We’re looking at Alex Rosenberg’s attempt to defend eliminative materialism from the charge of incoherence in his paper “Eliminativism without Tears.” Having set out some background ideas in an earlier post, let’s turn to the essay itself. It has four main parts: two devoted to arguments for eliminativism, and two devoted to responses to the charge of incoherence. I’ll consider each in turn.
Neuroscience and eliminativism
Rosenberg evidently supposes that eliminativism is so clearly correct that there must be some way of making it coherent. Hence he devotes the first half of the paper to setting out some arguments in its defense. If those arguments are powerful, then (he seems to think) the reader will have to concede that there mustbe some way of making eliminativism coherent even if it turned out that Rosenberg fails to show exactly whatthat way is in the second half of the paper. (Though again, he does try to provide a way.)
The first argument claims that “neuroscience makes eliminativism about propositional attitudes unavoidable.” A propositional attitude is a relation between a thinker and a certain proposition or content. When we say that Fred believes that it is raining, we are attributing to him the attitude of believing the proposition that it is raining; when we say that Ethel hopes that it is sunny, we are attributing to her the attitude of hoping that the proposition that it is sunny is true; and so forth. Neuroscience, Rosenberg tells us, shows that there is no such thing as believing, hoping, fearing, desiring, or the like.
Now in fact, it takes very little thought to see that “neuroscience” shows no such thing. For there is nothing in the neuroscientific evidence cited by Rosenberg that couldn’t be accepted by an Aristotelian, a Cartesian, a Wittgensteinian, a Whiteheadian, or an adherent of some other metaphysics. What Rosenberg should say is: “Neuroscience, when conjoined with the specific version of naturalism taken for granted by many (though by no means all) contemporary academic philosophers of an analytic bent, makes eliminativism about propositional attitudes unavoidable.” That claim is plausible, if, for obvious reasons, not quite as earth-shattering as Rosenberg’s way of putting it was.
The reason it is plausible is as follows. Suppose you assume -- as Aristotelians, Whiteheadians, Russellians, panpsychists, et al. would not, but most contemporary philosophical naturalists within analytic philosophy do -- that there is nothing more to matter than what natural science attributes to it. And suppose you assume also -- as Aristotelians, Cartesians, et al. would not but most contemporary philosophical naturalists within analytic philosophy do -- that if there is such a thing as thought then it must be entirely embodied in some sort of corporeal process. Then you will naturally suppose that thinking, if there is such a thing, must involve a corporeal process having no properties over and above those described by physics, chemistry, neuroscience, and the like. And if there is no plausible candidate for such a process, then you will have reason to conclude that there is no such thing as thinking.
Now perhaps the only plausible candidate for such a process -- not that it actually is plausible full stop (it is not) but plausible relative to the assumptions in question -- is something like the “Language of Thought” hypothesis (LOTH). This is the view that a thought is a sentence-like symbolic structure in the brain, and that thinking -- the process of going from one thought to another -- involves the transition from one such symbolic structure to another in accordance with the rules of an algorithm. The postulated symbolic structures are sentence-like insofar as they have syntax and semantics, just like the sentences of the languages we are familiar with. Only they are not sentences of English, German, or any other natural language, but rather of “Mentalese” -- a hypothetical language below the level of consciousness. The attraction of the idea is that it seems to give us something that has the characteristic features of thought (semantics and syntax) and yet is purely material (as in one sense a written or spoken sentence is).
An obvious problem with this theory is that nothing counts as a sentence apart from language users who form the convention of using it as a sentence. Sentences and the like are not natural kinds but artifacts. The hypothesis that there are naturally occurring sentences in the brain thus makes about as much sense as the hypothesis that there are naturally occurring sewing needles in the bones, naturally occurring purses in the skin, or naturally occurring money in the teeth. You could use teeth as money, skin as a purse, or a piece of bone as a sewing needle, but until you decide to do so the body parts in question don’t count as any of these things. Similarly, someone could decide to count some brain process as a sentence, but until someone does so it is not a sentence.
Putting that aside, though, Rosenberg says there is another problem with the idea that there are sentence-like symbols in the brain. Consider the well-known philosophical distinction between “knowledge how” and “knowledge that” -- that is to say, between the having of certain dispositions andabilities, and the grasping of certain propositions. Rosenberg holds -- correctly in my view -- that these are mutually irreducible. Having dispositions and abilities cannot be analyzed in terms of having propositional knowledge, and having propositional knowledge cannot be analyzed in terms of having dispositions and abilities.
But now consider relatively simple organisms like sea slugs, worms, and fruit flies. No one would attribute propositional knowledge to them, and unsurprisingly, their nervous systems exhibit only the sort of stimulus/response wiring needed for the sorts of dispositions and abilities we know them to have. Now our nervous systems are vastly more complicated than theirs, but the evidence seems to show that the difference from what is going on in sea slugs and the like is one of degree rather than kind. But if what is going on in them is just a rudimentary sort of “knowledge how” rather than “knowledge that,” then a (far more complex) “knowledge how” rather than “knowledge that” is all that can be going on in us. Neuroscience, Rosenberg concludes, simply doesn’t leave any room for propositional knowledge, and in particular no room for anything like the LOTH and its sentence-like structures in the brain.
Again, though, it isn’t really neuroscienceper se that rules out propositional knowledge, but rather neuroscience as seen through the lens of Rosenberg’s philosophical assumptions. Rosenberg’s conclusion is “So much the worse for the propositional attitudes.” But of course, it is perfectly open to anyone to conclude instead “So much the worse for the specific form of naturalism Rosenberg and his circle of friends in current academic philosophy are committed to.” And since the eliminativism that results is (as we shall see) incoherent, that is the conclusion we should draw. (Rosenberg has a bit of the Hegel complex -- supposing he has brought philosophy itself to a climax when in fact the most he has done is bring a certain culturally parochial, merely decades-old style of philosophy to a climax.)
It is obvious enough how the evidence Rosenberg cites might be interpreted by someone coming at the issue from a different philosophical perspective. A Cartesian, for example, would say that we shouldn’t be looking for propositional attitudes in the brain in the first place, but rather in the res cogitans. The neurological evidence, he might even say, is exactly the sort of thing we should expect given that (as Cartesians themselves have of course claimed for centuries -- no scrambling to deal with novel evidence here) res extensa is of its nature entirely devoid of thought.
For very different reasons, an Aristotelian or a Wittgensteinian would say that looking for beliefs and desires in neural structures is as silly as looking for foliage in chloroplasts or surface tension in individual H2O molecules -- and sillier still is denying that foliage and surface tension exist when one doesn’t find them there. It is only the human being as a whole who can properly be said to have propositional attitudes. Nor is having them like having a coin in one’s pocket, a crook in one’s spine, or a limp in one’s gait. That is to say, it isn’t a matter of possessing a kind of object, or a bodily attribute, or a behavioral tendency. Nor does the point have anything to do with “emergence,” if that is understood as the idea that “lower-level” features are as metaphysically fundamental as the reductionist supposes, but the “higher-level” ones “emerge” from them in a way that is either metaphysically exotic or simply too complex for us practically ever to know all the details. “Lower-level” features aren’t more fundamental in the first place. A brain and nervous system are if anything less fundamental than the organism of which they are a part, since they are what they are only relative to the whole.
From the point of view of Aristotelians, Wittgensteinians, and other radically anti-reductionist and metaphysically pluralist philosophers, Rosenberg’s position is a classic example of Procrustean dogmatism -- of forcing the richness of the real world into one’s simplistic ontology, rather than making one’s ontology fit the richness of the real world. To see how this works in the case at hand, consider the following analogy. Suppose that among the innumerable pebbles, bits of driftwood and seaweed, etc. on a vast beach there were here and there a number of such objects that by chance had appearances roughly like the following: A, B, C, D … No collection of these objects that arose through chance would amount to anything more than a meaningless string of shapes, not even if the collection happened to look like this: FOUR SCORE AND SEVEN YEARS AGO. Now of course, the beginning of the Gettysburg Address looks like that. But it is a sentence fragment rather than a random collection of shapes. You will never find the difference between them, though, if you look only at the two sets of shapes qua shapes. The shapes are only part of the story, and not the most important part.
Similarly, the neural similarities between sea slug and human being are only a part of the story, and not the most important part. Trying to understand human beings in terms of what they have in common neurologically with sea slugs and other lower animals is like trying to do linguistics exclusively in terms of what the Gettysburg Address has in common with the bits of matter one might find washed up on the beach. It is, the Aristotelian would say, to focus exclusively on material and efficient causes and to ignore formal and final causes, thereby simply ignoring rather than explaining the totality of the evidence. As with the difference between the shapes on the beach and the Gettysburg address, the difference between sea slug and man is not merely a matter of quantitative differences between aggregates of homogeneous elements.
Of course, Rosenberg would deny that there are formal and final causes. He would deny that Aristotelianism, Wittgensteinianism, Cartesianism, etc. constitute viable alternatives to naturalism. He would say that there is, ultimately, no deep metaphysical difference in kind between the Gettysburg Address and the random arrangement of shapes on the beach -- the former, he would say, is just what you get when meaningless shapes are arranged by highly complex but equally meaningless neural processes in a highly complex but equally meaningless cultural and historical context, rather than by relatively simple meaningless processes like tidal action. The point, though, is that his interpretation of the neuroscientific evidence is at best one interpretation among others, and he has given us no non-question-begging reason for preferring it to the others. You certainly won’t find such a reason in the paper, or in The Atheist’s Guide to Reality.
Rosenberg insists in his defense that natura non facit saltum. What counts as a saltus or jump is itself a metaphysically complicated question, but I would certainly agree with the more general point that you can’t get an effect that isn’t somehow prefigured in its total cause -- that is, after all, a hoary Scholastic principle. What that shows in the present context, though, is that since there is intentionality in us, there must be something in our total cause capable of getting it into us. If Rosenberg’s dogmatically anti-teleological metaphysics can’t handle that, so much the worse for it. Nor do you have to be a theist or an old-fashioned Scholastic Aristotelian like me to draw that conclusion. Just ask Thomas Nagel.
Teleosemantics and eliminativism
Rosenberg’s other argument for eliminativism about intentionality is this. All naturalistic theories of intentionality, including the Darwinian teleosemantic approach he thinks is the most plausible, founder on indeterminacy problems. First, for reasons of the sort familiar from Quine’s indeterminacy of translation argument, Fodor’s disjunction problem, and the like, no naturalistic theory can account for how a thought or utterance can have thisdeterminate content rather than thatone -- to use Quine’s example, for how it can be a thought about (say) rabbits rather than undetached rabbit parts or temporal stages of a rabbit. Second, he also notes that there is what he calls a “proximal/distal indeterminacy problem” insofar as there are indefinitely many links in a causal chain leading to any neural structure in which the naturalist would want to locate a thought, and there is no principled reason to think that the structure represents this link in the chain rather than that one. (Though Rosenberg doesn’t note the connection, this is a problem to which, as I have often noted, Karl Popper and Hilary Putnam have in different ways called attention.)
I have developed and defended both sorts of indeterminacy argument many times -- in the first case most systematically in my American Catholic Philosophical Quarterly article “Kripke, Ross, and the Immaterial Aspects of Thought,” and in the latter case most systematically in my Advances in Austrian Economics article “Hayek, Popper, and the Causal Theory of the Mind.” Interested readers are referred to those. (I have also addressed these issues here at the blog -- see the posts on intentionality, Kripke, Dretske, Popper, Putnam, etc. linked to here.)
Suffice it for present purposes to say that I think Rosenberg is absolutely right about this much: There is no way to reconcile naturalism with our having determinate thought contents. Where we differ is over the lesson to be drawn from this. Since he is a committed naturalist, Rosenberg concludes that we simply do not have any determinate thought contents. Since I maintain that it is impossible in principle coherently to deny that we have determinate thought contents, I conclude that naturalism is false.
Why is it impossible to do so? And how does Rosenberg try to get around this incoherence problem? I’ll turn to those questions in the next post.