Here are some preliminary thoughts about Alex Rosenberg’s Atheist’s Guide to Reality, particularly his claim that we do not think about things (hence the snarky title to this post). Sorry for the inordinate length. Once again, the writing is meant for the general, educated reader rather than the professional philosopher, though, naturally, I want to make sure that it is philosophically sound.
Rosenberg freely embraces the label “scientism,” generally regarded as a pejorative term. His tone is uncompromising and even flippant. Here is how he thinks science requires us to answer the “big questions” about the meaning of life:
Is there a God? No.
What is the nature of reality? What physics says it is.
What is the purpose of the universe? There is none.
What is the meaning of life? Ditto.
Why am I here? Just dumb luck.
Does prayer work? Of course not.
Is there a soul? Is it immortal? Are you kidding?
Is there free will? Not a chance!
What happens when we die? Everything pretty much goes on as before, except us.
What is the difference between right and wrong, good and bad? There is no moral difference between them.
Why should I be moral? Because it makes you feel better than being immoral.
Is abortion, euthanasia, suicide, paying taxes, foreign aid, or anything else you don’t like forbidden permissible or sometimes obligatory? Anything Goes.
What is love and how can I find it? Don’t look for it; it will find you when you need it.
Does history have any meaning or purpose? It’s full of sound and fury but signifies nothing.
Does the human past have any lessons for our future? Fewer and fewer, if it ever had any to begin with (2-3).
Wow. Atheist debaters often face religious apologists who make precisely some of the charges against atheism that Rosenberg enthusiastically affirms. Anyone trying to get a hearing for atheism might think that with “friends” like Rosenberg, we don’t need enemies! In fact, though he rejects theism he equally repudiates secular humanism (277-82). Clearly, Rosenberg has taken on a big task and has a heavy burden of proof. Since we have introduced the topic, let’s consider his arguments against free will.
Rosenberg says that the scientific case against free will is simple and direct:
The mind is the brain, and the brain is a physical system, fantastically complex, but still operating in accordance with all the laws of physics—quantum or otherwise. Every state of my brain is fixed by physical fact. In fact, it is a physical state. Previous states of my brain and the physical input from the world together brought about its current state. They were themselves the result of even earlier brain states and inputs from outside the brain. All these states were determined by the laws of physics and chemistry. These laws operated on previous states of my brain and states of the world going back to before my brain was formed in embryogenesis…When I make choices—trivial or momentous—it’s just another event in my brain locked into this network of processes going back to the beginning of the universe, long before I had the slightest “choice.” Nothing was up to me. Everything—including my choice and my feeling that I can choose freely—was fixed by earlier states of the universe and the laws of physics. End of story (236).
Someone might object that Rosenberg is assuming strict determinism, when we know that indeterminism rules at the quantum level. At the quantum level the old bumper sticker slogan applies (cleaning it up): Stuff happens. Might not quantum indeterminacy make room for free choice? However, Rosenberg notes that if a random, uncaused quantum event in the brain were to initiate a chain of causes resulting in an act of choosing, that act would be no freer than the effect of a deterministic causal chain. The “choice” would still just be something that happened to me, not something I did. Personal agency is a necessary aspect of free choice (237).
But what about introspection?
Introspectively it just feels like you choose; it feels like it is completely up to you whether you raise your hand or stick out your tongue. That feeling is so compelling that for most people it tips the scale against determinism. They just know “from inside” that their will is free (238).
However, for Rosenberg, introspection is not reliable.He cites famous experiments done by Benjamin Libet and replicated often since (152-4). In these experiments, subjects were asked to perform a simple task, like pushing a button whenever they wished. They also recorded when they first consciously decided to push the button. It takes about 200 milliseconds from the time of one’s decision to push the button to the actual flexing of the wrist. Yet 500 milliseconds before the wrist flexed, Libet detected activity in the subjects’ motor cortex that initiated the wrist flexing. In other words, the “choice” came after the process was already initiated by the brain! Apparently, the brain initiates both the “choice” and the movement! Hence, the subjects’ subjective perception of freely choosing was an illusion. We think that “we” make conscious choices but our brain makes them for us—and then makes us think that our conscious choices had something to do with it!
I think that many philosophers, myself included, would reply to Rosenberg as follows: “Quite unaccountably for a professional philosopher you seem to overlook the compatibilist position on free will, namely, that causal determinism is quite compatible with free will in the everyday sense, and, in fact, that this mundane sense encompasses all the freedom we need. In the everyday sense, being free to choose means that I am not compelled, either externally or internally, but get to decide a course of action based on my beliefs, my values, and my desires. This is quite compatible with determinism. Freedom consists not in being exempt from causation, but in the ability to deliberate, either with ourselves or with others on the proper courses of action to realize our ends. Philosophers going back to Aristotle have identified human autonomy—not with indeterminism—but with our ability to make and execute rational plans for the realization of our purposes.”
Rosenberg, however, will have none of this idea of freedom deliberation either:
Since the brain cannot have thoughts about stuff, it cannot make, have, or act on plans, projects, or purposes it gives itself. Nor, for that matter, can it act on plans that anyone else favors it with. There are no plans. That’s just more of the illusion Mother Nature exploited for our survival (238).
Is he serious? We never think about things? We never make plans? We never have purposes? This seems absurd. However, the history of philosophy is replete with ideas that appear quite absurd on their face (e.g. Berkeley’s claim that matter does not exist; Wittgenstein’s assertion that you cannot know that you are in pain; Quine’s claim that we can never really know what speakers of other languages are talking about) but these assertions were not gratuitous absurdities but offered on the basis of rigorous argument. Hence, even if, in the end, they are actually absurd, they cannot be dismissed but have to be argued out. So, we have to examine Rosenberg’s arguments for these extraordinary claims. As the saying goes, though, extraordinary claims require extraordinary evidence, and we are justified in putting a heavy burden of proof on Rosenberg’s arguments.
Thoughts appear to be about things. For instance, I seem to be thinking about my cat and the endearing but sometimes irritating way that she will climb onto my chest, even when I am trying to drink my morning coffee and read the newspaper. “Intentionality” is what philosophers call this “aboutness” of thought. Our thoughts seem to have this intentional quality even when we are considering things that do not exist. For instance, it certainly seems to me that I can think about mermaids, centaurs, Santa Claus, honest politicians, and other non-existent beings. Intentionality seems to just be a datum, a given of our conscious experience. Indeed, intentionality and qualia—the felt qualities of sensations, like the smoothness of silk or the richness of cream—seem to constitute the essence of consciousness. Conscious states just are those qualitative or intentional states. How could we be wrong about that?
I think that many—perhaps most—philosophers would at this point just dismiss the claim that intentionality is an illusion as a pathological aberration. I think that they would justify peremptory dismissal like this: That all thinking is thinking about is just a datum of consciousness. The idea that there could be a content-less thought seems just ridiculous, in fact, contradictory, like saying that there could be a colorless green nightgown. Therefore, in denying intentionality, Rosenberg is denying thought and asserting that we do not think. Such an assertion is no more worthy of philosophical rebuttal than the assertion that one is a poached egg..
Just dismissing Rosenberg’s claim therefore is therefore an eminently understandable response. Still, since Parmenides one purpose of philosophy has always been to push the limits of reason, to see whether reason can justify claims that seem false, even outrageously so. I think that this is an important, if, no doubt exasperating and fatiguing enterprise, that is worth doing only if, in the end, we return to our familiar beliefs with a deeper understanding of why we hold them.
Rosenberg’s answer, in a nutshell, is that thought is a physical thing—a configuration of neurons and their states in the brain—and no physical thing can be about another physical thing. Let’s consider the kinds of thoughts we call memories. My wife and I visited Paris in 2004 and we have many wonderful memories from that trip. Such memories must be encoded in highly complex neural connections, but neural connections are just physical states, like the wires of an old fashioned telephone exchange. Physical states and things may be related to each other in a number of ways, but none seems to be capable of being about another. We may use a key to open a lock, but the key is not about the lock. How, then, can my memories, a physical state, be about some other physical entity—the city of Paris?
But are not some physical things in fact about other physical things? Isn’t an octagonal red sign with letters spelling out “S-T-O-P” about the physical act of stopping your car? However, a stop sign, considered as a physical entity, is nothing but metal shaped and colored in certain ways. It is no more intrinsically about stopping a car than a pair of green trousers hung in the intersection. Rosenberg puts it this way:
Well, then, might not the brain serve as its own interpreter? Might not the brain interpret some of its own neural states as being about Paris? Yet the interpreter in the brain could be nothing other than another neural state, and we face the same problem all over again, namely how one physical state can be about another one. If neural state #2 is the interpreter of neural state #1, interpreting it as being about Paris, then there would have to be a third neural state to interpret#2 as being about #1! And then the same problem arises for #3! Clearly, rather than solving the problem, we are on the road to an infinite regress. Our proposed solution just recreates the same problem again, i.e. how one physical state can be about another physical state.
The upshot seems to be that we cannot have any memories about Paris, since memories are physical things and Paris is a physical thing, and physical things do not bear any intrinsic aboutness relations to each other. One would have to be interpreted as being about the other, but because the interpreter in the brain can only be another physical thing, we are just making the problem worse. Further, what holds for memories would hold for any other thoughts, including those that would be about plans or purposes.
This argument seems to rest upon an equivocation on the word “memory.” By “memory” we could mean either the physical traces in the brain that encode information and is passively awaiting retrieval, or we could mean the active, conscious memories that we create by accessing that stored information through a process we call “remembering.” Paul Thagard describes the process of recalling a concert you had once attended:
Retrieval of a memory works by reactivating a pattern of firing in a population of neurons. Suppose someone starts telling you about another concert that is similar to the one you went to, perhaps because the bands played the same kind of music. Hearing about the new concert may produce a pattern of firing in roughly the same population of neurons that encoded the various aspects of the old concert. The newly generated pattern of firing will then generate additional neural activity by virtue of synaptic connections, possibly producing a pattern of firing that is roughly similar to your original experience. That activation of a firing pattern of neurons constitutes your recalling the memory (49).
By reactivating a particular pattern of neuronal firings, the brain draws upon stored, encoded information to create an active, conscious memory. Memory is a creative process, by the way; remembering is not playing back a recording, but an active (and error-prone) reconstruction.
My memories of Paris in the sense of patterns of neuronal connections caused by a trip to Paris are there whether I am consciously remembering Paris or not. Memory in that sense—patterns encoded in the brain—are not intrinsically about Paris any more than the travel guide to Paris when nobody is looking at it. When nobody is looking at the travel guide it is just marks on paper, just another physical thing, and likewise for the stored information encoded in the neuronal connections in a brain. However, when we actively remember and that stored information is drawn upon to create a conscious recollection, then we have more than passive storage. We have a conscious (self-aware) process of remembering, actively reconstructing memories by drawing on stored information—a process physically realized in the reactivated patterns of neuronal firing. Memory in this latter sense can most definitely be about Paris or anything else.
At a more fundamental level, Rosenberg seems to conflate the doer with the doing. The singer is the doer, and hitting high C is what she does. She accomplishes this remarkable vocal feat entirely with her physical apparatus for creating and projecting sound. However, the accomplishment itself, hitting that true note, is not a physical thing, and you commit a grave category mistake if you classify it as such. Of course, to say that it is not a physical thing does not imply that it is a nonphysical thing. It is not a thing at all. It is something done by things. Of course, hitting high C is an event in space and time and it is entirely physically realized, but the aria, of which that note is a part, is the organized pattern of notes and words as arranged by the composer and librettist. The aria is not a physical thing, but a particular kind of order, a way of organizing things, that can be physically realized in innumerable ways. What is true of arias is true of memories and all sorts of other thoughts.
The upshot is that speaking of a brain’s relation to Paris is not like talking about the relation of one lump of clay to another. Lumps of clay are passive; they just sit there and one definitely cannot have any sort of “about” relation to the other. The living brain, however, is not a lump, but is constantly, fantastically active. It is doing things all the time, and Rosenberg has given us no reason whatsoever to think that it is incapable of generating thoughts about things. One of the many amazing things a brain can do is to think about Paris. If you can think about Paris, then you can think about plans, purposes, and the whole intentional shebang.
Rosenberg would not be impressed with the above line of argument. What the brain does, he indicates, is no different from what the brains of sea slugs or rats do, and no different, in principle, from what computers do. What sea slugs and rats do when they learn is to rewire their neuronal connections to create new input/output circuits that create a new habit. Thus, sea slugs and rats can be conditioned to acquire new habits and the conditioning works by effecting neuronal reconnections. Rats, for instance, can learn how to locate a life raft in a water tank. There is no reason to think that sea slugs or rats acquire new habits by thinking about them. Their brains simply change to correlate input with different output. The same thing seems to have happened with you when, in early infancy, you learned to recognize your mother’s face:
When the rat acquires and stores information “about” the location of the life raft in the tank, that’s just the neurons in its hippocampus being reorganized into new input/output circuits. They have changed in the same way the neurons in the sea slug have changed. Similarly, knowing what your mother looks like or that Paris is the capital of France is just having a set of neurons wired up to an input/output circuit (185).
Neither the sea slug, nor the rat, nor you need to think about these things to get them right. It is all input and output.
Computers can do very complex tasks that previously only humans could accomplish, like playing chess well enough to beat the human champion or even to excel at the TV game show Jeopardy! But computers, like Watson, the computer that plays Jeopardy!, do not think about things at all. Once again, it is merely input and output. Watson is cleverly programmed so that when the input is a Jeopardy! answer, the output is something we interpret as the appropriate Jeopardy! question (in Jeopardy! you are given the answer and required to give the appropriate question). Rosenberg says that the brain is just a computer:
[The brain is] composed of an unimaginably large number of input/output circuits, each one a set of neurons electrically connected with other through their synapses. The circuits transmit electrical output in different ways, depending on their electrical inputs and on how their parts—individual neurons—are wired up together. That’s how the brain works (188).
So, in answer to my suspicion that he is conflating the doing with the doer, Rosenberg could reply that we know precisely what the brain does, and the brain does just what a computer does. If a computer is incapable of thinking about anything, then the brain is incapable of thinking about anything.
Unquestionably many of the things our brains do can be explained in terms of input/output and intentionality does not enter into it at all. Recognizing your mom’s face is certainly one of these; you have been doing it automatically since you were a small infant, and doing it without any thought at all. The same goes for many other operations of our brains. For instance, as soon as I see certain politicians’ faces on TV, I immediately begin to mutter expletives. There must be an input/output circuit in my brain such that when the input is an image of politician P, the output is a string of monosyllabic words of Anglo-Saxon derivation.
On the other hand, some of our mental operations, prima facie, certainly are not easily explicable merely in terms of input/output, that is, stimulus/response reactions. Consider doing philosophy. Can we believe that Rosenberg’s production of The Atheist’s Guide to Reality was nothing but a complex concatenation of automatic inputs and outputs? Didn’t he have to think about it? This is not an ad hominem argument or a too-quick “hoist with your own petard” kind of maneuver. Rather, I am just pointing out that reflective thought, and innumerable other things we do with our brains certainly are not obviously merely input/output events. Let’s put it this way: If reflective and creative thought (recall Beethoven’s Eroica symphony again), for instance, are explicable in terms of inputs and outputs, we can say that such an explanation currently exists only as a promissory note. That puts it far too weakly. It is more like a third-party, postdated check drawn on a bank in Burundi.
I think that Rosenberg’s basic fallacy is not to mistake the doer with the doing, but to think that what the part is doing must be what the whole is doing. Invoking a musical analogy once again, no musician performs a symphony; symphonies are performed by orchestras. Each individual in the orchestra plays his or her part, and the result of a hundred people doing that correctly is a symphony. An individual neuron is just an input/output device. However, thinking involves the very complex, hierarchical, multiply-patterned, parallel-processed, and highly coordinated interactions of ensembles of millions or billions of neurons: the cerebral symphony.
With respect to any activity involving the coordinated interaction of numerous units, any question about what is being done can only be answered by specifying the level that we are talking about. Otherwise the question is meaningless. What is your car doing as you drive down the road? Well, what the car is “doing” depends on what level of organization you are talking about. The fuel injection system is doing one thing and the engine cooling system something else. Locomotion is an emergent property that comes in only when we are talking about the vehicle as a whole. By analogy, asking “What is your brain doing?” is unclear unless the level of organization is specified. The organization of the brain begins with individual neurons, which are organized into systems, which are organized into systems of systems, and then systems of systems of systems…all interacting in astonishingly complex ways with multiple feedback loops and cross connections. Reflective thought seems to be one of the higher level activities of the brain. Individual neurons cannot consider philosophical propositions, but millions or billions coordinated and interacting in the right way maybe can. Perhaps “Philosophy” is one of the tunes the cerebral orchestra can play.
But how can individual events that are about nothing add up to an event that is about something? Suppose that we arrange 10,000 people in a stadium in a 100 × 100 square. On each seat is a specific monochrome card for the occupant to hold up on cue. On cue each holds up his or her card, and when seen from across the stadium, the cards create a 10,000 pixel portrait of, say, Barack Obama. Each individual card is a portrait of nothing; it is just a solid color. However, when displayed all at once and in the correct order, they create a detailed and highly accurate portrait. Of course, this is just an analogy, but it does indicate how a picture of something can be constituted entirely by bits that are pictures of nothing.
It is not even clear that Rosenberg is right about computers and their capacities. Is it certain that no computer, or system of computers, could ever be programmed to think about things? Why not? Perhaps Rosenberg has an argument demonstrating the impossibility of intentional machines, but it is not to be found in The Atheist’s Guide to Reality.
Before continuing, a nagging question that has been waiting in the wings during this whole discussion and must finally be addressed: How can I think that I am thinking about P without being able to think that P? Conscious experiences have the peculiar property that appearing to be in a conscious state is to be in that conscious state. If it seems to me that I have a splitting headache, then I do have a splitting headache. If I seem to be hearing the opening notes of Eine Kleine Nachtmusik, then I am having the experience of hearing the opening notes of Eine Kleine Nachtmusik, even if I am having an auditory hallucination. How, then, can I seem to be thinking about, say, my cat, unless I am actually thinking about my cat? Well, maybe I have made a dreadful mistake and the cat that I am thinking about as mine is actually my neighbor’s cat and not mine. But even in this case I am still thinking about something, namely, my neighbor’s cat.
think that Rosenberg must at least admit that when I think that I am thinking about my cat, i.e. I seem to be thinking about her coloring, her temperament, etc., then my subjective experience—the way it seems to me—will be the same as if I were actually thinking about her. How could it be any different? What content would the state of really thinking about her have that only seeming to think about her would lack? But if a brain can achieve the one state—thinking that I am thinking about my cat—how can it not be able to achieve the subjectively identical state—actually thinking about my cat? How can one mental act be possible and a second phenomenologically identical one be impossible? Rosenberg may have an answer, but I have not found it.