Stephen Law has new arguments against physicalism (which is, approximately, the view that the account of the world given by physics is good enough to explain everything). He thinks conscious experience can’t be dealt with by physics alone. There is a well-established family of anti-physicalist arguments supporting this view which are based on conceivability; Law adds new cousins to that family, ones which draw on inconceivability and are, he thinks, less vulnerable to some of the counter-arguments brought against the established versions.

What is the argument from conceivability? Law helpfully summarises several versions (his exposition is commendably clear and careful throughout) including the zombie twin argument we’ve often discussed here; but let’s take a classic one about the supposed identity of pain and the firing of C-fibres, a kind of nerve. It goes like this…

1. Pain without C-fibre firing is conceivable
2. Conceivability entails metaphysical possibility (at least in this case)
3. The metaphysical possibility of pain without C-fibre firing entails ?that pain is not identical with C-fibre firing.
C. Pain is not identical with C-fibre firing

(It’s good to see the old traditional C-fibre example still being quoted. It reminds me of long-gone undergraduate days when some luckless fellow student in a tutorial read an essay citing the example of how things look yellow to people with jaundice. Myles Burnyeat, the tutor, gave an erudite history of this jaundice example, tracking it back from the twentieth century to the sixteenth, through mediaeval scholastics, Sextus Empiricus (probably), and many ancient authors. You have joined, he remarked, a long tradition of philosophers who have quoted that example for thousands of years without any of them ever bothering to find out that it is, in fact, completely false. People with jaundice have yellowish skin, but their vision is normal. Of course the inaccuracy of examples about C-fibres and jaundice does not in itself invalidate the philosophical arguments, a point Burnyeat would have conceded with gritted teeth.)

I have a bit of a problem with the notion of metaphysical possibility. Law says that something being conceivable means no incoherence arises when we suppose it, which is fine; but I take it that the different flavours of conceivability/possibility arise from different sets of rules. So something is physically conceivable so long as it doesn’t contradict the laws of physics. A five-kilometre cube of titanium at the North Pole is not something that any plausible set of circumstances is going to give rise to, but nothing about it conflicts with physics, so it’s conceivable.

I’m comfortable, therefore, with physical conceivability, and with logical conceivability, because pretty decent (if not quite perfect) sets of rules for both fields have been set out for us. But what are the laws of metaphysics that would ground the idea of metaphysical conceivability or equally, metaphysical possibility? I’m not sure how many candidates for such laws (other than ones that are already laws of physics, logic, or maths) I can come up with, and I know of no attempt to set them out systematically (a book opportunity for a bold metaphysician there, perhaps). But this is not a show-stopper so long as it is reasonably clear in each case what kind of metaphysical rules we hold ourselves not to be violating.

Conceivability arguments of this kind do help clarify an intuitive feeling that physical events are just not the sort of thing that could also be subjective experiences, firming things up for those who believe in them and sportingly providing a proper target for physicalists.

So what is the new argument? Law begins by noting that in some cases appearance and reality can separate, while in others they cannot. So a substance that appears just like gold, but does not have the right atomic number, is not gold: we could call it fool’s gold.  A medical case where the skin was yellowish but the underlying condition was not jaundice might be fool’s jaundice (jaundice again, but here used unimpeachably). However, can there be fool’s red?  If we’re talking of a red experience it seems not: something that seems red is indeed a red experience whatever underlies it. More strongly still, it seems that the idea of fool’s pain is inconceivable. If what you’re experiencing seems to be pain, then it is pain.

Is that right? There’s evidently something in it, but what is to stop us believing ourselves to be in pain when we’re not? Hypochondriacs may well do that very thing. Law, I suppose, would say that a mistaken belief isn’t enough; there has to be the actual experience of pain. That begins to look as if he’s  in danger of begging the question; if we specify that there’s a real experience of pain, then it’s inconceivable it isn’t real pain? But I think the notions of mistaken pain beliefs and the putative fool’s pain are sufficiently distinct.

The inconceivability argument goes on to suggest that if fool’s pain is inconceivable, but we can conceive of C-fibre firing without pain, then C-fibre firing cannot be identical with pain. Clearly the same argument would apply for various mental experiences other than pain, and for any proposed physical correlate of pain.

Law rebuts explicitly arguments that this is nothing new, or merely a trivial variant on the old argument. I’m happy enough to take it as a handy new argument, worth having in itself; but Law also argues that in some cases it stands up against counter-arguments better than the old one. Notably he mentions an argument by Loar.  This offers an alternative explanation for the conceivability of pain in the absence of C-fibre firing: experience and concepts such as C-fibre firing are dealt with in quite different parts of the brain, and our ability to conceive of one without the other is therefore just a matter if human psychology, from which no deep metaphysical conclusions can be drawn. Law argues that even if Loar’s argument or a similar one is accepted, we still run up against the problem that conceiving of pain without C-fibre firing, we are conceiving of fool’s pain, which the new argument has established in inconceivable.

The case is well made and I think Law is right to claim he has hit on a new and useful argument. Am I convinced? Not really; but my disbelief stems from a more fundamental doubt about whether conceivability and inconceivability can actually tell us about the real world, or merely about our own mental powers.

Perhaps we can look at it in terms of possible worlds. It seems to me that Law’s argument, like the older ones, establishes that we can separate C-fibre firing and pain conceptually; that there are in fact possible worlds in which pain is not C-fibre firing. But I don’t care. I don’t require the identity of pain and C-fibre firing to be true a priori; I’m happy for it to be true only in this world, as an empirical, scientific matter. Of course this opens a whole range of new cans of worms (about which kinds of identity are necessary, for example) whose contents I am not eager to ingest at the moment.

Still, if you’re interested in the topic I commend the draft paper to your attention.

 

 

35 Comments

  1. 1. Jochen says:

    Interesting piece, thanks for highlighting it. However, I still think there’s an easy out to conceivability arguments, which is that what we can conceive of may be subject to formal limitations that, however, do not dictate what is metaphysically possible. An example of that is the well-known fact that truth outruns provability in mathematics: in some sense, a given formal system F (or a reasoner governed by that system) can ‘conceive of’ its Gödel sentence G being both true and false, since it is consistent with both options; yet, G either is true or false, and hence, this sort of ‘conceivability’ does not imply (logical) possibility. Furthermore, there’s nothing ‘extra’ needed to fix G’s truth value: it is true if the axioms of G are consistent, which they either are, or fail to be.

    Likewise, it may be conceivable to us that the physical facts do not entail all the experiential facts in the same way—they may not be deducible from the physical facts, but are still fully determined by them. But this non-deducibility makes it appear conceivable that they could be absent (in the zombie argument) or different (in inverted spectra), while the physical facts remain the same. But that’s not in fact possible.

  2. 2. SelfAwarePatterns says:

    “but my disbelief stems from a more fundamental doubt about whether conceivability and inconceivability can actually tell us about the real world, or merely about our own mental powers.”

    I think this is the key weakness with any conceivability argument. We can conceive of all kinds of things, but that has no bearing on whether they exist. The entire history of science has been one of knocking down popular conceptions, and uncovering things that would have previously been very difficult to conceive of. For example, no one conceived of the bizarre nature of quantum mechanics before experimental results relentlessly forced the issue, and 19th century science’s conception of the aether turned out to be meaningless for whether it existed.

    In my mind, the answer to the c-fiber question is that what makes c-fiber signals pain is the brain’s interpretation of those signals. If the brain (particularly the anterior cingulate cortex and somatosensory cortex) are interpreting a painful situation when there is no signal coming from those fibers, then it’s still pain. That isn’t to say that pain is in any way a voluntary thing. Much of our brain’s lower level processing isn’t. Just that pain becomes pain in the brain. Signals along c-fibers and a-fibers are just the most common causes.

  3. 3. Richard Kennaway says:

    At a tangent to the jaundice tangent, I recall that a late colleague of mine had an operation to remove cataracts from his eyes. The normal procedure is to operate on one eye first, and if that goes well, only then operate on the second. So for a few weeks he had one eye with the cataract successfully removed and one still with a cataract. He realised that for years he had unknowingly been seeing the world tinged with yellow.

    To measure the effect he set up a computer screen with one half white and the other half filled with a colour he could adjust, and used a sheet of card so that the eye with the cataract saw only the white side and the other eye the adjustable side. Then he adjusted the colour until he could not tell the difference between them. He could thus show anyone what the difference was, including his eye doctor, who had never seen this done before.

  4. 4. John Kubie says:

    There’s irony in the c-fiber example. Perhaps it speaks of the remove of philosophy from science. C-fibers do not equal pain. Lots of examples. A subclass of c-fibers are now refered to as nociceptors, not pain fibers. Pain is a percept that occurs in the brain. Certainly, peripheral nerve afferents frequently lead to pain percepts. But in phantom limb pain or many chronic pains, there is pain without nociception firing. The converse, nociception firing without pain perception also common. Finally, the afferent inputs that commonly lead to pain are not just nociception firing; they are a balance between nociceptors and mechanorecptors. If concomitant mechanoreception activity is high, pain not perceived. That’s the likely mechanism for dorsal column stimulation (mechanoreceptors) relieving pain.

  5. 5. Jochen says:

    SelfAwarePatterns:

    We can conceive of all kinds of things, but that has no bearing on whether they exist.

    That’s not the argument being made. All the conceivability arguments need to establish is possibility; and once you get right down to it, it’s surprisingly hard to argue against that. After all, if you properly conceive of something, you have some sort of model of it in mind; but why, if that model can exist, should the thing it models be impossible? It’s just a different instantiation of the same sort of structure.

    It’s at least a very strong intuition that we can’t conceive of logically impossible things; but from there, it’s just a small step (laid out in great detail by Chalmers et al.) to conclude that what we can conceive of must be possible.

    In the end, I think the line of attack that simply questions the link between conceivability and possibility is too weak to make much progress. The possibility that we sometimes merely might think that we can conceive of something—i.e. we might fail to see the contradictory nature of complex issues, and thus, be essentially deceived into believing it to be conceivable—fares somewhat better, but in the end, the sorts of processes that ought to be accompanied by experience seem to be fairly simple to conceive of as lacking it, so this, too, doesn’t strike me as a very plausible option.

    That’s why I instead strike at the link between logical and metaphysical possibility: there are logically consistent states of affairs which are nevertheless impossible. It might be logically possible that a certain Diophantine equation has a finite number of solutions, as well as it having an infinite number; but only one of those is actually true. This sort of thing is, I think, where one needs to start chopping in order for the whole edifice of conceivability arguments to come tumbling down.

  6. 6. arnold says:

    Theoretical physics rest in no beginning no end…
    Cconceivability rest in observation…
    Suffering rest in interactions…

    …and of course their is more…

  7. 7. SelfAwarePatterns says:

    Jochen #5,
    I see your point. I suppose it depends on how rigorous the conception is. Astrophysicists mathematically derived the existence of black holes before there was any observational evidence for them. That was a rigorous conception. But even they didn’t accept the existence of these objects before there was at least some evidence to infer them.

    From what I’ve read about these philosophical conceptions, they don’t have anything like the rigor that the black hole hypothesis had. Indeed, per Sean Carroll, they seem to ignore what’s known about everyday physics, not to mention the lion share of evidence from neuroscience. But even if the rigor was there, we should be at least as cautious as the astrophysicists were with the black hole concept.

  8. 8. David Duffy says:

    More on those C-fibres 😉 Is this conceivable?

    “Panksepp (1998) found that low, nonsedative doses of the physical pain-killer morphine quelled the distress of separation from the caregiver in infants of multiple mammalian species (Herman and Panksepp, 1978; Panksepp et al., 1978a, 1978b). This data indicates that the endogenous opioid system is involved in signaling the distress of separation from conspecifics. According to recent neuroimaging data in humans, this role for the opioid system appears to apply to the distress associated with the severance of a social bond in adulthood as well.”

    http://scan.oxfordjournals.org/content/5/2-3/203.full

    Further to the jaundiced eye, I had not until recently heard of the story of John Dalton’s eye – gives new meaning to seeing things from another’s point of view.

  9. 9. James of Seattle says:

    The key here, which I think SelfAwarePatterns (SAP) meant, and Jochen picks out and then waves his hand at, is that there are some concepts which seem conceivable and yet are self-contradictory. It shouldn’t matter how hard it is to conceive the concept. Even simple concepts, like consciousness without physical activity, may be self-contradictory.

    Suppose, for example, the basic unit of consciousness is the interpretation of a (semiotic) symbol by a physical system. For example, when damage to a finger results in a burst of neurotransmitter at the ends of an axon (C-fiber?) deep in the brain, that neurotransmitter can be considered a symbol of the specific damage to the finger. But the agent doing the interpreting may be competent without comprehension. The agent may simply be generating a response without knowing why it is generating the response. The agent knows nothing of neurotransmitters. But we see that the agent is set up (physically) so that it’s response to that neurotransmitter is what you might expect as an appropriate response to finger damage. The agent is simply intuiting “damage”. If the response generated by the agent resembles “drop everything! Do everything possible to make the signal stop as soon as possible”, then we might call that intuition pain.

    Now the above may or may not be the correct explanation of the basis of consciousness, but it possibly (conceivably?) is. If it is, then the p-zombie is not actually conceivable without self-contradiction.

    *

  10. 10. arnold says:

    The cosmos the universe Sees Itself…nervousness abounds…

  11. 11. John Davey says:


    For example, no one conceived of the bizarre nature of quantum mechanics before experimental results relentlessly forced the issue

    That’s not ‘inconceivable’. That’s ‘not yet conceived’, which is totally different. Quantum mechanics was always conceivable. If it wasn’t, we wouldn’t know about it. What is not ‘conceivable’ is a linear algebra problem to a dog (cognitive scope). Or, how a quantity in a physical model based on scalars could lead to a mental quality ontological scope). There is no conceivable path from one to the other, as both mathematical models and mental qualities are in completely separate ontologies and never the twain shall meet. That’s inconceivable.

    That’s why I’m not sue what Law is adding. The physical model of the causes/correlates of a pain are ever going to be ontologically identical with the pain itself. As usual what it boils down to is trying to persuade some physicalists that conscious experiences exist. Isn’t that a waste of time ? Is there any point in doing a Hitchens and telling people not to believe in God ?

    J

  12. 12. SelfAwarePatterns says:

    John #11,
    “Or, how a quantity in a physical model based on scalars could lead to a mental quality ontological scope). There is no conceivable path from one to the other,”

    Can’t remember if we’ve had this discussion before, so apologies if you answered it on a previous thread. But what specifically about a mental quality has no conceivable physical explanation?

  13. 13. John Davey says:

    SelfAware


    “Can’t remember if we’ve had this discussion before, so apologies if you answered it on a previous thread. But what specifically about a mental quality has no conceivable physical explanation?”

    A mental quality cannot arise within the ontological scope of a physical model. Physics models are scalars – scalars in, scalars out. A closed loop, ontologically speaking. Metres per second in, meters per second out. Kg per kelvin in, Kg per kelvin out. There is no other way.

    That doesn’t mean to say that a mental quality can’t have a physical “explanation”, but that explanation requires an assertion of a cross-ontological correlation between the scalar world of physical metrics and the qualitative world of mental experiences. That is novel – in terms of the application of physics to scientific problems.

    But this ‘correlation’ cannot be the outcome of a mathematical process, or indeed any kind of process, as there is no way that numbers can leap into becoming qualities. It requires a leap of imagination as transformation is, literally, inconceivable.

    It also sets a question around “what is the origin of the correlation”. It’s a kind of epistemological hole in the process which the angelic types of computationalism – perfectionists who think humans will know everything because they aren’t animals – might think is is a permanent and and unjustified stain on the right of humans to know absolutely everything about how the world works.

    Now to some extent all physics is based upon a correlation. A correlation between a scalar concept of physics and cognitive concept of experience. They are not the same. For instance, there is an assumed correlation between the scalar concept known as ‘spatial extension’ in physics and our sensory and cognitive experience of space. A similar correlation exists with time. The Time of physics is not the same as the Time of experience. The Time of physics, for instance is just a scalar with no qualities and certain formal properties determined by mathematics (ie movement along time is one-dimensional). The Time of experience is felt and experienced as both a quality and a quantity. It has a feel, a certain form and shape, nothing like space. Yet in physics terms, space and time are slightly different variants of the same thing.

    So currently there is no problem in using the notion of identity between physics “time” and experience “time”, because we’re only ever interested in the quantitative component of time, which is entirely what the physics “time” is.

    It would be a lot more different if we were expecting our physics equations to deliver something about the semantic of time, the actual experience of time. There is nothing in the physics concept of “time” – a pure scalar – that is meant to represent it. There is no identity between physics “time” and experiential “time” in the semantic sense, so the standard correlation in physics can’t be used.

    So if we are looking at any mental quality – sense of space, time, matter, colour, pain, call it what you will – scalar models (otherwise known as “physical models”) cannot work in the conventional way. They need a “correlation” between a scalar state and a qualitative state, something that hasn’t – to my knowledge – been done before in science.

    But – I can’t see any alternative. It’s er .. inconceivable.

    JBD

  14. 14. SelfAwarePatterns says:

    John,
    I appreciate the explanation. In some ways, I agree with you, but the agreement isn’t complete, and in the incompleteness, I think, is the difference between how we see this question of mental qualities and physical systems.

    First, I think it’s accurate to say that we only ever have our own conscious experience. We never have anything else. This statement might seem tautological, but it’s a tautology that’s important to wrap our minds around.

    When we talk about anything outside of our consciousness, we’re always talking about models, theories, that we build based on the contents of conscious experience. I’m actually a scientific instrumentalist, in that I agree that scientific theories are things we invent to make sense of empirical observations / conscious perceptions, to relate them into a framework that allows us to predict future conscious experiences.

    But my instrumentalism doesn’t stop with the models we classify as scientific. It encompasses the model I currently hold for the room I’m typing this in, for the chair I’m sitting in, for my model of you, and for every aspect of the everyday world. In all cases, we can only judge the models by how predictive they are, or whether they are a productive component of those predictions. We have no other measure with which to judge them, because we never have access to anything but our own conscious experience.

    You made a distinction between physics time and experience time. I think all we have is experience time. Physics time, the scalar aspect of it, is simply us observing certain ratios between experienced times, such as the time it takes for our heart to beat (second), the sun to go across the sky (day), or for the moon to go through all its phases (month), or for the constellations to circle the sky (year). We break these up into ever smaller chunks, but it’s all based on ratios of experienced time.

    As you noted, all physics is based on correlation. Correlation between our theories, our models, and conscious perceptions. David Hume pointed out that we never actually observe causation. Causation is itself a theory. All we can ever do is tighten and isolate our knowledge of the correlations. If we manage to isolate the correlation down to the minimum number of variables necessary and sufficient for us to experience the correlated phenomenon, we declare causation, but that’s a theory based on the correlations.

    For me, the epistemic divide isn’t between conscious experience and physics, but between conscious experiences and our theories about what is behind them. The theories about how the sun produces the heat and light we consciously experience from it rests on correlations. So, in my mind, the correlations between the workings of the brain and conscious experience represent no unique difficulty. It’s fundamentally the same thing we’ve always worked with.

    That’s why I’m satisfied by a description of photons striking the photoreceptors on the retina and the resulting electrochemical cascade back to the thalamus and occipital lobe, along with the cascade of signals to the frontal lobes, as an explanation for the mental quality of redness. Granted, I want it mapped much more thoroughly than it currently is, but if we can find the physical correlates of red, of my memory retrieval about the concept of red, of the emotional feelings it triggers in me, and of any other aspect of the experience of red, I’ll be satisfied that we’ll have explained it as well as science can ever explain anything.

    Of course, there will be people who reject the explanations, just as there are people who still reject natural selection. Scientific explanations are often not intuitive, that is, often don’t meld well with our models of everyday life. But if the explanations produce a model that is predictive, at least more predictive than what we had before, then they’ll be a success.

  15. 15. John Davey says:

    SelfAware :

    I think we are in broad agreement but we have a couple of differences.


    Correlation between our theories, our models, and conscious perceptions.

    That is not the correlation I was on about. I was on about the correlation between experiential time and physics time, a necessary link if one is to extract semantic out of the results of physics calculations.


    That’s why I’m satisfied by a description of photons striking the photoreceptors on the retina and the resulting electrochemical cascade back to the thalamus and occipital lobe, along with the cascade of signals to the frontal lobes, as an explanation for the mental quality of redness.

    If you are happy with correlation as an account of mental experiences, then so am I. But the flow of photon->receptors->brain-internals I wouldn’t view as a cause of the quality of redness, as the first two steps aren’t essential to experience red. You don’t need light and you don’t need working receptors. The cause is wholly within the brain and is presumably linked to a certain neuronal configurations. All mental qualities are produced internally it seems to me, and the external incidents that trigger them are causally unnecessary antecedents.

    J

  16. 16. Shaikh Raisuddin says:

    Descartes’s conceivabillity argument is based on faulty premises.

    He ignored physical basis of thinking, actual physical meaning of thinking and memory. He ignored significance actual meaning of language. He had ignorance of what self (“I”) is. He ignored that a baby does not think. He was unaware of what mind is. And the most important fact he ignored that lexical meanings of words are not scientifically verified. He did not know that human mind is artificial and a social-implant like a software into computer.

    Can programming exist out of computer in vacuum?

  17. 17. Sergio Graziosi says:

    Ooh, this discussion has been inspiring, enough to bring me out of lurk-mode.

    I have been struggling to find a good way to explain why the “conceivability” argument fails to convince me. Reading Jochen’s #5 pushed me over the obstacle (even if, or because I disagree), so my sincere thanks to all!

    The bit I disagree with:

    The possibility that we sometimes merely might think that we can conceive of something—i.e. we might fail to see the contradictory nature of complex issues, and thus, be essentially deceived into believing it to be conceivable—fares somewhat better, but in the end, the sorts of processes that ought to be accompanied by experience seem to be fairly simple to conceive of as lacking it, so this, too, doesn’t strike me as a very plausible option.

    Jochen: I think it’s remarkable that you gesture towards maths, but didn’t spot the opportunity I’m trying to seize, perhaps I’m missing something…

    First of all, we can conceive things that are logically impossible. We have plenty of examples, from Maths and elsewhere, a famous one being the set of all sets that don’t contain itself. In fact, to find out that it entails a logical paradox, we had to conceive it (in an admittedly weak sense) first. QED. We can indeed conceive (in a weak sense) stuff that self-contradicts.

    Moreover, I think that Jochen implies the following: we may sometimes think that we can conceive something (in the strong sense of imagining it while knowing without reasonable doubt that it is not logically inconsistent) and be wrong, however this is rare enough to be irrelevant (that is: would be a weak rebuttal against the conceivability argument).

    I disagree. If the above were true, I would need someone to explain me what actual mathematicians do for a living.

    The story, as I understand it, goes this way:
    Someone notices an interesting trend/pattern. Produces a conjecture, which is an unproven statement (given A, B and […], C is necessarily true).
    In our terms, an idea is conceived.
    If the conjecture looks interesting, mathematicians will toil over it, trying to make a theorem out of it. That is: produce a proof that the conjecture is indeed true (or not).
    In our terms, this seems to be exactly the phase we are dismissing: proving that the original idea is indeed true or not.
    Thus, it seems to me that mathematicians, simply by existing and doing what they do, provide empirical evidence that the moving from weak to strong “conceivable” (not even to “possible”!) is indeed not trivial at all. It may be trivial in many cases, but it is non-trivial enough to support a whole profession.
    If we are still not convinced, there is an interesting additional thing to note: sometimes theorems that were accepted and taken for true will eventually be found to be incorrect. It’s not that frequent, but it happens.
    Some amusing stories in the following links:
    http://math.stackexchange.com/questions/139503/in-the-history-of-mathematics-has-there-ever-been-a-mistake
    http://math.stackexchange.com/questions/93063/how-rare-is-it-that-a-theorem-with-published-proof-turns-out-to-be-wrong
    http://mathoverflow.net/questions/35468/widely-accepted-mathematical-results-that-were-later-shown-to-be-wrong
    It follows that not only proving that “conceivability” in the strong sense (includes internal consistency) is not trivial at all, it is also evidence that some of the best minds in the world sometimes get it collectively wrong.
    Thus, I have three lines of attack:
    1. We can conceive stuff that is illogical, in fact, we could never find any illogical idea at all, if we couldn’t conceive it first. This could be enough: given any idea, you have to prove that it doesn’t contain/entail any internal contradiction before you can use the conceivability card. In terms of zombies and thereabout, I’d claim that you can’t do so convincingly: the subject is far too complicated, the terms involved not precise enough, etc.
    2. Even when dealing with exact definitions (Maths), the path from conception to proof of logical consistency is hard, laborious, and in some cases, hard enough to defeat all attempts made so far (plenty of unproven conjectures out there).
    3. Even when the path is found, we still can’t be absolutely sure that it is correct and it actually leads to our desired destination.

    Note: the conceivability argument is supposed to be absolute. If I can conceive A and A is logically possible, A is possible. To prove it is wrong, all I need is one exception, which is provided by the examples in the links, and line of attack #3 (the weakest). The other two lines of attack are even stronger: I can think that an idea is self-consistent when it’s not (1.) and I still need to demonstrate that it is (2.), which is, outside of formal systems, pretty hard to do in a convincing way.

    Overall, I’m with Peter again, and don’t think that “conceivability and inconceivability [arguments] can actually tell us [anything] about the real world”.

    More & trivia:
    The positive point that Jochen is trying to make is not touched by my disagreement above. It looks valid to me just as well.
    Re C-fibers: I’m with John Kubie (#4), that’s what the distant echoes of my neurobiology training whispered as well (I am not tempted to go an search for supporting references, though).
    Surprisingly, I also find myself to agree with the bulk of John Davey’s #13. Unusual, but it happens!

    I will try to propose this argument in extended form, sooner or later. All feedback appreciated, as always!

  18. 18. Jochen says:

    First of all, we can conceive things that are logically impossible. We have plenty of examples, from Maths and elsewhere, a famous one being the set of all sets that don’t contain itself. In fact, to find out that it entails a logical paradox, we had to conceive it (in an admittedly weak sense) first.

    I think that we may use the word ‘conceive’ differently, or your powers of conception outstrip mine—because I can’t conceive of any such thing. It would require me to hold, simultaneously, two contradictory beliefs, namely, that the set contains itself, and that it doesn’t.

    That doesn’t mean I can’t talk about it, or even reason. I can talk perfectly well about ‘the round square copula of Berkeley college’, but I can’t conceive of it—not as I understand the term, at least. That may be a language issue: I understand ‘conceive’ as German ‘vorstellen’, which, as its etymology (from ‘putting sth before sb’) suggests, has connotations of ‘imagine’ or ‘picture’. So I can’t picture, or imagine, something like Berkeley college’s round square copula, or the set of all sets that don’t contain themselves, because there can’t be any object such that it is both round and square, or any object that both contains itself and doesn’t contain itself.

    Consequently, I can’t conceive of logically contradictory objects, because there are no objects such that they have a given property, and don’t have it; and it’s this object I would have to picture in order to conceive of it.

    Still, I can talk meaningfully about such objects—as in saying ’round square copulas are impossible’. You seem to hold that such talk requires a conception of these objects; I disagree. We can view objects as collections of properties. In this case, it’d be something like:
    1. It’s round
    2. It’s square
    3. It’s the copula of Berkeley college
    I have conceptions of these properties; I know what ‘being round’ means, I know what ‘being square’ means, and so on. I can even use this knowledge to infer other properties: since ‘being round’ entails ‘having no corners’, I know the thing can’t have corners; since ‘being square’ entails ‘having four corners’, I know it has four corners. And so on.

    What I don’t need, and indeed don’t have, is a conception of the thing as a whole—simply because there is no thing that has both no corners and four corners, I also can’t imagine such a thing. Again, it would entail simultaneously holding contradictory beliefs, and if I could do that, I would’ve become a politician, it seems to be the success model there right now.

    You can view this analogously to the branch of mathematics known as model theory: a mathematical structure that fulfills a given set of axioms is said to be a model of those axioms. Now, any consistent set of axioms has a model; inconsistent axioms don’t. That doesn’t mean you can’t do math with such axiom sets—as indeed, you often have to, in order to unravel a contradiction. But in a sense, those axioms don’t talk about anything—there’s no mathematical structure that they describe. In the same sense, you’re not conceiving of anything if you’re thinking about something contradictory—there’s no object that actually has the properties you’re thinking of, so there’s in particular no object you could be thinking of.

    This also points toward what I meant regarding falsely believing to conceive of something. Contradictions aren’t always as obvious as in Quine’s example regarding the round square copula. Rather, it may be that you think of an object that has a given set of properties, which seem to be consistent at first blush, but on further investigation, turn out not to be.

    The solution here is, I think, again analogous to mathematics. There, you may have a set of axioms, and a proposed model—say, Fregean set theory, and the universe of sets (or the Peano axioms and the natural numbers, or what have you). Now, as Russell has shown, Fregean set theory isn’t in fact consistent. Does that mean that you’ve conceived of something impossible?

    In my view, it doesn’t: rather, you have conceived of the universe of sets, which, however, isn’t actually a model of Fregean set theory. That is, whatever thing you conceived of is not, as first supposed, described by Frege’s axioms—there is nothing that’s described by them. That thing, this universe of sets, is described by some other set of properties—as defined, for instance, by the Zermelo-Fraenkel axioms.

    Likewise, whatever object you conceive of when trying to conceive of ‘the round square copula of Berkeley college’ will not actually be ‘the round square copula of Berkeley college’. Maybe it’ll be something kinda round-ish and square-ish, or something oscillating between the two, or something like an optical illusion—all of which are perfectly consistent objects. You’re just wrong about the proposed description applying to it.

    So, in short: I’m going to be severely skeptical of any claims purporting to establish that we can conceive of anything that’s logically impossible—it’s not far short of claiming that logically impossible things could exist.

  19. 19. Sergio Graziosi says:

    Jochen (#18),
    you sent me to the dictionary pages!
    From oxforddictionaries.com:
    2. Form or devise (a plan or idea) in the mind.
    2.1 Form a mental representation of; imagine.

    I may be wrong, but it does seem that we have a language problem, and/or that you missed my point. My whole idea rests on the ambiguity of the term: what exactly does “form a mental representation” mean? This is why I’ve made the distinction between weak and strong “conception”.
    You may want to conceptualise weak conception as follows: it is whatever it is you do when you first think of Object-X with properties W, Y, Z. When the idea of Object-X is first generated in your mind, surely you have done something (produced the idea), but you also have not already established whether such an object is logically possible. To do so, you need to do some extra work. Not having a name for the first activity, I’m calling it “weak conception” (which seems consistent with the official definitions above) – I then proceed to argue that thought experiments in philosophy of mind are confined to the weak conception side.

    Weak conception is when I form the idea of the set of all sets that do not contain themselves. We can do it. Being an entirely abstract concept, we also have the tools to quickly realise the idea doesn’t hold water – it can’t be conceived-of in a strong sense. Maths and/or pure logic, allows us to realise this. Thus, in such a case, we can’t move onto strong conception.

    But let’s do another example: in chess, I can (and would have, a long time ago, when I did play occasionally) devise/conceive “plans”. A series of moves, where for multiple reasons, I expected the opponent to be compelled to comply with my expectations, or fail to do so and invariably give me an advantage (i.e.: every deviation initiated by my opponent would play to my advantage). I definitely did “conceive” such plans, and they were all, always, without exception, weak conceptions. Sometimes they worked, sometimes they didn’t, but each time I could never (and should never) assume they were strong conceptions: my plan could be wrong and allow the opponent some unexpected move which would play to their advantage, or unfold as planned, but fail to deliver the expected advantage, and/or be wrong in a gazillion of other ways. Even the plans that did work should be regarded as weak conceptions. Given that chess is regulated by abstract rules, in principle it would be possible, for each such “plan”, to figure out if it was correct (move to “strong conception”); however, I simply did not have the brainpower to do it in real-time (and never will): even chess has too many degrees of freedom for me to handle, thus, I couldn’t and shouldn’t have ever assumed that my plans were watertight.

    Moving on, it is not surprising that on the entirely abstract domain there is a whole profession sustained by the arduous task of figuring out whether a weak conception (a conjecture, in my previous comment) can be considered strong (by demonstrating it is indeed true, AKA writing a theorem or collection thereof). The existence of Maths as an academic discipline, along with how it is conducted, demonstrates that even when it is theoretically possible to demonstrate that a given conception is strong (logically consistent), doing so is neither easy nor 100% reliable.

    Question for you is: how do you do the same when we are dealing with ideas that are about the real world out there, not about pure abstractions?

    With pure abstractions our axioms can be well defined and unambiguous. With arguments such as zombie twins, no matter how hard we try, we will always have to deal with some fuzziness. The result is that moving from weak to strong conception is much, much harder. For me it’s an argument about burden of proof: given how hard it is to produce theorems, and given that we sometimes get it collectively wrong, I reckon that for someone to say “I can strongly conceive the existence of real-world-object-x” (i.e.: I can imagine object-x’s existence, and I can be sure I’m not mistaken: object-x, as defined, is indeed logically possible) is just not enough. They have to do the work: that is, the equivalent of writing down a theorem which demonstrates the conjecture (our weak conception) is indeed true.
    However, I don’t see how on earth one could even try: we don’t know what counts as physical (physically identical zombie twin), we don’t have an agreed understanding of causation, and we can’t agree on what we mean with “phenomenal consciousness / what is it like-ness”.

    Without uncontroversial definitions of your premises, and indeed, placing instead the object you want to better understand within your premises (what is it like-ness), how can anyone demonstrate beyond reasonable doubt that a particular conjecture (conception) is indeed logically possible?
    You can appeal to intuition, as in “it feels reasonable, I don’t know how it could not be”, but that’s just not enough. It is not enough because we know from the practice of Maths that conjectures which feel entirely reasonable are not always true.
    Thus, argument on conceivability end up being a rhetorical devise: they aim at convincing us at the level of intuition. But given our knowledge of how frequently our intuitions are mistaken, they cannot settle any argument.
    Or at least, that’s the argument I’m trying to make…

    In other words, to go back to your main rebuttal:

    I’m going to be severely skeptical of any claims purporting to establish that we can conceive of anything that’s logically impossible

    (Recall my definition of weak conception!)
    We can weakly conceive a lot of things – some are logically possible, some aren’t. Some are very obviously illogical, some are counter-intuitive but logical, some seem logical but aren’t, some are neither: we can’t know for sure, because they start from ambiguous premises. For many weak conceptions, we simply cannot be sure whether they are logically consistent or not: whether a given weak conception can be demonstrated to be true or not is undecided and undecidable. This is especially the case for thought experiments about philosophy of mind, where by definition we don’t know what we’re talking about, meaning that some level of fuzziness always creeps in our set of premises.

  20. 20. Jochen says:

    Sergio:

    My whole idea rests on the ambiguity of the term: what exactly does “form a mental representation” mean? This is why I’ve made the distinction between weak and strong “conception”.

    I think that what you call ‘weak conception’ is a trivial notion, as far as philosophical thought experiments are concerned—it seems to me we can ‘weakly conceive’ of just about anything, so obviously, it’s not going to tell us anything about possibility, whether logical or metaphysical.

    You also seem to shift back and forth between the dichotomies of ‘true/false’ and ‘consistent/inconsistent’. For instance, with the chess example, I think that you just ‘strongly’ conceived of (which, to me, really still just is what I mean by conceiving of something) a bad plan; there’s no requirement that you conceive of all of your opponents countermoves in order to conceive of the plan itself, at least not that I could see. The sequence of actions you had planned out is a perfectly possible course of events; as long as you follow the basic rules, there is a possible chess game sucht that it realizes this chain of actions. It just does not actually come to be realized in the real world.

    There are also inconsistent plans, however: for instance, one where you move your bishop in the next move, then your opponent moves, and then you plan a move that would only work if the bishop were still in its original place. Such a plan, I think, you can’t conceive of: for if you actually had had any clear conception, you would have noticed that the bishop isn’t there any longer. Perhaps you have conceptions of two different games of chess, that you confuse; perhaps you’re forgetting some rule, such that you’re not conceiving of a game of chess, but rather, of chmess; or perhaps, you’re just imagining a sequence of positions, which however can’t be connected by legal moves. In all of these cases, I think you have a strong conception of something; however, that something isn’t a chess plan, and you’re just mistaking it for one—that’s where the error really lies, not in having some conception of a logically impossible object.

    So I would argue that the typical flow of mathematical reasoning is the other way around: you start with a conception of something, say, the natural numbers, or the universe of sets. Then, you attempt to formalize this conception, using some set of axioms. Otherwise, what exactly are you trying to axiomatize?

    This can then go well, as in the case of the Peano axioms: you actually figure out a set of axioms describing your structure. Or, it can go belly up, as in Fregean set theory: the proposed axioms actually fail to describe anything; the universe of sets had to wait for Zermelo and Fraenkel, or Neumann, Bernais, and Gödel, or whatever your preferred set theory happens to be, in order to be appropriately formalized.

    As for zombies, I agree that nobody’s ever conceived of a zombie. But I don’t think we need to, in order to judge zombies conceivable. For instance, nobody’s ever conceived of a computer, how it performs billions of operations in seconds, etc. But still, it’s quite easy to judge that computers are conceivable: in principle, it suffices to understand the functioning of the NAND-gate; the rest is just more of the same. Nothing unexpected occurs there, in a sense, and a being capable of conceiving of the functioning of a couple of billions of NAND-gates could conceive of how a modern computer does its thing.

    That’s what shifts the burden of proof: for there doesn’t seem to be a reason why one can’t do the same thing with biological organisms. For instance, I can (strongly, in your verbiage) conceive of a mechanism that sounds an alarm bell when a certain amount of force is applied in a certain spot (that cries out when it stubs its toe, so to speak). In order to fulfill that function, no phenomenology at all is needed; indeed, postulating that it had any would at the very least seem odd (although these days, one has heard stranger things). Or, closer to the computer example, I may be able to conceive of the working of a single, strongly simplified, neuron—integrating impulses and changing frequency.

    So those that claim that if one could imagine the full functioning of the human brain, or the full set of functions of the human organism, phenomenology would necessarily emerge (i.e. zombie hunters), must claim that something goes wrong in the inductive step from starting with a single instance of behavior and generalizing to complex conglomerates, which seems to go so well with a single computer, or virtually all complex pieces of machinery ever created by humankind (if you think about it, nobody really can conceive of a modern car, or even a building, in its entirety; yet we feel supremely confident in asserting that if we keep piling on more complexity, nothing qualitatively unexpected occurs).

    It’s true: the above is not an absolutely compelling argument. It is possible (in the sense that it seems conceivable without encountering a contradiction) that once we’ve piled on enough neurons, poof, phenomenology. It’s also possible that once we’ve connected enough NAND-gates, Santa Clause pops out and starts handing out presents. But everybody trying to argue that this is what happens needs to provide a good reason: that is, the burden of proof is on their side. There needs to be some positive argument towards novel qualitative phenomena emerging from collections of neurons, such that their absence within the same collection is not, in fact, conceivable.

    Consequently, I think that a successful attack on conceivability arguments must dig deeper than merely gesturing towards possible limits of our conception, or to ill-defined emergence.

  21. 21. Sergio Graziosi says:

    Jochen,
    first of all, my apologies for the slow reply. It isn’t that I don’t want to engage, it’s just that I can’t find the time, not in conjunction with clarity of mind, at least. Thus, if we’ll continue sparring, it will have to be at a very slow pace; weekly, if not even slower :-(.

    We are inching forward, slowly, but perhaps we are. You agree that weak conception can’t tell us “anything about possibility, whether logical or metaphysical”. From here, I think all I need to do is to show how on the philosophy of mind domain, all we can do is propose some weak conception or other, and try to convince people that such a conception can be considered strong (what you do in the second half of #20). However, we can’t demonstrate this strongness beyond reasonable doubt, and thus such arguments can never be settled for good.

    You are right, I did mix dichotomies of ‘true/false’ and ‘consistent/inconsistent’, I was hoping a little charitable reading would allow me to skip a few trivial details. However, I think you’ve shown me a way to fortify my position, so I’ll give it a try.

    In the chess example, the kind of plans I conceived could be true or false, based on different levels of consistency. To be true, a plan should be guaranteed to work – that’s the definition I failed to make explicit. On the other hand, a plan could fail for multiple reasons – as you point out. It could be inconsistent with the rules, and thus be wrong at a very fundamental level. It could allow the opponent to gain an unexpected or unrecognised advantage, and thus be consistent with the rules, but still wrong/false (as a plan).
    The distinction you are trying to make thus becomes a practical example of how slippery ‘conceivability’ arguments truly are. For me, what you call a bad plan is a plan that was (weakly) conceived, but could not be considered strongly conceived as it did not fulfil the reason why it was conceived (guarantee some advantage in the match). For you, it is strongly conceived, but just a bad plan.
    An inconsistent plan (one that broke some chess rules), for me, is/was just as wrong (still only weakly conceived), for you, it fails to reach the “strongly conceived” threshold. Meh. This kind of fuzziness is precisely why we can’t say “yes, I can, without doubt, strongly conceive of p-zombies et al”.

    Still using chess: for me, the conceiver, a good(true/valid) plan (one that is guaranteed to work), a bad plan (one that can be defeated) and an inconsistent plan (one that can’t be actioned without breaking the rules), would, after reflecting about it and deciding to put the plan in action, feel indistinguishable. The decision of acting on the plan in question guarantees that I felt reasonably sure that it was a good plan (i.e. that my conception was strong enough!). However, such strong feelings were, demonstrably, unreliable. By following the plan, I was declaring my conclusion that the plan was strongly conceived, and I was thus finding out how easy it is to get it wrong.

    We can move to zombies: one can (weakly) conceive them, and feel absolutely sure that they can be conceived in a strong sense, and, despite such feeling, one can still be wrong (the proposed concept might be logically impossible, inconsistent, circular, etc.). I don’t need to demonstrate how a given concept fails. All I need to do is to show how, repeatedly and demonstrably, people can be wrong about such judgements. If they are, it follows that we can’t obtain certainty out of conceivability arguments. My use of the maths domain is intended to demonstrate how hard it is to ensure that a given set of ideas is indeed internally consistent – even when the domain in question is built in such a way that makes such demonstrations (theoretically) possible.

    Moreover, you show how it is possible to arbitrarily move around the boundary of what counts as “conception”.

    you start with a conception of something, say, the natural numbers, or the universe of sets. Then, you attempt to formalize this conception, using some set of axioms. Otherwise, what exactly are you trying to axiomatize?

    Oh, and can you “use” some set of axioms without first (weakly) conceiving them? See? How exactly are we supposed to decide what counts as conception and what doesn’t? If we can’t even decide this, how on earth are we supposed to distinguish between strong and weak conception in any domain that doesn’t afford formal proof? In maths and chess, we can – because they are organised to allow proofs, but philosophy of mind? No, we can’t!

    In other words, I don’t think your last comment helps your cause, it shows that we can’t even agree on an indisputable definition of what we are talking about. One of the required axioms to make any conceivability argument is itself too fuzzy: we can’t be sure that the idea of conception itself is coherent to start with.

    Thus:

    it’s quite easy to judge(1) that computers are conceivable: in principle(2), it suffices to understand the functioning of the NAND-gate; the rest is just more of the same. Nothing unexpected occurs there, in a sense(3), and a being capable of conceiving of the functioning of a couple of billions of NAND-gates could(4) conceive of how a modern computer does its thing.

    That’s what shifts the burden of proof: for there doesn’t seem(5) to be a reason why one can’t do the same thing with biological organisms(6). For instance, I can (strongly, in your verbiage) conceive of a mechanism that sounds an alarm bell when a certain amount of force is applied in a certain spot (that cries out when it stubs its toe, so to speak). In order to fulfil that function, no phenomenology at all is needed; indeed, postulating that it had any would at the very least seem odd (although these days, one has heard stranger things(7)). Or, closer to the computer example, I may(8) be able to conceive of the working of a single, strongly simplified(9), neuron — integrating impulses and changing frequency.

    No, the burden of proof doesn’t shift one inch – I am sure it feels like it does shift, but just as my feelings re chess plans, such a feeling is known to be unreliable. It’s easy to judge that computers are conceivable, yes, in hindsight.
    Funny how you used the word “judge”! In fact, I’ve counted 9 instances of expressions that are either fuzzy, questionable, or that inject uncertainty in your argument. Each one helps seeing why formal proof is simply out of reach. Thus, we may or may not be able to do the same from biological organisms, we just don’t know (despite our feeling that we can), and should have the decency to admit it. Or at least, that’s what I’m claiming! 😉

    Analytical philosophy tries hard to make philosophical domains amenable to formal proof. I’d argue that the state of the discipline reinforces my point: on one hand, it contains the highest concentration of Chmess variations. On the other, I don’t see how it helped to reach agreed-upon conclusions on anything but details. There is a reason: we may be able to agree on some formalisation, if and when we are looking at some minute detail (so to reduce the fuzziness of our axioms), but big-picture concepts such as “conceiving” or p-zombies? Nope, the project has already failed. Deploying such concepts requires some degree of vague hand-waving, but once you allow it, formal proof is moved solidly out of reach.
    Without formal proof, any weak conception that I may propose cannot be shown to be a strong conception. QED. The rest is persuasion (at which I clearly ain’t good! 😉 ).

  22. 22. Jochen says:

    Sergio:

    first of all, my apologies for the slow reply. It isn’t that I don’t want to engage, it’s just that I can’t find the time, not in conjunction with clarity of mind, at least. Thus, if we’ll continue sparring, it will have to be at a very slow pace; weekly, if not even slower :-(.

    No worries; do take your time. However, in the interest of trying to keep this discussion from ballooning beyond all reasonable measure (for once), I’ll not respond to all your points in detail, but instead concentrate on what I think is the main argument I brought forward, that seems to have passed you by somewhat.

    You say that:

    It’s easy to judge that computers are conceivable, yes, in hindsight.

    However, that isn’t the actual historical course of events. Computers were first conceived off, and, on that basis, judged (accurately, no less) to be possible, before they were built—it doesn’t matter here whether we think of Turing’s conception, Babbage’s Analytical Engine, von Neumann’s design, or whatever. The key point is that we went from conception to possibility to reality, by a process of reasoning that is exactly analogous to that used in judging zombies to be possible, and that moreover is used in basically every complex undertaking, and which we habitually rely on, and which has so far (to the best of my knowledge) never failed; yet, claiming that zombies aren’t possible is tantamount to claiming that this process is faulty, and thus, needs stronger justification than has so far been presented.

    Now, the process I mean is, basically, one of ‘absent defeaters’-reasoning: we start with a (‘strong’, in your terminology) conception of the individual parts of a complex system (say, the NAND-gates of a computer, or the levers, pulleys, and gears of a steam-powered machine), and from this conception, conclude that absent defeaters, a hugely complex conglomerate of these parts, which we cannot fully conceive of due to its complexity, will behave a certain way (a computer will compute, a steam-powered machine will fulfill its function, whatever it may be).

    This process of reasoning has proven reliable throughout human history. Whenever we carefully draw up plans for machines ranging in complexity from a simple snatch pulley to the Large Hadron Collider, in the end, the machine does what we planned for it to do, and nothing else. Of course, mistakes occur: but if they do, we can eventually pinpoint to where the plan failed, i.e., we can figure out what the defeater was.

    But the argument against zombies runs differently. Essentially, it is that since we can’t fully conceive of a zombie, anything at all might happen, so that in the end, we can’t claim that the zombie is conceivable, much less possible. And logically, that’s perfectly true: drawing up the plans for a computer, or the LHC, does not guarantee that either will work as planned, even if no mistakes were made. It is perfectly possible that there exists, say, a natural law such than when we pile up these-and-those ingredients, something irreducible to these ingredients occurs, and Santa Clause jumps out, handing out presents.

    Nevertheless, that doesn’t make it reasonable to believe that after we turn on the LHC, Santa Claus jumps out, handing out presents. There must be some aspect of the world that makes it so, and we have no justification in believing in such an aspect of the world; it would be arbitrary to postulate it.

    Likewise, it might be that once we assemble a physical duplicate of myself out of the component parts, something will spark up and imbue it with conscious experience. But if that is the case, then there must be something more to the world than we presently suppose, something that foils the inductive leap from our conception of the component parts to the full zombie. And I have no reason to believe in any such thing; rather, I believe, like the architects of the first computer did, like the constructors of the LHC did, like humanity has believed and been proven right with every complex project, that the inductive reasoning will indeed work out.

    Hence, one ought to expect from everybody claiming that it doesn’t, at minimum, that they give a reason why it wouldn’t. After all, nobody would listen to someone claiming that if we assemble a few billion NAND gates, Santa Claus will jump out; and there is presently no more reason to believe somebody when they claim that if we assemble a few billion neurons, consciousness will jump out. So if it was reasonable for the architects of the first computers to believe that they will succeed, it is reasonable to believe in zombies; and if it’s not reasonable to have these beliefs, then one stands before the great puzzle of how they turn out to be right time and again.

  23. 23. Sergio Graziosi says:

    Jochen,
    happy to proceed slowly, even happier to avoid tackling each and every point, it just doesn’t work: given the inevitable fuzziness (!) of our subject, you can always build objections to each and every point one of us may make. Better to concentrate on what we happen to perceive as the key points.
    I was trying to address those and I don’t think we are far away on that particular judgement, however, to some extent, we are talking past one another, I fear.
    I say so because once again, I feel you are not making the point I think you are trying to make. If I’m wrong, I guess I can be because I’m either misunderstanding your argument, or the aim of it all.

    You have two arms, conceivability of computing, and induction/Santa Claus. The first looks wrong to me, in more than one way, but I can see why it was worth trying it. The second is puzzling me, as it looks catastrophically wrong to me, meaning that I can’t figure out why you think it helps.

    Conceivability of computing: I think it fails for two reasons. Survivorship Bias and lack of fuzziness.
    Survivorship Bias: I claim it’s easy to move from strong conception of NAND gates to concluding that (time and tape limited) implementations of Turing machines are possible, but only in hindsight. You say no, this conclusion was reached in foresight, and indeed that’s how we make new technologies, all the time. We do, but most of the time, we get it wrong multiple times, sometimes because we had our theoretical details wrong (1), or because of some more practical implementation problem (2). Case (1) is similar to an “inconsistent” chess plan (impossible as it violated the rules), case (2) is analogous to a flawed plan (one that can be beaten). Because of Survivorship Bias, we do get to hear about successful stories, probably all of them (save for those kept secret for military/security reasons), but we certainly fail to hear about a good proportion of the failures. We might get to learn about a handful of failures which happen to be interesting, but I do not doubt that most of failed attempts become invisible, in hindsight. Thus, claiming that moving from conception (judged to be strong by the conceiver) to factual possibility (by actual implementation) is normal, doesn’t grant the conclusion that if a conceiver thinks she has strongly conceived something, then it’s likely that this conception is strong. This is simply not the case, people make mistakes all the time, and that’s final.
    Luck of fuzziness: computing is a branch of maths. It was therefore possible to formally demonstrate the strongness of conceptions. This objection is only important because when you move onto messier engineering practices, say for example flight, the stories that link conception to realisation are far less linear and full of recorded-because-interesting mistakes. QED: if we can’t formally demonstrate the soundness of a given conception, we can’t be sure it’s a strong one – we usually find out by trying, and we frequently fail.
    Interim conclusion:

    Computers were first conceived off, and, on that basis, judged (accurately, no less) to be possible, before they were built

    This is true. However, this does not grant that therefore whenever someone judges that a hard-to-formally-define conception is strong, it is indeed strong. No, the burden of proof that this is the case remains strictly on the conceiver/proponent.

    Induction/Santa Claus.
    The way I read it, it just doesn’t work. Two main reasons: we have lots of inductive reasons to expect mechanisms to produce consciousness. Your Santa Claus argument is misaligned.

    We produce new humans all the time, if they lack physical defects, they never fail to become conscious. If we damage them in well defined ways, consciousness goes away. Humans are physical structures that we can touch and bump on, also animals are, and we do suspect many animals are conscious. By induction, we should expect that physical structures are enough to produce consciousness. The fact that we don’t have a clue on how this may happen does grant your counterargument, also based in induction. However this counterargument rests on a weaker induction (the one you describe), because we do know the cardinality of our unknowns is much higher than the knowns. Thus, ignoring the first inductive strain (the one I’ve just summarised, accepts that we don’t know something) and jump to the second (yours: we can’t see how molecules bouncing around can make consciousness emerge. Dismisses our known ignorance as likely to be irrelevant) is precisely what I’ve been calling it all along. Epistemic arrogance.

    Re your own argument, it also fails because there simply is no parallel. Santa Claus does not pop-up every time a high energy particle accelerator is fired up, but consciousness does emerge every time a given class of physical structure is present. A better parallel, involving Santa Claus, can be built, but it happens to make my case.
    Take a child, surrounded by lots of stuff you can’t explain. Her main source of information tells her that Santa Claus will show up if certain rules are followed (write and send the letter, go to bed at the right time, etc.), and that he’ll bring presents. When the rules are followed, the child does find indirect evidence of the correctness of this story. Repeatedly. Child (in presence of many seemingly magical things she can’t explain) would be inductively correct to conclude that Santa Claus probably exists. She would be factually wrong.

    Thus, in the presence of many unknowns, it is indeed the case that correct induction can lead us to wrong conclusions, as the ordinary Santa story shows.

    If you prefer, whenever we do have pretty accurate copies of physical structures, if one is capable of being conscious, the other one is. This applies to all identical twins, as long as one of them is not physically damaged. There are no known exceptions. On the contrary, there are gazillion of cases where people have claimed to have made strong conceptions that were weak and mistaken, in fact, we can all probably find examples by looking hard enough at the history of our own conceptions (provided we don’t consistently show epistemic arrogance).
    This means that we don’t know for sure which inductive argument might be right. It could be that billiard balls bouncing around can make consciousness emerge (the inductive argument I’m making), or that they can’t and some unknown additional ingredient is needed (the manifestly weaker induction you are using to counter my argument – which is analogous to mine Santa Claus story). We should admit our ignorance, allow both routes to be explored. We can also hedge our bets, and make the educated guess that probably the inductive argument for which no counter-examples are known is stronger than the one that has been known to be weak for as long as declarative knowledge has been around.

    Conclusion:

    Likewise, it might be that once we assemble a physical duplicate of myself out of the component parts, something will spark up and imbue it with conscious experience.

    Err, no. We have every reason to expect that an accurate-enough copy of yourself will be as conscious as you are. We don’t expect an additional ingredient to mysteriously appear (spark up) so to make your copy conscious (we don’t even have a clue on how to detect such an appearance), neither we expect a molecule-by-molecule copy to be permanently unconscious. We have every reason to expect your component parts will suffice. We do know we could be wrong, but we don’t expect we are, precisely for the same reasons why we don’t expect Santa Claus to appear when we turn on the LHC. Why we could jump to the opposite conclusion is mysterious to me, perhaps even more than how can consciousness exist.
    I mean it!

    The key point is that we went from conception to possibility to reality, by a process of reasoning that is exactly analogous to that used in judging zombies to be possible, and that moreover is used in basically every complex undertaking, and which we habitually rely on, and which has so far (to the best of my knowledge) never failed

    Yes, the process from conception to possibility to reality is indeed analogous. It is a process that fails all the time, to the best of everyone’s knowledge (including yours!) and is the reason why R&D is so freaking expensive. It is also why we built the LHC: to verify if some of the conceptions made by theoretical physicists actually do happen to be realised (by nature, in this case). If supposedly strong conceptions could be straightforwardly assumed to be indeed strong and realisable, our societies would be organised in dramatically different ways.

  24. 24. Jochen says:

    Sergio:

    QED: if we can’t formally demonstrate the soundness of a given conception, we can’t be sure it’s a strong one – we usually find out by trying, and we frequently fail.

    I readily acknowledge this. But, if there’s some error to a conception, there is what I’ve called a defeater—a reason for why our conception failed. Give me a defeater to the zombie conception, and I’ll immediately quit my yammering; this is what I meant when I said that the burden of proof is shifted by this argument.

    However, despite many hours of work of very clever people, no such defeater has been found. That still doesn’t make the argument formally sound, but, as you yourself point out, aiming for formal soundness on such a topic is a futile endeavor.

    Moreover, pointing out such a defeater would likely help a great deal in forming a true conception—that is, we might expect actually finding out how phenomenology arises from physics, once we know why zombies aren’t conceivable.

    All I’m really saying is that, on the subject of zombies, we’re at the same point as with every complex endeavor, such as building a bridge: according to what we know, the bridge will hold; so we go ahead and build it. Our conviction serves as sufficient justification for the attempt of building a bridge. We might have made a mistake, and the bridge collapses—in that case, there will be a defeater to our earlier logic, and we learn why the bridge collapsed. Indeed, even before building the bridge, somebody might notice the flaw in our design and point it out.

    But absent such a defeater, it’s perfectly sensible to judge the bridge possible. Consequently, it’s similarly perfectly sensible to judge zombies possible, absent anybody bringing up a conclusive defeater.

    By induction, we should expect that physical structures are enough to produce consciousness.

    I may be misreading you here, but this simply doesn’t follow: either option of dualism, panpsychism, dual-aspect monism, neutral monism, and so on equally well explains the evidence you cite.

    Indeed, Chalmers’ belief in panpsychism/dual-aspect monism is fueled by just that evidence, and the zombie argument—and that’s a perfectly reasonable stance to take, if the zombie argument indeed works.

    Santa Claus does not pop-up every time a high energy particle accelerator is fired up, but consciousness does emerge every time a given class of physical structure is present.

    This is circular, as it takes the conclusion that physical structures suffice for consciousness for granted. In fact, we know of no single case—we have no conception of any case—in which physical structures suffice for conscious experience. That’s what all the fuss is about, after all. So the parallel is exact: just as we have no example of particles colliding and producing Santa Claus, we have no example of physical structures being assembled and producing consciousness. If we had, the zombie argument would already be defeated.

    If you prefer, whenever we do have pretty accurate copies of physical structures, if one is capable of being conscious, the other one is.

    Yes, but this tells us nothing about how the physical and the mental are connected. In other words, we may ask ourselves: is this a necessary or a contingent fact about out world? If it is necessary, physicalism is true, and there should be some reason why zombies fail to be conceivable after all. If it is, however, contingent, then physicalism is false, zombies are conceivable and possible, and we need to look toward something else (say, irreducibly mental properties) in order to bridge the gap. But merely noting that in every case that we know off, physical identity means mental identity simply doesn’t do any work at all.

    On the contrary, there are gazillion of cases where people have claimed to have made strong conceptions that were weak and mistaken, in fact, we can all probably find examples by looking hard enough at the history of our own conceptions (provided we don’t consistently show epistemic arrogance).
    This means that we don’t know for sure which inductive argument might be right.

    Again, I’ve acknowledged this angle of attack from the very beginning; you’re not saying anything new here. The key point, however, is that to every conception that’s faulty there is some fact that defeats it; and that absent this defeater, we are perfectly justified in believing our conception to be truthful—which, after all, we do all the time in ordinary everyday practice. Otherwise, your logic would require us to never get started on any complex project, as in principle its outcome must be unknown.

    So a successful attack on this argument will propose such a defeater, at least a plausible one; but merely saying that it’s possible such a defeater could exist buys you no ground at all, as that’s trivially always the case.

    It could be that billiard balls bouncing around can make consciousness emerge (the inductive argument I’m making), or that they can’t and some unknown additional ingredient is needed (the manifestly weaker induction you are using to counter my argument – which is analogous to mine Santa Claus story).

    Again, this gets things backwards. We have a strong conception of billard balls colliding without any consciousness being present; this conception simply is generalized. We have no conception whatsoever of billard balls colliding in such a way as to give rise to even the tiniest spark of consciousness. Consequently, we have no reason to expect more billard balls colliding in more complicated ways to give rise to consciousness—none, at least, that doesn’t ultimately rely on the unspoken premise that it simply must be the case that certain combinations of physical matter gives rise to conscious experience. To everybody not already marred by expectations of their favorite explanation working out, the very notion will appear utterly preposterous—and with good reason.

    To expect that consciousness will arise in this fashion ultimately rests on nothing but faith—faith that heaping on more of the same will yield a result that’s qualitatively different. And there is no reason at all to have such faith. Again, panpsychism could simply be right. Thus, in not admitting such faith, one must judge the zombie argument sound.

    Without the premise that consciousness must arise from certain combinations of physical matter, the case you’re making falls apart; but that premise is what’s in question, and hence, relying on it makes everything circular. I mean, I believe it just as strongly as you do, I’d guess; but I think that such circularity can only serve to weaken one’s stance.

    We have every reason to expect that an accurate-enough copy of yourself will be as conscious as you are.

    Of course. But we have no reason to believe that this is because of its physical properties. Strike out that assumption, and your case collapses.

    We do know we could be wrong, but we don’t expect we are, precisely for the same reasons why we don’t expect Santa Claus to appear when we turn on the LHC. Why we could jump to the opposite conclusion is mysterious to me, perhaps even more than how can consciousness exist.

    The reason is simple: we have a positive conception of the properties of the component parts. We extrapolate that conception into generalities. We derive, from these generalities, the properties of complex conglomerates.

    This process tells us that there will be no consciousness in a complex arrangement of pulleys, levers, and gears, or billard balls, or elementary particles.

    Yet as you point out (and as I’ve been saying all along), it’s a fallible process. It may be that something we haven’t thought of occurs, such that some assemblages attain consciousness. But we have no conception at all of something attaining consciousness in that way (as that would be a solution to the hard problem). So to say that something will attain consciousness that way is to believe something we have no justification for, while to say that no consciousness will arise is to merely extrapolate from our knowledge.

    So, we have reason to believe that no consciousness will arise; we have no reason to believe that it will. Yet, we might be wrong; if so, there is something we haven’t thought of, some factor we haven’t taken into account, something that makes us wrong—what I’ve called a defeater. And once such a defeater is brought up, of course, the burden of proof will be met, and the zombie argument will collapse. But until anybody proposes something along those lines, it’s just as reasonable to believe in zombies as it is to believe that the bridge you’ve constructed won’t collapse. It might, yes—but it’s not reasonable to believe that it will before we have further knowledge.

    Yes, the process from conception to possibility to reality is indeed analogous. It is a process that fails all the time, to the best of everyone’s knowledge (including yours!) and is the reason why R&D is so freaking expensive.

    I think you’ve misunderstood me here. I’m not saying that every conception we form is true by saying that the process never fails (that would be ridiculous); I’m saying that it’s never systematically wrong. It’s not the case that we’ve ever built a bridge that collapsed, without there being some reason why; some additional factor that we haven’t taken into consideration (for if we had, we’d not have had a conception of the bridge as working). And given that factor, we can conceive of that bridge as failing. And that factor is all I expect from the detractors of that zombie argument. Merely pointing out that maybe such a factor exists is like saying that maybe a factor exists that the bridge falls down: maybe, yes, but you’ve not given any reason thereby to believe that the bridge will fall down.

    In other words, once there exists some simple model system such that I can form a conception of how some subjective spark of awareness creeps into existence by the interrelations and interactions of its components, I’ll immediately and happily abandon this line of argumentation. Saying that something like that might exist, however, won’t do.

  25. 25. David Duffy says:

    Zombies is just “other minds” repackaged. It is logically conceivable but not particularly helpful.

  26. 26. Sergio Graziosi says:

    Jochen,
    we’re spiralling again, I guess a recap is due.
    Classic conceivability argument goes like:
    1. Pain [or p-consciousness, or qualia, or other such phenomenon we are interested in] without [your mechanism of choice] is conceivable.
    2. Conceivability entails metaphysical possibility (at least in [the chosen case]).
    3. The metaphysical possibility of [whichever phenomenon we are interested in] without [your mechanism of choice] entails that [whichever phenomenon we are interested in] is not identical to [your mechanism of choice].
    4. [whichever phenomenon we are interested in] is not identical with [your mechanism of choice].

    I’ve used the square brackets with generic ‘filling’ to highlight that the manoeuvre is used as an anti-physicalist argument because it’s expected to generalise: take any mechanism that someone is proposing to be generating [something], and you’ll be allowed to deploy the argument. This is the first indication of why I’m attacking the generic argument, not any particular version. Moreover, I do think that Law is proposing a new variation, and I’d like to invalidate it as well.
    My objection attacks the jump between points 1 and 2 above. Which I claim is not granted, on the basis of the difference between weak and strong conception.
    Premise 1. above can, in natural language refer to weak conception. However, in point 2., metaphysical possibility can be granted only if the conception introduced in 1. is strong[1].
    I claim that there is no way to guarantee that the proposed conception is indeed strong[2], and thus that we can’t claim the argument guarantees metaphysical possibility[3] (I hope we agree that the guarantee is crucial to the argument!). Moving from 1 to 2 therefore isn’t automatically acceptable, one needs first to demonstrate that a given conception is strong[4]. I also argue that demonstrating that much is exceedingly hard, and even in maths, where we enjoy the benefits of unambiguous definitions, demonstrating that a conception (conjecture) is strong (producing a theorem) is exceedingly hard. It follows that demonstrating the same when our initial definitions include stuff that we can’t define (’cause it’s our object of enquiry, as in “what exactly is p-consciousness?”), we have no way to establish beyond reasonable doubt if the conception is strong[5]. Thus, we can’t move from 1. to 2. and the whole argument fails, regardless of the chosen object and subject in point 1. I’d also claim the same failure applies to Law’s argument, but we don’t need to discuss this now.

    Your objections puzzle me, because you have agreed with much.
    Let’s see: #20

    I think that what you call ‘weak conception’ is a trivial notion, as far as philosophical thought experiments are concerned — it seems to me we can ‘weakly conceive’ of just about anything, so obviously, it’s not going to tell us anything about possibility, whether logical or metaphysical.

    Thus, you agree about [1] – I don’t think we need to revisit this bit.
    Then, #24:

    QED: if we can’t formally demonstrate the soundness of a given conception, we can’t be sure it’s a strong one – we usually find out by trying, and we frequently fail.

    I readily acknowledge this. But, if there’s some error to a conception, there is what I’ve called a defeater — a reason for why our conception failed.

    I’ll return to the defeater concept below. For now, I want to make sure that you accept my point [2]. You readily acknowledge my own quote, so I can’t see why I could think you disagree. I then proceed with [3], which merely joins [1] and [2] linguistically, it doesn’t do any work, not as far as I can see.
    This leaves me with just one point where we may be departing, singposted as [4] and [5]. I claim that given that weak conceptions are cheap and grant nothing, to accept a given conceivablity argument we need to be reassured that the specific conception deployed in a specific conceivablity argument is strong [4]. I also say that we “can’t formally demonstrate the soundness of a given conception” [5] (you agree: #24 “aiming for formal soundness on such a topic is a futile endeavor.”!) and that therefore conceivablity arguments can be rejected in bulk. They just don’t work.

    You disagree and ask instead for a defeater. However, you ask for a specific defeater: (#24)

    Give me a defeater to the zombie conception, and I’ll immediately quit my yammering

    This line told me we aren’t understanding one another, for two reasons.
    (a) I am aiming at defeating all conceivability arguments, not specific instances.
    (b) My points [1]-[5] are intended to be a defeater of the generalised version. They are generic, because they are supposed to work across the board.

    You disagree, but I truly can’t see how or why. You grant that we can’t automatically accept that a given conception is strong, you then agree that formally proving that any given conception (in the philosophy of mind domain) is hard enough to be probably impossible. However, you refuse to accept that therefore all conceivability arguments are all built on shaky foundations (to put it mildly!), instead you ask me for specific defeaters of specific instances (“the zombie argument”). To me that’s equivalent to ask “demonstrate me that the zombie conception is not strong” – see maths: that’s what happens when someone demonstrates that a conjecture is false. And how on earth am I supposed to do it on a purely conceptual basis? We’ve agreed that the symmetric demonstration (that a conception is strong) is formally impossible, so why would demonstrating the existence of a defeater be not only possible, but required? I don’t see the logic.
    It’s even worse than that. P-zombies are defined around p-consciousness, but p-consciousness lacks an agreed upon definition: we positively don’t know what it is. If we did, we could perhaps find defeaters and/or validate the strongness of conceptions built upon it, but right now, we can’t. Moreover, because conceptions like p-zombies entail epiphenomenalism (see below, and/or re-evaluate your position here), it is impossible to find a defeater via empiric methods.

    One possibility is that you are trying to say: “ok, formal proof of strong conception is impossible, we can’t even try. However for this specific instance, I still want a defeater, as I can’t see how it could be found”. This is wrong, because I’ve already broken all conceivability arguments – and you’ve agreed with all crucial passages. But never mind: I did play along and showed you why it’s reasonable to expect that defeaters exist even when we can’t see them. Indeed many possible defeaters have been proposed in the past, the one I’ve linked above (by me), Dennett’s Zimbo, and countless others? I don’t need to repeat them, or propose a new one, I did anyway, showing why I am convinced that the inductive chain on which the p-zombie conceivability rests is, or at least could be, weaker than the one which produces the requested defeater – when formal proofs are out of reach, all we can do is hedge our bets.
    This point is essential: once you accept the practical impossibility of formal proof, you can still look for defeaters, but you can’t formally prove they work. Hence, you have to use heuristic methods to evaluate defeaters. But this is unnecessary: conceivability arguments rest on their generalisability – to work, they have to apply across the board. The empirical observation of how Maths progresses tells us that this is not the case for easier cases, thus, we cannot expect conceivability arguments to be trustworthy.

    You don’t like my latest argument because it’s circular, according to you. I disagree, but we’ve played this game already. What I can offer instead is to show why merely accepting the conceivability of p-zombies sends you down a one way street that also happens to be a dead end. If you enter it, you can’t get out by finding a specific defeater, that’s because your already accepted assumptions make such defeater impossible.

    If you prefer (somehow, I doubt you do!), I think you are the one that has fallen into a circular argument. You want me to find a specific defeater (explain why the idea of a zombie is not a strong conception), by doing so, you imply that you have already accepted that the conceivability argument is valid (both in general and for p-zombies). Given all the agreements listed above [1-3, perhaps some parts of 4 and 5], this can only be if you accept that p-zombies can be strongly conceived, but before slipping into the particular, we’ve agreed that we can’t take it for granted that this is the case.
    It becomes even trickier, because you then follow our engineering detour and claim that defeaters, when are not spotted at the theoretical level, are typically uncovered empirically (via trial and error, and other such practical methods). Well, guess what? If you accept the strongness of the p-zombie conception, empirical falsification is possible and no causal explanation can be used to defeat it. That’s by definition: p-zombies are physically indistinguishable from non-zombies and p-consciousness is a strong epiphenomenon, so it can’t be detected in any way; you can’t demonstrate that X causes (epiphenomenal)p-consciousness because you can never demonstrate that (epiphenomenal)p-consciousness is present.
    This means that we need to defeat the argument before premise 2. is fully granted. Against zombies, any attack beyond that point will invariably fail. I saw a way to attack all conceivability arguments, by refuting to pass from premise 1. to 2. Absent a generic defeater of my own defeater (ugh!), I do not need to pass over that point, and act as if I could assume that p-zombies are possible. I don’t need to, and therefore I refuse to (beyond what I did already), especially because I know that playing along entails defeat.

    Perhaps you didn’t spot the generalised nature of my argument, and could thus slip into asking for defeaters of specific instances – in the case of p-zombies, this is formally impossible, no logic and no empirical observation can kill p-zombies, not once their conceivability is accepted. Moreover, I don’t care about specific instances, as long as you can’t explain why passing from 1. to 2. can be shown to be generally acceptable.
    I am more and more convinced that you think otherwise solely on the basis that you can’t see a defeater, which is a reason that cannot be accepted by a proponent of epistemic humbleness like yours truly. Stalemate?

  27. 27. Jochen says:

    It should be quite obvious that you haven’t shown the general conceivability argument to be fallacious. After all, I could conceive of this sentence, of the act of typing it, which ensured me of its possibility, that I then proceeded to realize; and it’s not by accident that indeed the sentence I intended to write materializes on the screen. So there’s manifestly something that conceptions tell us about possibility—and that’s all that we need.

    Every time we plan something, we rely on the general conceivability argument. We trust that, since we could formulate that plan, we can also enact it. And again, this just works—so arguments that purport to establish the contrary simply fail once tested against reality. You haven’t shown planning to be impossible; but that’s what you’re claiming to have done.

    Furthermore, even if your counterargument were successful, an entirely appropriate response would be—so what. I can still plan, and I can still rest assured that, on the whole, my plans will succeed far more often that they ought to, if there were no link between conceivability and possibility. It’s a bit like with Hume’s realization that induction is formally impossible: the best response is, so what, I still know with greater certainty than I know almost anything else that the sun is going to rise tomorrow, even if I can’t formally derive this from the fact that it’s risen every day so far.

    But still, I’m actually much more sympathetic to your argument than this, and I have, in the interest of defending against the strongest possible interpretation of your opponent’s arguments, given it far more credence from the start. I’m not merely saying that obviously, even if your argument is convincing in theory, in practice, I find that I can indeed plan, and realize my plans.

    My strategy is more akin to Popper’s response to Hume’s problem of induction. I propose that the inductive conclusion is to be accepted only absent defeaters. That, in other words, I have every right to be convinced of a plan, unless somebody points out an error. After all, this is also what happens in the real world: I make a plan, and it doesn’t work, or it doesn’t work on the first go. In this case, I’ve made a mistake somewhere—which may range from a typo while trying to write down the sentence I’ve previously conceived of, to accidentally running into and failing to recognize a logical impossibility.

    All of this doesn’t defeat the general fact that planning is possible, however. The only recourse is then to argue that a particular plan doesn’t work; but in order to convince anybody of that, you’ll have to do the work and show where the plan fails. Nobody planning to build a bridge is going to stop construction upon you pointing out that plans sometimes fail, that we cannot in general trust our conceptions to establish possibility. Everybody knows that; it’s a triviality.

    If you want to convince anybody that their plan will fail, point out the mistake with the plan. If you want to convince anybody that the zombie argument fails, point out the mistake with the argument. Just attacking the general conceivability argument will do no work at all—because again, we know that in general, conceptions do tell us about possibility. We rely on that fact in everything we do, after having formulated a plan for how to do it. This isn’t infallible, but it need not be; it need merely be better than chance, and that it is by a huge margin.

    Because then, emulating Popper again, we have justification for placing the burden of proof with those wishing to dissuade us from our plans—we may request a defeater, a falsification. And in the case of the zombie argument, that defeater must take the form of conscous experience emerging from physical systems. And this, and here’s something I’m almost as sure of as I am that the sun will rise tomorrow, is something nobody is ever going to provide. The fact that we cannot conceive of physical facts giving rise to experiential facts, after all, is the reason for the whole debate. That systems like gears and levers acting upon one another should give rise to conscious experience is, quite simply, inconceivable. Leibniz nailed it with his mill all those years back.

    So that, to me, is the bottom line. If you, as you claim, had defeated all conceivability arguments, you’d be faced with the task of explaining why plans work, ever, with greater than chance success. The simple fact of why they work is easily stated: brain-stuff is ultimately not different from world-stuff, and a conception is merely establishing a set of relations that supervene on brain-stuff, which then can be realized by means of world-stuff. What we can model, we can create (resources permitting and in principle, at least). This isn’t foolproof, but it need not be; it merely needs to be more reliable than chance, such that we can justifiably say that we succeeded in our actions because we had a good plan, and not just because things accidentally happened to fall into place, with our plans (conceptions) having nothing to do with it. That we sometimes fail does not tell us that our conceptions tell us nothing about possibility; it merely tells us that we sometimes make mistakes—either because our conception was flawed, or because we made an error realizing it, or for any of I’m sure dozens of other reasons. But if there is such a mistake, then it must be pointed out; merely saying that sometimes, plans don’t work does not suffice to show that any given plan won’t work. So I’ll be waiting for this pointing out, and it’s going to be the greatest surprise of my life if it ever happens.

  28. 28. Jochen says:

    Perhaps a slightly different analogy might help. The zombie argument essentially proceeds from an analysis of the building blocks to the conclusion that nothing assimilated from those building blocks has a certain characteristic. So, for instance, one may conclude that from sand, you can’t build a bridge spanning a river that can carry the weight of a car: the building block is simply entirely incapable of this.

    Then, you come along and point out that well, the argument made above relies on conceivability entailing possibility, which sometimes doesn’t work, and since we can’t conceive (or strongly conceive) of the bridge in its entirety, we can’t positively claim to know that there’s not something that’s gonna happen to imbue the bridge made from sand with properties capable of carrying the weight of a car.

    And in a sense, that’s right. Plans sometimes do fail, and we did, in fact, never conceive of the bridge in toto.

    But nobody is going to try and build the bridge; everybody is going to remain just as convinced as before that it’ll be impossible—unless you can come up with a way for the sand to acquire the necessary properties to build a load-bearing bridge. Perhaps you can show that a certain configuration of sand is self-supporting; perhaps you can propose baking the bridge at high temperatures to turn it into glass; perhaps you add a certain adhesive, or sand produces adhesion naturally under certain conditions. Anything like that, and people will be much more willing to be convinced.

    But merely saying that we can’t exclude that something like this happens, hence, we’ve never established that bridges of sand can’t carry the weight of a car, simply isn’t going to be enough.

  29. 29. Sergio Graziosi says:

    Jochen,
    my previous reply (was quick, for once) got lost somewhere in the interwebs, I’ll try again.

    You haven’t shown planning to be impossible; but that’s what you’re claiming to have done.

    There is a misunderstanding here. I was not trying to propose that weak and strong conceivability tell us nothing. I’m with you on using (weak) conception for planning: it is circular, as planning is a form of conception, but not all circular arguments are vicious ;-).
    Never mind, when we build something, we start by conceiving it, we then try building it, adjust if needed. We don’t have a requirement to ensure that our conception is strong (in my terms). We can just try, if resources at our disposal allow it.
    Thus, conception for planning is not only acceptable, it’s required, as it allows us to get something started without acting randomly on the world. It is also useful: by planning ahead, we could (and normally do) find a-priori defeaters of many plans, so we can discard all plans that we have good reasons to expect are destined to fail. We cut down the search space, if you wish.
    So yes, even weak conceivability does tell us something, because we can try to break it a priori.

    This gets us close to the work that conceivability arguments are supposed to do, in terms of metaphysical possibility. But importantly, it does not bring us to the desired destination. This is because “metaphysical possibility” needs to be interpreted in a special way for conceivability arguments, but it does not for planning.
    Planning: a weak conception tells me “this plan may work”. I can find out by trying. Here the word “possibility” means “it might be possible”. Fine. Let’s go on and try.
    Conceivability argument: for this to work, “(metaphysical) possibility” needs to become (A)”we know for sure that logic alone cannot provide any a-priori defeater”. This is required because:
    I – The argument has to generalise: in the absence of a fixed “mechanism of choice” (see point 1. on top of #26), generalisability is not optional. We are still searching for candidate “mechanisms of choice” (p-fibres are not good enough, we know that!), and conceivability arguments try to demonstrate that this search is futile (in bulk).
    II – “Metaphysical possibility” in itself implies logical consistency.
    (Both I and II are sufficient to establish the “must” requirement!)

    In comparable terms (given the planning example, with reference to I), conceivability arguments seek to build on the interim conclusion (B)”what I’ve conceived must be logically possible”. In the first case (planning), all that’s required is to reach the “may/might” threshold. In the second (conceivability argument), it only works if you can go all the way to “must”.
    Thus, unpacking the argument in this way makes it clear why it doesn’t work. You have conceded that strong conceivability is out of reach, but it happens that the phrase (A)”we know for sure logic alone cannot provide an a-priori defeater [for a given conception]” is equivalent to (B)”what I’ve conceived must be logically possible” and equivalent to (C)”[the given conception] is strong”. If we can’t concede (C), we can’t concede (B) or (A), which breaks conceivability arguments at their core.

    With reference to II, all I’m doing is pointing out that our ordinary conceptions are weak. Having produced an idea, it is not reasonable to say “since I’ve produced this idea, this idea must be logically consistent”. We don’t know it is, and you agreed.

    Importantly, all your attempts to show me why I’m wrong rely on specific instances of conceptions used in planning (or equivalent). Thus, all of them can stop at the weak “may work” threshold and therefore require a defeater to be dismissed a-priori (that’s agreed!). Actual conceivability arguments however rely on the stronger “must” threshold, that’s why they fail.

    If you must (;-) !), you can consider the above as the meta-defeater of generalisable conceivability arguments.

  30. 30. Jochen says:

    Sergio:

    I’m with you on using (weak) conception for planning:

    I don’t think weak conception is of any use in planning (or pretty much anything) at all. I can ‘weakly’ conceive of myself flying unaided, but it’s not actually possible for me to do so, and I also can’t strongly conceive of it (one might be convinced one could have such a conception, but defeaters—the lack of any force to counteract gravity—are easily found, and once those are included, completing the conception, one finds it isn’t possible to imagine oneself in unaided flight).

    for this to work, “(metaphysical) possibility” needs to become (A)”we know for sure that logic alone cannot provide any a-priori defeater”.

    I disagree. I’m not saying I can prove beyond a shadow of a doubt that zombies are possible (which would be the only case in which I need to know anything ‘for sure’), I’m merely saying that there’s an argument that’s convincing enough to shift the burden of proof towards having to explicitly provide a defeater in order to attack the argument.

    If you like, you can read me as making the following amended zombie argument:
    P1. Conceivability implies metaphysical possibility (in your terms, I would use ‘strong’ conceivability here, but I still think there’s too little to ‘weak’ conceivability to really take it into consideration)
    P2. We can conceive of the parts that make up, say, a brain, or a computer
    P3. Absent defeaters, conceptions of the parts imply conceptions of the whole
    P4. There are no defeaters to the conceivability of zombies
    —————-
    C1. (From P2, P3, P4) Zombies are conceivable
    C2. (From C1 and P1) Zombies are possible

    Of course, you can attack this argument, the same way as you can attack any argument—showing its premises to be false, or showing its logic to be fallacious. I’m reasonably sure the logic is OK, so you’ll have to attack the premises. So, what are your options?

    We’ll leave P1 for the moment. So, P2: you might hold that we can’t, in fact, conceive of the parts of a system we’d expect to be conscious. If so, we can just refine those parts: if neurons are too difficult to properly conceive of, just go to the simple interactions of atoms, if it comes to that. Or, if we’re talking about an artificial life form, we can just go to individual NAND-gates, which are perfectly well conceivable.

    P3 is the premise that allows for planning: typically, when we plan, we move to a conception of the whole from conceptions of the parts. There are no ‘hard’ surprises here: when a plan fails, it’s because our conception wasn’t accurate; it’s not the case that there are strongly emergent properties that foil us. In a way, denying P3 is to deny that planning is possible, and that the properties of the whole follow from the properties of the parts. I do not believe you’d want to give this up.

    Thus, we come to P4. The only defense in favor of P4 I have is really that in hundreds of years of thinking, nobody’s ever come up with a way to overthrow it. Furthermore, it seems utterly unthinkable that such could exist. It’s like saying that there is a way to explain what color looks like to a blind man, after all, even though anything we know seems to argue against it. But sure, I acknowledge that finding a defeater—a detailed explanation of how some sort of sensation, of subjective feeling emerges from a physical system—would bring the argument down. But merely pointing out that in principle, I can’t prove that no such defeater exists, does not do any work in actually showing the argument wrong.

    That leaves P1. It’s usually here that most think they could sink their teeth in, but it’s very much harder than it might seem at first. All P1 really needs, by way of defense, is 1) we can’t knowingly hold contradictory beliefs, and 2) there are no true contradictions in the real world. Given this, the set of conceivable things and the set of metaphysically possible things coincide, in the form of the set of logically possible things.

    Again, that doesn’t mean that everything we believe we have conceived of is necessarily possible. But if it isn’t, then the above means that there must be a defeater—something we haven’t included in our conception, but should have. Then, the way to defeat this argument is to point out the defeater.

    (Now, I of course do believe that P1 is wrong—the set of conceivable things and the set of possible things do not coincide, because we can only conceive of the computable, due to the nature of how conception works, and the real world isn’t computable; indeed, claiming it is, is a meaningless statement, amounting to what Whitehead called ‘the fallacy of misplaced concreteness’. But that’s a different argument.)

  31. 31. Sergio Graziosi says:

    Jochen,
    briefly (?):
    Any conception, in my own definition, can be either weak or strong – distinction is strictly binary.
    Strong conception: a conception that can be shown, beyond reasonable doubt, to be logically consistent.
    Weak: all other conceptions.
    For conceivability arguments to work, it is *necessary* to show that the proposed conception is strong.
    We agree this can’t be done. Thus, your argument stops at P1.
    You end up with something like this:
    P1. Strong conceivability implies metaphysical possibility.
    P2*. Nothing can guarantee the Zombie conception is strong.
    P3*. P-Zombies may or may not be strong conceptions, finding a defeater would guarantee the conception was weak.
    C*: We don’t know if zombies are metaphysically possible.

    Moreover, defeaters of the Zombie conception abound. For example (please do not answer on the merit of this example, the point is that they exist, but fail to convince the Zombie believers): I know I am conscious, I want to understand how. If I know I am conscious, understanding what caused this knowledge is required to identify what consciousness is, as consciousness is logically required to be a con-cause of this knowledge. It follows that consciousness (whatever it is) is not an epiphenomenon. Thus, P-Zombies are *logically* inconsistent.
    The above can be refuted if one assumes that P-Zombies are metaphysically possible, which is the same as assuming they are strong conceptions, or *logically consistent*. If you don’t make this assumption, the above stands. The same applies to most or all other proposed defeaters.
    Thus: accepting the strong conceivability of P-Zombies puts you in a dead end – which is why P-Zombie believers never acknowledge the logic of any proposed defeater.

    You may be trying to say that, by virtue of:
    P2. We can (strongly) conceive of the parts that make up, say, a brain, or a computer
    P3. Absent defeaters, conceptions of the parts imply conceptions of the whole
    It is therefore possible to say that the implied “conception of the whole” can be considered strong (as in: has been shown to be logically consistent, beyond reasonable doubt). This could be eventually granted as a function of the amount of effort placed in trying to find defeaters.
    But this can’t work if you keep it generic. In fact, everything we see is made of atoms, etc. Absent a proposed way of assembling such things, we cannot know if a given result is logically expected. If you wish, we cannot find specific defeaters because you haven’t specified anything.

    If you could produce a description of how to put together atoms is such a way that we would all accept it is indistinguishable from a given person, and you also had a theory to explain why such an artefact will behave like the original, but wouldn’t be conscious, then and only then we could start looking for defeaters of your specific conception. Given enough effort, if nobody could find a defeater, then we would need to have a discussion on whether the absence of defeaters is enough to consider a given conception strong. But you can’t do this, so the discussion is entirely moot.
    In other words, your original point P3 is vague hand-waving: “conceptions of the parts imply conceptions of the whole” only when you specify how to assemble the parts and have a theory to predict how such parts will interact. We have neither.

    Thus, the generic approach fails because strong conceivability cannot be demonstrated. Specific approaches (equivalent to planning) can be evaluated only when they are made specific; in such cases, the language of defeaters can be deployed, explaining why planning can and does work (within limits). However, defeaters start to apply once you fill in the details; asking for defeaters, absent all relevant details, is like asking to detect an epiphenomenon (i.e. a trick question!).

    Asides:
    1. You keep saying that “there’s too little to ‘weak’ conceivability”. That’s my point. Given that only demonstrable theorems entail (beyond reasonable doubt) strong conceptions, all this talk of metaphysical possibility is utter nonsense.
    2. “we can’t knowingly hold contradictory beliefs”, well, I’d suggest you to try debating a fervent brexiteer, then ;-).

  32. 32. Jochen says:

    Sergio:

    Strong conception: a conception that can be shown, beyond reasonable doubt, to be logically consistent.
    Weak: all other conceptions.

    The case you call ‘weak’ is just the case of ‘not having a conception’. Conceiving of something—fully having, say, a ‘mental model’ of it—implies its logical consistency in just the same way as building it does, since it’s nothing different—it’s just building it ‘in the mind’, so to speak.

    For conceivability arguments to work, it is *necessary* to show that the proposed conception is strong.

    No. Perhaps for a species of ‘idealized’ conceivability argument that an ideal reasoner could make, but not for the argument I am making. All I’m saying is that we can conceive of the parts of the whole, which, absent defeaters, implies that one may generalize to the whole. This does not require even being able to conceive of the whole in any form whatsoever.

    (You could frame things in terms of an ideal reasoner that could conceive of the whole, i.e. that could conceive of an atom-by-atom replica of a conscious entity; then my argument is that absent defeaters, we may conclude that this conception would not include conscious experience.)

    It follows that consciousness (whatever it is) is not an epiphenomenon.

    I think I’ve told you before that when you reach conclusions as strong as this after a paragraph of reasoning, you ought to check whether something’s gone wrong, particularly if there’s libraries full of writing on the subject.

    In this case, too, I don’t think your reasoning justifies your conclusion: it doesn’t follow in any way from the possibility of p-zombies that consciousness is epiphenomenal. For one, zombies might hold the same beliefs we do about consciousness, but simply be wrong—there’s nothing paradoxical about that. Furthermore, interactionist dualism may be true: in that case, we still had a physically identical copy with no consciousness, but the physical would not be causally closed, and thus, the copy might behave differently from the conscious original, at least in some cases. Or, there could be a Leibnitzian pre-established harmony: a supreme being has arranged to make our judgments about conscious experience come out right, despite there not being any sort of ‘ordinary’ interaction between the two. A neutral monist might hold that there are two sorts of stories that can be told, both perfectly sound, one which refers to physical causation, and another that refers to mental causation. And so on.

    If you wish, we cannot find specific defeaters because you haven’t specified anything.

    But it’s completely clear what a defeater would look like: a physical arrangement of parts that gives rise to even the tiniest spark of consciousness. In other words, any actual explanation of consciousness in physical terms will do, a reduction to physical properties, anything like that. Give me that, and the argument will be refuted, because then, the generalization from our conception of the parts to the whole will have shown to be false. Which, I agree, it may well be; but which we’ve got no grounds to judge false. It’s the same as saying that there are pink unicorns living on the dark side of the moon: sure, they may well do, but I’ll accept that only if you show me evidence. Likewise, I’ll accept that physical systems produce consciousness once you tell me how. It’s just the epistemic baseline one ought to take.

    If you could produce a description of how to put together atoms is such a way that we would all accept it is indistinguishable from a given person, and you also had a theory to explain why such an artefact will behave like the original, but wouldn’t be conscious, then and only then we could start looking for defeaters of your specific conception.

    In that case, we would no longer have to: we already know, if we have truly conceived of the complete system, whether it’s conscious or not, that is, whether there is a defeater or not. The whole defeater-business only enters because we cannot conceive of a whole human being at the atomic level!

    Absent a proposed way of assembling such things, we cannot know if a given result is logically expected.

    I do not need to specify how one builds a bridge to know that it can’t be made from jello: it’s the wrong material for the job. So this just misses its mark.

    Thus, the generic approach fails because strong conceivability cannot be demonstrated.

    The argument I’m making does not need such a demonstration. Its premises include that the conceptions we have of the part suffice, absent defeaters, and that there are no defeaters—that does the same work.

  33. 33. Sergio Graziosi says:

    Jochen,
    This conversation is probably gone beyond its usefulness…

    The case you call ‘weak’ is just the case of ‘not having a conception’. Conceiving of something—fully having, say, a ‘mental model’ of it—implies its logical consistency in just the same way as building it does, since it’s nothing different—it’s just building it ‘in the mind’, so to speak.

    This, as well as most of what follows it, is obviously wrong; so much so that I’m left wondering whether you are trying to wind me up.

    If that’s the case, you were not entirely unsuccessful: what I’ll write below will be unusually blunt, I fear – please accept my apologies if that’s the case.

    Back on topic: according to you, if you have a mental model (say: your plan of what to write in #32), it is either logically consistent or “not a conception” at all. What such an illogical plan should be called, is left unspecified, let’s call it a misconception.
    The problem I’m pointing at is that, from within, you can’t know if you had a conception or a misconception. What feels logically consistent to you (or me), does not need to be logically consistent at all.
    You obviously don’t believe me, but luckily(?!), we have a demonstration provided within #32 itself (more than one, but I’ll stick to the most obvious).

    I don’t think your reasoning justifies your conclusion: it doesn’t follow in any way from the possibility of p-zombies that consciousness is epiphenomenal. For one, zombies might hold the same beliefs we do about consciousness, but simply be wrong — there’s nothing paradoxical about that.

    This bit defies (my) comprehension, quite literally. In my understanding, a P-Zombie that holds beliefs is, by definition, not a P-Zombie. Therefore, what you wrote down is logically inconsistent, to my eyes. If I’m right, congratulations(!), you’ve made my point: we can indeed conceive of something (the quoted paragraph), be convinced it is consistent, and be mistaken. If I’m wrong, then my conception of your presumed mistake must be inconsistent in some way that I fail to recognise. The third possibility is that you wrote down the above even if you knew it’s wrong, which would be disappointing, so I’ll assume it is not the case.
    Excluding mischief, it follows that one of us was able to (proto?)conceive ideas that are not logically consistent, while being fully convinced that they are.

    If you or I can sincerely believe we have conceived something, but because of our disagreement we concur that one of us must be mistaken, how can we guarantee, lacking formal proof, that *any* idea *is* logically consistent? We can’t. What feels to us as a full “conception” (in your sense) can (indeed, in this case it must, for one of us) be a misconception. Thus, we can’t be sure that P-Zombies are logically consistent – our disagreement on whether P-Zombies can hold beliefs is testifying it, QED.

    Now, considering that I’ve made the point above a few times, and considering that what I perceive as the the quality of your replies has been decreasing steadily, I think we should let this conversation die.
    I am guessing you may feel the urge to reply. That’s 100% OK, but bear in mind that I will be trying to resist my own corresponding desire. As far as I’m concerned, you have vigorously tried to dismantle my initial idea. I the process, you have convinced me that my initial idea has some merit; of course I may be wrong, but that’s how I feel. Therefore I don’t believe you can change my mind by trying once again. In this circumstance, it would be a little mischievous for me to keep the debate going: if the possibility of changing my mind is off the table, the conversation would become a point-scoring exercise, not an honest debate.

  34. 34. Jochen says:

    Sergio,
    I feel like you’ve missed/forgotten about a couple of points in this conversation. For instance, in my very first post in this thread, I said that:

    After all, if you properly conceive of something, you have some sort of model of it in mind; but why, if that model can exist, should the thing it models be impossible? It’s just a different instantiation of the same sort of structure.

    So it’s not like what you decry as ‘obviously wrong’ is a new addition to my argument—it’s quite literally where I start from.

    Also, in that same post (and numerous other times), I’ve acknowledged that it’s possible for us to merely believe that we’ve got a conception of something; this is, indeed, what I’m driving at with the whole ‘defeater’-issue: if we wrongly believe we can conceive of something, then there’s something by virtue of which that belief is wrong. Pointing that something out shows we hadn’t in fact conceived of the thing in question.

    It’s a bit strange, to me, that if you’ve been following my posts to this point, this should cause you such trouble just now.

    In my understanding, a P-Zombie that holds beliefs is, by definition, not a P-Zombie.

    I think that differs from the usual understanding of zombies, then. Typically, a zombie does not have phenomenal consciousness; that does not mean they have no intentional states. It might lack beliefs, if something like phenomenal theories of intentionality hold true, but I don’t think it’s necessary to the concept of a zombie that it lacks them. (A quick google returns lots of hits talking about a zombie’s phenomenal beliefs, whether it can have any or not, so I don’t think I’m widely off-base here.)

    one of us was able to (proto?)conceive ideas that are not logically consistent, while being fully convinced that they are.

    At best, it shows that one of us believed to have a conception that turns out to be logically inconsistent, and thus, isn’t a conception at all. But this isn’t in contradiction with anything I wrote.

    how can we guarantee, lacking formal proof, that *any* idea *is* logically consistent? We can’t.

    And the entire volume of text I posted in this thread so far is basically concerned with explaining why we don’t need to. Let me try and provide the argument in all excruciating detail.

    P1 If an ideal reasoner can conceive of something, it is logically possible (as an ideal reasoner would spot
    any logical inconsistency).
    P2 If something is logically possible, it is metaphysically possible.
    C1 (From P1 and P2) If an ideal reasoner can conceive of something, it is metaphysically possible.
    P3 We can conceive of the functioning of simple systems without any attendant conscious experience.
    P4 A conscious being can, in principle, be built from these simple systems (e.g. a conscious robot from NAND-gates).
    P5 Absent defeaters, the conceivability of these simple systems without conscious experience entails the conceivability, to an ideal reasoner, of the whole system absent conscious experience.
    C2 (from P3-P5) Absent defeaters, a replica of a conscious being lacking conscious experience is conceivable to an ideal reasoner.
    P6 There are no defeaters.
    C3 (From C2 and P6) A replica of a conscious being lacking conscious experience is conceivable to an ideal reasoner.
    C4 (From C3 and C1) A replica of a conscious being lacking conscious experience is metaphysically possible.

    Nowhere do we need to prove the conception of a zombie to be 100% free from error; in fact, the entire argument basically starts with ‘we can’t prove the conception of a zombie to be absolutely consistent’. But given the way the facts lay, we should believe that the conception is possible in principle (to an ideal reasoner), until and unless somebody can show why it’s not.

    You’re of course free to continue believing that the idea that we can’t prove conceptions completely consistent somehow defeats the zombie argument; but you’re just barking up the entirely wrong tree there. After all, the fact that we can’f form such conceptions is where my argument starts out from—so in point of fact, if that’s really all that there is to your counterargument, you’ve never even acknowledged my point at all, much less argued against it.

  35. 35. Jochen says:

    I mean, really, in the end, the argument is just: We can’t conceive of zombies (in all detail). But we can conceive of the stuff they’re made from, and until somebody gives us a good reason to think otherwise, the reasonable thing to do is to believe that this stuff behaves in large (inconceivable) aggregate just the same way it does in small doses. After all, that’s how we know jello doesn’t suit bridge-building purposes; by just the same reasoning, NAND-gates don’t suit consciousness-building purposes. If you say there’s a flaw with the latter, you’ve got to explain why the former is wrong, since they’re exactly analogous (or else, build a bridge from jello).

Leave a Reply