Schliesser on the Evolutionary Argument against Naturalism

I commented recently on the remarks about Thomas Nagel’s Mind and Cosmos made by Eric Schliesser over at the New APPS blog.  Schliesser has now posted an interesting set of objections to Alvin Plantinga’s “Evolutionary Argument against Naturalism” (EAAN), which features in Nagel’s book.  Schliesser’s latest comments illustrate, I think, how very far one must move away from what Wilfred Sellars called the “manifest image” in order to try to respond to the most powerful objections to naturalism -- and how the result threatens naturalism with incoherence (as it does with Alex Rosenberg’s more extreme position).

The EAAN

First let me summarize Plantinga’s EAAN, which I think does pose a powerful challenge to naturalism, though I don’t think it shows quite what Plantinga thinks it does.  (Plantinga’s most recent statement and defense of the argument can be found in Where the Conflict Really Lies, which I recently reviewed for First Things.)

The EAAN begins by noting that what natural selection favors is behavior that is conducive to reproductive success.  Such behavior might be associated with true beliefs, but it might not be; it is certainly possible that adaptive behavior could be associated instead with beliefs that happen to be false.  In that case, though, there is nothing about natural selection per se that could guarantee that our cognitive faculties reliably produce true beliefs.  A given individual belief would have about a 50-50 chance of being true.  And the probability that the preponderance of true beliefs over false ones would be great enough to make our cognitive faculties reliable is very small indeed.

Now if evolution is only part of the story of the origin of our cognitive faculties, this is not necessarily a problem.  For example, if there is a God who ensures that the neurological processes generated by natural selection are generally correlated with true beliefs, then our cognitive faculties will be reliable.  But suppose that, as naturalism claims, there isn’t more to the story.  Then for all we know, our cognitive faculties are not reliable.  They may be reliable, but we will have no reason to believe that they are, and good reason to believe that they are not.  Now that means that we also have good reason to doubt the beliefs that are generated by those faculties.  For the naturalist, that will include belief in naturalism itself.  Naturalism, then, when conjoined with evolution, is self-defeating.  Evolution, concludes Plantinga, is thus better interpreted within a non-naturalistic framework.  

I think the basic thrust of this argument is correct, though I prefer the related argument that generally goes under the name of “the argument from reason” and has been defended in different versions by Karl Popper, Victor Reppert and William Hasker, and which I endorsed in Philosophy of Mind and The Last Superstition.  For one thing, I don’t think the basic point of the argument has anything to do with weighing probabilities, so that Plantinga’s tendency to state the argument in probabilistic terms needlessly muddies the waters somewhat.  The key point is rather that the logical relations that hold between thoughts cannot in principle be reduced to, supervenient upon, or in any way explained in terms of relations of efficient causality between material elements.  See the post on Popper just linked to for a summary of the argument as I would state it.

I also think that it is a mistake to suppose that the EAAN gives direct support to theism, specifically -- as opposed, say, to a non-theistic teleological view of the world (such as Nagel puts forward in Mind and Cosmos).  In Where the Conflict Really Lies, Plantinga acknowledges (rightly, in my view) that design inferences of the sort associated with William Paley and “Intelligent Design” theory do not constitute strong arguments for theism.  But he suggests reinterpreting the tendency to see design in complex biological phenomena as a kind of “perception” rather than an inference or argument.  Just as you can perceive that someone is angry from the expression on his face, so too, Plantinga suggests, can you perceive that an organ was designed from the order it exhibits.  And just as the former perceptual belief is rational despite its typically not involving an inference or argument, so too is the latter rational even if it does not involve an inference or argument.

There’s a lot that could be said about this, but the most important thing to say is that it is simply too quick.  As any Aristotelian can tell you, it is one thing to attribute a function to something, but quite another to attribute design to it.  That roots have the function of anchoring a plant to the ground and taking in nutrients may well be something we just perceive on close examination.  But that is precisely because having such functions is of the nature of roots -- something built into them, as it were.  In that respect they are very different from an artifact like a watch, whose metallic parts do not have a time-telling function built into them by nature.  That function has to be imposed on them from outside, which is why a watch requires a designer.  But precisely because natural objects are not artifacts, to perceive functionality or order in them is not ipso facto to perceive design.  And that means that while Plantinga’s EAAN and defense of the rationality of “perceiving” functionality in nature strike a blow against the naturalist’s dogmatic rejection of teleology, they do not by themselves constitute reasons to embrace theism, specifically.  (For more on the distinction between function and design, see this post, this post, this article, and other earlier posts dealing with the difference between an Aristotelian-Thomistic conception of nature and “Intelligent Design” theory.)

That is not to say that a divine intellect is not ultimately responsible for the order of things.  But for the Aristotelian (and for Thomists, who build on an Aristotelian foundation) that is a claim which certainly does require an argument, and an argument which does not conflate function and design, as too many Christian apologists have done at least since the time of William Paley, but which Aquinas’s Fifth Way -- though often mistakenly assimilated to Paley’s argument -- does not.  (I’ve defended the Fifth Way in several places, including in Aquinas.)

The bottom line is that what the EAAN/”argument from reason” shows, in my view, is that we cannot coherently trust our cognitive faculties unless we suppose that they are directed toward the attainment of truth as their telos or end.  But this does not by itself entail any extrinsic, artifact-like teleology of the Paleyan sort.  One could opt instead for an immanent teleology of the Aristotelian sort (and then try to resist a Fifth Way-style argument to the effect that this sort of teleology too ultimately requires a divine cause).  This is, in effect, Nagel’s strategy.

Schliesser’s response to the EAAN

Let’s turn now to Schliesser’s remarks.  Do read his entire post (which, never fear, isn’t as verbose as mine often are) in case I have missed any important elements of the context in interpreting the passages I’ll be quoting from it.  Schliesser begins as follows:

[L]et's grant -- for the sake of argument -- the claim [made by Nagel, following Plantinga] that "Mechanisms of belief formation that have selective advantage in the everyday struggle for existence do not warrant our confidence in the construction of theoretical accounts of the world as a whole."  What follows from this?

My quick and dirty answer is: nothing. For the crucial parts of science really do not rely on such mechanisms of belief formation.  Much of scientific reason is or can be performed by machines; as I have argued before, ordinary cognition, perception, and locution does not really matter epistemically in the sciences. 

End quote.  If I understand him correctly, what Schliesser is saying here is that even if the EAAN casts doubt on the reliability of our cognitive faculties (given naturalism), that is irrelevant to the question of whether science is reliable, for what is crucial to science can be done by machines, and the EAAN does not cast doubt on their reliability.  He also writes:

[Nagel] thinks that somehow there are "norms of thought which, if we follow them, will tend to lead us toward the correct answers" to "factual and practical" questions… Now… if this claim is true, it is utterly unsubstantive--none of the non-trivial results in physics or mathematics are the consequence of following the norms of thought. (I realize that there is a conception of logic that treats it as providing us with the norms of thought, but even if one were to grant this conception, it does not follow one obtains thereby mathematical or scientific results worth having.)

Schliesser’s point here, I think, is that the substantive results of science are not arrived at mechanically via the simplistic application of a set of more or less commonsense rules of the sort one finds in a logic textbook.  That is to say, scientists don’t proceed by saying: “OK, now let’s take the traditional Laws of Thought, the valid syllogism forms, inference rules like modus ponens, etc. and start cranking out some implications from what we’ve observed.”   Scientific practice is far more complex than that, especially insofar as it involves the use of computers following algorithms very unlike the patterns of reasoning we rely on in ordinary life.  Hence (again, if I am reading Schliesser correctly) if the EAAN shows that ordinary patterns of reasoning are unreliable on the assumption of naturalism, that is irrelevant to the reliability of science insofar as it does not rely on these patterns anyway.  Continuing in this vein, Schliesser writes:

Okay, let's assume -- for the sake of argument -- that it matters that humans are engaged in scientific practices that generate the building blocks of theoretical accounts.  In most of these the ordinary or average products of Darwinian evolution as such are not allowed near the lab.  In fact, the ordinary or average products of primary, secondary, and university education are also not allowed inside the lab.  Insanely high "achievement" over, say, twenty years of human capital formation is required before one becomes a little cog in the collaborative, scientific enterprise. (It's likely, in fact, that such achievement may just be a consequence of being a relatively rare freak of nature--a "monster" in eighteenth century vocabulary.) Parts of this achievement undoubtedly takes advantage of our selected for cognitive capacities and, perhaps, enhances these in subtle ways.  A large art of this achievement is the actual unlearning -- or generating the capacity for temporary disabling -- lots of our avarage Darwinian programming.  Moreover, much of the unlearning takes place after one's formal education is complete and inside the lab, where one's cognitive capacities are transformed into engagement with particular model organisms and particular specialized techniques. One does not need to accept all of Foucault, to see that the disciplining of scientific agents is as much an enhancement of human nature as a battle with pre-existing nature. So, "in science" our "cognitive capacities" are not used "directly." (Moreover, in so far as any human perception takes place in the epistemic processes of science much effort and skill is directed at making it entirely trivial.)

End quote.  Here I take it that Schliesser’s point is that the cognitive tendencies hardwired into us by natural selection are unlearned in the process of scientific training and practice -- the whole point of science being, as it were, to replace the “manifest image” that our natural cognitive tendencies generate with the “scientific image” (again to allude to Sellars) -- so that it doesn’t matter if those cognitive tendencies are unreliable.  

So, as I read him, the reliability of the cognitive tendencies put into us by natural selection is in Schliesser’s view irrelevant to the practice of science -- and thus to the defensibility of naturalism, which regards the scientific description of the world as either exhaustive or at least the only description worth bothering with -- for two reasons.  First, the relatively few human beings actually involved in scientific practice in a serious way do not rely on the cognitive tendencies in question in the first place, but seek precisely to resist and replace them.  And second, the modes of cognition they are engaged in can be carried out by machines anyway, which don’t have any hardwired human cognitive tendencies to resist.  So the EAAN fails, because it falsely supposes that it is the reliability of those hardwired human cognitive tendencies that naturalism presupposes.

Schliesser on our cognitive faculties

What should we think of all this?  Let’s consider first the claim that scientific practice involves radically moving away from our hardwired cognitive tendencies and their deliverances.  There is of course much truth in this, and I think Schliesser is right to suggest that any criticism of naturalism that does not factor it in is superficial.  However, this by no means suffices to disarm the EAAN.  

To see why, consider a couple of analogies.  Suppose you criticized a portrait or landscape artist for his poor drawing ability and he responded: “Drawing?  I don’t need no stinking drawing!  I’m a painter!  Hell, I haven’t done a complete line drawing since I was in school, and I rarely if ever even sketch out my subject before getting out the paints.  No, it’s all in the brushwork.  Obviously you don’t understand what we artists do.”  Or consider a dancer who suggested that the physiology of ordinary walking was irrelevant to understanding what she does, since she has over the course of many years had to acquire habits of movement that go well beyond anything the ordinary person is capable of, and even to unlearn certain natural tendencies.  (Think e.g. of the unusual stress a ballet dancer has to put on the foot, or the need to overcome our natural reluctance to move in ways that would for most people result in a fall.)  

The problem with such claims, of course, is that the fact painting or dancing involve going well beyond, and even to some extent unlearning, certain more basic habits does not entail that those habits are entirely irrelevant to the more advanced ones or that they can be entirely abandoned.  On the contrary, the more advanced habits necessarily presuppose that the more basic ones are preserved at least to some extent.  Even if drawing constitutes a very small part of producing a certain painting, and even if no sketch in pencil were made prior to getting out the paints, a painter without skill in drawing is going to produce a bad painting.  (The point has nothing to do with realism, by the way; a good painting done in the surrealist, impressionist, pointillist, or cubist style also presupposes the skills involved in drawing.)  Dancers have to have at least the muscles, bones, comfort with one’s body, ease of movement, etc. that are involved in ordinary walking even if they must also have much more than that.  The skills involved in ordinary drawing and walking constitute a framework for the more advanced skills, a framework that can be so covered over and modified that it may go virtually unnoticed in the course of painting or dancing, but which nevertheless cannot in principle be altogether abandoned.

Now by the same token, the ordinary patterns of reasoning as familiar to common sense as to the professional logician -- modus ponens, disjunctive reasoning, conjunctive reasoning, basic syllogistic reasoning, basic arithmetic, etc.  --  are, as Schliesser implies, a “trivial” part of science, but only in the sense that being able to walk over to the barre is a trivial part of being a ballet dancer, or the ability to draw a line or circle is a trivial part of being a painter.  While being able to walk over to the barre is obviously very far from sufficient for being a good ballet dancer, it is nonetheless absolutely necessary for being one; and while being able to draw a line or circle is obviously very far from sufficient for being a good painter, it too is still absolutely necessary for being one.  Similarly, while having the ability to reason in accordance with modus ponens, basic arithmetic, etc. is very far from sufficient for being able to do serious science, it is still an absolutely necessary condition for doing it.  The reason scientists don’t make a big deal of these “norms of thought” is the same reason ballet dancers don’t make a big deal out of their ability to walk and painters don’t go on about their skill in holding a pencil.  It is not that basic inference rules, walking, and drawing are irrelevant to science, dancing, and painting, respectively; it is rather that their relevance is so blindingly obvious that it goes without saying.

As Hilary Putnam pointed out in Representation and Reality, if you are going to call “folk psychology” into question -- which is what Schliesser is essentially doing (at least in the context of scientific practice, if not in other contexts) -- then you are going to have to call “folk logic” into question as well.  But we have nothing remotely close to an account of how this can coherently be done.  However far removed from ordinary cognition scientific modes of reasoning might be, they will presuppose fundamental logical notions like truth, consistency, validity, and the like, and our ability to recognize them when we see them.  And that means that they will presuppose the very abilities that even uneducated, untrained, pre-scientific “folk” possess.  (The fact that such “folk” sometimes make basic, systematic logical errors doesn’t change anything.  Pointing out to undergraduates that “This inference seems valid, but it is not” requires that they be able to see validity somewhere, and in particular in the argument that tells them that the inference in question is not really valid after all.)

Something similar is true of our perceptual faculties, which modern physics (with its account of the world as made up of colorless, odorless, soundless, tasteless particles etc. -- think of Eddington’s two tables)  might seem to have moved beyond altogether.  That this cannot be the case is obvious from the fact that physical theory, in the name of which perception is said to be misleading, is itself empirically based and thus grounded in perception.  Science can supplement or correct what perception tells us, but it cannot coherently deny the reliability of perception wholesale.  That it is at the very least difficult to see how it could coherently do so has, as I have noted in several places (e.g. here), been noticed by a number of thinkers from Democritus to Schrödinger.

We might also note that the degree to which the actual practice of science really does involve moving beyond ordinary modes of cognition is itself a matter of controversy (as the work of thinkers like Michael Polanyi illustrates); and that equally controversial is the question of whether the methods of physics really do reveal to us the whole nature of objective material reality in the first place.  Nor need one take a purely instrumentalist view of physics to doubt that they do.  To appeal to an analogy I’ve used in earlier posts, when aircraft engineers determine how many passengers can be carried on a certain plane, they might focus exclusively on their average weight and ignore not only the passengers’ sex, ethnicity, hair color, dinner service preferences, etc., but even the actual weight of any particular passenger.  This method is very effective, and is effective precisely because it captures real features of the world, but it hardly gives us an exhaustive description of airline passengers.  Similarly, the methods of physics, which focus on those aspects of a system that are susceptible of prediction and control and thus abstract away aspects which cannot be modeled mathematically, are extremely effective, and effective precisely because they capture real features of the world.  But it simply does not follow that the description of physical reality they afford us is exhaustive, any more than the engineer’s description is exhaustive.  And thus the fact that that description is radically different from the picture afforded by perception does not entail that it falsifies the latter.  To assume otherwise is (as I have noted before) to commit what Alfred North Whitehead called the “Fallacy of Misplaced Concreteness.”  

In any event, whether we think our ordinary, pre-scientific perceptual and rational faculties are unreliable to only a minor extent or to a significant extent, we cannot coherently regard them as fundamentally unreliable.  And that they are fundamentally reliable is all the EAAN requires.  Even science at its most rarefied presupposes that at some level our senses tell us the truth in a systematic way, and that basic arithmetic, modus ponens, conjunctive reasoning, etc. are valid modes of inference.  EAAN claims that naturalism is inconsistent with this presupposition, and nothing Schliesser has said shows otherwise.

Schliesser on machines

But couldn’t Schliesser now appeal to the suggestion that the role of human beings in science is irrelevant anyway, since what they do could just as well be done by machines?  

No, one reason being that the machines in question must be designed and constructed by human beings -- they don’t grow on trees after all!  That means that, however it is they get the results they do, the machines will be reliable only if the cognitive faculties of those who designed and constructed them --namely, human beings -- are reliable.  (Nor would it help to suggest that machines that were constructed by other machines rather than by us wouldn’t face this problem; for the machine-constructing machines, or their ancestors anyway, would have been constructed by us, so that the problem is only pushed back a stage or several stages.)

But a deeper problem is that however they get here, machines in fact cannot carry out the cognitive tasks associated with scientific reason.  What they can do is merely serve as instruments to assist us as we carry out those tasks, as telescopes, microscopes, electrometers, scales, slide rules, pencil and paper, etc. do.  Schliesser is essentially taking for granted the computationalist theory of mind, on which cognitive processes in general are computational processes (in the sense of “computational” associated with modern computer science), so that they could be carried out by a machine as well as by us.  The “machine scientists” Schliesser is describing would accordingly be characterizable in terms of a kind of “android epistemology,” or perhaps in terms of what Paul Thagard calls “computational philosophy of science.”  But you don’t have to be an anti-naturalist to think that this whole idea is wrongheaded.  You just have to “get your Searle on,” as it were.

I am alluding here not to John Searle’s famous Chinese Room argument, but to the less well-known but more penetrating argument of his paper “Is the Brain a Digital Computer?” (restated in The Rediscovery of the Mind), according to which computation is not intrinsic to the physics of a system, so that it makes no sense to regard anything as carrying out a computation apart from the designers and/or users of a system who assign a computational interpretation to its processes.  (I discuss Searle’s argument in more detail in the post on Popper linked to above, since it is related to Popper’s argument.)  Saul Kripke has presented a similar argument, to the effect that there is nothing in the physical properties of any machine that can determine precisely which program it instantiates.  Any set of processes could, as far as their inherent physical properties alone are concerned, be interpreted either as the carrying out of one program or as a malfunction in the carrying out of some different program.  (I’ve discussed this argument too in greater detail in another earlier post.)

Now I think that it is in fact too strong to conclude on the basis of Searle’s, Popper’s, or Kripke’s arguments that there is nothing like computation inherent in physical processes, full stop.  The correct thing to say is rather that there is nothing like computation inherent in physical processes given an essentially materialist, anti-teleological conception of the physical.  However, if we allow that there is teleology of a broadly Aristotelian sort immanent to physical systems, then (as I’ve noted in earlier posts like this one and this one) we can make sense of the idea that certain physical systems are inherently directed toward the realization of this computational process rather than that one.  And if Nagel’s brand of naturalism is correct (though of course I don’t myself think it is), then such teleology can be made sense of without reference to a divine cause.  But what we would be left with in such a case is precisely Nagel’s form of naturalism -- a form that acknowledges the force of the EAAN and affirms teleology so as to get around problems of the sort the argument raises -- and this can hardly help to salvage Schliesser’s objection to the EAAN.

And of course, even if some computational processes are inherent to nature, that wouldn’t include those exhibited by the machines we use in our scientific endeavors, which are man-made and have only a derived teleology and thus a derivative status as “computers.”  We would be the true computers, with the machines serving as mere enhancements to our computational activity, just as binoculars enhance our vision but do not themselves see anything.  The reliability of the machines’ processes would, again, presuppose the reliability of our cognitive processes; and if the reliability of the latter is grounded in immanent teleology, then the force of the EAAN has been conceded and Nagel’s position will have been embraced rather than rebutted.

The bottom line is that we cannot altogether get outside our cognitive skins, even if we can modify, supplement, or even eliminate parts of those skins.  Schliesser’s position seems to suppose otherwise insofar as it implies that we could coherently practice science and accept its results while simultaneously denying the reliability of our cognitive faculties.  In fact, however we spell out the details of their relationship, Sellars’ “scientific image” is ultimately a part of the “manifest image” itself, so that the former cannot coherently be appealed to as a way of undermining the latter.  To quote Putnam quoting William James, “the trail of the human serpent is over all” -- or if not over all, then at least all over science, which is no less essentially human a practice than dancing, painting, machine-building, or philosophizing are.