Culture & Ethics Icon Culture & Ethics
Faith & Science Icon Faith & Science

C. S. Lewis and the Argument from Reason

Photo: C. S. Lewis, by Asar Studios/Alamy (Photo by Hans Wild/The LIFE Picture Collection/Getty Images).

Editor’s note: To mark the release on November 3 of the new C. S. Lewis biopic, The Most Reluctant Convert, we are running a series of articles exploring C. S. Lewis’s views on science, mind, and more.

SPECIAL LIMITED-TIME OFFER: Get a FREE chapter exploring C. S. Lewis’s views of intelligent design from the book The Magician’s Twin.

To see Lewis’s genius, I’d like to focus on one of his best-known arguments — often called the “argument from reason.” The purpose of the argument is to show that naturalism and reason are incompatible, that believing in naturalism is self-defeating. That is, if naturalism is true, then we ought not to trust our capacity for reason, and so, ought not to trust arguments in favor of naturalism. 

Philosopher Victor Reppert describes the argument (and several versions he develops from Lewis’s original) as “beginning with the insistence that certain things must be true of us as human beings in order to ensure the soundness of the kinds of claims we make on behalf of our reasoning.” This argument gained attention when Lewis proposed it in the first edition of Miracles. Philosopher Elizabeth Anscombe critiqued the original formulation of the argument, so Lewis corrected it in a subsequent edition of Miracles. It is this revised version of his argument that millions of readers have encountered. (He also discusses the argument in some lesser-known articles published in Christian Reflections and God in the Dock.)

Lewis taught philosophy in his first year as a lecturer at Oxford, but he wasn’t a professional philosopher. Moreover, he made the argument from reason in a small book written for public consumption. And yet it has been remarkably resilient and fruitful. Its philosophical offspring still play a role in contemporary philosophy. The most rigorous form of the argument is the “Evolutionary Argument Against Naturalism” developed and refined by philosopher Alvin Plantinga.

The Basic Argument

Miracles is not a historical defense that miracles have actually occurred. It is a preliminary defense of their possibility and propriety. One of its central arguments is that we cannot determine the antecedent probability of a miracle without first deciding what reality is like. If you think that a transcendent exists, for instance, you will assess the evidence for a miracle differently than if you are a naturalist who believes that the closed, interlocking system of nature is all that exists. As a result, Lewis spends a good deal of time in Miracles evaluating the competing claims of what he calls supernaturalism and naturalism. 

It is in this context that Lewis takes up the so-called “cardinal difficulty of naturalism.” Naturalists in Lewis’s day were very much like naturalists in our day. They normally imagine that their philosophy is the result of sound reasoning and solid evidence, and assume non-naturalists are ignorant and irrational. Lewis argues quite the opposite: naturalism is not compatible with knowledge and the reliability of reason. 

Naturalists, like everyone else, generally trust their reason to lead them to truth. We all take it for granted that we can learn about the world around us through our senses. We experience heat and sound and color and other people. We somehow synthesize and take account of these things with our mind. From these experiences we make inferences about the world: “We infer evolution from fossils: we infer the existence of our own brains from what we find inside the skulls of other creatures like ourselves in the dissecting room.”

But what is an inference? Clearly it’s not an object of the senses, such as a bullfrog or a forest fire. An inference, we might say, is a logical structure. We see dinosaur fossils with our eyes; but we infer the prior existence of dinosaurs with our minds. We simply understand that if all men are mortal and Plato is a man, then Plato is mortal. When we consider this argument, we’re not observing the world around us. We’re perceiving logical relations between propositions that would obtain in any possible world. Propositions are claims about states of affairs, but not the states of affairs themselves. That Columbus sailed the ocean blue is a state of affairs that obtained sometime in 1492. “In 1492, Columbus sailed the ocean blue” is an English formulation of a true proposition affirming that state of affairs.

When we conclude that Plato is mortal, we are certain that if the propositions that form the premises are true, then the conclusion must be true. Unless such mundane inferences are possible and reliable, we can’t have knowledge:

All possible knowledge… depends on the validity of reasoning. If the feeling of certainty which we express by words like must be and therefore and since is a real perception of how things outside our own minds really “must” be, well and good. But if this certainty is merely a feeling in our own minds and not a genuine insight into realities beyond them — if it merely represents the way our minds happen to work — then we can have no knowledge. Unless human reasoning is valid no science can be true.

Naturalism doesn’t contain such ingredients as minds, propositions, perceptions and logical relations. It contains elementary particles and attractive forces, chemical reactions, quantum fields, and the like, in a closed and impersonal system of cause and effect. And all of those causes are material and non-rational. Naturalism doesn’t countenance immaterial entities such as persons, with thoughts and beliefs, persons that can infer from the proper ground of a propositional belief to a valid conclusion, which can then guide behavior and so cause things to happen in the world. If naturalism is true, then all these “things” either don’t exist or must have some non-rational physical cause. And we have no reason to think that such causes would provide us with a way of inferring correctly from a ground to a consequent (as Lewis puts it). 

Naturalism Refutes Itself

But naturalists normally trust the conclusions of natural science and typically believe they have arrived at their naturalist convictions by following evidence and sweet reason to their inevitable conclusion. If Lewis is right, however, then naturalism as a belief refutes itself. Consider any argument for naturalism. If it is sincerely offered, it will presuppose that people have beliefs and rational faculties that can affect their action because they can perceive the validity of the argument, or lack thereof, and act accordingly. 

Even prominent naturalists have admitted this dilemma. Lewis quotes famous naturalist J. B. S. Haldane to this effect. “If my mental processes are determined wholly by the motions of atoms in my brain,” said Haldane, “I have no reason to suppose that my beliefs are true… and hence I have no reason for supposing my brain to be composed of atoms.” (Haldane’s reference is to materialism — the idea that the fundamental constituents of reality are bits of matter. One could be a naturalist but believe nature consists of something other than matter. For our purposes, however, we’ll treat materialism and naturalism as synonyms.) 

In fact, even Darwin himself admitted the worry:

With me the horrid doubt always arises whether the convictions of man’s mind, which has been developed from the mind of the lower animals, are of any value or at all trustworthy. Would any one trust in the convictions of a monkey’s mind, if there are any convictions in such a mind?

The “cardinal difficulty of naturalism” doesn’t depend on a debatable theistic assumption. It emerges from the lack of causal tools in the naturalist toolkit. Strictly speaking, Lewis’s argument doesn’t show that naturalism is false, so much as it shows that naturalism can’t be rationally believed. Again, if naturalism were true, then beliefs, purposes, and inferences either wouldn’t exist or wouldn’t have any obvious power to transmit truth and so wouldn’t give us real knowledge of the world. That would apply to naturalistic belief as well. If naturalism were true and we believed in naturalism, we would lack the rational, truth-conducive faculties to consistently believe that it’s true, even less to know it. 

If, in contrast, the fundamental reality is mind or reason, then we would expect that our reasoning is at least sometimes reliable with respect to those matters that our reason was designed to grasp. 

Invoking Charles Darwin

The naturalist who is unfamiliar with this argument invariably invokes Darwin at this point. Naturalists tend to believe that Darwin’s account of the evolution of life is roughly correct. And they think the evidence establishes it. According to the Darwinian story, the adaptations of living things to their environment are not the result of purposeful design, but are the result of a blind process of natural selection acting on random variations within a population. Natural selection preserves and then propagates those variations that provide organisms with a survival advantage, and weeds out those that don’t. While there are other factors in evolution — genetic drift, bottlenecking, and so forth — this process of selection-and-random-variation largely creates these adaptations, according to Darwinian theory. This process is not mere chance or randomness; but it is blind and unconscious. There is no agent choosing variations, such as genetic mutations, based on the survival advantage that they confer on an organism, or for any other reason. 

If this story is roughly correct, then there would seem to be a survival advantage in forming true beliefs. Surely our ancestors would have gotten on in the world much better if they came to believe that, say, a saber-tooth tiger, is a dangerous predator. And if they believed that they should run away from dangerous predators, all the better. In contrast, those early humans who had false beliefs, who believed that saber-tooth tigers were really genies who would give three wishes if they were petted, would tend to get weeded out of the gene pool. So wouldn’t the Darwinian process select for reliable rational faculties, and so, give us faculties that would produce true beliefs?

Lewis argues that this process — which preserves survival-enhancing features — is nevertheless non-rational, and so cannot be expected to produce rational faculties. Again, if naturalism were true, then one would not expect minds and agents, choices and intentions to exist at all. If these things did exist, surely they would be mere epiphenomena of physical states. But let’s grant their existence, and even allow the naturalist the luxury of assuming that beliefs can guide our behavior. The naturalist will then want to argue that our reason and belief-forming faculties have been shaped by natural selection over eons, and so should be quite reliable.

The problem is that there are millions of beliefs, few of which are true in the sense that they correspond with reality, but all compatible with the same behavior. Natural selection could conceivably select for survival-enhancing behavior. But it has no tool for selecting only the behaviors caused by true beliefs, and weeding out all the others. So if our reasoning faculties came about as most naturalists assume they have, then we have little reason to assume they are reliable in the sense of giving us true beliefs. And that applies to our belief that naturalism is true.

This argument wasn’t original with Lewis. it appears in the Gifford Lectures given by British statesman Arthur Balfour in 1914. At the time, Balfour’s lectures were well known. They were even reported individually in the newspaper, and eventually published as the book Theism and Humanism, which Lewis credits as one of the ten books that most influenced him. But it is Lewis’s form of the argument that is still published, and read, in the twenty-first century.

This essay was adapted from Jay W. Richards’s chapter, “Mastering the Vernacular,” in The Magician’s Twin, edited by John West. For reference notes and sources, please consult the book version.