The Argument from Reason (1998)
Victor Reppert

from http://homepages.wmich.edu/~mcgrew/ReppertAFR.pdf

When we hear of some new attempt to explain reasoning or language or choice naturalistically, we ought to react as if we were told that someone had squared the circle or proved the square root of 2 to be rational: only the mildest curiosity is in order—how well has the fallacy been concealed?1

There are various phenomena that believers in the existence of God can appeal to to support the claim that God exists. One can look to the beginning of the universe, the design of the universe, to our moral experience, to evidence in support of miracles, and religious experience to ground belief that God does exist. One phenomenon that is sometimes neglected in the development of theistic arguments is the existence of rational thought. Does our very thinking provide evidence that the universe is more likely to be the kind of place that theists say it is than the kind of place that atheists say it is. The argument I will be presenting attempts to answer that question in the affirmative. This argument is often advanced against materialism or determinism, and as such it has Kantian2 roots. However, it was developed as an argument for accepting theism as opposed to naturalism in the last century by the British Prime Minister Arthur Balfour3 and in the 1940s by C. S. Lewis.4 It was this line of argument that was criticized by Elizabeth Anscombe,5 and as a result Lewis revised his argument in the second edition of his book, Miracles. Contemporary philosophers who have employed the argument against physical determinism include James Jordan and William Hasker,6 and those who have turned it into an argument for theism include Richard Purtill7 and J. P. Moreland.8 Following John Beversluis,9 one of the argument’s critics, I will refer to this argument henceforth as the argument from reason.

In presenting this argument, it is necessary to contrast the atheist’s view of the world with that of the theist. I will present a model of the atheist universe which I will call mechanistic materialism. Although worldviews other than mechanistic materialism are compatible with atheism, mechanistic materialism seems to be the worldview held by most atheists. So if I can show that the existence of reason makes sense in a theistic universe but not in a mechanistic materialist universe, I will have given some good reasons for preferring theism to atheism. If an atheist wishes to propose a form of atheism that differs from mechanistic materialism, I would be happy to discuss that worldview as well.

According to theism, the universe is a rational place because it is the creation of a rational being, namely God. Reason is, so to speak, on the very ground floor of reality. Given that God creates creatures, it is a least possible that God might wish to provide those creatures with the some measure of the rationality which God himself possesses. And human beings reflect God’s rational character by having the capacity think logically. If we make the further supposition that God has created human beings in such a way that they consist of both a soul and a body, we might be able to say that while the body’s activities are determined by the laws of physics, it is possible for human beings, through their souls, to perceive not merely physical activities in the environment, but logical and mathematical truths that apply throughout all that God has created. If someone in a theistic universe who had a spiritual as well as a physical nature were to reason were to reason to a conclusion logically, it might very well be that the person reached the logical conclusion because the conclusion follows logically from the premises, and not because the laws of physics mandated that the physical particles in the brain move to such and such places.

Now let us consider the universe from the point of view of mechanistic materialism. The universe may have begun with a Big Bang, but what results from that Big Bang are material substances of various kinds. And material substances go where they go, not because it would be a good idea to go there, but because such motion is mandated by the laws of physics. If rocks fall down a mountain due to an avalanche, they will not stop because they don’t want to hit and kill any people. On the mechanistic view of the world, material particles can, through evolution, organize themselves into complicated systems that work together to further the survival of the organism and the species. So, for example, your eye might be said to have the “purpose” of allowing you to see. That is, your eye might be structured, as a result of centuries of evolution, in such a way that it serves the purpose of your seeing. But the particles that make up your eye are just as mechanistically determined as the particles of a rock falling down a mountain. What we call “drawing a rational inference” must be accounted for in the same way. Perhaps our brains are structured in such a way that the activity we call rational inference will be performed, and that this capacity contributes to our survival individually and collectively. But the description of this activity as rational inference is not the description of this activity on the most basic level of analysis. The most basic level of analysis is that of physics, which makes no reference to purposes or logic whatsoever.

Daniel Dennett presents the commitment to mechanistic explanations as follows:

Psychology of course must not be question-begging. It must not explain intelligence in terms of intelligence, for, instance by assigning responsibility for the existence of intelligence to the munificence of an intelligent creator, or by putting clever homunculi at the control panels of the nervous system. If that were the best psychology could do, then psychology could not do the job assigned to it.10

Similarly, Dennett explains the appeal of Darwinian theory on the grounds that

Darwin explains a world of final causes and teleological laws with a principle that is entirely independent of “meaning” or “purpose.” It assumes a world that is absurd in the existentialist’s sense of the term: not ludicrous or pointless, and this assumption is a necessary condition of any non-question-begging account of purpose.11

I think that the materialist’s commitment can be described by three theses. One, causation within the physical order is mechanistic, that is, non-purposive. Two, the physical order is closed, that is, nothing apart from the physical order can cause anything to happen within the physical order. Third, all states supervene on physical states. By this I mean that given the state of the physical, there is only one way that anything not physical (biological, psychological, sociological, moral) can be.

Now, the question that I want to pose is whether reason is possible in a mechanistic universe. Let’s consider what reasoning amounts to for a moment. If you are arguing with an atheist, it is reasonable to expect that the atheist will present an argument of the following type:

  1. If God exists, then there can be no gratuitous evil.

  2. Probably, there is gratuitous evil.

  3. Therefore, probably, God does not exist.

Presumably he hopes that some of you will be persuaded by this argument. But what does that mean? First of all, the atheist is expecting us to apprehend the content of the premises and the conclusion. He expects us to accept the premises as true and will no doubt provide arguments as to why we should accept them. Then he expects us to notice that, by logical law, if you accept premises 1 and 2, you must accept premise 3. So in order for rational argument to meaningfully proceed one must presume that

  1. Sentences can be meaningful and are not just a series of marks.

  2. Human beings can apprehend the propositional content of written sentences.

  3. Human beings can be in the condition of either accepting, rejecting, or suspending belief concerning propositions.

  4. Logical laws exist.

  5. Human beings are capable of apprehending logical laws.

  6. The state of accepting the truth of a proposition can play a causal role in producing other beliefs, and that propositional content is relevant to the playing of this causal role.

  7. The apprehension of logical laws plays a causal role in the acceptance of the conclusion of an argument as true.

  8. Persons exist throughout the moments of time required for a rational inference to be performed.

In short, what the advocate of the argument from evil, or indeed any argument in favor of atheism, hopes that we will do is accept the argument in virtue of is legitimacy as an argument. This involves a lot of assumptions about what human beings are like.

It follows from this that anyone who hopes to defend atheism, naturalism, or materialism by argument needs to assure us that an atheist, naturalist, or materialist universe can house within itself the necessary conditions of logical inference. If she cannot do this, then we might say that these worldviews refute themselves insofar as they are presented, not merely as bald assertions, but as reasoned conclusions of logical arguments. Suppose I tried to argue for the claim that no one is ever persuaded by an argument. If I make my case and persuade you that this is true, then, of course, my position cannot be true. This shows that we can see that something is wrong with this kind of a position. If it’s true, it can’t be defended, if it can be defended, it’s not true.

The simplest way to generate this kind of argument against materialism is to say that since all thoughts are determined (apart from pure quantum chance) by the nonpurposive motion of atoms in the brain, it follows that such a process cannot be a rational inference. But such an argument would be, of course, rather too simple. As Anscombe pointed out in response to C. S. Lewis’s early version of the argument:

Your idea appears to be that ‘the explanation’ is everywhere the same one definite requirement, so that we can know, when it is filled, that, if it has been correctly filled, the whole subject of ‘explaining this fact’ has been closed.12

But of course, many explanations for the same event can be compatible with one another. If I want to know why the soda-can is on the bookshelf, I can say “because I put it there yesterday,’ or “ because I wanted it recycled” or “because of the law of gravity” or “because it is cylindrical and not spherical.” But in the case of an instance of rational inference, we want to know whether an event can be at the same time, the motion of brain matter in a mechanistic universe, and, at the same time, the inference to a conclusion from its premises.

We know from science that the same process can be given distinct explanations depending on the level at which we are analyzing it. Physics looks at subatomic particles, while biology looks at something as an entire system. If I were to punch someone in the mouth, one could give an explanation for this punching in terms of subatomic particles, biochemistry, or gross anatomy. And it might be that the punch could also be explained in psychological terms, and sociologist could come along and tell me that members of such and such a social group are more likely to engage in this kind of aggressive behavior under such and such circumstances. These levels of explanation involve purpose, however, and the compatibility of mechanism and purpose is just what is at issue here. But what is clear enough is that different explanations, at different levels, can be given for one and the same causal transaction.

So most rebuttals to the argument from reason make the claim that explanation of a mental act of rational inference in terms of particle physics that makes no reference to reasons, and the explanation of that same act in terms of propositional attitudes, intentionality, truth, and laws of logic, are compatible with one another. Given a proper relationship between the explanation given at the physical level, the explanation at the mental level can obtain also, and one needs no more to choose between a mental and a physical explanation than one needs to choose between a biological and a physical explanation.

But let’s take a look at what explanatory compatibility amounts to in different cases. Let us say we have an explanation, at the level of particle physics, for a round peg going into a round hole. In this case, the explanation in terms of physics fails to make mention of pegs or holes, but the physical structure of the peg and the hole, (as well as the force putting the peg into the hole), is what makes the causal transaction possible. It could be pointed out that one could have a peg and a hole made of different types of material, and that the statement “a round peg fits in a round hole” is multiply realizable-it can be a wooden peg, a metal peg, or a plastic peg, etc. But any such peg or hole would have to satisfy a certain structural specification to play a role in a transaction described thus, in particular, it would have to be round.

Another example would be the webbed feet of a duck. The natural webbed feet of a duck will permit the duck to move in the water as well as on land, but perhaps veterinary medicine might be able to provide ducks whose webbed feet have been injured or destroyed with artificial webbing that is molecularly very different form their natural webbing. Nevertheless even though its microstructure is different from natural webbing, its macrostructure is similar, and it is the macrostructure that enables it to do its job. Putting steel in place of webbing will result in the duck not being able function in water as effectively as they currently do. We might call this kind of explanatory compatibility structural compatibility. That is, even though the upper level explanation uses terms which are not found in the lower-level explanations, the structure of lower-level items guarantees that the upper-level explanation will obtain. So, for example, one can describe each brick of a wall without mentioning a wall at all, we know very well that, given the position of the bricks, the wall must exist. Where there is structural compatibility, an ideal scientist, looking at large-scale physical patterns, could know what higher-level compatible explanations could be given.

But explanatory compatibility can be of a different type entirely. Jerry Fodor has argued that monetary exchanges cannot plausibly be analyzed in terms of particle physics, nevertheless no one is inclined to suppose that there is anything nonphysical about money.13 Once again it is possible to fully describe a monetary transaction from the standpoint of particle physics without mentioning that it is a monetary transaction. But the property “being money” is a matter, not of physical structure, but of human convention. In the United States presenting slips of paper with green ink on them depicting various leaders from American history will result in a person’s receiving goods and services commensurate with the indicated value of those slips of paper. Presenting maroon and gold slips of paper with the pictures of Arizona State University football players will not produce this result, even at the concession stands at Sun Devil Stadium. The structure of money makes no difference whatsoever, what makes a difference is its relation to a background of mental states. What that means is that economic explanation is compatible with physicalism just in case mental explanation is, and if not then not. As Geoffrey Madell puts it:

Given our understanding of human interests and wants, we can understand how those wants can be furthered by the institution of money, by the business of endowing a range of very different items with the same conventional significance. But our understanding of this presupposes the background of human psychology.14

So contrary to Fodor, there may be something nonphysical about money, if it turns out that our convention-making capability is something nonphysical. So, although this example is often offered as an example of an ontologically innocuous instance of irreducibility, it is nothing of the sort. Economic exchanges are compatible with physicalism just in case the kinds of mental states listed above are compatible with physicalism, and that’s the very thing we are trying to ascertain at the moment.

We might call the type of explanatory compatibility we are talking about here conventional compatibility. But conventional compatibility between mechanism and reason is not going to help us solve the problem of whether reason is possible in a mechanistic universe, because it leaves open the question of how the convention-maker can be rational in a mechanistic universe.

Computers are often offered as clear and decisive cases where physical and mental explanations can be offered for the activities of undeniably physical systems. But computers are clear cases of conventional compatibility. In the case of the computer, a “full” explanation of its activities can, of course, be offered in physical terms. But the computer has the characteristics it does because reasoners built it to model their own rational thought patterns. So the computer is not a clear example of a system for which the total causal story is clearly mechanistic. As William Hasker points out,

Computers function as they do because they have been constructed by human beings endowed with rational insight. A computer, in other words, is merely an extension of the rationality of its designers and users; it is no more an independent source of rational thought than a television set is an independent source of news and entertainment.15

So the compatibility of mechanism and rationality found in computers hardly shows that the mechanistic world view is compatible with the actuality of human reason. If humans are physical systems, then computer activity can be reconciled with a physicalist view of the universe. If not, then computer states, while themselves physical, rely for their mental characteristics on the existence and intentions of nonphysical entities. But the existence of “mental” states in computers helps not at all in showing that our mental states are or can be physical states, given that the universe is purely physical.

Further, it seems clear enough that conventionally established characteristic cannot be causally relevant. Structure, after all, is what counts in causal transactions. If a baseball breaks a window, this window breaking can be explained both at the level of particle physics and at the level of macro-mechanics. Some properties of the baseball, such as its size and weight, and its velocity, are relevant to the question of whether it will break the window. But other properties are clearly not relevant. If the ball was used in Sandy Koufax’s fourth career no-hitter, this may explain why the owner, (presumably a Dodger fan) would prize it over other baseballs he owns. But it is irrelevant to whether the ball will or will not break the window upon impact. Whether a computer’s activity is interpreted as a chess game or a word-processing program will not affect the actual output of the computer, though it will no doubt affect the input that its users generate. What this means is that if I want to say that I accept naturalism because there are good reasons for believing it, then it cannot be true that my acceptance of naturalism is a matter of conventional compatibility. What is more, conventionally established characteristics of a person are also irrelevant in evolutionary explanation. After all, it cannot be a matter of convention whether I survive long enough to pass on my genes. So any attempt to use evolution to account for the reliability of my belief-producing mechanisms is bound to fail if mental states are conventionally established characteristics.

Yet many materialist theories of mind seem to provide compatibilist accounts of mechanism and purpose which are, in the final analysis, based on conventional compatibility. According to some versions of functionalism, not only is it the case that mental states are multiply realizable, but that those mental states can be realized in just any medium. On this view mental states are to be distinguished by their causal role, and these roles can be played by anything from Swiss cheese to a Cartesian mind. What this means to me is that nothing about physical structure is relevant to whether some particular causal role is played or not played. But unless there is something about the structure of those states that makes them capable of playing the relevant causal role, it must be the case that these states play that role relevant to a set of conventions adopted by thinkers. To illustrate this we can compare two different game devices. A game played with frisbees requires those frisbees to have a certain structure. I can’t play a frisbee game if I replace the frisbee with just anything, such as a manhole cover or a basketball. There may be conventions surrounding a frisbee game, but a frisbee still requires one particular kind of structure. On the other hand, if I am playing chess, and I am missing a pawn, I can get a penny, a button, a half-eaten carrot, or just about anything else to play the role of a pawn. This is because unlike the frisbee, the pawn’s role is purely symbolic, determined by convention. It seems clear from these examples that causal roles can be assigned to physical things with total medium independence just in case those roles are assigned conventionally. If the causal role is not assigned conventionally, then something about the structure of the system is required for it to play the causal role it does.

An even more straightforward case of conventionally based mental-physical compatibility is found in Donald Davidson’s anomalous monism.16 An essential feature of anomalous monism is the claim that there are no psychophysical laws, that is, laws linking mental states with physical states. Mental states, according to Davidson, cannot be ascribed apart from considerations of coherence, consistency, and rationality. If these considerations are set aside, then we are not interpreting mental states at all. So talk about mental states can never be reduced to talk about physical states. Nevertheless mental states are physical states, because if they were not physical states they could not be causally connected to physical states. Since, on Davidson’s view, causality is grounded in strict laws, and since there are no strict psychological laws and no strict psychophysical laws, mental events must be connected by physical causation. But as Jaegwon Kim points out: 

Davidson’s anomalous monism fails to do full justice to psychophysical causation in which the mental qua mental has any real causal role to play. Consider Davidson’s account: whether or not a given event has a mental description (optional reading: whether or not it has a mental characteristic) seems entirely irrelevant to what causal relations it enters into. Its causal powers are wholly determined by the physical description or characteristic that holds for it; for it is under its physical description that it may be subsumed under a causal law.17

The problem with Davidson’s position is that, in order for rational inference to be possible, beliefs must cause other beliefs, but also, beliefs must cause other beliefs in virtue of their propositional contents. And this seems impossible on Davidson’s model. In physical causation what are relevant are physical properties, although this can be conceived of broadly to include properties like “being round,” being square,” “being webbed.” What clearly cannot play a role in physical causation are properties ascribed as a matter of convention. Thus it is impossible to see how, on Davidson’s model, I could possibly believe that anomalous monism is true because I accept Davidson’s argument that it is true.

So it seems clear enough that one cannot accept both a materialist world-view and the claim that someone has reached the conclusion that materialism is true on the basis of argument only if mental states exist as a matter of physical structure. The difficulty with this suggestion is that in reasoning the mental states involved are supposed to be connected by logical necessity, but one cannot given an account of this logical necessity by describing contingently connected physical states. There is, to paraphrase Lessing, a big ugly ditch between the necessary truths of reason and the contingent facts of neurochemistry. It is part of what it means to have a thought or a belief that we perceive it as part of a network of logical connections. If I believe that the cat is on the mat, I must also be capable of recognizing that I reject the belief that the cat is not on the mat. But how can a physical state be the perception of a logical connection. A physical state is determined, not by the laws of thought, but by the laws of physics. And the laws of physics sometimes allow people to form beliefs contrary to the laws of thought. What is more, if we are purely physical systems, then whatever we know must come to us through causal connection with particular states in the physical world. Thus, if I see a computer monitor before me, the monitor affects the rods and cones in my eyes in such a way as to produce in me the belief that there is a computer monitor before me. But principles like the laws of non-contradiction apply universally. Contradictions cannot be true in Phoenix, in Houston, in Antarctica, on Venus, in the Virgo cluster, in the furthest nebulae, or even in Southern California. From what particular experience could our knowledge of this have come? One can describe physical states to one’s heart’s content, without being able to infer form the physical system itself what logical laws are being obeyed in the course of, say, a person performing a Barbara syllogism. If we come to a physical system armed with the laws of logic, then we can, perhaps, interpret some physical arrangement as the representation of a logical law. But then one is still left with the problem of how interpreters could arise in a purely physical world.

I conclude, therefore, that the Argument from Reason stands untouched by challenges based on the possibility of compatible explanations. It does not follow from the fact that compatible explanations can sometimes be given for the same one and the same event that a brain process in a purely material universe can be given both a true physical explanation and a true explanation as a rational inference. Materialist accounts of reasoning typically presuppose the existence of the very thing that they are trying to explain. According to materialism, the universe begins with no mental states and somehow evolves them into existence through the shuffling and reshuffling of material particles. Suppose, however, that rationality were, so to speak, on the ground floor of reality. Suppose the universe were the result of the activities of a rational being. If that were the case, then we could understand how such a rational being could bestow beings in the universe with a measure of its rationality. In response to Dennett, I would want to say that explaining reason in terms of reason is no more question-begging than explaining physical states in terms of physical states. If the foregoing argument is correct, then explaining reason in terms of unreason explains reasoning away, and undercuts the very reasoning on which the explanation is supposed to be based.

How does this argument contribute to the case for God? What we call arguments for the existence of God do not demonstrate the existence of every attribute of God, but only certain attributes of God. The attribute of being a rational being is an attribute that believers ascribe to God, and it is an attribute that atheists deny to the fundamental structure of the universe. So if successful this argument establishes that something with at least one of the characteristics of God is fundamental to the universe, and it also establishes that materialism, as ordinarily understood, cannot be true.18

1 Peter Geach, The Virtues (Cambridge: Cambridge University Press, 1977), p. 52.

2 For a discussion of the Kantian foundations of this argument, see Henry Allison, “Kant’s Refutation of Materialism,” The Monist 79 (April 1989), pp. 190-209.

3 Arthur Balfour, The Foundations of Belief: Notes Introductory to the Study of Theology, 8th ed. Rev. with New Introduction and Summary (New York: Longmans, 1918), pp. 279-285.

4 Lewis’s best version of the argument is to be found in C. S. Lewis, Miracles: A Preliminary Study (New York: Macmillan Paperbacks Edition, 1978), pp. 12-24.

5 Anscombe’s critique of the first edition is to be found in her The Collected Papers of G. E. M. Anscombe, 3 vols. (Minneapolis: The University of Minnesota Press, 1981) Vol. 2 Metaphysics and the Philosophy of Mind, pp. 224-231. For a discussion of this exchange see Victor Reppert, “The Lewis-Anscombe Controversy: A Discussion of the Issues” Christian Scholar’s Review 19 (September 1989), pp. 32-48.

6 William Hasker, Metaphysics (Downer’s Grove, IL: Inter-Varsity, 1983), 47-50, and “The Transcendental Refutation of Determinism,” Southern Journal of Philosophy 11 (1973), pp. 175-83.

7 Richard Purtill, Reason to Believe (Grand Rapids: Eerdmans, 1974), pp. 44-46.

8 J. P Moreland, Scaling the Secular City (Grand Rapids: Baker Book House, 1987), pp. 77-105.
9 John Beversluis, C. S. Lewis and the Search for Rational Religion (Grand Rapids: Eerdmans, 1985) 58-83. For a response to his criticisms see Reppert, “Lewis-Anscombe,” above.

10 Daniel Dennett, “Why the Law of Effect Will Not Go Away,” in Brainstorms (Cambridge: MIT Press, 1977), pp. 46-47.

11 Ibid.

12 Anscombe, Metaphysics, p. 228.

13 Jerry Fodor, “Special Sciences” in his Representations (Brighton: Harvester Press, 1981), pp. 127-145.

14 Geoffrey Madell, Mind and Materialism (Edinburgh: Edinburgh University Press, 1988), p. 40.

15 Hasker, Metaphysics, p. 49.

16 Donald Davidson, “Mental Events,” in Essays on Actions and Events (Oxford: Clarendon Press, 1980).

17 Jaegwon Kim, “Epiphenomenal and Supervenient Causation” ch. 6 of Supervenience and Mind (Cambridge: Cambridge University Press, 1993), p. 106.

18 I would like to thank Bill Hasker and Jim Lippard for helpful comments and suggestions on earlier versions of this paper.