Saturday, June 12, 2010

Dawkins is not a neurologist, and neither am I, so let's have at it!

Richard Dawkins has a new featured article on his website that contains some musings about a subject in which he is very much not an expert: neurology. Since I am not an expert in neurology either, I suppose that makes me eminently qualified to play the other side. And in any case, it touches on a subject that is highly relevant to my most recent post: qualia. So let the battle of the Armchair Neurologists commence!

Dawkins main idea is that, if we assume that there is no absolute binding between the physical wavelengths of light and the qualia of color (and this seems a highly reasonable assumption), then it also stands to reason that a red-green colorblind person has "qualia receptors" (my phrase, not Dawkins') that they have never used. If one could electrically stimulate that part of the brain, so Dawkins suggests, the colorblind person would have the unique experience of "seeing" a totally new color that they could never have imagined, thus shedding light on some significant philosophical issues. (I'm surprised Dawkins never mentioned to explicitly cite Mary's Room, since that's basically the core thought experiment of the philosophical problem he is describing)

I'm mostly with Dawkins on all of this -- and I will give my consonant opinion on Mary's Room at the end of this post, which is weird timing because I was going to put it in the last post but decided it was too long and rambling already -- but from what little I know of neurology, I suspect his idea about the neurological experiment on a colorblind person will be unable to produce the desired results.

As I mentioned before, I think it's a quite safe assumption that there is no absolute binding between physical phenomena and the qualia they stimulate in the brain -- though in practice I imagine that the qualia I experience when I see the color "red" is pretty damn similar to what most humans experience. The alternative is certainly possible, i.e. that each person's wavelength/qualia mapping is clean-sheet original, but it seems far more likely that, within a species, those mappings are going to be pretty similar from individual to individual. Admittedly I am basing this mostly on intuition -- but I do think there is one shred of evidence, namely that we all tend to mostly agree on which colors are similar to each other -- if our individual wavelength/qualia mappings had no relationship to each other whatsoever, then maybe we'd all agree which wavelength was called "red", but we might argue vehemently about whether that color was more similar to "magenta" or "chartreuse". On the other hand, Dawkins' speculation about bats' "seeing" color (by co-opting "qualia receptors" that are used visually in our brains to receive audio stimuli) is eminently reasonable. While I imagine that organisms with nearly identical genetic makeup have very similar stimuli/qualia mappings, all bets are off when you are comparing two different species.

Where my understanding of neurology diverges from Dawkins is the idea that a red-green colorblind person has significant patches of unused "qualia receptors". This seems highly unlikely to me based on the way the brain tends to self-organize. It seems likely to me that the visual qualia receptors for these colors would have simply been "invaded" by connections for nearby colors.

Allow me, if you will, to engage in a bit of wild speculation and story-telling here, which is probably complete bullshit, as it is nothing more than blind intuition based on a layman's understanding brain development, paired with a hazy recollection of a brief primer on the neurology of vision I got during a 3-day crash course in color science I took several years ago. With that caveat out of the way I imagine that the development of color vision -- at least in terms of these theoretical "qualia receptors" -- goes something like this:

I imagine there is a a field of cells somewhere which serves a function analogous to the cortical homunculus, except that it represents the qualia of color (and in fact, it seems not unlikely that this field of cells could be a part of the neurological correlate to the cortical homunculus itself). In early life, our genetic program establishes a binding at certain key points on this "visual homunculus" if you will. Since our sensation of hue is two-dimensional (red-green/yellow-blue) perhaps our DNA is "hard-coded" (I analogize liberally here) to wire the signal for 100% red to one region, 100% green to a region spaced as far away as possible, and 100% yellow and 100% blue to regions in between. This establishes four "calibration points" which, since they are hard-coded in our genetic program, are more or less the same among all normal individuals.

As we experience color as infants, the brain automatically populates the gaps in the "visual homunculus" in relation to the calibration points. For instance, a color sensation that mildly stimulated both the 100% red and 100% blue cells could cause a new, stronger wiring to be built at 50% red and 50% blue. As more and more colors are experienced, the gaps are gradually filled in -- according the calibration points, but also with the sloppiness and imprecision that is typical of biological systems. There's nothing in the genetic program that says, "The visual signal for 75% red/25% yellow shall be routed to this approximate region of qualia receptors" -- it just winds up (approximately) there over time, as new stimuli/qualia bindings are populated by recognizing their proximity to existing stimuli/qualia bindings.

I'm sure if any neurologist reads this they are already cringing, but at the risk of making myself look even stupider, let me plow further ahead into territory in which I am completely unqualified to speak: In this fable I have concocted, what happens to a red-green colorblind individual? Their initial four calibration points get wired up just like the genetic program says, but then no stimuli with a red or green component are received. This has two effects: First, the intermediate qualia receptors get populated with wirings based strictly on stimuli from the yellow-blue spectrum. Second, since we know that the strength of neural connectivity is primarily a function of use, the initial red-green calibration bindings would have a tendency to atrophy, with those regions of the "visual homunculus" perhaps even being invaded and overtaken by intermediate yellow-blue spectrum bindings.

In this fantasy world, what would happen if Dawkins' experiment were performed? What if we stimulated in a red-green colorblind person the region of the "visual homunculus" that was supposed to have been hard-wired to 100% green? He would simply see something pretty similar to what he normally sees when shown the color green -- a hue somewhere intermediate between yellow and blue. Counterintuitively, the qualia he experiences might actually be akin to the qualia I experience when I see the color green -- but he would have no frame of reference to recognize that this was distinct from the qualia he and I experience when we see the color red. In fact, given the plasticity of the brain, maybe colorblind individuals experience the qualia for the color red and green simultaneously when they see either green or red. Who knows, I'm just making shit up anyway, right?

Okay, so this concludes my idle speculation. All of the above is likely to be complete bullshit. But from what little I know about neurology and fetal/infant brain development, it would seem it has to be something like that, and it's easier to tell a highly-specific fable than to speak in general and guarded terms for several paragraphs straight.

As evidence that my "something like that" hypothesis seems reasonable, I would point to the experiences of those who have disorders involving the cortical homunculus (e.g. phantom limb, body integry identity disorder). It seems that, if a section of the somatosensory cortex ceases to receive stimulation, it is not uncommon for those areas to be "invaded" by sensory cells for the adjacent body regions (if you aren't a Wikipedia hater, see the last section of this paragraph for some descriptions of this). This takes place even in adults -- how much more pronounced would we expect the effect to be in newborn infants, especially those who never experienced the missing stimulation to begin with!

So Dawkins' experiment is a cool idea, and the article probably serves as a good introduction to the idea of qualia. Moreover, the idea about bats "hearing" in color and rhinos "smelling" in color is pretty cool, and quite plausible I would say. (Though given that the mammilian somatosensory cortex spent most of its time evolving in tiny rodents for whom smell was far more important than vision, it might be more fair to say that humans "see in scents") But I'm guessing the answer to his final question is: No, it doesn't really work that way. Ah well.

Okay, now, as promised, since it's highly relevant to this discussion, my opinion on Mary's Room: Neglecting for the moment some serious technical implausibilities in the scenario as described, I think the entire issue has been clouded by ambiguity over the word "knowledge". So rather than mess with that semantical can of worms, let me take an uber-functionalist position on this and rephrase the question as: When Mary emerges from the room and sees color for the first time, does it stimulate regions of her brain in a profoundly new and novel way? The answer here is obvious: Yes, of course! And thus, qualia do exist, not as some dualist phenomenon separate from the physical world, but rather as a certain type of mental stimulation within the brain.

Like the problems with the Chinese Room, I think the trick with Mary's Room relies on getting the listener to accept that two concepts are analogous, and to then assert that an extremely complex phenomenon can be embodied in an extremely simple one. The latter seems absurd, but that's because it is absurd -- not because there is something in the analogy we are missing, but because a complex thing and a simple thing are not equivalent even if they are in most ways analogous.

In the case of the Chinese Room, the complex thing is a Turing machine programmed to speak and understand Chinese, with enough storage capacity and processing power to carry on a conversation; and the simple thing is a guy with a fucking stack of books. Because humans are bad at thinking about scale, we don't realize the stack of books is the size of the moon and the guy is dead before he even comes up with a single sentence. So the two are analogous in the sense that a guy who is executing instructions from a book is more or less "Turing complete", but who cares?1

In the case of Mary's Room, the complex-to-simple shenanigans is in the premise of Mary being a "super-scientist", and what exactly that implies. We tend to subconsciously parse the phrase "she knows everything there is to know about color vision" as meaning that she has abstract knowledge of how it all works. In other words, her brain has produced an abstract representation of the brain's response to color stimuli.

But then the dualist proponent does a switcheroo, and says, "Ah hah! You said she's a super-scientist and knows everything there is to know about it. Therefore, she must know what it's like to see the color, right??!" But this is stupid if you put it in terms of brain stimuli. What we are really asking for when we say she has learned what it's like to see the color is that she has not only created an abstract representation of the color in her brain, but that she has "magically" stimulated regions of the somatosensory cortex just by knowing those regions of the cortex exist. That's not just a "super-scientist", that's some kind of magic wizard. A "super-scientist" as we imagine it is no more comparable to this version of Mary than a guy in a room with a book is comparable to the world's most epic supercomputer.

The word "knowledge" is really what clouds the issue, because at the start of the thought experiment we implicitly define it as abstract knowledge of facts, while at the end of the thought experiment we have defined knowledge as any sort of brain stimuli. By analogy, let's say I see an object on the other side of the room and I decide I'm going to go pick it up. One might argue that have all of the "knowledge" there is to know about the location of the object. But obviously the pattern of neurons firing in my brain will be very different depending on whether I walk over and pick up the object, or if I continue to sit on my lazy ass writing an over-long blog post on philosophy.

If you restate Mary's Room with "knowledge" explicitly and consistently defined, then it becomes absurd. If we define "knowledge" broadly as any sort of pattern of neural firings, then either the answer becomes an obvious "yes" (but with qualia being nothing more than a pattern of neural firings) or else the experiment would be better called "Magic Mary" since we are asking her to do the impossible. On the other hand, if we define "knowledge" narrowly as being abstract factual information, then the answer is "no, but so fucking what?" Clearly there are things that exist in the physical world other than neural firing patterns representing abstract knowledge of facts. You might as well argue that since Mary doesn't gain any knowledge when something happens behind her back, that therefore all things that happen behind Mary's back exist in some separate epiphenomenal supernatural world independent of physical reality. Um, no.

Yes, qualia exist, but they are nothing more than patterns of neurons firing in the brain. Whether or not you want to call that "knowledge" is boring-ass semantics.

1On a side note, as I understand it one of the main objections to this counter-argument against the Chinese Room is that by relying on speed and complexity as the thing that distinguishes an understanding mind from a guy executing instructions from a book, one has failed to define a sharp quantum boundary between the conscious and the mechanistic -- to which my reply can only be, "No shit Sherlock, get over it!" Seriously, what philosopher worth her salt hasn't figured this out already? Does a rock have consciousness? Does a prion have consciousness? Does a virus have consciousness? Does a bacterium have consciousness? Does a patch of moss have consciousness? Does a mushroom have consciousness? Does a stalk of broccoli have consciousness? Does a worm have consciousness? Does an ant have consciousness? (What about those species of ants in which some individuals are born as immobile storage vessels for food? Do they have consciousness?) Does a trout have consciousness? Does a hummingbird have consciousness? Does an eagle have consciousness? Does a mouse have consciousness? Does a lemur have consciousness? Does a pig have consciousness? Does a dog have consciousness? Does a dolphin have consciousness? Does a chimpanzee have consciousness? Does a newborn human infant have consciousness? Does an adult human have consciousness? Does a severely brain-damaged person have consciousness? Does a person in a persistent vegetative state have consciousness? If you didn't answer at least one of the previous questions with a "no," at least one with a "yes," and at least one with an "I'm not sure," then you are a fucking idiot. It's a continuum, Searle, deal with it!


  1. I haven't read all your post, but apparently non-colorblind people are relatively colorblind to shades of brown that coloblind (or some type of colorblindess) people can more easily distinguish. So it may be the case that colorblind people don't have "unused color-qualia" but rather it's somehow shifted/compressed, distributed differently.

    It's also interesting to note that it doesn't happen only with "physical" colorblindness, but color discrimination is "learnt"/developed, and living in different environments, with different exposure to color will lead people to develop better acuity/discrimination of some shades. The most famous case are the Himba people of Africa, who can easily sort out shades of lime-green that most people see as outright the same, but they're dumbfounded with cyan and blue (or cyan and some greenish cyan). To a lesser degree, Norwegians will also have some differences in color vision depending on the latitude and season of birth. Most strikingly some research suggests that our colour vision is substantially "rewired" as we grow up, and affected by the color naming/pallete of one's native language

  2. This subject is really clouded by some counter-intuitive terminology sometimes. I think it may be the first time I saw this rather obvious point that learning everything about optics and neurology of color vision *most likely* wouldn't make one able to voluntarily control one's own brain to make the neurons "fire" in the same way they'd do when stimulated by nerve inputs from the eyes. Mary would need to be not only the most accomplished neuroscientist, but also some have an uncanny skill in a sci-fi-worthy hybrid of meditation and self-hypnosis, to the point of having voluntary control over her endocrine system among other things. She would actually have been able to see color-qualia before leaving the room.

    Actually Paul Churchland makes a similar point, even though I think he still makes the sort-of-mistake of speaking of qualia in terms of "knowledge", even though he does make the point that it would be a different type of knowledge than the "textual" knowledge that's more commonly understood from the word. This is one tricky subject I've many times found myself "facepalming" in the middle of "debates" with an insane denial of qualia in one side, and some people using Mary argument to try to prove some sort of "supernatural" in the other, while to me it seems so obvious that neither are true from the experiment that I've found hard to express how, it's so "not even wrong" that makes you stumped.

    Churchland also has an interesting comeback: Mary's twin sister or something is the most knowledged scholar of *ectoplasm* (or "un-matter", whatever), she's on the same room, and leaves, sees color, and she also would have seen something she couldn't before have known, therefore we need yet something more than immaterial/ectoplasm, perhaps another whole layer of supernatural! Not dualism, but trialism! (Something along these lines).

    I've found simply mind boggling that Dennett apparently said something along the lines that perhaps one could "explain" a color, but it would take "three million years of talking to do that", or something like that. That's ridiculous.