Monday, December 14, 2009

Cogitating Ross

Longtime commenter Cogitator alerts me that he has made James Ross’s important article “Immaterial Aspects of Thought” available in HTML format here. Now no one has an excuse not to read it!


  1. Just wondering--how do naturalists/materialists/physicalists typically respond to this paper?

  2. I cannot answer your question aletheist, but according to the recent survey, the majority of philosophers who specialize in the philosophy of mind are physicalists.

    That's no argument for physicalism. Although Dr. Feser encourages us to "trust the experts".

  3. I'm really finding it bizarre how many people are taking away the point of Ed's post as simply "trust the experts".

    Maybe I'm wrong, and if so Ed, please clear this up for me. But wasn't the point that if we're going to "trust the experts", then we should assent to theism given the representations in PoR? But if we reject the "expert opinion" of PoR, then that casts doubt on whether we should "trust the experts" in other philosophical specialties?

    Also, in both TLS and past blog posts, Ed has argued strongly that "physicalist" explanations of mind tend to (knowingly or not) reintroduce teleology and a broadly Aristotlean worldview. I honestly half expect some "physicalists" to one day simply assert that formal/final causes fit in with physicalism, just as panpsychism has now become - apparently, by magic - some variant of materialism.

  4. If Ross's paper is as definitive and important as Dr. Feser and others believe, one would expect the physicalist majority to have made every effort to refute it over the last 17 years. However, so far, I have not been able to track down any rebuttals whatsoever; but then, I may not be looking in the right places or using the right search keywords.

  5. aletheist,

    Are you saying: I can't find a refutation of it; therefore it must not be a very good argument. Very curious reasoning, if so!

    Frankly, I doubt you'll find any reply to it, because most physicalists either ignore dualist arguments altogether -- "We already know they can't be right" etc. -- or attack stock caricatures (ectoplasm, etc.) Ross's argument is so different from what they're used to that I imagine on encountering his article most physicalist-minded readers of J of P thought "What the hell is this? Whatever..." and flipped to the next one.

  6. Yep. I take it as a case of cognitive dissonance. Ross is, IMO, as unjustly ignored by relevant targets of and invitees to his work as Stanley Jaki. On the one hand, Ross's argument in the essay just restates what Goodman, Wittgenstein, Kripke, Quine, Duhem, Putnam, et alii, have said about physical under- and indeterminism, so it might seem derivative. On the other hand, the conclusions he draws from it -- by updating Thomistotelian/Scholastic accounts of intellection -- are so radical and lucid that it's just "dualist" for serious discussion. It all stems, IMO, from the devastating collapse of "nature" into "physical nature."


  7. "Are you saying: I can't find a refutation of it; therefore it must not be a very good argument."

    Oh, not at all! Quite the opposite, in fact. It just surprises me that no one seems to have seriously challenged it after all this time, especially now that Thought and World has been published. I am really interested in knowing how a committed naturalist/materialist/physicalist would respond to it.

    By the way, TLS was eye-opening for me; I am reading it a second time now that I know how it ends. :-) I have already ordered Aquinas.

  8. I should also point out that in Michael Gorman's ND Philosophical Review of Thought and World, he directs the reader to Joel Steinmetz's 2006 dissertation for a detailed discussion of Ross's "immaterial" argument. I also discovered a lecture outline (PDF) by Steinmetz, when he was in the Gonzaga Socratic club, about the problem intentionality poses for physicalism, though it does not refer to Ross, as far as I can see.

    I checked in Thomas Polger's recent defense of the mind-brain identity thesis (both Natural Minds and "A Posteriori Physicalism") and I find no reference to Ross at all. Dodge? If so, I would take it as a typical case of such evasiveness.

    Br. Fadok's MA thesis ("Looking Forward...") does explicate Ross's argument and reminds me to add that this sort of argument is not exclusive to Ross. Mortimer Adler argues the same thing in Mind Over Matter and David Braine's The Human Animal develops the same idea. I think G. Klima makes similar points here and there, as does Fr. Jaki in various places. And let's not forget ol' Plato, Aristotle, Aquinas, et alii themselves!

  9. Crude: "I'm really finding it bizarre how many people are taking away the point of Ed's post as simply "trust the experts"."

    I'm not, because I've realized that most people don't read carefully, most people don't attend (and no one attends at all times).

  10. Truth be told, while this may (or may not) be an excellent argument, I got lost in the verbiage.

  11. "Truth be told, while this may (or may not) be an excellent argument, I got lost in the verbiage."

    Same here, I could not understand it at all.

    I think it is reprehensible how so many academics write like this - I am never impressed, but I don't think they are really wanting their thoughts to be clearly stated.

  12. Ross has a somewhat notoriously from-the-hip, aphoristic style, but it can be defogged with a little persistence. What he consistently avoids doing is masking straight talk in logical symbols, as he feels modal apparati themselves have generated metaphysical fictions just because philosophers like toying with the symbols. (A Lego house is no more a domicile than a symbolic modal operation is a guide to reality.) Anyway, as I said in my post containing "Immaterial", don't be put off if a first reading leaves you with little or nothing. As Ross says in Thought and World, expertise allows for not attending to every little detail. His argument in "Immaterial" is pitched at and for the professional level, so it assumes a LOT of specialized knowledge. I'll probably post an excerpt from a book I'm working on, which might help unpack Ross's exposition. The argument basically asks, "Does addition actually happen when leaves pile up?" and concludes No.


  13. "What he consistently avoids doing is masking straight talk in logical symbols, as he feels modal apparati themselves have generated metaphysical fictions just because philosophers like toying with the symbols."

    For that he has my gratitude.

  14. Going through Ross's paper a second time helped me feel like I got a better grasp of his arguments. I have posted a link to it on Philosophy Forums, where I interact regularly with atheists and physicalists, in the hope that some will read it and (attempt to) respond to it. Nothing substantive so far; anyone who wants to monitor the discussion can do so at

  15. I am starting to get a few rebuttals, and some of them go beyond my own meager philosophical resources. If anyone can take a look and chime in, I would appreciate it.

  16. aletheist:

    You've probably seen it by the time you see this comment, but I have posted a rather lengthy comment at that thread. If binary voltage readings we the iceberg to this "Immaterial"'s Titanic, a thinker as conceptually agile and scientifically literate as Ross would have seen it and ran loooong ago. ;)

  17. Good stuff on the message board Codgitator.

    You're not, by chance, a philosophy major are you? Maybe studied alittle Aquinas?

  18. Bobcat:

    Glad my glosses in that thread were helpful.

    Me, a philosophy major? Nahhh, just got a minor 6-7 years ago at a state university and have-done/always-do tons of reading on my own. (I was a German major.) As for me and Aquinas, well, you could say I've dipped my nose in his scant writings now and then. ;)

  19. And here's V. Reppert's treatment of the argument in the Blackwell Companion to Natural Theology.

  20. Man, I really wish Victor had a better proofreader on that. Those typos are not in my original...

  21. Dr. Feser:

    Yeah, the typos in the p. 367 quotation are pretty egregious. Did they scan a copy of his essay and then print it in the book??

  22. Isn't Ross, in his section II, really making an epistemological point? Given a putative adding device that generates a numerical output given two numerical inputs we can't be sure what process is going on inside to produce the result. Hence the concerns about infinite sequences of input pairs and arbitrarily long periods of observation. I agree, but only if we treat the device as a 'black box'. If we can look inside, see how it's constructed, consult its designers and their technical drawings, apply the relevant physical principles, etc, we can see that every output will be the perfectly determinate sum of the inputs, provided the device operates within its design range of ambient conditions. Now, clearly, the device isn't likely to be understanding what addition is in the way that we do, but that's not Ross's argument. He's saying that it's indeterminate what the device is doing. I'm afraid I flatly disagree. And so, I'm sure, would the people at Intel and AMD. What are we missing in Ross's argument?

  23. David,

    Ross is not saying that it is indeterminate full stop. He is saying that it is indeterminate from the physical facts alone. And you have essentially conceded that he is right about this when you say that we need to "consult the designers." That is exactly his point: There is nothing inherent in physical processes that gives them whatever meaning they have (at least on the modern cocneption of what counts as "physical"). Any determinate meaning necessarily derives from outside.

    Of course, one could always dig one's feet in and insist that the meaning really is inherent in the physical after all, thus giving up the modern conception in question. But in that case one has essentailly conceded that an Aristotelian conception of nature, and in particular the theory of formal and final causes, is right after all. As Ross has argued elsewhere (as have I) that is exactly what all the computational metaphors used by contemporary scientists really imply, if they are taken seriously.

  24. Hello Ed, and thanks for the reply. I wouldn't claim it's *necessary* to consult the designers, just helpful. The chaps at Bletchley Park didn't have access to the Enigma machine's German engineers but were still able to work out what it did from possession of a machine alone. But I'm not sure Ross is interested in the "meaning" of a physical process---he doesn't use the word---rather the abstract function of which the process is a concrete realisation. His claim, if I've got him right, is that any finite or infinite sequence of x,y,z triples that the device produces (x and y being the inputs and z the output) is compatible with quus as well as with plus. Even as an epistemological claim I think this is wrong because if we choose the inputs at random there is a small probability that we will choose an x or y that *does* exceed the critical value (57 in his example) or "differentiating point" as he calls it. But as an ontological claim about the structure of the device I'm convinced he's utterly wrong. If the device implemented quus rather than plus one would find evidence of this within its physical organisation. For otherwise we are being asked to imagine a device that looks inside just like a conventional adder but whose behaviour is that of a quuser. This would be so anomalous as to call for a revision of our physical understanding. Surely Ross would be claiming that our physics is not up to the job of producing reliable machines? I'm left with the feeling that I must have missed something essential from his argument.

  25. Hello again, Ed. I note you give your own version of Ross's argument on page 204 of your Philosophy of Mind. You consider two calculators, one of which performs ordinary addition and the other 'quaddition', a variant of Ross's quus where the differentiating point occurs at some value N, say, that exceeds the highest number that either calculator can display. You say that "there would be, not only no way of knowing which of either of the machines was doing quaddition instead of addition, but no fact of the matter at all about which was performing which."

    Am I right in thinking that you have in mind here that though the calculators can internally represent numbers greater than N, only numbers up to some value D, less than N, can be displayed and entered? The externally visible behaviour of the calculators is thus identical, and hence we can never tell which is which merely by observing inputs and outputs. But if we look inside we must find a difference in the physical structure of the calculators. In the quaddition calculator, for example, we would expect to find a circuit making an explicit comparison of the inputs with N. Such a difference surely counts as part of the 'physical facts'? If it's your position that the calculators are of identical structure I think you owe us an explanation of the sense in which one can be said to be implementing quaddition and the other plain addition.

  26. David B.:

    I recommend you check the PF thread that aletheist initiated as well as my post of gloss-responses to numerous objections to Ross's argument (at my blog).


    QUOTE: Isn't Ross, in his section II, really making an epistemological point? … we can see that every output will be the perfectly determinate sum of the inputs, provided the device operates within its design range of ambient conditions. … He's saying that it's indeterminate what the device is doing. I'm afraid I flatly disagree.

    As Dr. Feser replied, the epistemological point hinges on an ontological difference between a plus- and a quus-machine (pm and qm). To wit, that the latter is “informed” in a way formally incompatible with how the latter is informed. In fact, the argument about physical indeterminacy wouldn’t even require differentiating outputs, since the outputs for the pm could be triggered by auditory inputs of the form “five,” “fünf,” “cinco,” etc., whereas the identical outputs on the qm could be triggered by—and therefore represent—auditory inputs of the form “John,” “Mary,” etc. Let us then imagine that the outputs for the pm were recorded by a digital video camera and set off fireworks, whereas the outputs of the qm triggered a video camera to shut off an idling engine down the street. In this way, both pm and qm would be “doing different things,” even though their physical composition and input/output array were identical.

  27. CONT.

    I imagine the objector will say pm and pm are, on my hypothesis, actually just parts of larger physical systems—call them S(pm) and S(qm)—which are determinate in their own ways. The problem is that this objection already grants the essential point, namely, that, in and of themselves in purely physical terms, pm and qm are formally indeterminate. For all we know—and literally, for all their doing physically—they could always be running different functions. Indeed, even if we established the “forms” of S(pm) and S(qm), we could just rig one of them to a new video camera system and trigger some different physical outcome, in which case, even the larger systems would be indeterminate with respect to their endless formal possibilities. It’s not just that we don’t know which function a physical/material system is “running”; the problem is that nothing about the systems themselves in purely physical/material terms restricts—determines—their being instances of a single formal operation. Pure functions, however, can never be indeterminate in this way, and therefore crucially differ from physical systems. A pure function—say, addition—can never even possibly be “running” a different incompossible operation than what it is, nor can it even possibly “mutate” to alter its formal parameters without ceasing to be the same formal operation. Add to this that pure functions exhaustively include every possible instance of themselves, whereas any single case of a function in purely physical terms is just that—a single case of some function—and therefore the single-physical case is intrinsically formally-incommensurate with the instance-exhaustion of any pure function.

    Presumably, the objector would claim that all physical functions are determinate in the sense that they all “tie in” to the entire cosmos. In this way, all physical systems would be like massive Rube Goldberg devices (e.g., S(qm) triggers a video camera to shut off a car, which traps a chicken inside, which kills the chicken and release noxious fumes, which float into the atmosphere, which deflect photons back into space, which eventually get sucked into a black hole, etc.”). The problem is, no matter how Byzantine one made his Goldberg cosmos, it would still be intrinsically formally-indeterminate, since it could suddenly advert to running an incompossible somewhere down the spatiotemporal road (“amplified grueness”). For that matter, the cosmos could collapse and cease to be—would we then be justified in saying any formal functions also ceased to exist? Purely formal functions cannot ever advert to running a different function, nor can they be limited to a subset of their instances. Hence, even the cosmos as a purely physical system is intrinsically indeterminate in a way formal operations cannot be. If the human mind is a purely physical system, it follows that our minds are always just as intrinsically indeterminate with respect to cases of formal truth, which means we never actually and formally-determinately perform pure functions, which is absurd. That is Ross’s argument.


  28. Hello Ed and TC,

    I've looked at the PhilosphyForums pages but none of the commenters there appears to address the point I'm making. Let me try again. Ross says

    "No physical process is so definite as to determine among incompossible abstract functions that one rather than another is realized, and thus to settle for every relevant case what the "outcome" is to be."

    If this were true surely it would be easy for him to give us an example of a physical process and two distinct abstract functions of which it was a realisation? My claim is that the examples he gives simply don't cut the mustard. I will quote his first example:

    "For instance, if the function is x(*)y = (x + y, if y < 10^40 years, = x + y + 1, otherwise), the differentiating output would lie beyond the conjectured life of the universe."

    Isn't y one of the two arguments to a putative adding device? What has it to do with time? My *guess* is that he's suggesting that if the machine has been working for less than 10^40 years then it outputs x+y, and if for more than 10^40 years then it outputs x+y+1. But this is just to say that for the first 10^40 years of its life the device realises the perfectly determinate function f(x,y)=x+y and ever after that the equally determinate function g(x,y)=x+y+1. The machine simply undergoes a change.

    My challenge to you both is to come up with a convincing example of Ross's 'indeterminacy of the physical'.

  29. Mr. Brightly:

    I would like to take a quick stab at this. Elsewhere, Crude has suggested the example of a calculator that can only add whole numbers, with a maximum output value of 2. As he pointed out, the same exact machine could instead indicate which kind of meal--carnivore, vegetarian, or vegan--to serve to two guests, depending on their individual preferences. Here is how the different interpretations map to each other:

    0 = carnivore
    1 = vegetarian
    2 = vegan

    No doubt we could come up with infinitely many alternative schemes. Thus the meaning of the inputs and outputs--i.e., the abstract function of the device--is indeterminate from its physical structure alone.

  30. David:

    A formally determinate function can't undergo a change. (Can addition per se "become" quaddition?) But a physical instance (even a ginormous series of cases) of a function can undergo a change. Ergo, formal operations and physical systems are not identical. Further, pure functions, in every single case, "capture" every possible case at the exclusion of any other incompossible function. But a physical simulation of a function, in a single case, is always amenable to numerous contrary, variant function (points on a curve, etc.). Therefore they fundamentally differ. Physicalism is pressing for the IDENTITY of all formal operations (and everything else) and physical systems. But clearly the two domains have different IDENTITY CRITERIA; ergo, physicalism is false.


  31. Hello A and TC,

    A: We may be using the word 'function' in slightly different ways. Ross talks of 'abstract functions' and gives examples based on addition so I take him to be using 'function' in the mathematical sense. I agree that we can assign unlimited arbitrary meanings to a machine's inputs and outputs but the same underlying pattern remains and it's this pattern that the abstract function (addition, for example) captures. The question is Can the machine be a realisation of two incompatible patterns? Ross says Yes, I say No.

    TC: I agree that (math) functions are not the kind of thing that undergo change---they are descriptive elements, like colours, through which we can express change, just as an apple ripens from green to red, say. But I don't think Ross (or myself) are *identifying* functions with physical objects---the former being abstract and the latter concrete. The question is about what it means for a concrete thing to be a *realisation* of an abstraction. For me, Ross's claim is rather like saying something can be green all over and red all over at one time.

  32. DB:

    Ross talks of 'abstract functions' and gives examples based on addition so I take him to be using 'function' in the mathematical sense.

    He also gives the examples of modus ponens and conjunction, which are not mathematical; he is using "function" in the (broader) formal sense.

    I agree that we can assign unlimited arbitrary meanings to a machine's inputs and outputs but the same underlying pattern remains and it's this pattern that the abstract function (addition, for example) captures.

    No, it does not. The pattern correponds to an infinite number of different functions, addition being only one of them. We have to assign meaning to the pattern in order for it to represent a specific function by simulating it.

    The question is Can the machine be a realisation of two incompatible patterns? Ross says Yes, I say No.

    Ross is talking about formal functions, not physical patterns. The two are not the same--formal functions are always incompossibly determinate, but physical patterns never are; at least, not in the same way.

    The question is about what it means for a concrete thing to be a *realisation* of an abstraction. For me, Ross's claim is rather like saying something can be green all over and red all over at one time.

    Actually, it is more like saying that something can be green all over and grue all over at the same time--which it can, if we define grue as "green until 01/01/2011, and blue thereafter."

  33. aletheist:

    Thank you for addressing the "red and green" point. David B., the point is that a physical system CAN be construed as formally "red and green" (and grue and gru*, etc.) at the same time. Nothing about the physical outputs themselves dictate which of the countless "competing" incompossible functions it is actually and exclusively "running." In contrast, no formal function can be metaphysically 'dichotomous' (let alone tri-, quadri-, omnichotomous) like that, so the formal is determinately instantiated in a way the purely physical cannot provide.


  34. Hello A,

    he is using "function" in the (broader) formal sense. Yes, he does. But my beef is with his section II where he says "... if the function is x(*)y = (x + y, if y < 10^40 years, = x + y + 1, otherwise)..." and "...the function to show the indeterminacy of the single case: quus, symbolized by the plus sign in a circle, is defined by: x ⊕ y = x + y, if x, y < 57, =5 otherwise". This is unmistakably mathematical function talk.

    Ross is talking about formal functions, not physical patterns. So am I, if by *physical* pattern you mean something like a piece of wallpaper. My understanding is that we are talking about the *abstract patterns* of which concrete objects are realisations, eg, symmetry groups of wallpaper

    I'm happy to accept that something can be green and grue at the same time. At 7/1/2010 green and grue are quite compatible. But Ross claims that a physical thing can (indeed must) realise an infinity of *incompatible* abstractions. I'd be delighted to see a clear example of just two!

    Perhaps I could ask Ed to say a little more about what he thinks Ross is driving at in this section?

  35. DB:

    Let me try another illustrative example, this time in the form of a riddle of sorts.

    Suppose that you come across a device with only two buttons, labeled "J" and "N". You press "J", and nothing happens. You press "J" again, and suddenly the device spits out a small piece of paper with "J" on it. You press "N", and nothing happens. You press "N" again, and out comes "N" on a piece of paper. You press "J", then "N", and this time you get a paper with "S" on it. Pressing "N", then "J", also produces a paper with "S" on it. What is the device doing?

    Without changing its physical characteristics or behavior in any way whatsoever, I can suggest at least five different (and equally viable) abstractions of what the device is "really" doing. No doubt there are infinitely many others. It all depends on what meaning we assign to the inputs and outputs--something that the device itself cannot do.

    Take a stab at it, and I will come back and post my answers later. :-)

  36. A: One way we might describe your machine is to say that it translates sentences of length two from an alphabet of two symbols into sentences of length one from an alphabet of three (possibly four) symbols:
    There's an ambiguity concerning the translation of AB and BA: is it an essential part of the machine that they map to the same symbol? We can only resolve this by looking inside.

  37. No takers? I must admit to being a bit disappointed, although I suppose that I should not be surprised, given that the original blog post with which these comments are associated is nearly a month old. Anyway, here are my answers:

    1. Assign "J"="J", "N"="N", and "S"="S". F(x,y) = x, if x=y; = "S" otherwise.
    2. Assign "J"=1, "N"=2, and "S"=3. F(x,y) = x, if x=y; = x+y otherwise.
    3. Assign "J"=1, "N"=2, and "S"=3. F(x,y) = x, if x=y; = x*y+1 otherwise.
    4. Assign "J"=1, "N"=3, and "S"=2. F(x,y) = (x+y)/2.
    5. Assign "J"="All A is B", "N"="All B is C", and "S"="All A is C". F(x,y) is a syllogism.

    There is nothing inherent to the symbols "J", "N", and "S" that rules out the meanings assigned to them in 2-5 and the corresponding (incompatible) interpretations of what the device is (formally) doing; i.e., it is incompossibly indeterminate.

  38. A: I guess our last comments crossed in the post. I'm not saying that the relation of the inputs to outputs of your machine uniquely determines how the machine derives the outputs from the inputs. Your examples clearly refute this. We can agree that the workings of the device are indeed indeterminate *from this relation alone*. What I am saying is that the physical structure of the machine fixes this relation and that this structure (or enough of it) is knowable, in the ordinary sense in which we 'know' the physical world, and that we can therefore know what this relation is. Ross has not explained how this leaves room for any indeterminacy in how the machine works and I don't see how there can be.

    Ross mentions certain results of Quine, Kripke, et al, but doesn't develop these ideas for us. Rather, he asserts the thesis of the 'indeterminacy of the physical' and attempts to bolster it with a couple of examples. As I've urged in earlier comments here, these, in my view, do not demonstrate what he needs to show. One example, in particular, is incoherent as it stands (has nobody else noticed this?) and needs restating before it can be criticised (see here). Later in the same section he twice misquotes the Newtonian formula d=1/2gt^2 for distance travelled from rest under constant acceleration. I'm afraid these blunders, plus the lack of a recognisable argument, don't lend this reader confidence in his thesis.

    I'll go out on a limb perhaps and say where I think he goes wrong. He sees an analogy between, say, Quine's thesis of the indeterminacy of translation and his own thesis of the indeterminacy of the physical. Quine argues that there is no fact of the matter as to what a native speaker means by 'gavagai'. All we can do is observe that he utters this in certain circumstances. Ross notes, correctly, that the workings of a calculating device are similarly undetermined by observation of its inputs and outputs. What he ignores is that, though we may be unable even in principle to obtain a law-based account of the relation between a speaker's mental processes and the goings-on in his brain, the analogy fails for the calculating device. For the gadget operates entirely on the physical level, and if we open it up and analyse its structure we can understand what it's doing and how it works against our background knowledge of physical processes. I appreciate that this criticism may be doing Ross a serious disservice. He may well do better in his book-length treatment of the topic, and I may have grossly misunderstood him anyway. But so far nothing that Ed has said here or in his earlier references to Ross, or anyone's comments here, has set me right. Could I also add that I'm not coming at this as a dogmatic materialist, though I may sound like one. It may well be true that thought is immaterial. It's just that I don't think Ross has made a good case for concluding this.