other way.)

Thus, your argument may be broken down as follows:

        (1)  Sentence A is true.

        (2)  If A is true, then B is true.  (This is said to be based on the appeal to "structural similarity.")

        (3)  If B is true, then the ability to remember redness is predicated upon phenomenal knowledge of colors.

        (4)  If the the ability to remember redness is predicated upon phenomenal knowledge, then phenomenal knowledge is
              epistemically prior to and cannot be explained in terms of the ability to remember redness.

After demonstrating that A is true, you conclude that the ability hypothesis fails to explain phenomenal knowledge and so cannot properly answer the knowledge argument.

The only one of those four points I challenge is (2), which seems highly problematic.

You say (2) is based on the requirement of structural similarity.  This isn't so simple.  It is apparent that the abilities to ride a bike and wield chopsticks are structurally more complicated than the ability to identify colors.  In fact, it seems possible that the ability to identify colors is basic, and not composed of sub-abilities.  (It may be composed only of capacities, and not learned skills.)  However, I grant that both sets of abilities involve phenomenal knowledge and phenomenal experience.  So we can stipulate that there are some similarities, but we should not ignore the differences.

(1) is not problematic.  The ability to ride a bike involves a whole set of other abilities, including the ability to balance on a bike.  The relationship between the two abilities (bicycle riding and balancing on a bicycle) is observable and easy to define.  Thus, sentence A is true.

But now consider B again.  How are we to understanding the relationship between remembering redness and knowing what redness is like?

Knowing what redness is like is the phenomenal knowledge in question, which the ability hypothesis regards as the ablity to recognize/identify the color red.  Thus, B can be restated as follows:  Part of being able to remember redness is being able to recognize/identify the color red.

This looks backwards.  We should rather say that part of being able to recognize/identify the color red is being able to remember redness.  (How can you identify red if you do not remember that it is red you are seeing, and not some other color?)  So the correct conclusion to draw here is that phenomenal knowledge is constituted by abilities, as argued by the ability hypothesis, and not vice versa.  This forces us to reject (2).

You might wish to salvage your argument by reformulating it in terms of some other ability or set of abilities, and not specifically the ability to remember redness.  You might thus replace B with some other sentence, such as the claim that part of being able to recognize/identify redness is knowing what redness is like.  But this takes us right back to the issue of identity.  Knowing what redness is like is the ability to recognize/identify redness.  So we should remove the "part of" from the beginning of the sentence, and in so doing deny the possibility that phenomenal knowledge is prior to the ability to recognize/identify colors.

I think the above is sufficient to undermine your paper's argument.  I will now address the date-cake argument.

You say, "That will depend on whether making a date cake is a nut-free operation. And this is left, so far, an open question. So the claim that what is happening is the making of a date-cake is just irrelevant to (doesn't advance) the original claim of nut-freeness."

First, the argument you are making with this analogy is quite different from the argument you make in your paper, which is the argument I critiqued above.  This date-cake argument is an argument for an agnostic position regarding the ability hypothesis.  It is made on the grounds that the ability hypothesis has not been successfully established.  It is thus quite distinct from your argument that phenomenal knowledge is epistemically prior to the relevant abilities, and so even if it is valid, it does not force us to conclude that the ability hypothesis is hopeless.

Second, it is only your conjecture that the case has not been made for the ability hypothesis.  You have not demonstrated that this is "an open question."

Third, as I have argued (see the blog post I referenced earlier), it can be successfully demonstrated that the relevant abilities do not involve new propositional knowledge.  Thus, the agnostic position is not correct.  The case for the ability hypothesis has been made. 

In conclusion . . . If you wish to take on the ability hypothesis, you will have to survey the various developments it has undergone over the past 30 years.  You might be able to show that the case for the ability hypothesis has not been made yet, and that more argumentative or evidentiary support is required before it can be unequivocally adopted.  However, I do not think you will be able to dispense with the ability hypothesis in the manner you attempt in your paper.  I do not see any hope for that argument, as it is predicated upon an error.

I don't really think your breakdown represents my argument accurately.

In the paper I give an appealing analysis of the elusiveness of ability knowledge. On this analysis, that knowledge is founded on knowing what certain phenomenal states feel like. In which case the hypothesists have the explanatory priority of abilities and phenomenal knowledge inverted, with the result that their reduction doesn't touch the question of the factuality of phenomenal knowledge. Now my analysis of ability can be rejected by the hypothesist. But I note that the analysis has explanatory power in its favour. Any rejection of it will simply seem to be motivated by a desire to save the hypothesis as a means of defending physicalism. My argument, by contrast, is neutral about physicalism.

To repeat, the claim that AH leaves the factuality question open is not an agnostic position, it is damaging to AH. AH is meant to advance, even decide, the claim of the non-factuality of phenomenal knowledge. If we can end up agnostic about whether abilities are fact free, then this shows that AH is irrelevant to the issue. The hypothesist should simply stick by their fact-free claim without purporting to support it via abilities.

I am unpersuaded by your argument that abilities do not involve new propositional knowledge.

Finally, a note on etiquette: if you've read the bibliography of my paper, you'll see that it takes in quite a lot of the AH literature, right up to the present day. And I still fail to see what 'error' it is that my paper is making. Lastly, words like 'salvage' used in the context that you do are all well and good when used in journal articles or a formal setting, but are less appropriate to personal communication.

Sam,

I do not see where my criticism of your paper fails.  If you have time, I would appreciate a few words about where you think I have gone astray.

Also, thanks for taking the time to comment on my argument for the ability hypothesis.  I'm sorry you didn't find it persuasive.  I would like to clean it up a bit and perhaps make it stronger, though I think that, even in its current formulation, its strengths outweigh its weaknesses.

About the date-cake argument, it seems to be saying, among other things:  (a) one could argue that making a date cake is a nut-free operation; (b) so far, we do not know if date cakes are really nut-free--i.e., the argument that making date cakes is nut-free has not been proved.  We can only be agnostic about whether or not date cakes are nut-free.  I interpreted this as meaning that we can only be agnostic about whether or not Mary's new abilities are free of factual information.  I must have been mistaken, since that is clearly not what you meant.  In any case, I don't see this as having any consequence for the argument in your paper.

Finally, about etiquette.  I am sorry if any of my comments seem inappropriate or offensive to you.  I'd like to clarify my intentions, in case there was any misunderstanding.

First, you made a point about your bibliography, which must have been a response to my comment about the need to survey the literature.  While you do mention several instantiations of the ability hypothesis, you do so only to construct a general framework for discussing it, and then proceed to argue that the ability hypothesis, as a general strategy, does not address the relevant issues raised by the knowledge argument.  You seem to take it as a given that the ability hypothesis merely claims, and does not demonstrate, that Mary's newfound abilities do not involve new factual knowledge.  But haven't arguments have been put forward aimed at demonstrating that Mary's new abilities do not involve new factual knowledge?  Perhaps the arguments are all flawed, and a different survey of the literature might show that.  (And, if they are all flawed, wouldn't the correct conclusion be the agnostic one, until it could be shown that new factual knowledge was in fact implied by Mary's new abilities?)

Second, you said you still do not see what 'error' your paper is making.  I was referring to the one evidenced in your transition from page 12 to page 13, as you illicitly shift from the phrase "phenomenal experience" to the phrase "phenomenal knowledge."  Your analysis seems to confuse phenomenal experience with phenomenal knowledge, leading to an incoherent distinction between phenomenal knowledge and abilities. I thought I had made that clear, and thus did not see the need to point it out again.

Third, you find my use of the word "salvage" inappropriate.  I suppose it can be difficult finding the right tone for discussions such as these.  I thought the word "salvage" was both accurate and respectful, and I am sorry if you found it offensive.  I will try to avoid using it in the future.

Regards,

Jason

Hello to both, 
Your discussion has been quite helpful. I think I have had similar worries about your paper, Sam, as Jason, although I am not as confident as Jason that the problem is so serious. 

Your argument is, I think, that all abilities of the relevant kind - that is, abilities which we call 'skills', partly because one can be better or worse at them, abilities which often take practice, or at the very least hands-on 'experience', abilities which are, in some important way, different from being able to unlock a combination lock - depend on, or are at least partially constituted by phenomenal knowledge. And, I think you intend 'phenomenal knowledge' to be 'knowing what something is like'. 

You defend this thesis from potential counterexamples and build quite a strong case for thinking that abilities of this special kind all involve phenomenal knowledge. 

You conclude from this that abilities to imagine, recognise, remember experiences must also depend on or be partially constituted by phenomenal knowledge. 

Then you argue that it follows from this that the AH has no impact on the knowledge argument. This is because, even if knowing what it is like to experience red just is being able to imagine, reocognise, remember an experience of red, as Lewis and co claimed, because these abilities - like all abilities of this kind - are partially constituted by some phenomenal knowledge or other (that is, by knowing what something is like) the nature of phenomenal knowledge in general remains an open question. 

I have two issues with this argument. 

1) I think it is open to a supporter to AH to simply say: yes, most abilities of the relevant kind depend on phenomenal knowledge, most abilities depend on knowing what some experience is like - but being able to imagine, recognise and remember an experience is an exception.These abilities depend on having had the relevant experience - this is a brute psychological fact - but they don't depend on knowing what any experience is like. It is reasonable to expect these abilities to be an exception because, unlike all the other abilities in the relevant kind, imagining, recognising and remember are basic actions, i.e. not actions one does by doing something else. If you don't do anything else in order to imagine, why do you need to be able to do anything else to be able to imagine? (Recall: if the AH is true - something you assume for the sake of argument - then phenomenal knowledge is ability to imagine, recognise, remember so saying that being able to imagine, recognise, remember something depends on phenomenal knowledge just is to say that being able to imagine, recognise, remember something depends on being able to imagine,recognise, remember something - presumably something else.) 

2) I think you have to concede that if all abilities, including the ability to imagine, recognise, and remember an experience, are partially constituted by phenomenal knowledge, then the phenomenal knowledge the ability to imagine, recognise and remember an experience is partially constituted by is phenomenal knowledge of that very experience. Being able to imagine, recognise, remember an experience does not depend on knowing what that experience is like, knowing what that experience is like just is having the relevant abilities. One thing cannot be partially constituted by itself. I thus assume the claim you were arguing for is: being able to imagine, recognise, remember a certain experience, e.g. of seeing red, depends on knowing what some other experience is like, e.g. of imagining objects. But now I think a defender of the AH can just say: even if being able to imagine, recognise, remember a certain experience, e.g. of seeing red, depends on knowing what some other experience is like, e.g. of imagining, knowing what that other experience is like is also just being able to imagine, recognise, and remember it.

You might worry this generates a regress. But I'm not certain it's a vicious regress. In any case, the nature of phenomenal knowledge does not remain an open question, if the AH is true. 

I did wonder though, whether some of the things you discuss in your paper could be used to formulate an argument against the AH, i.e. an argument that the hypothesis Lewis and co put forward just cannot be right. This would be to take a stronger line than the line you endorse in your paper, but maybe some of the considerations about how abilities depend on phenomenal knowledge could be put to this alternative task. 

My hunch is that this wouldn't work, that you couldn't argue that phenomenal knowledge just can't be a set of abilities because all abilities depend on phenomenal knowledge because the objections I raise above could be redeployed. 

I'm not sure about this though. What do you think? 

Best wishes, 
Andrea
 

Hello Andrea,
I think proponents of AH can accept at least one of Sam's points:  that AH still leaves open the question of how phenomenal knowledge works.  But that is not a mark against AH, as far as I can tell.  As you observe, AH says that phenomenal knowledge "just is" the ability to recognize, remember and imagine an experience.  To fully account for that kind of knowledge, we would need some understanding of the way our brains (or perhaps a larger class of systems) can perform such functions.  But AH was never meant to fully account for phenomenal knowledge.  It was meant to provide an intuitively appealing response to the knowledge argument against physicalism.  And in that respect, I still think it is successful.

I think David Lewis, back in 1988, was speaking directly to arguments of Sam's sort when he wrote:  

"A friend of phenomenal information will agree, of course, that when we learn what an experience is like, we gain abilities to remember, imagine, and recognize.  But he will say that it is because we gain phenomenal information that we gain the abilities.  He might even say the same about other cases of gaining know-how: you can recognize the C-38 when you have phenomenal information about what it's like to see that shape, you can eat with chopsticks or wiggle your ears when you gain phenomenal information about the experience of doing so, and so on.  What should friends of the Ability Hypothesis make of this?  Is he offering a conjecture, which we must reject, about the causal origin of abilities?  I think not.  He thinks, as we do, that experiences leave distinctive traces in people, and that these traces enable us to do things.  Likewise being taught to recognize a C-38 or to eat with chopsticks, or whatever happens on first wiggling the ears, leave traces that enable us to do things afterward.  That much is common ground.  He also interprets these enabling traces as representations that bear information about their causes.  (If the same traces had been caused in some deviant way they might perhaps have carried misinformation.)  We might even be able to accept that too.  The time for us to quarrel comes only when he says that these traces represent special phenomenal facts, facts which cannot be represented in any other way, and therefore which cannot be taught in physics lessons or even in parapsychology lessons.  That is the part, and the *only* part, which we must reject.  But that is no part of his psychological story about how we gain abilities.  It is just a gratuitous metaphysical gloss on that story."  ("What Experience Teaches," in Mind And Cognition: A Reader, 1990, p. 517).