Thursday, October 13, 2011

David Lewis, Anesthesia by Genocide, and a Materialistic Trilemma

In his famous 1980 essay, "Mad Pain and Martian Pain", David Lewis tries to thread the needle between a flat-footed functionalism and a flat-footed neural-state identity theory about the mental. Flat-footed neural-state identity theory equates mental states, like being in pain, with possession of particular neural states. Thus, counterintuitively, it implies that beings who are behaviorally similar but internally very different, such as (hypothetically) Martians, can't feel pain. Flat-footed functionalism equates mental states with causal/functional roles. Being in pain, on such a view, is just being in a state that is caused by things like tissue stress and that tends to cause things like wincing, avoidance, and self-ascriptions of pain. This view, counterintuitively, implies the impossibility of "madmen" who feel pain for unusual reasons and have unusual reactions to it.

Lewis's solution is to say that some entity X is in pain if and only if X is in the state that occupies the causal role of pain for the "appropriate population". The "appropriate population", he says, might be (1.) us, since it's our term, (2.) a population that X belongs to, (3.) a population in which X is not exceptional, and (4.) a natural kind such as a species. In the normal case, all four criteria are met. In the Martian case, 2-4 are met though 1 is not, which is good enough. In the mad case 1, 2, and 4 are met though 3 is not, which is also good enough. Since mad Martian pain also seems possible, 2 and 4 alone will be sufficient for pain on Lewis's account.

Now the funny thing about these criteria is that they are all extrinsic or relational , and you might have thought that whether X is in pain or not should depend entirely on what is going on within X; you might have thought that pain would, in today's jargon, "supervene locally". The weirdness can be made vivid with further thought experiments. Criterion (3), for example, can be altered by genocide. Suppose that X is in a state that plays the causal role of pain for most of the population but the causal role of hunger for him and maybe a few others -- a "madman" case. On Lewis's account he will be experiencing pain. Now suppose that X is desperate to end his pain. On Lewis's account he might end his pain by perpetrating genocide upon all the non-mad people of the world. Voila, condition (3) flips, and X's pain has changed to hunger! This is anesthesia by genocide. We could similarly produce anesthesia by reproduction or speciation.

Real advocates of physical-state identity theory are hardly ever as flat-footed as those imagined by Lewis (as Lewis explicitly acknowledges). Like Lewis, they tend to embrace accounts on which to be in pain (or any other mental state) is to be in a state of a certain physical type, where the relevant physical type can vary between different types of being. What type of physical state is identical to what type of mental state, for beings of your type, then depends on facts about the particular causal or functional role of that state in members of your group or on the causal or functional history of that physical state in members of your group and/or in your own evolutionary or developmental past. Such type classifications are extrinsic or relational. Thus, such views have the bizarre consequences that flow from the denial of local supervenience. They allow anesthesia by genocide, or by speciation, or by hypothetical differences in past history that have no neural trace in the present.

We might thus see the mad pain-Martian pain issue as a trilemma in which each horn has bizarre consequences: Either accept the bizarre consequences of a strict functionalism (no mad pain), accept the bizarre consequences of neurobiological chauvinism (no Martian pain), or accept the bizarre consequences of denying local supervenience (anesthesia by genocide or speciation). Can a plausible materialist metaphysics dodge this trilemma? (I set aside hand-waving appeals to yet-to-be-identified intrinsic properties, a la John Searle.) I'd be very interested if you think you can point me to an example!

If all the options are bizarre, as I think, then something bizarre must be true. (Yes, dualism is also bizarre.) The problem is in figuring out which bizarre view to accept! If none of the various bizarre options merits credence, then crazyism follows.


Anonymous said...

Though this may be considered a hand wave à la John Searle, I just don't see what would be wrong with Rejecting all three consequences. Couldn't we say that, in human beings and like biological animals, pain is identical to the possession of certain neurological states, while deny that Martians without such states do not have a functional equivalent to what we call pain? Strictly speaking, Martians wouldn't then have "pain", but we may call it something like "quain". And these different sensations which are each identical to some biological mechanism, could share many improtant functional similarities (such as exclamations, wincing, yelping, etc.) that their only identifiable difference is the mechanisms. In this way "pain" and "quain" would not be identical, but linguistically and categorically indistinguishable.

If the problem is with strict identity to a particular system, then I wonder why the problem of neurological chauvinism wouldn't apply to each individual. We may go so far as to say that every pain ever felt by any individual was entirely distinct. With what we know of the brain and it's constant reinforcement of connections between each neuron during any brain process, we ought to assume that no neurological state is exactly similar to any other, despite how apparently indistinguishable the conscious experience is. If I hit my thumb with a hammer and feel a pain, that pain is identical to some neurological system at that time. Durring that process, the stimulation of the clusters of neurons will both reinforce and weaken certain connections. If a moment later I hit myself in the exact same spot with the same force, I will likely describe it as a qualitatively identical pain to the one I had previously felt. This pain would not actually be identical, however, because of the slightly different neural-state. So what justifies me in calling it the same name?

Of course the objection is that the martians pain (or quain) is completely unlike mine because of it's identity with such different biological mechanisms, but if we are willing to call each individual pain felt by me by the same name, why can't that be extended to the martian as well? Maybe we should concede that we use the word and concept of pain in many ways given certain resemblances, and since this is where we derive our concept we ought to apply it to martians who ellicit similar reactions to particular stimuli.

My attack has been mainly on our idea of neurological identity, but that is simply because I find it the most plausible way of defining pain, provided that it is revised in some Wittgenstinian way.

William said...

@Anonymous I think you may have missed the point here, if everything is behaviorally indistinguishable, there could be no Mad Pain. Also, isn't that simply an elaborated explanation of what was already described here,

"Like Lewis, they tend to embrace accounts on which to be in pain (or any other mental state) is to be in a state of a certain physical type, where the relevant physical type can vary between different types of being. What type of physical state is identical to what type of mental state, for beings of your type, then depends on facts about the particular causal or functional role of that state in members of your group or on the causal or functional history of that physical state in members of your group and/or in your own evolutionary or developmental past."

All the objections still apply then.

Richard Marshall said...

William, I think that's right. I was with a bunch of Wittgensteinians rently and Anonymous's response to the trilemma seems genuinely Wittgensteinian. I think this exchange shows that Crazyism isn't accessible to Wittgensteinians. This seems to add weight to rejecting the Wittgensteinian approach as Crazyism seems much more subtle and sensitive to the actual messyness of psychological states used in metaphysics that Wittgensteinian options are. Crazyism seems to be a kind of scientistic approach (reappropriated a la Rosenberg as a badge of honour). A Wittgensteinian move made against scientistic approaches is to claim, as Anonymous does here, that its reductionism makes it incapable of tracking the subtle varieties of linguistic discourse and that a therapeutic approach will show that the reductionism is unecessary because there is no genuine problem for any such reductionism to solve. Ironically, the Crazyism seems to be doing a much better job at theorising to the target issue by allowing for the discovery of inconsistencies that are embedded as part of the phenomena to be explained rather than revised away a la Ludwig.

Anonymous said...

I don't see that anesthesia by genocide is really possible. The appropriate population for comparison can be an extinct one. So an unusual specimen of a certain type of thing can remain unusual as a specimen of that type even if it is the only one left. In fact, I think it could be unusual even if it were the only one ever to have existed. E.g. if a car manufacturer designs a new type of car, the model-C, and produces a prototype P which is just as the finished article would be but without seatbelts, but then goes bust and never makes another model-C, P is still an unusual model-C (for not having seatbelts) despite being the only one ever made.

Eric Schwitzgebel said...

Thanks for all the thoughtful comments, folks!

@ Anon 9:37/William: I agree with William's response. This would be to deny the possibility of mad pain. Maybe that's the best bullet to bite, but it is a bullet. And as seems to be coming out in your comments, there will be more weird choices than *just* denying mad pain.

Richard: Thanks for those thoughtful comparisons with a Wittgensteinian approach.

Anon 11:54: That's one choice you could make. It doesn't seem to me the most natural interpretation of Lewis's view, construed literally. It transforms the current comparison view into a historical view, and so enables the different bizarreness of pain depending on events at a historical distance even if they have left no neurological trace in the present, yes? That's because you're still going to be committed to denying local supervenience, yes?

Anonymous said...

I think we often and reasonably deny local supervenience even for natural properties. E.g. if a mutant grizzly bear is born, which has white fur and an instinct to hunt seals, and in fact the exact constitution of a polar bear, still it isn’t a polar bear because it has the wrong history - it was born to grizzly parents. I don’t think that this itself is odd. In the case of pain, does it perhaps lead to odd consequences? Well, it allows there to be two identically constituted beings in the same state (physically, locally), only one of which is in pain. I’m actually OK with that. Pointing out that they’re in the same state skews our perspective by emphasising their similarities over their differences. If they come from different populations in which pain is experienced differently and have different histories, the fact that we’re interrupting them at an instant when they’re both passing through the same physical state shouldn’t count for a lot. (Though, in fact, they could be in identical physical states for an extended period - I’m not even sure if an instantaneous state can be a pain state anyway - and still be in different degrees of pain.)

Eric Schwitzgebel said...

Right, that's one option. What if there was a speciation event right in the middle of the organism's pain, due to nonlocal events (say reproduction and migration several miles away)? Would that end the pain? It would be strange if speciation were sudden rather than gradual, but the problem remains: Would a nonlocal sort-of shift in species reduce the pain a bit? Maybe it depends only on the organism's own past ancestral history? But then a nonlocal speciation event in that history could change that organism's feelings of pain even if there was no direct narrowly causal trace on that organism or its ancestors. Right?

Richard Marshall said...

Would something like Edouard Machery's argument about concepts help? He argues that concepts as used in psychological categorisation go to three diferent kinds of explanation - prototypes, exemplaries and theories. So a whale is a fish if prototypicality organises it, although obviously its also not a fish because its a mammal. He argues that this psychological concept is not the one that philosophers target when they talk about concepts, e.g. Jerry Fodor. The Crazyism would then be about the way this kind of categorising shifts between the three concept formulas. perhaps in unsurveyable or at least as yet unsurveyed, ways). Mad pain would be pain derived from one source, alien pain another etc.

Richard Marshall said...

Just be clear - Machery's approach denies that the concepts are natural kinds.

Eric Schwitzgebel said...

Richard, I'm not getting the connection. Could you clarify?

There may be different definitions of "fish", but you can't make a fish disappear by recategorizing it as a mammal. I'd have thought the same about pain. (But maybe if we're forced to countenance bizarrenesses, this could be among them.) I'm not sure if this is relevant to your point.

Richard Marshall said...

Sorry Eric, I guess I was being a bit quick. We might take a higher order thought theory of consciousness and propose a first order state pain and then a higher order consciousness of that pain. We might presume that higher order states have conceptual content that makes it like something eg taste of whiskey changes by learning word oaky and applying it to the taste of whisky. (This is a David Rosenthal example). Then we have a link between concepts and phenomenal consciousness. The different concept changes that consciousness. In a case like pain approaching it using human prototypes of human pain may change how it feels if I discover that I am not human. It may be like having the concept of human pain extracted from the higher order state. I might still feel something, but not pain understood in terms of prototypical human pain concepts. (Just like extracting knowledge of the concept of oaky changes the taste of the whisky. The weirdness here is that if you extract all concepts from the higher order state and ask what it is like to be in that state it seems we have to say the pain is not like anything at all. But being in a state of pain without it being painful is just Crazyism I guess.

Eric Schwitzgebel said...

Richard: Thanks, that's clearer to me! I agree that conceptual learning can alter phenomenology. The theory you mention seems to have the bizarre consequence that you point to -- which I think of as a dilemma about animal consciousness, either ascribing seemingly-too-sophisticated concepts to non-human animals or denying consciousness beyond a narrow circles. Also, I think we don't dodge the local supervenience trilemma: Concepts probably don't supervene locally, and yet it seems bizarre to think that the experience of pain does not supervene locally.