From PhilPapers forum Philosophy of Mind:

2009-05-05
The 'Explanatory Gap'
Reply to Stevan Harnad

Hi Stevan.  I enjoyed your response. We're actually much closer in our views than you realize.  I've read your papers on the symbol grounding problem (years ago) and agree with your view.  In fact, I've reformulated that concept just slightly (from my engineering perspective) and plan on referencing the symbol grounding problem in a paper I'm working on.  

We talk a different language - philosophy versus engineering, and I think that difference is what may be confusing.  The only real difference we seem to have as far as I can tell, is that you feel there is no fundamental problem with there being a reliable correlation plus no mental causation.  I'm not suggesting there is no reliable correlation, nor am I suggesting there is mental causation.  I'm only suggesting that the two are incompatible.  I don't really care which one gets discarded, but I don't see any way we can have both reliable correlation AND no mental causation.  In contrast, you seem to be claiming that there is both a reliable correlation AND mental causation.

To your points.

SH: (1) The most important is the ontic/epistemic distinction: Distinguish been what there really is (ontic) and what we can know about what there really is (epistemic), e.g., what we can observe or measure. Although it was fashionable for a while (though one wonders how and why!), it will not do to say "I shall assume that what I can observe and measure is all there is and can be." Not if you want to address the question of the explanatory gap, rather than simply beg it! 
 

Agreed.  Just to clarify however when you quote others, "I shall assume that what I can observe and measure is all there is and can be." I agree this begs the question regarding the explanatory gap.  Just to be sure we agree, I would claim there is no way to come up with a measure (or to impliment a physical interaction as you put it) that can discern feeling.  Looking at all the possible methods of physical interaction, I can't think of a single one that might for example, interact with feeling matter while not interacting in an identical mannor with matter that is not feeling.  Putting a chunk of matter into a physical machine that can test for feeling and report it is not possible because there is no physical interaction that can tell us that feeling exists and what the feeling is like.  If there was, there would be no explanatory gap.  I believe this is what you mean by incommensurable, is that correct?  If so, I agree.

SH: (2) Observation and measurement . . .

SH: (3) Your third equivocation . . .

I fully agree with 2 and 3.  I think we're on the same page.  The arctic sensors have no feeling (and no meaning), nor the computer that records them.  We obviously agree there is an explanatory gap.  Physical interactions don't explain feeling, and the arctic sensors are not capable of meaning. 

SH: It is perfectly natural, indeed universal, to believe and feel that our feelings matter, and that most of what we do, we do because we feel like doing it, and not just because functing is going on, of which our feelings are merely correlates -- correlates of which we do not know the causes, and, even more important, correlates which themselves have no effects of their own, and we cannot explain how and why they are there at all. (That, yet again. is the f/f problem and the explanatory gap.)

Yes, I agree there is a correlate.  The issue I have is only that it seems inconsistent to suggest that feelings correlate AND are not part of any causal chain.  I don't have the answer to how mental causation might influence physical phenomena, and I have no solution to the explanatory gap.  I only want to explore the possibility that; if feelings correlate then they must be in some sense, part of the causal chain.  By "causal chain" I mean there is a series of causal influences, whether they be classical or quantum scale. 

Just one note on what is meant by causal chain.  A causal chain can:
(1) be all on one physical level.  Transistors on a chip represent a causal chain at a single level in which one transistor influences another. 
(2) move up or down levels such as transistors and transducers causing computers to change state.  The computer's change of state causes valves and pumps to operate which causes fluid to flow in a predetermined mannor, eventually leading to a rocket being launched. 

It seems to me we're much closer to agreement than you realize from the remainder of your response.  I would also like to point out that your symbol grounding problem is perfectly acceptable to me.  However, I might have a slightly different perspective.  In my view, there is a meaning in my head that must be converted to symbolic form in order for there to be a physical transmission of that meaning.  Neurons in my brain convert the meaning into a symbolic vocalization.  That vocalization has no meaning itself, it is simply a symbolic representation of the meaning I had in my head.  There is a transmission of symbolic acoustic pulsations in the air that travel out away from my mouth in a spherical wave front.  A pair of ears converts these pressure waves in the listener's ear to neuron interactions and then back into meaning in someone else's head.  However, if one claims mental causation is not part of the causal chain, I will claim that the above interpretation is unacceptable.  For anyone who refuses mental causation, they must deny the above is true because feeling has no causal influence and thus can not make itself known.  I can not reliably report my feelings! 

Perhaps using Kim's framework for a causal chain, and how mental causation is problematic, might help ground the conversation.  Here is my very brief overview as I understand the salient points:
We will assume a physical basis P at time t produces mental state M (or any higher level property) at that time (mind-body supervenience).  Any identical physical basis P should have an identical mental state M. 

Similarly, at time t + dt, physical basis P* causes mental state M*. 

Kim says, "M* is instantiated on this occasion: (a) because, ex hypothesi, M caused M* to be instantiated; (b) because P*, the physical supervenience base of M*, is instantiated on this occasion. "

In other words, if we want to claim that M causes M* we have two causes for M*.  We have P causing P* on which M* supervenes and we have M causing M* directly.  We can't have 2 causes, so we rule out mental causation.  Similarly, we might suggest M* causes P* (downward causation) which again results in 2 causes for M*.

If we refuse mental causation, we can't accept that M causes M* nor can we accept that M* causes P*.  All we can accept is for P to cause M and P* to cause M*.  Nothing more, right?  No wait, there's one more thing you want but you don't say it.  You've hidden it behind the 'telekenisis' door.  ;) 

You want M to correspond to P AND do so reliably.  You want there to be a purely functed report of M by P. 

The problem as I see it isn't that you refuse M causing M*.  That much is fine, I agree.  That is one mental state causing another.  We don't need that since the supervenience base P is sufficient to produce M.

The problem as I see it is that in order for there to be a RELIABLE report of M (or M*) by P (or P*), P must not only have knowledge of P, it must ALSO have knowledge of M.  However as soon as you allow that P can reliably report M, you've admitted to M being part of the causal chain.  M is being acted upon when reported by P.  In order for P to reliably report M, P must funct a description of M in some way.  How does P create this description of M?  How can P funct a description of M if P can't implement a physical interaction - a physical interaction that is able to correlate that functed description with M? 

It's no good saying P reports M because P is reporting about itself because M can not be physically measured.  If you want to now claim that if we measure the momentum or pressure or some other physical property closely enough then we'll be able to measure feeling, I disagree.  M can not be found out about by observing physical interactions, so P can't find out about M by monitoring it's own physical interactions, P can only (at best) report it's own physical interactions.  Also, there is no meaning associated with those physical interactions, and when I say meaning here I mean that the physical interactions are not meaningful per your symbol grounding problem (which I agree with).

One might contest that by monitoring neural interactions we can determine what felt experience is occuring, but to do this we need emperical correlations.  We need to know that when certain neurons fire, this firing of neurons correlates to reported feelings, so this is no help for us.  We are not measuring feeling, we are measuring something physical and we are making the assumption there is a reliable correlation.  If there were any way of monitoring neural interactions and determining what feeling those neurons were experiencing, there would be no explanatory gap and we could stop arguing about feelings and explanatory gaps.

Just to point out where IMHO Kim goes wrong on all this, Kim said:

KIM: Suppose that pain could be given a functional definition -- something like this: being in pain is being in some state (or instantiating some property) caused by tissue damage and causing winces and groans.  Why are you experiencing pain?  Because being in pain is being in a state caused by tissue damage and causing winces and groans, and you are in neural state N, which is one of those states (in you, or in systems like you) that are caused by tissue damage and that cause winces and groans.  Why do people experience pain when they are in neural state N?  Because N is implicated in these causal/nomic relations, and being in pain is being in some state with just these causal/nomic relations.  It is clear that in this way all our explanatory demands can be met.  There is nothing further to be explained about why pain occurs, or why pain occurs when neural condition N is present.

Just as Kim tries to convince us that pain correlates with tissue damage, winces and groans, our intuition tells us this is true.  However, we need to stop thinking our intuition is leading us only to true conclusions.  If we can't logically support this intuitive deduction, we must stop using our intuition. 

Best regards,
Dave.

(Posted 5/3/09, 9:30 PM EST)