Blogging Alex Rosenberg’s An Atheist’s Guide To Reality: A Kind of Conclusion

0. Underlying Rosenberg’s project is the claim, made but not defended, that science revels that God does not exist. Obviously, no well-confirmed theory of empirical science implies anything about God and thus science implies nothing about God’s alleged non-existence. Rosenberg assumes that if there were good reasons to believe that God exists, they would reside in theistic explanations of things that in fact science explains better. This is one of many false beliefs he shares with religious fundamentalists. However, Rosenberg can hardly be blamed for taking so many Christian theists at their word when they insist that what contemporary science tells us of the world and our place in it is antecedently improbable from the perspective of the Christian faith. They are wrong about this: the naturalistic implications of science are antecedently probable on Christian theology. But it’s not Rosenberg’s job to figure that out.

1. On Rosenberg’s account introspection reveals little or nothing of explanatory relevance about the human mind/brain because conscious states (a) are (something like) epiphenomenal, (b) no more than post hoc rationalizations of the unconscious goings on in the brain that actually cause human behavior, and (c) lack intentionality so they aren’t about anything anyway. Indeed, unconscious mental states, those that do cause behavior, also lack intentionality. There are experimental results (famously Libet’s) that support (a) and (b), though there’s plenty of distance from this evidence to these sweeping conclusions. The status of deflationary claims about intentionality, or to change the idiom, about the mind/brain having no semantic properties, is a central issue in the philosophy of the last century. I take it that all physicalists take it for granted that the facts about meaning are ontologically superficial, that in some sense the brain is a syntactic device that simulates a physically impossible semantic device. Where the mainstream sees this as setting the task of locating meaning in the world science knows, Rosenberg sees it as reason to elimiate serious talk of meaning. This is a manifestation of a general strategy, see #4, infra.

It would be nice to see at least a gesture toward the obvious issues of self-reference that arise from the claim that none of our thoughts are about anything. How do we manage to have scientific knowledge when our scientific beliefs are not about the world that makes science true?

2. Physics, Rosenberg reasonably asserts, fixes all the facts. (We theists who agree of course acknowledge an implicit scope restriction: physics fixes all the facts about the created world.) However, he appears to slide effortlessly from this to the stronger claim that all the facts are physical facts, or at least that all the facts are physical facts or facts reducible to physical facts. Most humanly important “facts,” e.g., those of folk-psychology or morality, thus turn out not to be facts at all. Here, as in other places, Rosenberg’s claims seem finely balanced between interpretations on which they are true but uncontroversial and interpretations on which, if not false, they are at least very controversial.

3. On the epistemological side there’s a similar move from the true, “Science is our best way of knowing,” to the highly problematic, “Science is our only way of knowing.” Such inferential leaps are all too familiar, but surely they are not mere mistakes on Rosenberg’s part. His texts in the philosophy of science are admirably clear and carefully reasoned introductions to many of the subjects he addresses in An Atheist’s Guide. The main disappointment of this book is that it seems to have been written by someone else! Writing, or trying to write, for a popular audience, one cuts all kinds of philosophical corners, and we see this going on left and right in An Atheist’s Guide. Yet it often seems that the wrong corners have been cut and that the effect on the philosophically innocent reader might well be the opposite of what Rosenberg intends: to instill the idea that the implications of taking science seriously are simply crazy. (I’m tempted to blame some editor’s notion of what’s appropriate for a book buying public.)

4. Along the way I’ve complained about Rosenberg’s strategy of reasoning from the fact that nothing in the reality known to science precisely answers to a familiar human concept, e.g., meaning, freedom, purpose, to the conclusion that that concept applies to nothing and should be eliminated from serious thought about the world. Why in at least some of these cases isn’t it more reasonable to regard our concepts as being corrected and revised by science and thus accorded their secure place in the world? We need not suppose that our pre-scientific concepts are infinitely malleable, but it’s hard to accept that they cannot sometimes be retained in versions improved by the encounter with science. And this is especially hard to accept in instances where the concepts at issue are adaptations that enable us to navigate the natural or social environment. But, of course, if all the facts are facts of (and not just fixed by) physics, there’s no place for such things in reality.

Here it’s worth pointing out Rosenberg’s reliance on folk-philosophical conceptions as though they are in perfectly good order. For example, he regularly tells us that since science shows that human behavior is causally determined, we are not free. But, of course, this is valid only with the addition of the premise that no causally determined action can be free. That premise remains unreflectively embraced by many, including many of the scientifically well-informed, and it might be true. The centuries of philosophical attack on it may be misguided. But the mere fact that this seems simply obvious to so many is no more a good reason to accept it than, say, the still-popular belief that heavy objects fall faster than light objects…or that our choices cause our actions.

5. An abiding question is about the supposed connection of all this—Rosenberg’s radically eliminative materialism—to atheism. Does he think that run of the mill naturalistic philosophers, engaged in the project of trying to explain how mind, rationality, meaning, purpose, freedom, and so on might fit into the physical universe best known by way of the sciences are some sort of crypto-theists, clinging to concepts that make sense only if there is a God? It would be interesting to see a case for that. (Perhaps Rosenberg and folks like Al Plantinga are on the same page: if no God then no meaning, no rationality, etc.) Insofar as anything like this appears in An Atheist’s Guide to Reality, it seems to be no more than one more fast inference, this one from the fact that there is no purpose in nature to the further claim that there is no purpose for nature. (For Rosenberg the claim that there is no purpose in nature encompasses not just the obvious fact that there is no final causation in nature, that evolution’s trajectory is blind, and that the etiology of human action is entirely a matter of efficient causation, but the not at all obvous claim that human beings do not really do things for reasons, i.e., that beliefs and desires are not the causes of behavior.)

Rosenberg’s Atheist’s Guide to Reality, Chapter 11

Rosenberg tells us that there’s little hope of ascertaining the meaning of any individual human action because the particular motives expressed in thoughts about ends and means cannot be known.  If I understand the claim, it’s that our intentional folk-psychology has little value for limning the causes of human action.  Even if we knew the reasons for which someone thinks he did something, these reasons are at best “rough indicators” of the real, non-intentional causal structures in the brain that underlie behavior.  “Biography” (folk-psychology) is a blunt explanatory instrument at best and mostly just storytelling.

It’s hard to know what to say here.  On physicalist assumptions, the proximate causes of human action are neural events, so any explanation of what we do that invokes beliefs, desires, and other intentional mental states succeeds only if they connect to the neural reality in some reliable way.  But there’s plenty of room to maneuver here.  At one extreme, the folk-psychological states ascribed to human beings are type-identical to neural states, but if this classical reduction of the mental to the physical fails (as, presumably, it does) it might be true that any given, particular belief, desire, etc., is identical to some particular neural state.  If this token-identity fails as well, we can proceed to more loosely conceived connections of intentional states to brains. For example, we ascribe mental states holistically to minds construed as engines of massively parallel, non-symbolic, computation.  So someone believes that P just if his brain realizes a particular pattern of weightings distributed across large assemblies of neurons; thus (as in the Churchlands) intentional states roughly correspond to points in multidimensional representational spaces.  Or (following Dennett) we say that the behavior of very complex organisms (and other devices) exhibit real patterns that we usefully and economically detect when we adopt the intentional stance.  And so on.  It’s not obvious where in all this we are entitled to say that the intentional idiom affords a rough approximation of the causes of behavior, and when we’re entitled to say only that it provides at most a blunt explanatory instrument.  Rosenberg denies the former and accepts the latter, but the distinction is not clear.

One thing that seems obvious and, in light of the continuing preeminence of computationalist theories of the mind, important, is that we can couch powerful explanations in intentional terms, at least when the “minds” in question are artificial. We regularly and reliably explain how a device got from one computational state to another by reference to their representational content.  The machine entered into the state that represents Q as an effect of being in the state that represents If P then Q, and the state the represents P.  The clever wiring makes causation mirror logical inference. It is possible, though not for mere humans, to explain the computer’s behavior in terms of physics alone.  And for most practical purposes, explanations couched in the higher-level vocabulary of electrical engineering are similarly out of human reach. Yet we happily explain computation in intentional terms, relying on the token-identification of representational and physical states.  Extant artificial computation is, of course, much simpler and neater than what transpires in brains, but it does not seem obvious that this precludes intentional explanations of human minds and behavior.

We build machines that realize a secure connection of the normative realm of reasons to the natural, physical, realm of causes.  The intentional, by design, piggybacks on natural causation, ultimately on physics, and intentional explanations are the best explanations we can hope for here.  How likely is it that we, products of natural selection, manage this with our handiwork but natural selection didn’t manage it in putting us together?  Is it plausible that the result of eons of evolution is an organism that has a high degree of inclusive fitness because it does a good job of explaining, predicting, and (sometimes even) controlling the behavior of other organisms—predators, prey, and one’s conspecifics—by means of a folk-psychological theory of mind, if all this affords it is a blunt explanatory instrument? How blunt is blunt?  Our ancestors caught food, avoided being food, and seduced mates so as to become our ancestors because, along with some luck, they were endowed with an instrument that worked.

We may agree with the Churchlands and praise folk-psychology as a marvelous instrument that reached maturity millennia ago and is currently largely moribund, being refined on the margins by our artists but long since having come up against the gross limits of what it can explain.  Whether its elimination in favor of something better, something that can be smoothly integrated into the natural sciences, is a practical possibility remains to be seen. But—and this seems the crucial point against Rosenberg—the demise of propositional attitude psychology is not the demise of the intentional. The Churchlands’ eliminationist program is, as they call it, neurosemantics, aspiring to explain human behavior as the effect of neural states that represent things.

The focus of Rosenberg’s denigration of folk-psychology might lie elsewhere.  If the mental states to which we have conscious access and which we cast as beliefs and desires are something akin to epiphenomena, then explanations that invoke them might simply be false, the intentional states playing no real role in the causation of behavior.  This is not an attack on intentionality per se.  The true explanations might, as in Freud, be intentional yet accessible not at all or only by way of therapy.  Rosenberg (for reasons that do not seem adequate) contends that brain states that cause behavior could not possibly mean anything, so folk-psychological “explanations” wind up explaining nothing.  But even if they could, it might be that they lie beyond our introspective powers, powers whose feebleness Rosenberg highlights earlier in the book.

The other possible ground to dismiss folk-psychological explanations is a view of causation, roughly the idea that true claims about cause and effect must be formulated either in terms of physics, or in terms reducible to it. When we assert that Marvin went to the store because he wanted some beer and he believed that in the circumstances the best way to obtain some was by going to the store, these reasons can be the causes of his going to the store only if they are identical—token wise or maybe even type wise—to the neural states that actually cause the bodily movement.  I suspect that we’re entitled to be more generous with ourselves when it comes to describing causation, and that reasons can be identified as causes without being so closely tied to the underlying physical reality. (Here I ignore what seems to me the fact that in any event the neural stuff cannot be reduced to the physical stuff. The neuroscientific and evolutionary baby are at risk of going down the drain with the folk-psychological bathwater.) To say that one thing causes another is, as is well known, in part a practical matter, dependent on human interests, and thus not relegated to the God’s-eye point of view of physics.  Physics fixes the world’s facts, including the facts about causation, but this does not imply that all the facts about causation are facts of physics.

But if the only truths about causation are those couched in the vocabulary of physics and whatever special sciences are reducible to it, then Rosenberg properly dichotomizes scientific explanation and mere storytelling.

It is striking how seemingly unselfconsciously Rosenberg proffers explanations of the kind he tells us are useless. For example, in the process of informing us that all history, being no more than folk-psychology on a grand scale, provides are entertaining stories, not real explanations, he offers the example of Henry Kissinger, who became a power broker by convincing people that knowing about the 1815 Congress of Vienna was going to help Nixon deal with the Soviets.  Whether what Kissinger convinced people of was true or false, here we have a folk-psychological explanation of how he came to pre-eminence.  Later in the chapter, Rosenberg offers a perfectly reasonable, yet patently folk-psychological, explanation of the Chinese practice of binding female feet.  It’s all about what people wanted and what they believed about how to get it. (I’m enough of a humanist to suspect that this explanation, though correct, is missing something, viz., the rich symbolism that was part and parcel of the cultural practice.) And it’s worth noting that this folk-psychological explanation is at the same time a satisfying selectionist explanation.

Rosenberg’s discussion of innovation, discovery, creativity, etc., is an instance of a familiar pattern. We do not explain it by learning what it really is, i.e., that it’s the product of random variation and selection in a brain that operates on stochastic principles.  To learn this is to find that there is no such thing.  It’s not the mysterious, magical thing some might be inclined to imagine it is, so, since they are right about what it is, they are wrong about there being such a thing.  More scorched earth eliminativism.

Rosenberg’s claims about history are interesting.  There’s a familiar account on which explanation and prediction are essentially the same thing; explanation is “postdiction” in contrast to prediction.  If this is true and we see that history cannot manage much in the way of prediction, maybe we should doubt its ability to explain anything.  However, it seems plausible that a system, even one governed by deterministic laws, that is chaotic, as we might reasonably take the course of history to be, would be explainable after the fact but not reliably predictable.  I assume, for example, that had the world’s leading historians known everything going on on 21 June 1914, they could not have predicted that a world war would be precipitated one week later.  That Gavrito Princip and his pals would, in fact, carry off the assassination in Sarajevo and that this event would, in fact, cascade into world war was something no human being could have reasonably predicted.  God, if his world really were governed by deterministic laws, would have known with certainty what was coming, but no human being could have reliably predicted the events that started the Great War.  Nonetheless, retrospectively we know what started the warPrincip’s assassination of the Archdukeand this is a true, even if quite obviously very incomplete, explanation.  The fine-grained knowledge of initial conditions needed for precise predictions isn’t needed for explanations.  Any number of things going on in the summer of 1914 can be ignored as unknown and now unknowable, yet we still can explain what happened.  There’s a big difference between knowing that the butterfly will flap its wings and thus predicting the hurricane, and knowing that it did, since we know that the hurricane ensued.

The unsurprising fact that historians seem fixated on the past and offer no predictions of the future is due to there being no laws of history.  As Rosenberg tells us, history is large scale folk-psychology, and account of why human beings did what they did that invokes their beliefs and desires on the supposition that they are more or less rational.  This is not a matter of knowing the (non-existent) laws that govern human behavior.  It’s the result of having been supplied by natural selection with a mental module dedicated to making inferences about other minds, inferences we could not typically justify on the available evidence. That is to say, the folk-theory of mind is not a theory in the scientific sense; it involves no subsumption of events under genuine laws.  I imagine that our evolved but quite specialized theory of mind isn’t particularly well-suited to explaining the behavior of remote and unobserved strangers. We can know, say, that Princip fired his gun at Ferdinand because he wanted to kill him and that he believed that in the circumstances that was the best way to bring about his death.  Beyond this truism, I suppose that things get murky fairly quickly, and that a good explanation of why he wanted to assassinate the Archduke and believed that it was a good idea to act on that desire might not be easy to come by, even on the debatable assumption of more or less intact rationality.  And to get from the shooting to the outbreak of hostilities one month later involves all manner of folk-psychological explanations, some trivial and others hard to come by.  This is all hard enough after the fact and impossible beforehand.

Rosenberg’s Atheist’s Guide to Reality Chapter 10

Chapter 10

If, as a materialist might reasonably believe, a brain is a mind in virtue of the structure of its neuronal interconnections, and a mind is the mind of a particular person for the same reason, then one need not suppose the Freaky Friday scenario presupposes that the mind is an immaterial substance. We leave the two bodies and their brains in place, but we re-organize the neural interconnections in the brains, so each realizes the computational state the other initially realized. It’s possible that the truth is more complex than a straightforward computationalism, and that whether a brain is the brain of a particular person is not settled at so abstract a level, but depends also on features that some other brain might be incapable if instantiating. In that case, persons are tied to particular brains in ways that makes the body switching scenario impossible. And of course one might contend-implausibly, I think-that sameness of mind does not suffice for personal identity and what we have here is not a body swap, but two persons becoming very confused and losing their memories.

The fact that a human person, a self, persists through all the changes in our body over the course of our lives hardly implies that it is an immaterial thing. If it did, the implication is either that the car I have today is not the one I bought in 1988 or that it has an immaterial component. But no: it’s the same car even though quite a few parts have come and gone. This does not mean that a person survives all possible changes to a body over time.  Whether a person at one time is the same person at another time depends on what kinds of changes the body has undergone: a person survives some changes and not others.

Alex Rosenberg’s An Atheist’s Guide to Reality Chapter 9

There is no question but that particular goings on in the frog brain play an indispensable role in explaining how frogs catch flies. If there were something in there that had what is for Rosenberg the magical property of being about flies, it would be of explanatory interest only if it played the causal role that the neural state in question plays, i.e., if it typically made its appearance only when a frog is in the vicinity and typically caused frog-catching behavior. Granting that nothing in the frog has the mysterious aboutness Rosenberg rejects, the temptation to point to the neural state that plays this causal role and say this is what aboutness really is is hard to resist. The Rosenbergian claim that it cannot be about anything falls flat absent some account of what it lacks that’s essential to really being about something.

Is there much of anything to the folk-concept of aboutness for science to subvert?  In my experience, if someone is asked, “What makes the frog’s idea of the fly an idea, and what makes it an idea of a fly?” the notion that it has to with some sort of resemblance of the mental item to what it represents falls apart quickly, leaving not much of anything that we are entitled to say no mere brain stuff can satisfy.

It’s interesting to ask what could be going on when someone mistakenly believes that something is about something.  Someone sees a combination of letters and in error takes it to be a word in a foreign language when in fact it’s just a meaningless combination of letters. Is it conceivable that we are systematically wrong in such judgments, or are errors of this kind parasitic on a background of correct judgments? Or one could encounter a description that fails to refer, e.g., when the student refers to Martin Luther, the German king who was an activist for civil rights for black Protestants in the 1960’s.

The crucial fact lurking here is that nothing in nature fixes reference. There is, perhaps, no objective matter of fact as to what the frog’s brain represents when it catches flies. The frog can be fooled, capturing BB’s as though they were flies, so do the goings on in its brain mean small black thing rather than fly? In some quarters much has been made of the indeterminateness of translation and the inscrutability of reference. This tells us something important about the place of meaning in the natural order, but the conclusion that what it tells us is that there is no place for it, i.e., that nothing is about anything, is hard to believe, or even grasp. I’m no fan of facile objections that charge a view with self-referential incoherence, but it’s hard not to want to ask Rosenberg if the words in his book are about anything, whether reading it doesn’t cause us to think about frogs, and so on. Physics fixes all the facts, and facts about meaning are not facts of physics. But why say this implies there are no facts about meaning? As a scientific naturalist, I want to say that we need to identify neural states as representing things in order to explain the behavior of organisms. (Not to say that there must be a local mapping of the intentional onto the neural, an identification of something in the brain as the representation of something; the brain might be in the state of representing something in virtue of rather global features, on analogy with non-symbolic, parallel computation.) The fact that there is an alternate, lower level, description, couched solely in terms of physics, does not imply that the intentional description is scientifically dispensable.

Sometimes The Atheist’s Guide reminds me of early 20th-century popularizers of science, informing us that there are no solid objects, since what we take to be solid objects are mostly empty space containing atomic nuclei surrounded by mostly empty space.  But such a claim strikes us (or at least me) as anachronistic. Of course solid objects exist, and it’s often important to distinguish them from objects that are not so solid, or not objects at all but volumes of liquid or gas. Physics reveals not that there are no solid objects but what solid objects are really like. As Wilfrid Sellars famously said, the task of philosophy is to figure out how the manifest image, i.e., the world as humans experience it fits together with the scientific image.  Sometimes this means dismissing something as illusory, a mere artifact of how we represent things, but sometimes it means revising concepts to fit into the world where physics fixes the facts.  Recall the issue that led Rosenberg to moral nihilism: our natural concept of a moral fact is of a state of affairs that provides those who detect it with a reason to act. If, as physicalists and naturalists tend to think, there can be no such practical reasons, because all reasons to act are rooted in the agent, not in objective reality, then either moral facts need not have this feature, or there are no moral facts and no moral truth.

Or consider our concept of time.  Late in his career Gödel showed that there are physically possible universes in which solutions to the equations of general relativity imply that there are closed, finite time-like loops, which is to say that in those worlds time travel is possible. He saw this result not as implying that time-travel is possible, but that general relativity reveals that what we think of as time is not captured by the fourth, time-like, spatial dimension, and thus that what we think of as time has no objective reality. Talk of time as a fourth spatial dimension is not just a way to interpret general relativity, but the sober truth: there is no such thing as time. Did Gödel draw the right conclusion? Or is the implication that time is real yet quite different than we imagined?

Rosenberg stresses the implausibility of mapping propositional attitudes onto the brain’s physical structure, and the correlative implausibility of its computation being  linguiform, or symbolic at all. On the most likely account the brain’s computation is at bottom parallel and non-symbolic, not serial and symbolic. We still need to explain human linguistic competence, the fact, e.g., that anyone who makes sense of, “The dog bit the boy” can makes sense of, “The boy bit the dog,” and it’s at least not obvious how to do so unless some serial computation goes on in one’s head. (In principle a virtual serial machine can run on a parallel machine and a virtual parallel machine can run on a serial machine.) We may have good reasons to reject the idea that in general the computation that makes a brain a mind involves propositional representations of the world, and to reject the idea that mental representations can be identified, even token-wise, with discrete brain events. But there is no reason to think the scientific explanation of human beings can get by without identifying some state of the brain as a representation, as being about goings on in the world outside the brain, even if that information bearing state consists of an activation pattern realized in some ensemble consisting of some very large number of neurons. (The Churchlands make this sort of thing sound very plausible.)

A general question that arises here is how what looks a lot like some version of eliminativism coheres with Rosenberg’s scientism. It’s hard to see how these claims cohere:

(i) Physics fixes all the facts

(ii) Physics fixes all the facts about human beings

(iii) Insofar as human beings as such are subject to scientific explanation it is by means of subsuming them under biological concepts

(iv) Biological concepts cannot be reduced to those pf physics

(v) What cannot be reduced to physics is not real; it’s at best an illusion provided by natural selection

(vi) Science alone provides the sober truth about human beings.

The only way I can make out a coherent view is by rejecting (iv) and claiming that there is an intertheoretic reduction of biology to physics.  One the one hand, it might be that natural selection can be understood as a law of physics, though it seems to me that this is arbitrary, in contrast to regarding it not as a law of nature but as an abstract principle true of anything, irrespective of its physical constitution, of which a few things are true, e.g., it makes imperfect copies of itself, properties of the copies have a bearing on the odds of their making more copies, etc.  But even if we can include natural selection as part of physics, biological explanation relies on many other concepts and some of these, e.g., gene, seem obviously irreducible to physics in virtue of being functionally defined. Further, if the scientific understanding of minds proceeds largely in computational terms, then while the facts about the mind are fixed by physics, their explanation depends on concepts that cannot be reduced to those of physics. The God Rosenberg finds obviously non-existent could make sense of the world using only the concepts of physics, but probably no finite mind, and certainly no human mind, could do so. We need the special sciences, and probably quite a few other things, even some chastened folk-psychology and history.

Misguiding Atheists

Richard Dawkins, and some others who see Christian faith and science in intractable conflict, take it for granted that the fundamentalist interpretation of the Bible is the correct one, and that when biblical scholars advise us to read Scripture in ways that do not make it at odds with science, e.g., they tell us not to think the Bible teaches that there was a literal Adam, Eve, talking snake, magical trees, etc., all on a flat Earth under the turning metal dome of heaven, this is a dishonest dodge to avoid the plain implication that the Bible teaches these things and is wrong.  It seems to me that Rosenberg adopts an analogous, and no less hard to justify, position on our common sense concepts. Whatever the man in the street, untainted by philosophy, believes about the nature of freedom, consciousness, meaning, purpose, moral truth, and so on, is correct, and any naturalistic philosophers who try to explain how these things can, with some conceptual re-engineering, find their place in the world science describes, engage in a bad-faith effort to avoid the obvious unsettling news. Rosenberg’s account leaves the uninformed reader, whether atheist, agnostic, or theist, who accepts the idea that physics fixes all the facts about this world ,with the impression that if any such project exists, it is a marginal and futile effort, not the mainstream of contemporary philosophy.

Rosenberg’s Atheist’s Guide to Reality V Chapters 7 & 8

There doesn’t seem to be much to say about Chapter 7.  Taking heed of scientific knowledge of the human mind/brain, including what it reveals about the fallibility of introspection, does significantly reshape our image of what we are. (I find the degree to which our minds are modularized and its modules encapsulated the more radical discovery.)  But Rosenberg’s quick move from the fact that introspection is not infallible to the conclusion that it is not reliable is a bit much.  Even when an apparent way of knowing fails in some dramatic way, as in the case of blindsight, this does not justify the inference to unreliability. “If it can fail in this way then it must be overall unreliable!”  Reliable ways of knowing may come up against unexpected limits. By analogy: it was extremely surprising news when Russell in 1902 wrote to tell Frege that the set theory in which he sought to ground mathematics is inconsistent, or 30 years later when Gödel proved that any formalization of arithmetic is incomplete.  That whatever cognitive mechanism underlies mathematical intuition is not infallible is a conclusion in some ways as surprising as the discovery that introspection can fail in spectacular ways.  But no one draws from this the conclusion that it is not a reliable means to obtain knowledge. Nor, when it comes to it, does the fact that we are subject to visual illusions–ones that persist when we know they are illusions, a manifestation of perceptual encapsulation–imply that vision is not a reliable way to know.

Chapter 8 is a different story:

Of course: no physical thing inherently means (represents, is about, etc.) anything, i.e., it’s not about anything in virtue of its intrinsic physical features. From this truism Rosenberg infers that no physical thing is about anything. He is entitled to this inference only if it is also true that if a physical thing is about something, then it is in virtue of its intrinsic physical features. A physical thing also has non-intrinsic properties, its various relations to other things.  Among these are causal relations.  It exists as an effect of other things, and other things exist as its effects. In a particular setting it plays a characteristic causal role: it has typical causes and effects. These causal relations are, of course, manifestations of the laws of physics recruited by natural selection in the construction of brains that navigate organisms through their natural and social environments. There is, say, a neural state typically caused by there being a cat in the vicinity, and this state in turn has, in conjunction with other mental states, typical effects. There is a mental state in Karen caused by the cat being on the table, via her visual system, and this state, together with Karen’s desire for the cat not to be on the table, causes the behavior that consists in Karen evicting the cat from its illicit location. Whatever plays this causal role in the economy of the nervous system, irrespective of its physical constitution, is a representation of the cat. This is how mere physical things can be about things.

What seems very tendentious is the claim that this is not an account of how neuronal things can be about things, but the revelation that they are not about anything. A familiar pattern: start with a folk-concept of something, show that nothing in the world where physics fixes all the facts perfectly answers to it, and announce that nothing answers to it. The alternative, that we have discovered that the concept needs revision, remains out of sight. The reiterated assertion that nothing in the brain can be about anything invites us to wonder what is being denied. The causal role theorist claims that chunks of grey stuff are about something, but that this is not in virtue of their intrinsic properties, Rosenberg says, “That’s not genuine aboutness!” But why isn’t it? What is it that those clumps of grey matter lack that precludes them being about anything? We’re entitled to ask Rosenberg to complete the sentence, “A chunk of grey stuff would be about something if_______.” Is there no way to finish the sentence because what’s required is some non-physical property? I suppose that he would regard it as equally obvious that if there were a non-physical thing, it could not be about anything either. So is it that there’s no way to complete the sentence because the concept of aboutness is incoherent; nothing can be about anything just as nothing can be a square circle? But then there’s a burden to reveal the contradiction allegedly implicit in, “That clump of neurons is about Paris.” There’s no sign of this burden being discharged.

If there’s an argument lurking here, I imagine that it’s something like:

Aboutness must be either intrinsic or derived.

Nothing possesses intrinsic aboutness.

Something possesses derived aboutness only if something else possesses intrinsic aboutness.

Therefore: Nothing possesses aboutness.

There is, I suppose, some intuitive appeal to the third premise. But is it really true? The idea is something like: the word “cat” is about cats, but not in virtue of any intrinsic characteristic; its aboutness derives from human beings, who establish the convention that makes it about what the idea cat in their minds is about, and that is about cats not derivatively but intrinsically. But then we’re left with the truistic: nothing is about something by convention unless something else is about something not by convention. Clumps of grey matter in the brain, like the word “cat,” lack intrinsic aboutness, but this does not imply that any aboutness they could have must be conventional. It might simply be natural, having its antecedents in neural structures like those of the sea slug, in which there is no good candidate for being a mental representation. Between the sea slug and humans there need be no sharp phylogenic line where aboutness appears. Eventually, though, we can economically explain a creature’s behavior only by describing some of its neural structures as representations. That which is about something arises gradually from what is not about anything; in this sense, its aboutness is derived, but not from something that has intrinsic aboutness.

This said, it might be that the upshot of all this is that we’ve found that the concept aboutness (intentionality, representation, meaning, etc.) is useless; on analogy, we’ve come around to the view that the concept witch has no application in the real world. But this strikes me as implausible. Representation, e.g., seems to me precisely how we want to describe that mental state in the mind of Karen caused by the cat and causing her cat controlling behavior.

Blogging Rosenberg’s Atheist’s Guide to Reality IV (Ch 6)

1. Whatever experiments such as Libet’s imply about free will or its lack is indirect. At face value, what they support is some kind of epiphenomenalism, i.e., that conscious states, at least the introspected experience of making a choice, don’t have the causal powers we are apt to ascribe to them. Thoroughgoing epiphenomenalism is, I assume, incoherent: if these states of what we call consciousness exist, then they do have effects, at least our reports of them. But it is surprising if, as the evidence suggests, one’s conscious choice to perform an action is not its cause, but also an effect of whatever unconscious brain event causes it. Libet’s results undermine a particular conception of free will, one on which a choice is free only if (a) it is the effect of a conscious act of will, and (b) that conscious act of will is not the effect of some antecedent event. Which is to say it subverts a robust libertarian idea of freedom. (Maybe it is an effect of the person that chooses, in contrast to some event in her, but no one has any idea how we could manifest such “agent causation.” And we really have no idea how a material being could be an agent cause, even if we believe that God is free in this radical sense.) Materialists either accept this libertarian idea of free choice but say that we do not have it, or reject libertarianism in favor of a compatibilist conception of freedom: choices are free so long as they have the right kind of causal history, i.e., they deny condition (a). However, compatibilists have traditionally accepted condition (b), assuming that the proximate cause of a free act is a conscious choice to do it, one which is a free choice in virtue of having causes of the right kind. This is typically spelled out in terms of practical rationality: a choice is free just if it is caused by the chooser’s very own, reflectively examined reasons, i.e., her desires and beliefs. Can compatibilism survive the demotion of conscious acts of choice to a semi-epiphenomenal status?

2. I don’t find this a threat to compatibilism. First, we all know that the standard account on which something or other causes one to have various desires and beliefs, these cause a conscious act of choice, and that in turn causes the bodily motion that constitutes the action, is an idealization. Most of our free and responsible behavior proceeds without conscious deliberation and choice. Libet’s experiments bring into focus what we are already introspectively aware of: Conscious deliberation is relatively rare; mostly, we just do things, without consciously examining reasons. The essential compatibilist claim is that our free choices are caused by our reasons, and compatibilism has no interest in denying the obvious fact that we are only episodically conscious of them. At this moment you have a great many beliefs and desires but almost all of these mental states are dispositional, not occurrent.

3. There is a familiar worry that our conscious awareness of the reasons for which we act need not be entirely accurate, and we act for reasons other than the ones we think we act for. This is plausible enough to have kept psychoanalysis in business for years. Compatibilism is not committed to any given free choice being an effect of conscious deliberation, but it is in general committed to the idea that a high degree of freedom and responsibility does depend on a disposition to conscious deliberation under certain conditions, e.g., when the best course of action is not obvious but getting it right matters. This is, I assume, arises from the connection of self-consciousness to personhood, and of personhood to moral responsibility.

4. It might be that our concept of freedom is incoherent, that it combines the idea of free choice as uncaused and as having causes of the right sort. (There are empirical studies that support this.) If so, we may ask just what science leaves us and whether what’s left is worth having. Is it a sufficiently robust idea of freedom and responsibility to sustain the practices where we put it to use? As I noted earlier, when we discover that something lacks a feature we’ve long regarded as essential to it, we must either continue to regard it as essential and conclude that there is nothing to which the concept applies, or revise it, acknowledging that it was a mistake to take that feature of essential. Science sometimes shows us that things do not exist, e.g., witches, caloric, phlogiston, and sometimes that what exists is radically different than we thought, e.g., atoms, gravity, the mind. Rosenberg, for reasons not fully clear to me, seems strongly inclined to elimination over revision.

5. With many philosophical naturalists, I think a good case can be made for compatibilism, but difficulties lie ahead. One worry, which I suspect figures in Rosenberg’s rejection of it, is due to the fact that at face value it depends on the idea that human behavior generally can be explained as the effect of reasons, of beliefs and desires. Explaining, and predicting, and making sense of what human beings−others and ourselves−do by appeal to these mental states manifests our innate folk psychology, a presumed product of natural selection. Suppose, as may well be true, that folk psychological concepts cannot be smoothly mapped onto the physical, or even information-theoretic, structures of the human brain. These mental things can be identified with nothing in the domain of neuroscience. In what sense can beliefs, desires, and other intentional states be said really to exist? (“Intentional” here in the sense of having semantic properties, being about something.) When kinds of thing that we want to invoke in causal explanations cannot be reduced to the things physics refers to, the status of those explanations can be problematic. Is there are place for these mental states in the objective, physical universe, and if there is, is it a place that affords them a role in causal explanation? (It’s worth recalling that it’s not reduction, but reduction’s failure, that raises doubts about the reality of things. It’s the irreducibility of mind to matter that challenges the materialist.)

6. The other worry arises only for theists. Even if compatibilism is in general true−there’s no insurmountable conceptual difficulty in human action being at once caused and free−free choice might still be impossible if God is the First Cause. After all, if God created the universe, i.e., he brought about its initial conditions ex nihilo, and authored its causal laws, then his free creative act is the remote cause of every human action. Even if our actions could be free and responsible despite being the effects of natural causes, if what one does is caused by, foreknown by, and intended by someone else, then we are not free and responsible. If you choose and act as you do as result of the microchip a clever neuroscientist surreptitiously implanted in your brain last year, then you are not free and responsible. It will seem to you that you satisfy the compatibilist conditions for freedom, but this will be an illusion. Whatever investment a theist has in free will and moral responsibility as such−maybe not all that much−a Christian theist has a profound investment in human beings being capable of relating to God interpersonally and this requires that vis-à-vis him we be free, that what we do is distinct from what he does, and that we can do other than what he wants us to do. We might be free even if everything we do is causally determined, but not if it is causally determined by God. For this reason, the Christian hopes that nature’s fundamental laws are indeterministic. If they are, then God remains the most remote cause of all our actions, but he does not specifically foreknow or intend them. So long as the creation is indeterministic God and creatures can share responsibility for what creatures do, a view that lies at the heart of the Christian faith, which confesses the crucified God who takes responsibility for his wayward and ruined creatures. Whether nature’s laws are deterministic or indeterministic makes no difference so far as the issue of compatibilism goes, but it matters a great deal when our aim is to integrate a naturalistic account of the human condition into the Christian faith.