Wednesday, October 23, 2013

C.S. Lewis's Argument against Naturalism, part 6

In earlier posts in this series I presented C.S. Lewis’s argument from reason, G.E.M. Anscombe’s objections to it, and my response to Anscombe. In this post I’ll go over Lewis’s response to Anscombe

The Second Edition of Miracles
Lewis changed the title of the third chapter of Miracles, from “The Self-Contradiction of the Naturalist” to “The Cardinal Difficulty of Naturalism.” His revision of the argument from reason deals with several of Anscombe’s criticisms, such as his conflation of the nonrational with the irrational (which he accommodates by simply substituting the former term for the latter), and the paradigm case argument. His primary revision deals, appropriately, with Anscombe’s primary criticism: the claim that explaining a belief in terms of grounds or causes are two distinct types of explanations that are not in competition with each other.

Causes and grounds
As Lewis puts it, we use the word “because” in two different senses: to indicate a cause-effect relation (“He cried out because it hurt him”) and to indicate a ground-consequent relation (“It must have hurt him because he cried out”). The former is a dynamic connection -- his being hurt is what caused him to cry out -- while the latter is a logical one -- his crying out is our ground for believing that it hurt him.

He further emphasizes this by giving two illustrations of it: first, just as we can use the term “because” in two different senses, so we can use the word “follow” in two different senses. We can use it in a temporal sense (“B followed A”), which corresponds to the cause-effect relation; and we can use it in an eternal or logical sense (“B follows from A”), which corresponds to the ground-consequent relation. The first describes an accidental relation, while the second describes a necessary one.

The other illustration he offers is to point out that acts of thinking, i.e. inferences, are unique: they are about something other than themselves, and as such, can be either true or false. This is not true of other events, or even of other acts undertaken by a subject. Inferences, then, can be understood in two different senses: they can be seen as subjective physical, physiological, and psychological events in the brain (cause-effect), or as insights into something other than themselves (ground-consequent).

The problem comes in when we recognize that if a belief could be fully accounted for by a cause-effect relation, there would be no room left for the ground-consequent relation to play a role in reaching the belief. How could having or not having a ground-consequent relation have any bearing on the belief? It would be held regardless, because

… if causes fully account for a belief, then, since causes work inevitably, the belief would have had to arise whether it had grounds or not. … But even if grounds do exist, what exactly have they got to do with the actual occurrence of the belief as a psychological event? If it is an event it must be caused. It must in fact be simply one link in a causal chain which stretches back to the beginning and forward to the end of time. How could such a trifle as lack of logical grounds prevent the belief’s occurrence or how could the existence of grounds promote it?{1}

Thus, contra Anscombe, Lewis argues that these two types of explanation are in competition with each other. If the grounds of a belief have nothing to do with one’s coming to hold that belief, then the grounds are simply irrelevant. One would hold the belief whether it had grounds or not, because it has been caused.

Yet Lewis also recognizes that both relations need to apply to a belief in order for it to be valid. It needs a ground-consequent relation in order to be epistemically justified; without it the belief would not be derived from a reason, and so would not be valid. The ground-consequent category must be valid and accepted in order for inference and human knowledge to be valid and accepted. However, the belief also needs a cause-effect relation in order to take place at all; even if we ignore the principle of causality for the sake of argument, any belief that just spontaneously appeared in the mind without any cause could not make any claim to being valid, since it would not be derived from a reason.

In order to resolve this, Lewis argues that the ground-consequent relation and the cause-effect relation must coincide. They must be united into a single explanation in order for a belief to be valid. In other words, the ground of the ground-consequent relation must also be the cause of the cause-effect relation -- not merely by being the ground for it (because then every possible conclusion would be caused) but by being seen to be the ground for it. “If you distrust the sensory metaphor in seen, you may substitute apprehended or grasped or simply known. It makes little difference for all these words recall us to what thinking really is.”{2}

This “seeing to be” is essentially any truth-tracking element that connects the knowledge of something to that which is known. For Lewis, it is similar to the correspondence theory of truth: the degree to which a belief corresponds to what is known is the degree to which what is known is known. If the belief were totally explicable by something other than what is known, it does not qualify as knowledge. In the same way, “the ringing in my ears ceases to be what we mean by ‘hearing’ if it can be fully explained from causes other than a noise in the outer world -- such as, say, the tinnitus produced by a bad cold.”{3} Once we have factored the tinnitus out of the equation, whatever is left over is what qualifies as real, valid hearing. Similarly, valid knowledge is what is left over once we have factored out causes of a belief other than what is known (as a cause). Thus, any account of our reasoning capacities that does not provide for them to be connected to what is known is essentially an argument that no argument is valid, and is therefore self-refuting. It is comparable to saying, “I heard that everything we hear is produced by tinnitus.”

After this analysis, one could be forgiven for taking Lewis’s point to be that naturalism is somehow inconsistent with the correspondence theory of truth,{4} or with his account of how the ground-consequent and cause-effect relations apply simultaneously to the same belief. However, his argument is much more basic than this: naturalism is the view that everything can be accounted for by natural processes. But natural processes only provide cause-effect relations, never ground-consequent ones. This further explains Lewis’s rejection of Anscombe’s claim that more than one type of explanation can apply to the same belief: according to naturalism, naturalistic explanations are the only ones available. It is all well and good to argue that one type of explanation of a phenomenon does not rule out another, but only as long as one accepts both types of explanation. Lewis’s point is that naturalists do not. For the naturalist, natural explanations are the only ones allowed; and natural events progress according to cause-effect relations rather than ground-consequent relations. As such, naturalism cannot account for ground-consequent relations. Yet without them, no belief, including the belief that naturalism is true, could ever be epistemically justified, and could only be true by chance.

It is not enough simply to say that different “full” explanations can be given for the same event. Of course they can. … The question that is still open is the question of whether the kinds of mental explanations that must be offered in any face-saving account of rational inference are compatible with the limitations placed on causal explanations by materialism. If not, then there is a conflict between the existence of rational inference and materialism. This means that materialism refutes itself if it presents itself as a rationally inferred belief (or a belief that depends crucially on the existence of rational inference).{5}

The evolution of reason
Having reworked his argument, Lewis finds it necessary to rework his response to one of the possible objections to it: namely, that evolution can account for our reasoning processes to be (generally) valid. Those early human ancestors whose beliefs accurately corresponded to the world were those who stood a much greater chance of survival. So over time, those who reasoned more and more correctly passed on these capacities to their offspring more readily.

Leaving aside the assumption this scenario makes -- that a capacity for abstract thought would have a positive impact on an organism’s chances of surviving and producing offspring -- this requires us to believe that thoughts, before natural selection touched them, were not originally rational. They were, instead, merely subjective events, responses to stimuli. However, the relation between response and stimulus is not the same as the relation between knowledge and the truth known. An organism with a fully developed eye is not any closer to knowledge of light than an organism that merely has a light-sensitive spot. No improvement of a response to a stimulus could ever lead to genuine knowledge.

Perhaps adding more to the equation will help. In addition to a mere biological capacity, there is also experience and expectation, instruction and tradition. So our ancestors could learn from their experiences to expect things to be a certain way, and could then pass this gleaned information on to their descendants. They learned from experience that “where there’s smoke, there’s fire.”

But expectations like this may only be fulfilled accidentally. As such, they do not connect what is known to the knowing of it; beliefs formed in this way would not be epistemically justified. Reason and inference come in precisely when we look for the nature of the connection, to see if the fulfillment of an expectation is essential to it or not. Moreover, such a scenario could not apply to logical axioms. “My belief that things which are equal to the same thing are equal to one another is not at all based on the fact that I have never caught them behaving otherwise. I see that it ‘must’ be so.”{6}

A third potential escape route might be to reject truth and affirm pragmatism. Reason is merely useful as aid to practice, and is not meant for speculation. But of course, naturalism is the product of speculation. It goes far beyond our experiences, both individually and collectively, to say that the universe is all that exists. Nature is an abstraction, not something presented to the senses. Naturalism is something some people infer, not something some people experience or practice. Moreover, it is a universal negative, and a universal negative is not something that can ever be experienced.

Lewis concludes his rewrite of chapter three by contrasting the naturalist with the supernaturalist. The latter is guilty of many of the same things as the former: one cannot experience supernature anymore than nature. The supernatural is also an abstraction, going beyond our sensory experiences.{7} The difference between the two is this: first, the supernaturalist is not guilty of a universal negative, like the naturalist; and second, the supernaturalist is not advocating a worldview which calls the validity of such abstractions into question.

The paradigm case argument
What of Anscombe’s (and Antony Flew’s) criticism that by erasing the distinction between valid and invalid reasoning, Lewis has emptied these concepts of meaning? That he is posing a skeptical threat argument akin to radical skeptical claims like Descartes’s evil genie or Nozick’s brains in vats? Victor Reppert argues that Lewis’s argument in the second edition of Miracles is no longer in a skeptical threat format, but a best explanation format.

Neither the first edition argument nor the second is, I believe, a pure case either of a Skeptical Threat Argument or of a Best Explanation Argument. However, the earlier edition seems to correspond more to the Skeptical Threat argument than does the second. In the first edition we have the suggestion of an argument like this: “If thoughts are produced by irrational causes, then beliefs are likely to be false. What if all beliefs were produced by irrational causes? Then we would have no knowledge. And if Naturalism is true that has to be the story. So we’d better not accept Naturalism.” In the later edition we are simply told that in order for rationality to be possible two systems of connection, the Ground-Consequent system and the Cause and Effect system, must coincide, and this is possible only if naturalism is false. Thus the revised edition corresponds more closely to the Best Explanation Argument.{8}

In other words, the version of the argument from reason in the second edition argues what is necessary for our knowledge to be valid, and then shows that naturalism cannot meet these requirements. Thus, it is not presenting the absurdity that all of our beliefs might be invalid. If this were a problem (and I have argued in the previous post that it is not) it can be, and has been, resolved by reforming the argument to meet it. Reppert concludes, “It seems to me that Anscombe’s Paradigm Case argument is ineffective against this sort of argument.”{9}

Anscombe argued that, “if a man has reasons, and they are good reasons, and they are genuinely his reasons, for thinking something -- then his thought is rational, whatever causal statements we make about him.”{10} However, the argument from reason is precisely the claim that, if naturalism is true, these conditions do not hold.

1. A man could not have reasons, because his mental processes are dictated by purely natural processes that care not a whit for logic and rationality.

2. Even if a man could have reasons, he could not have good reasons. Whatever beliefs he reached would be brought about by cause-effect relations (such as association of ideas) because naturalism precludes the ground-consequent relation. Even if the beliefs were true, they would only be accidentally true.

3. Even if there were good reasons for a man to hold a belief, they could not be his reasons. They played absolutely no role in his coming to hold that belief, and if they did not exist, he would hold the belief anyway, since the natural processes responsible for his coming to hold any given belief would be operable regardless. As such, how exactly could any good reasons that might exist for the belief be rightfully called “his”?

Anscombe’s claim that reasons are not what bring about beliefs, but “are what is elicited from someone whom we ask to explain himself”{11} seems plainly false. At least some of the time, we arrive at a belief as a result of a reason. This is what we usually think reasoning consists of. We can, of course, distinguish reasons and causes in general; but to completely disconnect reasons from the occurrence of a belief is not only wrong, it is fatal for our claims to be reasoning beings.

Any adequate account of the relation between reasons and causes must provide an account of the role that convincing plays in our cognitive economy. The idea of being convinced by something seems to imply that reasons are playing a causal role. Anscombe is attempting not merely to distinguish, but to divorce reasons-explanations from causal explanations, considering the former to be noncausal explanations. And insofar as she is divorcing these types of explanations, her critique of Lewis is faulty. If reasons cannot be part of the explanation of how we come to hold beliefs as a matter of personal history, then human rationality as we ordinarily understand it simply does not exist.{12}

That Lewis thought the argument from reason survived Anscombe’s criticism is demonstrated by his inclusion of it in his final book, published posthumously:

No Model yet devised has made a satisfactory unity between our actual experience of sensation or thought or emotion and any available account of the corporeal processes which they are held to involve. We experience, say, a chain of reasoning; thoughts, which are ‘about’ or ‘refer to’ something other than themselves, are linked together by the logical relation of grounds and consequents. Physiology resolves this into a sequence of cerebral events. But physical events, as such, cannot in any intelligible sense be said to be ‘about’ or to ‘refer to’ anything. And they must be linked to one another not as grounds and consequents but as causes and effects -- a relation so irrelevant to the logical linkage that it is just as perfectly illustrated by the sequence of a maniac’s thoughts as by the sequence of a rational man’s.{13}

Anscombe herself thought that the second version was “much less slick and avoids some of the mistakes of the earlier one; it is much more of a serious investigation. … The argument of the second edition has much to criticize in it, but it certainly does correspond more to the actual depth and difficulty of the questions being discussed. … The fact that Lewis rewrote that chapter, and rewrote it so that it now has these qualities, shows his honesty and seriousness.” She even acknowledges that how the grounds of a belief are connected to its actual occurrence remains an unresolved problem.{14}

Several years after Lewis’s death, a “rematch” of sorts was arranged by the Socratic Club, with Anscombe defending her own position, and J.R. Lucas defending Lewis’s. When the smoke cleared, the argument from reason was still standing tall.{15} And perhaps the most ironic twist in the Lewis-Anscombe debate is that Anscombe’s husband, Peter Geach, seems to have agreed with Lewis.{16}

{1} C.S. Lewis, Miracles: A Preliminary Study, 2nd ed. (London: Collins, Fontana Paperbacks, 1960), 20.
{2} Ibid., 21, italics in original.
{3} Ibid., 22.
{4} Of course, according to Lewis’s argument, naturalism is inconsistent with the correspondence theory of truth. By disallowing the ground-consequent relation, naturalism disallows the connection between what is known and the knowing of it.
{5} Victor Reppert, C.S. Lewis’s Dangerous Idea: In Defense of the Argument from Reason (Downers Grove, IL: InterVarsity, 2003), 68-69.
{6} Lewis, Miracles, 2nd ed., 24.
{7} However, this does not mean that the effect of a supernatural event cannot be experienced with the senses. “Miraculous wine will intoxicate, miraculous conception will lead to pregnancy, inspired books will suffer all the ordinary processes of textual corruption, miraculous bread will be digested” (Lewis, Miracles, 1st ed., [London: Bles, 1947], 72; 2nd ed., 64).
{8} Victor Reppert, “The Lewis-Anscombe Controversy: A Discussion of the Issues,” Christian Scholar’s Review 19 (1989): 37 n. 21.
{9} Ibid., 39.
{10} G.E.M. Anscombe, “A Reply to Mr C.S. Lewis’s Argument that ‘Naturalism’ is Self-Refuting,” in The Collected Philosophical Papers of G.E.M. Anscombe, vol. 2: Metaphysics and the Philosophy of Mind (Oxford: Basil Blackwell, 1981), 229.
{11} Ibid.
{12} Reppert, C.S. Lewis’s Dangerous Idea, 65.
{13} C.S. Lewis, The Discarded Image: An Introduction to Medieval and Renaissance Literature (Cambridge: Cambridge Univ. Press, 1964), 165-66. For another post-Anscombe version of the argument, see A Grief Observed (1961; San Francisco: HarperSanFrancisco, 1989), 41.
{14} Anscombe, Collected Philosophical Papers, vol. 2, ix-x.
{15} J.R. Lucas, “The Restoration of Man,” Theology 98 (1995): 451.
{16} Peter Geach, The Virtues: The Stanton Lectures 1973-4 (Cambridge: Cambridge Univ. Press, 1977), 26-28, 48-53; Reppert, C.S. Lewis’s Dangerous Idea, 45 n. 1.

(cross-posted at Agent Intellect)

Discuss this post at the Quodlibeta Forum

Thursday, October 03, 2013

C.S. Lewis's Argument against Naturalism, part 5

An Analysis of Anscombe’s Criticisms
In the previous post I looked at G.E.M. Anscombe’s critique of C.S. Lewis’s argument that naturalism is self-defeating and culled six specific arguments from it. In this post, I’ll go over these arguments one by one.

Point 1: Conflating nonrational causes
Anscombe argues that Lewis had given examples of nonrational causes which lead to false beliefs in order to demonstrate that such causes are unreliable. It does not follow, however, that just because some such causes do so, all do. Indeed, we only know that his particular examples are problematic because we have observed them causing unreliable beliefs.

This point is correct. In order to move from the nonrational cause of a belief to the falsity of that belief, we would have to include another premise: namely, that beliefs caused by nonrational processes are false. This, however, is not the case: sometimes they are, sometimes they aren’t. But to simply point to the cause as nonrational does not by itself demonstrate that the belief is therefore false.

However, Lewis is not merely arguing that nonrational causes would lead to false beliefs, but to unjustified beliefs. Saying that some of our beliefs so caused may turn out to be correct does not address the point that this would be an accidental aspect of the belief instead of an essential one.

Anscombe’s point seems to be that we need an example of a physical process leading to a false belief before we can therefore conclude that that particular process is inconsistent with reasoning. This spills over into point 3, addressed below. For now, I will just point out that Lewis’s argument is that we need a foundation for our reasoning capacities that is unimpeachable. By showing examples of nonrational causes bringing about false beliefs, he is demonstrating that such causes are not unimpeachable, and thus cannot function as this foundation. It would be akin to finding a contradiction in basic logic or mathematics; such a discovery would demonstrate that logic or mathematics cannot be the foundation of our rationality, it does not provide us with the solid base upon which we can base our claims to knowledge. We would not conclude that just because mathematics only leads us to false beliefs occasionally that this does not prove to be much of a problem.

Point 2: Conflating nonrational with irrational
Antony Flew follows Anscombe in this criticism, and takes it to be a knockdown refutation of Lewis’s argument:

Lewis is too carefree in his talk of “rational” and “irrational.” Why must atoms, or systems of neurons, or whatever may be the terms of the scientific explanation of my mental processes, be either rational or irrational? Can they not be just non-rational -- things to which the rational/irrational distinction does not apply? Lewis would surely not say that atoms were immoral. But then, must they be moral? Of course not. Lewis would say the distinction did not apply. He would be quite right. In the same way, the rational/irrational distinction does not apply to the sort of things in terms of which “naturalists” would give their causal explanations of mental processes. But since atoms are neither rational nor irrational, the argument breaks down, for the causes by which the “naturalist” explains his own thinking are no longer irrational and the “naturalist” thesis no longer refutes itself. A chain of argument is as weak as its weakest link.{1}

It is certainly true that Lewis conflates irrational and nonrational causes in his treatments of the argument from reason; Arthur Balfour does as well. If Lewis’s argument were that a belief is irrational if it has an irrational cause, then this criticism would indeed refute it. But I do not think this is Lewis’s argument. He is, rather, arguing that in order for a belief to qualify as rational, it must have a rational cause. Therefore, any cause that is not rational, whether it be irrational or nonrational, would not lead to a rational belief. Even if the belief were true, it would not have been arrived at by a rational process. Thus, this objection is irrelevant; as long as nonrational causes are not rational causes, then beliefs caused by them would not be rational. In other words, Lewis is inferring the nature of the cause from the nature of the effect. If the effect is rational, the cause must be as well. Anscombe and Flew mistakenly think he is inferring the nature of the effect from the nature of the cause. They have it precisely backwards.

It should also be pointed out that Lewis had already noted the difference between an irrational and a nonrational cause elsewhere. A physical event is not rational in a different sense than a paralogism; it “does not rise even to the dignity of error.”{2} That is, it is not about anything, and so the appellations of truth and falsehood simply cannot be applied to it (although propositions about it obviously could). This reflects Lewis’s second form of the argument from reason (see part 1). So it seems that nonrational causes are in an even worse state than irrational causes. Far from refuting Lewis’s argument, appealing to the difference between irrational and nonrational causes increases the difficulty.

Point 3: The paradigm case argument
Lewis had argued that, because we can ask whether a particular belief is valid, we can also ask it of all of our beliefs taken together. However,

it is a complete mistake to think that if it is sensible to ask a question about a particular case of something (perceptions, pieces of thinking, etc.) it follows that it is sensible to ask the same question of all those particular cases taken as a class. It is wise and proper to ask of any given piece of reasoning, “Is this valid?” But it is not profound, but preposterous, to ask, with Lewis, “Is human reason valid?”: for some pieces of reasoning are, and some are not, sound.{3}

According to this objection, the claim that all reasoning is invalid is nonsensical, because “we could have the concept ‘valid argument’ only if we had drawn a contrast between valid and invalid arguments, and in order to do that we would have to come in contact with at least one instance” of each.{4} According to Flew, suggesting that all reasoning is invalid is like suggesting that all sensory perceptions are hallucinations. This would empty the concept of hallucination of all meaning, since it obtains its meaning by the contrast between it and a real perception.{5} This is very similar to certain skeptical problems, such as Descartes’s evil genie or Nozick’s brain in a vat.

Victor Reppert, a strong proponent of Lewis’s argument, thinks this objection hits home. He calls it a “Skeptical Threat Argument,” and he is “not optimistic” about them.{6} If Lewis’s argument is that

If naturalism is true, nonrational processes cause all of our beliefs.
If nonrational processes cause all of our beliefs, none of them are rational.
Naturalism is a belief.
 Belief in naturalism is not rational.

then this objection applies. A naturalist could simply respond that it is not necessary to show that our reasoning processes are rational, for the same reason that it is not necessary to show that we are not constantly being deceived by an evil genie, or that we are just brains in vats being manipulated to think that there is an external world. Such radical skepticism is refuted by the fact “that we have overwhelmingly strong reasons for acknowledging the ‘validity of reasoning’ -- that is, for acknowledging that people do sometimes reach conclusions because of good reasons they accept, and that they are rational in doing so -- and that, therefore, any argument to the contrary must be based on a mistake or trick of some kind.”{7} As such, “no absolute security against such doubts is available from any quarter, and … even if it were available it is not needed.”{8}

There are two responses that can be made to this. The first is that this objection does not apply to Lewis’s argument, and the second is that, even if it does, Lewis’s argument can be reformulated to meet it.

The first response was made by Ernest Gellner in his reply to Flew’s paper.{9} Gellner concedes that just because a question can be asked of a member of a class it does not follow that it can also be asked of the class itself. Yet the opposite is not true: that because a question can be asked of a member of a class, it cannot be asked of the class. Perhaps it can, perhaps it cannot. Flew’s example of hallucination is an example where it cannot. If all perceptions were hallucinations, the concept of hallucination would become meaningless, because its meaning is derived from its contrast with a real perception. This “contrast theory of meaning” is sometimes true, but we must beware of “the dangers of applying it indiscriminately.”{10}

The question is, does it apply to our beliefs? Gellner thinks not. The claim that if we can ask whether one belief is determined by nonrational causes, we can therefore ask it of all of them, is a perfectly coherent claim. This is because the concept of our beliefs being determined is not based on the contrast between determined and undetermined beliefs. Rather, it is based on “the presence of a causal mechanism; hence not by contrast but by a correlation that might be absolute and universal,” and “the tests for confirming this would not become unworkable by being applied to the totality.”{11}

Gellner thus makes two points: contrast is not the only way we can understand concepts in general; and contrast is not how we understand the concept of determinism in particular. Unfortunately, however, he has chosen the wrong target: the question is not whether we can understand determinism without reference to contrast, but whether we can understand the nonrational without reference to contrast. From an etymological standpoint, it looks as though we cannot: “nonrational” obtains its meaning from its contrast with “rational.” It is the negation or absence of the rational.

Another example may shed some light on this. As Flew writes, we would not call brute physical events moral or immoral because this distinction simply does not apply to them.{12} They are amoral or (to be consistent) nonmoral. Now the question is: does the concept of the nonmoral obtain its meaning from its contrast either with the moral or with the moral-immoral distinction? I am not convinced that it does. We are simply not applying the concept of morality to nonmoral events or entities. We understand what such entities are, and if someone were to ask whether they are moral or immoral, we would know that this concept has nothing to do with them. Our recognition that morality has nothing to do with nonmoral things does not increase or alter our knowledge of them.

Now let us return to Lewis’s argument. He states that nonrational processes could never give rise to rational processes. Here, he is clearly contrasting the nonrational with the rational. However, the question is not whether we can contrast the rational and nonrational, but whether the nonrational obtains its meaning from this contrast. Again, I am not convinced that it does. His second version of the argument from reason is that physical events are brute facts that are not “about” anything. The true-false distinction does not apply to them. So if they make up the whole of reality, the true-false distinction does not apply to anything. This is not only problematic; it is incoherent, since one would then have to ask whether it is true that the true-false distinction does not apply to anything.

Since the issue of contrast and meaning is likely to be controversial, let us grant for the sake of argument that we do need the contrast between the rational and nonrational in order to understand what we are talking about. Let us assume that it is exactly parallel to Flew’s analogy of hallucination. What then? Say that Lewis’s argument is that, if naturalism were true, it would lead to the conclusion that none of our perceptions are real, all are hallucinations. However, this is incoherent; a hallucination does not mean anything without the contrast of a real perception. Could not Lewis say, “Precisely! If naturalism were true, it would lead to an absurd conclusion like this. Therefore, naturalism is not true”?

Or put it in Anscombe’s terms: she argues that we need examples of valid and invalid reasoning in order to understand what we are talking about. Lewis’s argument, though, is that if naturalism were true, we would not have examples of valid reasoning. If we need an example of it in order to make sense of the concepts, and naturalism does not provide us with an example, it would follow that naturalism does not allow us to make a distinction between valid and invalid reasoning. Since it is obvious that we do have such examples and can make such a distinction, naturalism is false.

Let me this another way: the point to the skeptical threat argument is that the suggestion that our reasoning faculties are completely unreliable is like other radical skeptical claims, such as that we are brains in vats being made to think there is a physical world, or there is an evil genie who makes us add numbers incorrectly. Such claims are so outlandish that they need not be addressed. We do not feel threatened by them because we do not take them seriously.

But Lewis is not asking us to take radical skeptical claims seriously. He is telling us that the reason why we cannot accept naturalism is precisely because it leads to a radical skeptical claim. Of course we cannot seriously consider the possibility that all of our beliefs are invalid. That’s the point.

If it is suggested that radical skeptical claims are not merely incredible but incoherent, and therefore an argument cannot refer to them, the response would be that it is a valid argument to say that a position leads to an incoherent situation and should be rejected as a result. This is the very definition of a reductio ad absurdum argument. Lewis’s claim is that naturalism leads to an incoherent scenario in which none of our beliefs would be rational, and should therefore be rejected.

This leads us to the second response to this objection. Even if we grant that Lewis’s argument is invalid because of the paradigm case / skeptical threat objection, it can be reformulated to accommodate it. Reformulating the argument, however, is the subject of the following post in this series, and so will be left until then.

Point 4: Naturalism does not preclude logical explanations
By this, I take Anscombe to be saying that a person having a belief and expressing that belief and arguments for it in speech or writing is not inconsistent with naturalism. Which step does the naturalist deny? That people have beliefs? That people speak? These are both “natural” processes. Lewis seems to be arguing against a form of naturalism that no one actually holds.{13}

In the second form of the argument from reason, Lewis does question whether physical processes, which are not about anything, could ever give rise to processes that are, such as beliefs. As such, it is questionable whether naturalism can accommodate the occurrence of beliefs, since beliefs are about things. Even if he were to grant the possibility of beliefs in a naturalistic world, such beliefs would never be the result of rational processes. Given naturalism, if someone were to give a rational argument for a belief, that argument played no role in their coming to hold the belief in question. That is the problem naturalism presents for our knowledge.

To state this another way, it might be possible, in a naturalistic world, for someone to state a valid reason for a true belief, but it would be akin to a zombie giving a reason. The reason played no part in the formation of the belief, and it would be difficult to say that the zombie truly “believes” anything. As William Hasker writes,

… consider a possible world that is physically exactly similar to the present world, but in which the natural laws establishing psychophysical connections do not obtain. In such a world all the physical facts, and with them the entire physical course of events, are exactly as in the actual world: the complete absence of mentality makes no difference whatever. Similarly, we may consider a possible world physically identical with the actual world, but in which mental properties are redistributed in as bizarre a fashion as one might wish: this world is still indistinguishable from our own in all physical respects. Could there be a more dramatic demonstration of the fact that, given the closure of the physical, mental facts are irrelevant to the physical course of events? … The entire process [of reasoning] makes no sense at all, except on the assumption that a person’s awareness of reasons and her knowledge and application of principles of rationality make a difference to the conclusions that are accepted.{14}

I think Lewis would also argue that a naturalistic explanation could provide for everything in a belief except the meaning. For example, a poem can be analyzed as black marks on a white sheet of paper, but it is questionable whether such an analysis would ever arrive at the meaning of the poem.{16} To go back to one of Lewis’s examples, we could analyze a news broadcast in terms of the functioning of the television set. Yet if that was all there was to it, we would not be able to put any stock in the message it conveys.

Point 5: Back to Bulverism
Lewis had denounced those who dismiss beliefs allegedly derived from irrational sources. Yet his whole argument is dependent on the claim that in order for beliefs to be rational, they must have rational sources. However this is not how we judge whether an argument is valid. This criticism is a striking one, and on the surface seems perfectly justified. Lewis does appear to commit his own fallacy of Bulverism, i.e. the circumstantial ad hominem fallacy, when he writes that all people dismiss a belief once we know it is the result of a nonrational cause.

If the additional premise -- that nonrationally caused beliefs are invalid -- is accepted by Lewis’s opponents, then his argument still works. The Freudians and Marxists, for example, hold (according to Lewis) that one can refute a belief by showing it to have nonrational sources. As long as they still maintain that such beliefs are invalid, then their own beliefs fall along with them. Not all naturalists think that having a nonrational cause of a belief thereby invalidates it, though. In order to hold his ground, Lewis would have to demonstrate how having a nonrational cause for a belief makes that belief invalid. Lewis, however, would probably argue that he has demonstrated this. His argument is that in order for a belief to qualify as rational, it has to have a rational cause. So by definition, a nonrational cause would not allow the belief to be rational.

This, nevertheless, just avoids the problem: the formation of a belief has nothing to do with its validity. It simply does not matter whether it has rational, irrational, or nonrational causes. To dismiss a belief because of its causes is precisely the Bulverism fallacy. What matters is whether the belief can be demonstrated true upon further examination.

There are two responses to this: first, the Bulverism fallacy argues that we cannot determine a belief’s truth-value based on its cause. However, it has nothing to do with whether a belief is epistemically justified. Justification, warrant, or any truth-tracking element, connects a true belief to what makes it true -- and the claim of the argument from reason is that this connection is an inherently rational one. Therefore, a nonrational cause could never bring about an epistemically justified belief, regardless of whether that belief was true or not. Nonrational causes could, at best, bring about true belief. Yet all epistemologists acknowledge that we need more than just true belief in order to have knowledge. Insofar as a belief that is just accidentally true does not qualify as knowledge, nonrational causes can never bring about knowledge.

Second, the reason Bulverism is a fallacy is because we are able to take a belief out of its historical context (where it may have nonrational causes) and into a logical context. If this abstraction were impossible, the assessment of a belief independent of its causes could not be made. The only standards by which we could then judge the rationality of a belief would be the inherently nonrational standards of how the belief was caused.

To put it in Flew’s terms, the Bulverism fallacy applies to individual beliefs because of the possibility of this movement from “historical” causes to “logical” grounds. However, it does not apply to our beliefs taken as a whole, because, by definition, such a whole could not be taken out of its context into another one. Any given context would be a part of the whole already. In which case, we would never be able to examine a belief’s validity, because any test would also have nonrational grounds, and any test of the test would as well, ad infinitum. So if rational processes never enter into the equation, how can any belief ever be rational?

This latter point, I take it, is Eric L. Mascall’s argument in his defense of Lewis against Anscombe.{16} He thinks Anscombe’s contention, that an argument’s validity is independent of its formation, “is, I think, a good one, but only so long as we exclude from the sphere of application of the naturalistic theory the examiner’s conviction of the validity of his examination.”{17} He illustrates this with a parable about a man who has a deep hatred for a particular bishop. The man justifies his hatred with a syllogism, that some churchmen are alcoholics; the bishop he hates is a churchman; therefore, the bishop he hates is an alcoholic. A psychoanalyst examines the man and determines that his hatred is based on an unpleasant event in his childhood. The man can then say that the cause of his belief is irrelevant; what matters is whether he can prove it upon further examination, and his syllogism does just that. However, the psychoanalyst is also a logician, and he proceeds to point out that the man’s syllogism has an undistributed middle, and such syllogisms are invalid. The man, however, responds that, “the widespread belief in the invalidity of syllogisms with undistributed middles is simply caused by something in people’s genetic inheritance.”{18} In other words, this belief also has a nonrational cause. At no point can we step out of the circle of nonrational causes in order to test a belief’s validity, because any proposed test would be produced nonrationally as well.

Mascall concludes that any plausibility such determinism may have “is due, I would maintain, to the fact that when it is asserted an escape-clause is either explicitly included or, more often, implicitly assumed. It is held to apply to volitions and attitudes, but not to ratiocination; or, if it does apply to ratiocination in general, it does not apply to the ratiocination which its propounder makes use of in arguing for its truth.”{19}

Augustine Shutte later comments on “the sequence of Lewis’s article followed by Anscombe’s reply and then Mascall’s comments on both.”{20} He quotes Mascall’s parable in its entirety, and argues that if our convictions about the validity of logic are the product of nonrational causes, we cannot use logic to verify the validity of a given argument. Anscombe might respond, however, that she is not speaking of convictions. This, however, “assumes that the rules determining validity can be defined in total abstraction from real events, psychological or otherwise, and yet must be regarded as normative for events and processes in the real world, namely those that constitute thinking and arguing.”{21}

Whether we think the arguments of Mascall and Shutte can avoid the skeptical threat objection mentioned above is another issue. Reppert argues that they do not.{22} But they can avoid the charge of Bulverism. The circumstantial ad hominem fallacy assumes that we can abstract a belief from its nonrational causes and judge its truth on purely rational grounds. Insofar as this abstraction is impossible, this fallacy cannot be applied. Plus, it has no bearing on whether a belief is epistemically justified. Therefore, appealing to Bulverism cannot refute Lewis’s claim that if all of our beliefs are derived from irrational or nonrational sources they are all invalid. Shutte concludes that, “Lewis’s argument against determinism has been vindicated.”{23}

Point 6: Distinguishing between causes and grounds
By equating the irrational with the nonrational, Lewis had confused the grounds of a belief (which are rational or irrational) with the causes of a belief (which are nonrational). This is partially because both grounds and causes are given to answer “why” questions, and begin with “because.” However, this is a confusion of thought. These are two completely different types of explanation. By confusing them, Lewis has mistakenly thought that they are in competition with each other. They are not.

It was to this objection, Anscombe’s primary one, that Lewis addressed his revision of the argument, which will be presented in the next post. For now, I will just reiterate Lewis’s comments after Anscombe read her paper at the Socratic Club. He argued that the grounds of a belief could function as its cause. In fact, unless the grounds do so, the belief would not have been reached rationally. This foreshadows a similar claim made by Donald Davidson in 1963.{24} As Shutte puts it,

Lewis is entitled to claim that if the causes of a person’s holding a belief do not include the actual grasping of the logical link between premise and conclusion as holding between that belief and another then his holding of the belief will be unreasonable. Hence, if all the causes of all beliefs are of a natural or deterministic kind, then no rationally held beliefs will occur. And, hence, no one will ever be justified in claiming any belief to be true, determinism included.{25}

Another point is noteworthy: this criticism is based on Anscombe’s Wittgensteinian belief that different types of explanations are different language games that do not conflict. However, another follower of Wittgenstein disagreed with this assessment. In his essay “The Conceivability of Mechanism,” published after Lewis’s death, Norman Malcolm defended a view very similar to Lewis’s, and according to the final footnote of that article, Anscombe herself reviewed it before its publication.{26} The point being that a fellow Wittgensteinian did not think the concept of language games allowed us to make such a sharp dichotomy between causes and explanations.

A biographical note
From the foregoing discussion, it should be evident that I think Lewis’s original argument largely survives Anscombe’s criticisms. Biographers have sometimes argued, though, that Lewis did not think so, and that he felt personally humiliated by her analysis. Some go so far as to suggest that he abandoned writing apologetics altogether,{27} and took up children’s literature as a result.{28}

There are a few points to make in response. First, John Beversluis, one of Lewis’s most trenchant critics, who had repeated this claim himself,{29} later concluded that it was inaccurate, irrelevant, and presumptuous.

… the myth that Lewis abandoned Christian apologetics overlooks several post-Anscombe articles, among them “Is Theism Important?” (1952) -- a discussion of Christianity and theism which touches on philosophical proofs for God’s existence and their relevance to the religious life -- and “On Obstinacy in Belief” (1955) -- in which Lewis defends the rationality of believing in God in the face of apparently contrary evidence (the issue in philosophical theology during the late 1950s and early 60s). It is rhetorically effective to announce that the post-Anscombe Lewis wrote no further books on Christian apologetics, but it is pure fiction. Even if it were true, what would this Argument from Abandoned Subjects prove? He wrote no further books on Paradise Lost or courtly love either.{30}

Second, such speculations about Lewis’s motives are extraordinarily tone-deaf. One of his most popular essays is “Modern Theology and Biblical Criticism” in which he makes the point that reviewers of his own writings and those of his friends have often tried to reconstruct their motives. According to Lewis, such attempts were universally incorrect; he could not recall a single accurate statement.{31} For biographers of Lewis to make such attempts themselves in light of Lewis’s explicit claim that “the results are either always, or else nearly always, wrong,”{32} either demonstrates that they were unfamiliar with this essay or that they chose to ignore it. Neither case is responsible. He even explicitly asks readers of Mere Christianity to refrain from speculating about his motives.{33}

Third, Anscombe herself was not aware of any such reaction on Lewis’s part, and neither were their mutual acquaintances. She recalled the meeting as “an occasion of sober discussion of certain quite definite criticisms,” but suggests that those who think Lewis felt shocked or humiliated by the encounter are evincing “an interesting example of the phenomenon called ‘projection.’”{34}

Fourth, Lewis obviously did not abandon apologetics or the argument from reason: he recognized that Anscombe’s criticisms deserved a full response,{35} and to this end, rewrote the third chapter of Miracles, publishing it in 1960. To this I turn next.


{1} Antony Flew, “The Third Maxim,” The Rationalist Annual 72 (1955): 64.
{2} C.S. Lewis, The Abolition of Man (1943; New York: Macmillan, 1947), 30.
{3} Flew, “Third Maxim,” 65.
{4} Victor Reppert, “The Lewis-Anscombe Controversy: A Discussion of the Issues,” Christian Scholar’s Review 19 (1989): 37.
{5} Flew, “Third Maxim,” 64-65.
{6} Reppert, “Lewis-Anscombe Controversy,” 37.
{7} William Hasker, The Emergent Self (Ithaca, NY: Cornell Univ. Press, 1999), 68.
{8} Reppert, “Lewis-Anscombe Controversy,” 37-38.
{9} Ernest Gellner, “Determinism and Validity,” The Rationalist Annual 74 (1957): 69-79. Although Gellner’s essay is essentially a critique of a critique of Lewis, he never mentions Lewis by name. Flew responded to Gellner (Flew, “Determinism and Validity Again,” The Rationalist Annual 75 [1958]: 39-51), and brought up the Lewis-Anscombe debate elsewhere in his writings as well (idem, “A Rational Animal,” in A Rational Animal and Other Philosophical Essays on the Nature of Man [Oxford: Clarendon, 1978], 92-99; idem, The Logic of Mortality [Oxford: Basil Blackwell, 1987], 84).
{10} Gellner, “Determinism and Validity,” 72.
{11} Ibid., 70-71.
{12} Flew, “Third Maxim,” 62.
{13} Actually, eliminative materialists hold that beliefs are just a part of “folk psychology,” and can therefore be eliminated. So they do deny that people have beliefs. I will address this in the final post of this series.
{14} Hasker, Emergent Self, 71, 73, italics in original.
{15} Cf. Arthur Stanley Eddington, Science and the Unseen World (New York: Macmillan, 1929), 63-67.
{16} E.L. Mascall, Christian Theology and Natural Science: Some Questions in Their Relations (London: Longmans, Green & Co., 1957), 212-19.
{17} Ibid., 215.
{18} Ibid., 216.
{19} Ibid., 216.
{20} Augustine Shutte, “The Refutation of Determinism,” Philosophy 59 (1984): 481. Although he is writing nearly a quarter century after the second edition of Miracles was published, Shutte is working from the first edition.
{21} Ibid., 487.
{22} Reppert, “Lewis-Anscombe Controversy,” 37-38.
{23} Shutte, “Refutation of Determinism,” 487.
{24} Donald Davidson, “Actions, Reasons, and Causes,“ in Essays on Actions and Events, 2nd ed. (Oxford: Clarendon, 2001), 3-20.
{25} Shutte, “Refutation of Determinism,” 484-85.
{26} Norman Malcolm, “The Conceivability of Mechanism,” The Philosophical Review 77 (1968): 72 n. 14. Malcolm was later critiqued by Alvin Goldman (“The Compatibility of Mechanism and Purpose,” The Philosophical Review 78 [1969]: 468-82), defended, with qualifications, by Jaegwon Kim (“Mechanism, Purpose, and Explanatory Exclusion,” in Supervenience and Mind [Cambridge: Cambridge Univ. Press, 1995], 237-64), and then defended without qualification by William Hasker (Emergent Self, 64-68).
{27} Humphrey Carpenter, The Inklings: C.S. Lewis, J.R.R. Tolkien, and Their Friends (1978; New York: Ballantine, 1981), 238-39; Michael White, C.S. Lewis: The Boy Who Chronicled Narnia (London: Abacus, 2005), 174-75.
{28} A.N. Wilson, C.S. Lewis: A Biography (New York: W.W. Norton & Co., 1990), 210-15, 220, 225-27.
{29} John Beversluis, C.S. Lewis and the Search for Rational Religion (Grand Rapids, MI: Eerdmans, 1985), 66; idem, “Beyond the Double Bolted Door,” Christian History 4/3 (July, 1985): 29.
{30} John Beversluis, “Surprised by Freud: A Critical Appraisal of A.N. Wilson’s Biography of C.S. Lewis,” Christianity and Literature 41 (1991-92): 192. The essays he cites are C.S. Lewis, “Is Theism Important?” in God in the Dock: Essays on Theology and Ethics, ed. Walter Hooper (1970; Grand Rapids, MI: Eerdmans, 1996), 172-76; and idem, “On Obstinacy in Belief,” in The World’s Last Night and Other Essays (1960; New York: Harvest Book Paperback, 1973), 13-30.
{31} C.S. Lewis, “Modern Theology and Biblical Criticism,” in Christian Reflections, ed. Walter Hooper (1967; Grand Rapids, MI: Eerdmans, 1977), 158-61. Alternatively titled “Fern-seed and Elephants.” This apologetical essay was a lecture delivered in 1959, over a decade after Anscombe’s lecture at the Socratic Club, further demonstrating that Lewis did not abandon apologetics in the wake of Anscombe’s critique.
{32} Ibid., 161.
{33} C.S. Lewis, Mere Christianity (1952; London: Collins, Fontana Paperbacks, 1955), 6-7.
{34} G.E.M. Anscombe, The Collected Philosophical Papers of G. E. M. Anscombe, vol. 2: Metaphysics and the Philosophy of Mind (Oxford: Basil Blackwell, 1981), x.
{35} C.S. Lewis, “Rejoinder to Dr Pittenger,” in God in the Dock, 179.

(cross-posted at Agent Intellect)

Discuss this post at the Quodlibeta Forum