Why Believe That There Is a God?

Document Type: Research Paper

Author

Emeritus Professor of Philosophy, University of Oxford

Abstract

This article presents an argument for the existence of God, showing that the evident phenomena are best explained by supposing that a God causes them. The argument is based on the inductive force of four very evident general phenomena: that there is a physical Universe; that it is governed by very simple natural laws; that those laws are such as to lead to the existence of human bodies; and that those bodies are the bodies of reasoning humans, who choose between good and evil.

Keywords


The activity of producing arguments for the existence of God from premises “evident to the senses” has been part of the Christian and Islamic traditions from their earliest centuries—even though many in the Christian tradition have rejected it for the last two centuries. To adduce such arguments is to do “natural theology.” My own natural theology is inductive, that is it seeks to show that the evident phenomena are best explained by supposing that a God causes them, and that makes it probable that there is a God. In this article, I shall be able to consider only the inductive force of four very evident general phenomena: that there is a physical Universe; that it is governed by very simple natural laws; that those laws are such as to lead to the existence of human bodies; and that those bodies are the bodies of reasoning humans, who choose between good and evil. Due to length limitations, I shall not be able to discuss arguments against the existence of God, such as the argument from the existence of pain and other suffering.1

Theism, the claim that there is a God is an explanatory hypothesis, one which purports to explain why certain observed phenomena (that is, data or evidence) are as they are. There are two

Basic kinds of explanatory hypothesis: personal hypotheses and inanimate (or scientific) hypotheses. A personal hypothesis explains some phenomenon in terms of it being caused by a substance (that is a thing), a person, acting with certain powers (to bring about effects), certain beliefs (about how to do so), and a certain purpose (or intention) to bring about a particular effect, either for its own sake or as a step towards a further effect. Thus, the motion of my hand on a particular occasion may be explained by I (a substance), in virtue of my powers (to move my limbs), my belief (that moving my hand will attract attention), and my purpose (to attract attention) cause that motion. An inanimate (or scientific) explanation is usually represented as explaining some phenomenon in terms of it being caused by some initial state of affairs and the operation on that state of laws of nature. The present positions of the planets are explained by their earlier positions and that of the sun, and the operation on them of Newton’s laws. But I think that this is a misleading way of analysing inanimate explanation—because “laws” are not things; to say that Newton’s law of gravity is a law is simply to say that each material body in the universe has the power to attract every other material body with a force proportional to mm// r2 and the liability to exercise that power on every other such body. So construed, like personal explanation, inanimate explanation of some phenomenon (e.g., the present positions of the planets) explains it in terms of it being caused by substances (e.g., the sun and the planets) acting with certain powers (to cause material bodies to move in the way codified in Newton's laws) and the liability always to exercise those powers. So, both kinds of explanation explain phenomena in terms of the actions of substances having certain powers to produce effects. But while personal explanation explains how substances exercise their powers because of their purposes and their beliefs, inanimate explanation explains how substances exercise their powers because of their liabilities to do so.

I suggest that we judge a postulated hypothesis (of either kind) as probably true insofar as it satisfies four criteria. First, we must have observed many phenomena which it is quite probable would occur and no phenomena which it is quite probable would not occur, if the hypothesis is true. Secondly, it must be much less probable that the phenomena would occur in the normal course of things—that is, if the hypothesis is false. Thirdly, the hypothesis must be simple; that is, it must postulate the existence and operation of few substances, few kinds of substance, with few easily describable properties correlated in few mathematically simple kinds of ways.3 We can always postulate many new substances with complicated properties to explain anything we find, but our hypothesis will only be supported by the evidence if it is a simple hypothesis which leads us to expect the various phenomena that form the evidence. And fourthly, the hypothesis must fit in with our knowledge of how the world works in wider fields—what I shall call our “background evidence.”

I now illustrate these criteria at work in assessing postulated explanations. I begin with a postulated personal explanation.  Suppose that there has been a burglary; money has been stolen from a safe. A detective has discovered these pieces of evidence: John’s fingerprints are on the safe, someone reports having seen John near the scene of the burglary at the time it was committed, and there is in John’s house an amount of money equivalent to the amount stolen. The detective puts forward as the explanation of the burglary the hypothesis that John robbed the safe, using his normal human powers, in the light of his belief that there was money in the safe, with the purpose of getting the money. If John did rob the safe, it would be to some modest degree probable that his fingerprints would be found on the safe, that someone would report having seen him near the scene of the crime at the time it was committed, and that money of the amount stolen would be found in his house. But these phenomena are much less to be expected with any modest degree of probability if John did not rob the safe; they, therefore, constitute positive evidence, evidence favouring the hypothesis. On the other hand, if John robbed the safe, it would be most unexpected (it would be most improbable) that many people would report seeing him in a foreign country at the time of the burglary. Such reports would constitute negative evidence, evidence counting strongly against the hypothesis. I assume in my example that there is no such negative evidence. The more probable it is that we would find the positive evidence if the hypothesis is true, and the more improbable it is that we would find that evidence if the hypothesis is false, the more probable the evidence makes the hypothesis.

But a hypothesis is only rendered probable by evidence insofar as it is simple. Consider the following hypothesis as an explanation of the detective’s positive data: David stole the money; quite unknown to David, George dressed up to look like John at the scene of the crime; Tony planted John’s fingerprints on the safe just for fun; and, unknown to the others, Stephen hid money stolen from another robbery (coincidentally of exactly the same amount) in John’s house. If this complicated hypothesis were true, we would expect to find all the positive evidence which I described, while it remains not nearly as probable otherwise that we would find this evidence. But this evidence does not make the complicated hypothesis probable, although it does make the hypothesis that John robbed the safe probable; and that is because the latter hypothesis is simple. The detective’s original hypothesis postulates only one substance (John) doing one action (robbing the safe), which leads us to expect the various pieces of evidence; while the rival hypothesis, which I have just set out, postulates many substances (many persons) doing different unconnected actions.

 

But as well as the evidence of the kind which I have illustrated, there may be “background evidence”; that is, evidence about matters which the hypothesis does not purport to explain, but comes from an area outside the scope of that hypothesis. We may have evidence about what John has done on other occasions; for example, evidence making probable a hypothesis that he has often robbed safes in the past. This latter evidence would make the hypothesis that John robbed the safe on this occasion much more probable than it would be without that evidence. Conversely, evidence that John has lived a crime-free life in the past would make it much less probable that he robbed the safe on this occasion. A hypothesis fits with such background evidence insofar as the background evidence makes probable a theory of wider scope (e.g., that John is a regular safe-robber), which in turn makes the hypothesis in question more probable than it would otherwise be.

The same four criteria are at work in assessing postulated inanimate (or “scientific”) hypotheses.  Consider the hypothesis that Newton's theory of gravitation explains many phenomena known in 1687 when Newton proposed his theory: evidence about the paths taken (given certain initial positions) by our moon, by the planets, by the moons of planets, the velocities with which bodies fall to the earth, the motions of pendula, the occurrence of tides, and so forth. Newton’s theory consisted of his three laws of motion and his inverse square law of gravitational attraction. These laws were such as to make it very probable that previous observed phenomena, such as the positions of the sun and planets in 1677, would be followed by various phenomena observed in 1687, such as the positions of the planets in 1687. It would be very unlikely that the latter phenomena would occur if Newton’s theory were not true. There was no significant negative evidence. The theory was very simple, consisting of just four laws, the mathematical relations postulated by which were very simple (F=mm’/r2 being the most complicated one). Yet innumerable other laws would have satisfied the first two criteria equally well. Within the limits of accuracy then detectable, any law in which you substitute a slightly different value for the ‘2’ (e.g., ‘2.0000974’) would have satisfied the first two criteria as well as did the inverse square law; so too would a theory which postulated that the inverse square law held only until AD 2969, after which a quite different law, a cube law of attraction would operate; or a theory containing a law claiming that quite different forces operated outside the solar system. But Newton’s theory, unlike such theories, was rendered probable by the evidence, because it was a very simple theory, because it involved simpler mathematical numbers and relations. One number or mathematical relation is simpler than another if you can understand the former without understanding the latter, but not vice versa. Thus, ‘2’ is simpler than ‘2.0000974’; and note -  '0' (zero) is simpler than all numbers apart from ‘1.’ A law holding that only one mathematical relation operates is simpler than a law containing two different mathematical relations between material bodies—for example, one holding before AD 2969 and a different one holding thereafter. There was no relevant background evidence, because there was no evidence outside the scope of Newton’s theory making probable any explanatory theory (e.g., a theory of electromagnetism) with which Newton’s theory needed to fit. Hence, Newton’s theory was very probable on the evidence available in the seventeenth century, because it satisfied our four criteria; and so, therefore, is the hypothesis that it, together with the initial positions of the sun and planets, explains the present positions of the planets. Rephrased in a more satisfactory way, that hypothesis is the hypothesis that the sun and each of planets have simple powers and liabilities (as codified by Newton's laws) and initial positions, which explain the present positions of the planets. Although, unlike the hypothesis that John robbed the safe, it postulates, as an unexplained starting point, several substances (the sun and planets) with certain positions and velocities, it ascribes to each of them the same simple powers and liabilities, the operation of which explains their subsequent behaviour.

I stress again the importance of the criterion of simplicity. There are always an infinite number of mutually incompatible theories, which could be constructed, which predict any finite number of observed data when these would not otherwise be expected, yet make different predictions from each other about what will happen tomorrow. Without the criterion of simplicity, it would be impossible to predict anything beyond what we immediately observe.2 If the hypothesis is concerned only with a narrow field, it has to fit with any background evidence. But for many hypotheses, there may be no relevant background evidence, and the wider the scope of a hypothesis (that is, the more it purports to tell us about the world), the less background evidence there will be. For a very large-scale theory of physics (such as Quantum Theory), there will be few physical phenomena apart from those within its scope (ones which it purports to explain), and so little—if any—background evidence.

Such are the criteria for the probable truth of some postulated explanatory hypothesis. I now spell out the hypothesis of theism. Theism is clearly a personal hypothesis. God is supposed to be one person, who is essentially omnipotent, omniscient, perfectly free, perfectly good, and everlasting. (I note that Christianity claims that God is a Trinity—three persons of one substance. For this reason Christians must regard arguments such as mine as arguments to the existence of God the Father, on whom the other two persons ultimately depend.) A person is a being who has powers (to perform intentional actions; that is, actions which he or she means to do), beliefs, and purposes (choosing among alternative actions which to perform).  It is simpler to suppose that the cause of the universe has zero limits to his power (that is, is omnipotent), rather than that he can only make a universe of a certain size and duration; and it is simpler to suppose that he has zero limits (backwardly or forwardly) to the length of his life, rather that he came into existence only a trillion years ago or will only live for another trillion years. And it is simpler to suppose that the above properties belong to God essentially, rather than that he has them only by a fortunate accident. An omnipotent person can do any logically possible action, any action which can be described without contradiction; and so he cannot make me both exist and not exist at the same time. But since it makes no sense to suppose that I could both exist and not exist at the same time, a logically impossible action is not really an action at all, any more than an imaginary person is really a person. A truly omnipotent person would not be subject to irrational forces in forming his purposes, as so often are the choices of humans; he would be influenced by reason alone, and so by what he believes is good to do. In that sense of “perfectly free,” an omnipotent person is necessarily perfectly free. A truly omnipotent person would know all the possible actions open to him, and so know whether they are good or bad, and so, being perfectly free, would do actions only insofar as they are good. Therefore, he would be perfectly good.

But what does God's perfect goodness involve? So often, there must be before God, as there are before us, a choice between equally good incompatible actions. And, since God is omnipotent, the range of incompatible equal best actions available to him would be so much greater than the range available to us. Further, God must often be in a situation where we cannot be, of having a choice between an infinite numbers of possible actions, each of which is less good than some other action he could do. For example, elephants are good things; they can be happy and loving. So, the more of them the better (given that they are spread out among an infinite number of planets, so that they do not crowd each other out).  So however many elephants God creates, God must know that it would be better if he had created more. It may be, however, that when there is no best or equally best action available to God, there may be a best kind of action available to God, such that it would be better to do some action of that kind than to do any number of actions of any other incompatible kind. For example, God can create creatures of many different types, including angels, humans, and animals. If it were the case that it would be better to create at least some humans (even if he creates no angels or animals) than to create any number of angels and animals and no humans or to do an action of any other incompatible kind, then it would be a best kind of action for God to create some humans, although there would be no best number for him to create. In that case, I suggest, God, being influenced by reason alone, would inevitably create some humans. And if there are two or more equal best kinds of action available to him, he will inevitably do some action of one of these kinds. Since God's omnipotence only requires him to do the logically possible, the obvious way to understand God being “perfectly good” is that he will inevitably always do the best or equally best action; or if there is no such action, any action of the best or equally best kind; and if even that is not possible, some good action; but he would never do a bad action.             

God's perfect goodness thus follows from his omnipotence. Given the logical impossibility of backward causation, God will not be able to cause past events, but he will be able to cause any logically possible future event.  But in order to be able to know which future events are logically possible, he needs to know everything that has happened in the past—for what has happened in the past puts logical limits on what can happen in the future. For example, God cannot now immediately bring about a second world war, because such a war has already happened; and he cannot immediately bring about a fourth world war, because a third world war has not yet happened. But of logical necessity, his knowledge will be confined to knowledge of the past and of any necessary truths. He cannot know the future, for that will depends on his future choices.  Hence, just as omnipotence is to be understood as the power to do anything logically possible, so omniscience should be understood as knowledge of everything logically possible to know.

I conclude that theism is a very simple hypothesis indeed.  It postulates just one substance (God), having essentially the simplest degree of power (omnipotence), and lasting for the simplest length of time; all the other essential divine properties—including his omniscience, his perfect freedom, and his perfect goodness—follow from his everlasting omnipotence. God, being what he is in virtue of these essential properties, makes God a “person,” in a sense somewhat analogical to the sense in which we are persons. Theism is such a wide-ranging hypothesis (it purports to explain all the most general features of the universe) that there is no background evidence; all the evidence (whether positive or negative) is within its scope. Therefore, the hypothesis of theism satisfies the third criterion superbly well, and does not need to satisfy the fourth criterion. Hence, whether the hypothesis of theism (that there is a God) is probable on the evidence of the phenomena which I listed earlier turns on how well that evidence satisfies the first two criteria.

So, first, are the phenomena such as, if there is a God, it is probable that he would bring them about? If there is a God, he will seek to bring about good things. It is good that there should be a beautiful universe. Beauty arises from order of some kind—the orderly interactions and movements of objects in accord with natural laws is beautiful indeed, and even more beautiful are the plants and animals which evolved on Earth. Animals have sensations, beliefs, and desires, and that is clearly a great good.  Humans have the power to reason and understand the universe, and that is an even greater good.  But all these kinds of goodness are kinds of goodness which God himself possesses. God is beautiful and has beliefs and desires (and in my view, also sensations), and the power to reason and understand. But there is one kind of great goodness which God himself does not possess: the power bring about good or evil. God can only bring about good. Yet it would be very good indeed that there should be persons who have the free will to make this all–important difference to the world, the power to benefit or harm themselves, each other, and other creatures. So, if there is a God, we have very good reason to suppose that there will be persons who have, as I believe humans have, that freedom.3 but clearly there is a bad aspect to the existence of such persons; they may cause much evil. So it cannot be a unique best action to create such persons, but in view of the unique kind of goodness which they would possess, surely it must be an equal best action to create such persons. And, if so, it is as probable as not that if there is a God, there will be such persons; that is, persons like us humans. But if God is to create us, he must provide a universe in which we can exercise our choices to benefit or harm ourselves and each other. We can only do that if we have bodies, through which we can learn about the world and make a difference to it, and places where we can get hold of each other and escape from each other. But only if there are comprehensible regularities which we can discover, will there be ways in which my doing this or that will make a predictable difference to me or you, and so we can have a choice of how to treat each other. Only if humans know that by sowing certain seeds, weeding and watering them, they will get corn, can they develop an agriculture. And only if they know that by rubbing sticks together they can make fire, will they be able to burn the food supplies of others. But comprehensible, observable regularities are only possible if the fundamental laws of nature are simple ones. Further, if God is to create embodied humans, the laws must be such as to allow the existence of human bodies, either brought about by an evolutionary process or created directly by God. And finally, human bodies only have a point if they are controlled by conscious persons. God could have brought about the existence of humans, and so the necessary conditions for their existence which I have just been describing, either directly in one day or one week (as an over-literal reading of the Hebrew Bible Book of Genesis would suggest), or by an evolutionary process, which brought about other good things on the way, such as the beauty of the movements of the galaxies, stars, and planets, and the existence of plants and animals. And evidently, if God caused humans, the latter is the method which he adopted. So he acted by creating matter with inbuilt powers and liabilities such that in the course of time they produced organisms in an environment which led to the natural selection of those best fitted for survival, and so ultimately to human beings. So the four phenomena to which I have referred are to be expected (that is, it is quite probable that they will occur) if there is a God; the hypothesis of theism satisfies the first criterion of an explanatory hypothesis very well.

But if there is no God, it is immensely improbable that these phenomena will occur. It is enormously improbable that each of the innumerably many fundamental particles, or rather chunks of compressed energy, immediately after the Big Bang, should just happen to exist. And it is even more improbable that each such chunk should behave in exactly the same fairly simple way as each other chunk (the way codified in the laws of Relativity and Quantum theory and the four forces). So while there are fairly simple laws, their instantiation in each of innumerably many chunks of matter/energy would be an enormous coincidence unless caused by some external agent. And even if such an enormous coincidence occurred by chance, it is immensely improbable that those laws should be such as together with the boundary conditions of the universe (which are its initial conditions if the universe had a beginning) should have given rise to human bodies. They are “fine-tuned” to do so, in the sense that if the constants in the laws or the values of the density or momentum of the Big Bang had been different within at most one part in a million, humans would never have evolved. And even if this too occurred by chance, as far as any plausible scientific laws are concerned, the laws might just as easily have given rise to robots.

To explain the fact that human bodies are the bodies of conscious beings, there must be innumerable different laws connecting different brain events with different conscious events (different sensations, different occurrent thoughts, different intentions). One reason why there are (probably) fairly few fundamental laws of physics from which all other physical laws follow, is that these laws correlate measurable quantities of one variable (such as the position and velocity of one particle) with measurable quantities of another variable (such as the position and velocity of another particle). But, while brain events differ from each other in measurable ways, in general, conscious events do not differ from each other in measurable ways. My thought that today is Tuesday does not have a measurable value, such that it could be a consequence of some general law that a brain event with properties having certain numerical values would cause the thought that today is Tuesday, whereas a brain event with properties with slightly different values would cause the thought that today is Wednesday.  And because the properties of conscious events do not have numerical values, there must be separate laws stating that a brain event with certain numerical values causes the thought that today is Tuesday, and a different law stating that a brain event with certain other numerical values causes the thought that today is Wednesday, without there being a general law from which these two lower-level laws could be derived. And, to generalize that result, there must be a very large number of separate psychophysical laws. If we add these laws to the laws of physics, the overall psychophysical law system becomes very complicated indeed, and so again such as would be most unlikely to arise by chance. But, of course, God has reason to bring about such complicated laws in order thereby to bring about human beings.  Human-type consciousness is totally improbable unless there is a creator who created physical objects inbuilt with such complicated powers to produce particular thoughts, feelings, and intentions, under particular circumstances.

Some contemporary physicists will tell you that we live in a multiverse such that many different possible universes (with different laws of nature, and different initial conditions) will eventually occur, and so it is not surprising that there is one like ours. But we could only have reason to believe what the physicists tell us, if the most probable explanation of phenomena observable in our universe is that the most general laws of nature are such as to bring about these many universes; and to postulate that is to postulate that all the particles, not merely of our universe but of the vastly bigger multiverse, behave in accord with the same very general laws, which throw up particular variants thereof in different universes—which is to postulate an even bigger coincidence. And the laws of that multiverse would have to be such as to produce at some stage a universe like ours, which in turn produces us, when almost all possible multiverses would not have this characteristic. So even if our universe does belong to a multiverse, it is immensely improbable (if there is no more ultimate explanation thereof—e.g., God) that that would be a multiverse of a kind to bring about the existence of humans. So the possible existence of a multiverse makes little difference to the force of the arguments which I have discussed.

So these four general phenomena are such as it is moderately probable will occur if there is a God, and almost certainly will not occur if there is not a God. Theism is a very simple hypothesis indeed, and simpler, I suggest, than any inanimate hypothesis which could be constructed. I conclude that arguments from the phenomena which I have discussed are strong, cogent arguments to the existence of God.

1. For the fully developed account of my natural theology (including my treatment of arguments against the existence of God), see my The Existence of God, 2nd ed., Oxford University Press, 2004; for a shorter version, see is There a God? Oxford University Press, 1996, Persian translation published subsequently.

2. For a full account of the nature of simplicity, see my Simplicity as Evidence of Truth, Marquette University Press, 1997; or my Epistemic Justification, Oxford University Press, 2001, chapter 4.

3. For my defence of the claim that humans have libertarian free will—that is, freedom to make choices, either good or evil, despite all the influences to which they are subject—see my Mind, Brain, and Free Will, Oxford University Press, 2013.