Thursday, September 9, 2010

What's Wrong with Rhodopsin?

Looking up info for another blog post, I see that Adams and Aizawa gave another example trying to indicate what the coupling-constitution distinction might be (rather than trying to give a philosophical account of what the distinction might be).   The process is the isomerization of rhodopsin upon photon capture.
    Both Rockwell, (2005), and Hurley, (forthcoming), express skepticism about the coupling-constitution distinction, although neither express serious reasons for doubting the distinction.  There is, it seems to us, some reason for their suspicion, namely, that it is hard to make out this distinction for the case of cognition.  What is the difference between things that merely cause cognitive processes and things that constitute cognitive processes?  If we restate the question slightly, the source of the difficulty should be clear.  What, we should ask, is the difference between things that merely cause cognitive processes and things are cognitive processes?  The problem lies in the uncertainty about what exactly cognitive processes are. In support of this diagnosis, consider a case where we do have a well-established theory of what a given type of process is.  Consider again the process of nuclear fission.  The process of nuclear fission is constituted by the process of a large atomic nucleus being broken into smaller atomic nuclei.  Nuclear fission can be caused by bombardment of the nucleus with neutrons.  The process of neutron bombardment causes nuclear fission, but does not constitute nuclear fission.  Consider the isomerization of the retinal component of rhodopsin in the human eye.  This process is constituted by a change in the molecular structure of the retinal component from 11-cis retinal to all-trans retinal.  It is typically caused by absorption of a photon.  Now the distinction is intuitively clear, although possibly difficult to explicate philosophically.  Where we have a clear theory of the nature of a process, we have a very fair idea of the difference between what might cause it and what might constitute it. (Adams & Aizawa, 2008, p. 101).
They object to the nuclear fission example because "the literal description of nuclear fission is mathematical and incorporates no such simple intuition".  (Ross and Ladyman, 2010, p. 163).  Ok.  So, what's the problem with the rhodopsin example?

16 comments:

  1. What's wrong is you've got an incomplete description of the process. Rhodopsin doesn't just isomerise: it isomerises when exposed to enough energy, typically a photon's worth. So the process is actually "a change in the molecular structure of the retinal component from 11-cis retinal to all-trans retinal typically caused by absorption of a photon." The photon seems to be doing quite critical work and is thus a sensible candidate for being a constituent of the process.

    ReplyDelete
  2. Well, let's distinguish two processes.

    There is the process of isomerization, which is the change of shape of the molecule. This process is caused by photon capture.

    Then there is this other process, "a change in the molecular structure of the retinal component from 11-cis retinal to all-trans retinal typically caused by absorption of a photon." That process is just by its description a process of which the photon capture is a constituent.

    By, I'm talking about isomerization, not your new fangled "more complete" notion.

    ReplyDelete
  3. But, your move--What's wrong is you've got an incomplete description of the process--iterates.

    Think about the process "a change in the molecular structure of the retinal component from 11-cis retinal to all-trans retinal typically caused by absorption of a photon." To make a "complete" description of the process, don't you have to add in the process of the emission of the photon, say, from the sun? That seems to be a road to madness. Or, if you say that the photon emission caused your process, you get the causation-constitution distinction.

    ReplyDelete
  4. Why can't you just say: you can make the photon any way you like (with a star, with a light bulb), so none of those specific processes need to be accounted for here; but you need a photon, no question, so it deserves a mention in your analysis. Implicit (or even explicit) is "the photon was created by some process", but that's all you need to say.

    Apparently the photon is necessary for isomerisation; given that, my new-fangled version seems more appropriately complete.

    ReplyDelete
  5. 1) But,isomerization can likewise be caused by many things. It just so happens that *rhodopsin* isomerization is caused by photon capture. (And maybe rhodopsin isomerization does happen spontaneously due to quantum mechanical fluctuations or something other than photon capture. Maybe change of pH.)

    So, if we adopt your strategy for blocking mention of stars, light bulbs, etc. in isomerization, then we can use your strategy for blocking photon capture in talk of isomerization.

    2)Setting aside 1), it remains that case that there is a "complete" process and and an "incomplete" process. Photon capture is a constituent, but not a cause of the complete process, but photon capture is a cause and not a constituent of the incomplete process. That gives you an example of what is up.

    3) "Apparently the photon is necessary for isomerisation;" But, there are, at least, two types of necessity in play in this area. A) being causally necessary and B) being constitutively necessary. It's not enough to say that there is a necessity involved.

    4) What about distinguishing the process of melting from the process of heating? Isn't heating a cause of melting?

    ReplyDelete
  6. Re 1:
    But,isomerization can likewise be caused by many things. It just so happens that *rhodopsin* isomerization is caused by photon capture.
    This gets into something that hasn't really come up yet but is part of my analysis style here; there are no general solutions. There are only task-specific solutions generated by general principles. (Feel free to disagree on this, but if you do then we're always going to have a fairly fundamental incommensurabiity).

    So you have to identify whether you are talking about isomerisation, or whether you are talking about the isomerisation of rhodopsin; and what counts as a complete description depends entirely on which you are interested in. Pick whichever is most appropriate to the question your interested in. This is the 'task analysis' stuff I was banging on about - you have to ask the right question to get the right answer.

    2. By what I just said, there's actually two different 'complete' process descriptions. I was focused on the specific example in what I said earlier.

    I'm starting to get the feeling that I don't really get the cause/constitution distinction, because I'm becoming less concerned by it every day. Things that constitute part of a system presumably cause things in that system; why not the reverse?

    ReplyDelete
  7. Well, the causation/constitution distinction doesn't work for everything. It doesn't work for your example of the "complete" rhodopsin isomerization process just because you've defined the process to consist of a cause.

    There are also certain understandings of "system", such as the auditory system, is which the system appears to be whatever (e.g. pinnae, bones of the inner ear, cochlea, etc.) causally contributes to hearing.

    But, there are plenty of cases where the distinction is as clear as any philosophical distinctions ever get. Here are a couple of more (in addition to the nuclear fission case).

    1) The process of NaOH dissolving in water causes, but does not constitute the warming of the water.

    2) The process of irradiation of human skin by ultraviolet light causes, but does not constitute, the production of extra melatonin.

    3) The process of irradiating the human crystalline lens by ultraviolet radiation causes, but does not constitute, the process of cataract formation.

    4) The process of piercing a Bridgestone Potenza RE-11 195/50R15 ties causes, but does not constitute, the process of rapid air loss in the tire.

    But, if you don't like the word "constitute", bear in mind that there is another way to speak of what is going on. (See http://theboundsofcognition.blogspot.com/2010/08/ambiguity-of-constitution.html.) You cans say,

    1*) The process of NaOH dissolving in water causes, but isn't the same thing as the process of the warming of the water.

    ReplyDelete
  8. Can I say instead

    1*) The process of NaOH dissolving in water is a constitutive element of the 'water being warmed by NaOH dissolving in it' system

    You may or may not want to talk about that system; that's fine. But if you add NaOH to water and it warms because of that and not some other reason and you want to talk about that then it seems unfair to relegate NaOH to the sidelines.

    I don't think you're describing the correct process, nor do I think this entails 'bloat' or iterates. In the cognitive examples, my descriptions are bounded by considerations of information (which may or may not be right, but that is a different question). Gibsonian ecological information can serve as a principled, theoretical constraint on this kind of analysis, but it will end up with these broader task descriptions that I think are much more suitable.

    I have a blog post about this brewing, I think. I've been reading your 2001 BoC paper and I have some criticisms along these lines I need to work up. Not 100% sure when I'll get time but hopefully soon :)

    ReplyDelete
  9. I) Well, yes, you can define new fangled processes and systems in which the causes are constituents. But, what of the old fashioned ones? That was my comment #2.

    II) But, the NaOH example was meant to exploit one of your principles. In your comment #4 you wrote, "you can make the photon any way you like (with a star, with a light bulb), so none of those specific processes need to be accounted for here;" So, when I offer

    1) The process of NaOH dissolving in water causes, but does not constitute the warming of the water.

    it seems to me that by your own principle you shouldn't offer your 1*) because there is more than one way to warm water. Dissolve in HCl or place over a bunsen burner, for example.

    ReplyDelete
  10. Basically I think you're old fashioned process description is incomplete and to do things properly you need the new fangled one. Your process begs the question 'why is the water warm?' which you have to answer with a second causal process. Mine just includes everything you need to understand this specific instance of the water warming, and I think you need to be this specific.

    ReplyDelete
  11. I) It sounds like, protestations notwithstanding, you get the causation/ constitution distinction. In fact, you seem to have a theory that uses it. Your theory is something like "When you have an apparent case of process X causing process Y, you should introduce a more complete process (or system) of "X's causing Y""

    In other words, you use some such theory to guide you in the replacement of 1) by 1*).

    ReplyDelete
  12. Indeed, go back to your first comment, "So the process is actually "a change in the molecular structure of the retinal component from 11-cis retinal to all-trans retinal typically caused by absorption of a photon." The photon seems to be doing quite critical work and is thus a sensible candidate for being a constituent of the process." You describe causation in your complete description and you understand constituents of the thing description.

    ReplyDelete
  13. Apparently I'm learning something as we go :)

    The other thing I think I've learned is why I don't think it's a general fallacy. Your approach to cognition is good old fashioned cognitive psychology: in, for instance, vision, light impinges on the retina, causing a cascade of events in the nervous system. For you, and cognitive psychology in general, the second process is perception, which is a cognitive process. It is caused by the first process, which ends at the retina and is not cognitive. But this is just the orthodoxy rejected by Gibson for all the reasons laid out in the '66 and '79 books. Your fallacy is only a fallacy from the orthodox view; from my perspective it's merely a risk you run if you don't do your science carefully.

    ReplyDelete
  14. Well, the point of the fallacy is that you can't in general just go from "X causes Y" to "X is a constituent of Y". It depends a lot on why X and Y are. This is why part of avoiding the fallacy involves spelling out what Y is. In the case at hand, it means spelling out what one means by a "cognitive process".

    So, there are two principal ideas in our Bounds of Cognition book. You need to dodge the coupling-constitution fallacy and you need a "mark of the cognitive". The two interact.

    Now, your approach of creating more complete processes enables you to avoid the c-c fallacy on a kind of technicality. You hypothesize these more complete things processes that have causes built into them.

    Now, this seems to me to be a bizarre move that cannot be worked our consistently. (I think that your demand for "completeness" either iterates or fails. That was II) in the more recent comments.) But, the more I read in the Extended Cognition literature, the more I see something like this idea implicitly at work. But, it seems to me to be such a strange idea, I never thought of it.

    But, the causation-constitution distinction is not a bit of cognitivist baggage. It seems to me to be a big part of lots of sciences, such as chemistry and physiology. Adams and I have tried to motivate it by what we take to be unproblematic cases outside cognitive science (although I guess one might say that the biochemistry of phototransduction is a part of cognitive science).

    ReplyDelete
  15. I agree entirely that my hypothesis doesn't simply work; I need a theory to constrain it from iterating and to distinguish it from things that aren't cognitive.

    But Gibson, affordances and information do that work. The iteration critique was the Fodor & Pylyshyn move, which was rebutted by Turvey et al and which I talked briefly about here. Then affordances, information for affordances and the ability to perceive affordances via information then work towards a mark of the cognitive.

    Frankly, my point has always been that it's cognitive science that lacks the theory you note correctly is important; Gibson is really the only game in town on this front. Hence our blog. There have also been developments since Gibson (the Turvey et al ecological laws paper, Turvey on affordances as dispostions, Bingham on task specific devices, the entire dynamical systems approach) that have all been pointing in the same direction; the work is not yet done, but the plan is, in essence, a principled theory of behaviour that meets your criteria but that involves redefining what cognition is.

    This last is a big part of it; your defence rests fairly heavily on the current state-of-the-art and it's precisely this that I think this the problem we can solve.

    This is all useful: it's helped clarify a few things :)

    ReplyDelete
  16. Yeah, I need to re-read the Fodor-Pylyshyn piece. I read that many years ago before I had read anything by Gibson, and the Turvey piece is on my list.

    ReplyDelete