The Argument from Evil and the Existence of God
The idea that at least some of the evil that is present in the world constitutes an objection to belief in the existence of God is both an ancient idea - going back at least to Job, and presumably beyond - and a very natural one. How best to develop the basic idea, however, and to convert it into an explicitly formulated argument in support of the non-existence of God, has only recently become clear. For, first of all, there is a crucial choice between, on the one hand, purely deductive formulations of the argument from evil, according to which the proposition that God exists is logically incompatible with the existence of evil - or, at least, certain types, or certain instances, of evil - and, on the other hand, inductive formulations of the argument from evil, according to which certain facts about the evil in the world render it more or less unlikely that God exists. Secondly, if one opts - as I think one should - for an inductive formulation of the argument from evil, then it turns out, as has become clear in recent discussions, that arriving at the correct account of the logic of the argument is rather more problematic than one might initially have expected. Thirdly, there is also an important choice between axiological formulations of the argument, and deontological ones.
The structure of my discussion is as follows. I shall begin, in sections one and two, with some preliminary discussion of the meaning I shall be assigning to the term "God", and with a contrast between the concept of God, thus understood, and God as conceived of in some historical religions. Next, in section three, I shall draw a very familiar contrast between deductive (or incompatibility) versions of the argument from evil, and inductive (or evidential, or probabilistic) versions. From that point on, the focus will be on arguments of the second type, and in section four I shall, first, briefly describe some of the types of states of affairs to which formulations of argument from evil typically appeal, and then, secondly, I shall draw a contrast between axiological formulations of the argument from evil, and deontological ones. Next, in section five, I shall briefly describe an important objection to evidential versions of the argument from evil - an objection that involves the claim that humans suffer from cognitive limitations, and that, in particular, humans probably do not possess complete moral knowledge. Then, in section six, I shall set out one very natural response to that objection - a response advanced, in effect, by William Rowe in his article, "Ruminations about Evil". But though the response is a very natural one, I shall argue in section seven that it is unsound. Next, in sections eight and nine, I shall consider an alternative, Bayesian formulation of the argument from evil, as offered by Rowe in a later article, "The Evidential Argument from Evil: A Second Look". But here too I shall contend that the argument that Rowe has offered does not succeed.
Is it, then, time to rejoice? Can we conclude that the evidential argument from evil fails, and thus that the existence of God is, perhaps, not unlikely after all? Unfortunately, this does not seem to be the case, as will emerge in the last two sections. In the first of these, I shall argue that, on pain of being able to demonstrate that a very unwelcome form of moral skepticism is true, there must be an answer to the above objection. But what is the answer? That will be the subject of the final main section of this paper.
1. The Meaning of the Term "God"
Before turning to the argument from evil, I want to comment briefly upon the interpretation that I'll be assigning to the term "God" - since this term is used in very different ways by different people. The interpretation that I favor is based, I believe, upon things that are crucial to the use of the term in religious contexts, as contrasted with certain purely metaphysical contexts.
What is central to the notion of God, thus understood, is, I suggest, the idea of an entity that is an appropriate object of certain characteristically religious attitudes, particularly the attitude of worship. If so, then the question of how the word "God" might most appropriately be employed in such contexts comes down to the following question: "What characteristics should an object possess if it is to be an adequate object of worship, and other religious attitudes?"
Let us now consider what answer might be given to this question. In the first place, then, many people have felt - and this would seem to be a very natural and plausible view - that an impersonal object or force would not be an appropriate object of worship, and for this reason they have insisted that it should be part of the meaning of the term "God" that God is a person.
Similarly, many people have felt that if worship of God is to be morally commendable - if it is not to be merely an abject bending of the knee before awesome power - then it must be part of what is meant by the term "God" that God is supremely good, or morally perfect.
Is it necessary to include any other characteristics in the definition of "God"? Perhaps not, if one were concerned only with an adequate object of worship. But there are other religious attitudes that need to be taken into account. One that is especially important is the religious person's faith that in the end good will triumph - that life is not, ultimately, either meaningless or tragic. One of the functions of God is to guarantee that this is no empty hope. What characteristics must God possess, then, if his morally perfect purposes are ultimately to be realized? The answer seems clear: God must possess both very great power and very great knowledge, at the very least, and ideally there should be no limitations at all either upon his power or upon his knowledge. Thus, if there is to be no room for doubt whether God's morally perfect purposes will ultimately be achieved, God must be omnipotent and omniscient.
The conclusion, then, is this. If one is using the term "God" to refer, not to some purely metaphysical entity - such as a first cause, or the ground of being, or being itself - but to a being that will be an appropriate object of religious attitudes, one should require that, in order to be God, a being must be (1) personal, (2) morally perfect, (3) omnipotent, and (4) omniscient. This, accordingly, is how I shall be using the term "God" in this paper.
Is this crucial with respect to the argument from evil? In the case of some deductive formulations of the argument from evil, it is, since the existence of a deity that is not morally perfect, or not omnipotent, or not omniscient, may well be logically compatible with the existence of at least some evil in the world. But it is a mistake to think that the argument from evil only applies to a deity that is omnipotent, omniscient, and morally perfect deity. Some deductive versions, and most, if not all, inductive versions of the argument from evil, will bear not only upon the existence of deities that are not omnipotent, omniscient, and morally perfect, but also upon the existence of deities that fall considerably short with regard to the possession of each of these properties.
2. God Versus the Deities of Some Historical Religions
Another preliminary issue that calls for at least two brief comments concerns the relation between God, thus conceived, and the deities believed in by people in various historical religions. The first comment is this. On the one hand, Christians usually want to say that the god that they worship is, if he exists, identical with God as I have defined him. Similarly, Muslims would, I think, identify Allah with my conception of God, while Jews would do the same for their deity, Yahweh. But, on the other hand, these identifications are by no means unproblematic, since it is not clear that the deities associated with these historical religions can be characterized as perfectly good. Suppose, for example, that Allah, as some Muslims have believed, approves of holy wars against infidels - that is, against people who reject Islam. Is Allah, then, plausibly described as perfectly good? Or suppose that, as the Roman Catholic Church teaches, and as Protestant Fundamentalists believe, Hell exists, and that most of the human race will wind up there, suffering unending torment. Can this god of Catholicism and Protestant Fundamentalism be characterized as morally perfect?
It is important to keep this issue sharply in focus. For though I'm an atheist, I would very much like it to be the case that I am mistaken, and that it turns out that God, as I have defined him, does exist. But on the other hand, the existence of the god of Protestant Fundamentalism, or of Roman Catholicism, is not something that I would welcome, for it would mean that the world, while certainly not the worst imaginable, would be very bad indeed.
Secondly, suppose that a believer in Judaism insists, contrary to what I've claimed, that Yahweh, understood as a deity who performed all of the actions attributed to him in the Torah, is morally perfect, and also omnipotent and omniscient. Or that a Christian insists, contrary to what I've claimed, that the deity whose actions are described in the Old Testament and the New Testament, is morally perfect, and also omnipotent and omniscient. Or that a Muslim insists that Allah, understood as the deity whose actions are described in the Koran, is morally perfect, and also omnipotent and omniscient. How does this effect the argument from evil? Part of the answer is that such a believer is now confronted with an ad hominem version of the argument from evil - that is, a version that appeals to claims that the believer accepts about the deity in question. In particular, rather than appealing, as in the case of standard versions of the argument from evil, only to cases where, if there is an omnipotent and omniscient deity, that deity has, by failing to act in certain ways, done something that is prima facie wrong, one can also appeal to cases where actions that the deity is believed to have performed are prima facie wrong. Thus, for example, in the case of Orthodox Judaism, or of Protestant Fundamentalism, one can appeal to such things as Yahweh's act of drowning all men, women, and children in a great flood, with the exception of a single family - as described in the story of Noah's ark - or to Yahweh's command to Saul, to kill all of the Amalekites: "Go now and fall upon the Amalekites and destroy them, and put their property under ban. Spare no one; put them all to death, men and women, children and babes in arms, herds and flocks, camels and asses." (1 Samuel, 15:3) Similarly, in the case of forms of Christianity that are not too casual about rejecting troubling passages in the New Testament, one can appeal to the action of the Christian deity in creating a place where most of the human race will undergo eternal torment.
3. Two Very Different Types of Formulation of the Argument from Evil: Deductive/Incompatibility Arguments versus Inductive/Evidential Arguments
The argument from evil focuses upon the fact that the world appears to contain states of affairs that are bad, or undesirable, or that should have been prevented by any being that could have done so, and it asks how the existence of such states of affairs is to be squared with the existence of God. But the argument can be formulated in two very different ways. First, it can be formulated as a purely deductive argument that attempts to show that there are certain facts about the evil in the world that are logically incompatible with the existence of God. One especially ambitious form of this first sort of argument attempts to establish the very strong claim that it is logically impossible for it to be the case both that there is any evil at all, and that God exists.
So understood, the argument typically runs as follows. If God exists, he will want to eliminate evil, since he is by definition morally perfect. Being omniscient, he will know about any evil that happens to exist. And being omnipotent, he will have the power to eliminate any evil. So if God exists, he will be willing and able to eliminate any evil that there is. Therefore, if God exists, there will not be any evil. But the world does contain evil. Therefore, God does not exist.
This form of the argument initially has a striking and perhaps impressive quality. It seems very doubtful, however, that it is sound. The reason is that the claim that a morally perfect being would want to eliminate any evil that exists is not unproblematic. For what this claim can be seen to rest upon, when it is scrutinized, is the assumption that no evil is ever logically necessary for some good state of affairs that outweighs it. Is this claim true? Some people have argued that it is isn't. For example, some people have argued that the world is a better place if people develop desirable traits of character - such as patience, and courage - by struggling against obstacles, including suffering. But if this is right, then the elimination of all suffering would actually make the world a worse place, by depriving people of the chance of developing desirable traits of character by responding appropriately to suffering that they undergo.
The examples that are usually advanced of cases where some evil is logically necessary for a greater good that outweighs the evil do not seem to me convincing. But, on the other hand, I do not think that one can establish, without appealing to some substantive, and controversial, moral theory, that there cannot be cases where some evil is logically necessary for a greater good that outweighs it. Consequently, it seems to me that the argument from evil should be formulated in a different way - not as a deductive argument for the very strong claim that it is logically impossible for both God and evil to exist, (or for God and certain types, or instances, of evil to exist), but as an inductive (or evidential or probabilistic) argument for the more modest claim that there are evils that actually exist in the world that make it very unlikely that God exists.
4. The Evidential Argument from Evil
So understood, the argument from evil claims that there are undesirable states in the world that provide strong evidence against the existence of God. In this section, I shall briefly set out some of the types of undesirable states of affairs to which advocates of the argument from evil appeal, and then I shall go on to draw a contrast between axiological formulations of the argument, and deontological formulations.
4.1 A Brief Catalogue of Some Notable Evils
First, then, what are some of the undesirable states that one might have in mind here? The answer is that there are a number of different types, including at least the following:
(1) Extreme moral evil. Here I am thinking of the suffering brought about by very naughty individuals, such as Hitler and Stalin, by acts of attempted genocide, by great wars - including religious wars - by the Inquisition, and so on.
(2) The suffering endured by innocent children Here one has the suffering caused both by diseases such as muscular dystrophy, leukemia, cerebral palsy, and so on, and by abuse inflicted upon children by adults. (In a famous illustration, William Rowe refers to the case of a young girl, Sue, who was brutally raped and murdered by her mother's boyfriend.)
(3) The suffering of animals. This will include such things as suffering at the hands of other animals, and suffering due to natural disasters. (Here Rowe cites a Bambi-type case, where a deer suffers an agonizing death due to a forest fire.)
(4) The suffering that adults endure as a result of terrible diseases - such as cancer, mental illness, Alzheimer's disease, and so on.
(5) 'Design faults' in human beings that contribute greatly to human suffering and unhappiness - such as, for example, the following:[1]
(a) The sinuses are misdesigned: the lower sinuses open upward, and thus they do not drain properly, with the result that they may become infected, and cause, in some cases, severe headaches. (Evolution provides an explanation of both good "design" and bad "design". Thus, in the case of our sinuses, they would be fine if we were four-legged, rather than two-legged. But if the theist attempts to appeal to this explanation, he or she needs to say why evolution is a good way of designing intelligent beings. Why leave things at the mercy of a probabilistic process that has had, as one would have expected, bad results?)
(b) As in the case of the sinuses, so with the human spine: while its design is not too bad in the case of four-legged animals, it is not a good piece of engineering in the case of two-legged animals. This bad design, in turn, means that many humans suffer from back problems, some of which are very debilitating and painful.
(c) Wisdom teeth. Most people today in affluent societies have their wisdom teeth extracted. That operation can itself lead to complications, in the form of sinus infection, and damage to nerve pathways leading to the lower lip and the tongue. But one needs only to go back to the 18th century to find a time when the presence of wisdom teeth had much worse consequences, since impacted wisdom teeth, by becoming infected, could then lead not only to considerable pain, but to septicemia, and to death.
(d) Childbirth in the case of humans. The size of the human head relative to the size of the birth canal has three unfortunate consequences. First, humans are born in a much more underdeveloped, and more vulnerable state, than newborns of other species. Secondly, childbirth is often a very painful experience. Thirdly, childbirth is potentially a very dangerous event for the woman. Today, of course, a much smaller proportion of women die in childbirth, especially in more affluent countries, because of the use of birth via Cesarean section, etc., and because women tend to have far fewer children than in the past - though various religious groups, such as Orthodox Jews, still have many children. In the past, however, many women died in childbirth.
(e) Men and women differ in various ways. One interesting way which has recently been discovered involves a gene called gastrin-releasing peptide receptor - or GRPR for short - which is linked to abnormal growth of lung cells. It had been noted earlier that women were more likely to develop lung cancer than men, without smoking more, and it turns out that the explanation is that while the GRPR gene in not active in men unless they smoke, it is active in 55% of non-smoking women. (The reason for this is connected with the fact that the gene is on the X chromosome, of which women have two, and men only one. So greater susceptibility to lung cancer is programmed into women.)
(f) Declining hormone levels as one grows older. This affects men as well as women, but the effects are especially dramatic in the case of women, where the fall in the level of estrogen increases greatly the probability both of death due to heart attacks, and of osteoporosis, the latter of which often leads to significant suffering and loss of mobility due to bones that break easily.
(g) The body is equipped with sensors that detect injury, and that then announce the presence of injury via painful sensations. These injury detectors are not well designed, in at least two ways. First, they are not sensitive to the presence of many life-threatening bodily changes. In particular, the presence of cancer is often detected only after it is too late to do anything about it. Secondly, there is no way of shutting down these injury detectors in situations where, rather than providing the individual with a useful warning of bodily damage, they only contribute to the person's misery by producing pain sensations, often of an unbearable level of intensity. So the injury detectors, which failed to warn a person of the early presence of cancer, swing into action when it is too late, at which point they fill the dying person's last weeks with excruciating pain.
(h) Fat and appetite. When people become overweight, there is no reduction in appetite, nor is the mechanism that enables one to make use of stored fat an effective and well-designed one, nor does the body cease extracting and storing calorie-rich compounds, such as fats, from the food that it is processing. The result, once again, is enormous suffering because people are overweight - with much overeating resulting from the fact that food can provide comfort when one is under stress, etc.
(i) The body contains a variety of defense mechanisms to deal with threats posed by bacteria, viruses, toxins, etc. But viruses often are capable of countermeasures, sometimes of quite a sophisticated sort, that enable them to foil the body's defense mechanisms.
(j) A disease that has caused many deaths is malaria. There is a gene, however, that provides an effective defense against malaria, which it does by destroying any red blood corpuscles that have been occupied by any of the types of parasitic protozoans that cause malaria. So far, so good. But if one has inherited the relevant gene from both of one's parents, the result is sickle cell anemia, in which sickle cells crystallize within red blood corpuscles, distorting them, and clogging blood vessels, which then results in intense suffering.
(k) Humans are sexually mature some time before they exhibit sufficient emotional maturity, and the upshot is that quite young girls can bear children long before they have the emotional responsibility and commitment needed to care for children satisfactorily.
(l) The association of intense pleasure with sexual activity also appears to be a design fault. For while sexual pleasure can certainly contribute to human happiness, it appears that when one totes everything up, the world might well be better off if people reproduced simply because they wanted to have children, and if people were not led, by the pursuit of pleasure, into bad decisions in this area.
(m) Hume's example of the capacity to experience pain. (Rather than using decline in pleasure to motivate, as Hume proposes, I favor awareness of the area that is being threatened with damage, together with automatic withdrawal, etc., mechanisms.)
(n) The susceptibility of humans to aging, involving both physical impairment and the destruction of the mental capacities that make one a person, is another obvious design fault.
(o) The unavoidability of bodily death, which is generally unwelcome, and which makes impossible the pursuit of a variety of interesting and satisfying projects.
(6) 'Design faults' in the world outside of human beings that contribute greatly to suffering, either by humans or by other sentient beings, such as:
(a) The existence of carnivorous animals.
(b) The existence of resources that are too limited to provide for populations of humans, and other animals, that are expanding at natural rates. The world could instead be an infinite plane, or have inhabitable planets that were easily accessible.[2]
4.2 Hypothetical Evils and the Evidential Argument from Evil: The Problems of the Hiddenness and Silence of any Deity that Exists
The list of some notable, undesirable states of affairs that I have just set out involve only actual states of affairs. It is important to note, however, that there are what might be called hypothetical evils that are relevant to the argument from evil. Specifically, there are states of affairs that can only exist if there is a being with certain properties, but which are such that if they do exist, then, first of all, they appear to be highly undesirable, and secondly, they bear upon the existence of God.
To illustrate this, suppose that the world does contain an omnipotent and omniscient person. Then given that assumption, the following states of affairs exist, and appear highly undesirable:
(1) The Hiddenness of an Omnipotent and Omniscient Being
If there were an omnipotent and omniscient being, such a being could make his existence much more evident than it is. Knowledge of the existence of an omnipotent and omniscient being is, however, surely a very valuable thing.
(2) The Silence of an Omnipotent and Omniscient Being
Human history involves enormous suffering. An omnipotent being could have prevented much of that suffering. If there is such a being, that being has thus intentionally refrained from preventing the evils in question, and that fact naturally leads not only to the question of why the omnipotent and omniscient being in question intentionally refrained from preventing those evils, but also to the question of why that being says nothing about why he intentionally refrained from preventing those evils. Why does such a being remain silent in the face of evils such as the Holocaust?
In short, the point is that it is not merely doing nothing to prevent great evils that is morally problematic: it is also morally problematic to remain silent, to say nothing about why one is allowing the evils in question.
4.3 Two Versions of the Evidential Argument from Evil
Secondly, given the impressive array of evils mentioned in the preceding two subsections, let us now consider how an inductive or evidential version of the argument might be formulated. One way of formulating such a version of the argument from evil is as follows:
The Evidential Argument from Evil: An Axiological Formulation in terms of the Goodness and Badness of States of Affairs
(1) Cases where animals die agonizing deaths in forest fires, or where children undergo lingering suffering and eventual death due to cancer, are states of affairs which are intrinsically bad or undesirable, and highly so, and which an omnipotent and omniscient person could have prevented without thereby either allowing an equal or greater evil, or preventing an equal or greater good.
(2) An omniscient and morally perfect person would prevent the existence of any states of affairs which were intrinsically bad or undesirable, and highly so, and whose prevention he could achieve without either allowing an equal or greater evil, or preventing an equal or greater good.
Therefore:
(3) There is no omnipotent, omniscient, and morally perfect person.
There are three comments that I want to make on this argument. First, it involves only two premises, and of these, the first is the one that is most likely to be challenged. What grounds are there for thinking that it is likely to be true?
One very natural line of thought is as follows. First, events that involve animals dying agonizing deaths in forest fires, or children undergoing lingering suffering and eventual death due to cancer, are surely cases of states of affairs which are intrinsically bad, and highly so. Secondly, it is also surely true that an omnipotent and omniscient person could prevent such states of affairs. So the crucial question is whether an omnipotent and omniscient person could do so without thereby either allowing an equal or greater evil, or preventing an equal or greater good.
How might a defender of the above version of the argument from evil argue that it is reasonable to believe that an omnipotent and omniscient person could prevent such occurrences without thereby either allowing an equal or greater evil, or preventing an equal or greater good? One very natural line of thought, it would seem, involves holding that not only are each of the following two claims reasonable, their conjunction is also reasonable:
(a) A person's preventing the existence of the occurrences in question would not make possible the existence of an equal or greater evil;
(b) A person's preventing the existence of the occurrences in question would not rule out the existence of an equal or greater good.
The idea would then be that if it is reasonable to accept the conjunction of (a) and (b), then it is also reasonable to accept the conjunction of the following two claims:
(c) If an omnipotent and omniscient person were to prevent the existence of the occurrences in question, then that would not make possible the existence of an equal or greater evil;
(d) If an omnipotent and omniscient person were to prevent the existence of the occurrences in question, then that would not rule out the existence of an equal or greater good.
But if this is right, then there is good reason for accepting the crucial, first premise in the core argument set out above.
The second observation is that the argument, as formulated above, is a purely deductive argument. But the idea is that when the claim that is made in the first premise is challenged, it will turn out, first, that it can only be supported by arguing for claims (a) and (b) - or something rather similar - and secondly, that the latter claims can only be supported, in turn, by some sort of inductive argument.
Thirdly, a feature of the above statement of the argument which seems to me important, but which is not often commented upon, is that the argument is formulated in terms of axiological concepts, in terms of the goodness or badness of states of affairs. I am not convinced that this is the best way of setting out the argument. For one thing, such a formulation fails to make explicit how a failure to bring about good states of affairs, or a failure to prevent bad states of affairs, entails that one is acting in a morally wrong way, and because this is not explicit, the question arises as to whether one is in fact implicitly appealing to some form of consequentialism - such as, for example, the view that an action that fails to maximize the balance of good states of affairs over bad states of affairs is morally wrong.
In short, an axiological formulation may either be incomplete in a crucial respect, or it may appeal to controversial ethical claims, such as the claim that the right action is the one that maximizes expected value. The result in turn may be that discussions become sidetracked on issues that are not really relevant - such as, for example, whether God is morally blameworthy if he fails to create the best world that he could.
I would favor, then, a different sort of formulation:
The Evidential Argument from Evil: A Deontological Formulation in terms of Rightmaking and Wrongmaking Properties
(1) The property of intentionally allowing an animal to die an agonizing death in a forest fire, or a child to undergo lingering suffering and eventual death due to cancer, is a wrongmaking characteristic of an action, and a very serious one.
(2) Our world contains animals that die agonizing deaths in forest fires, and children who undergo lingering suffering and eventual death due to cancer.
(3) An omnipotent being could prevent such events if he knew that they were about to occur.
(4) An omniscient being would know that such events were about to occur.
Therefore:
(5) If there is an omnipotent and omniscient being, then there are cases where he intentionally allows animals to die agonizing deaths in forest fires, and children to undergo lingering suffering and eventual death due to cancer.
(6) In many such cases, we do not know of any rightmaking characteristics that we have good reason to believe would apply to the case in question, and also be sufficiently serious to outweigh the wrongmaking characteristic in question.
Therefore:
(7) If there is an omnipotent and omniscient being, then there are specific cases of such a being's intentionally allowing animals to die agonizing deaths in forest fires, and children to undergo lingering suffering and eventual death due to cancer, that have a wrongmaking property that is not, as far as we know, outweighed by any rightmaking properties.
Therefore it is likely that:
(8) If there is an omnipotent and omniscient being, then that being intentionally refrains from performing certain actions in situations where it is morally wrong to do so, all things considered.
Therefore:
(9) If there is an omnipotent and omniscient being, then that being both intentionally refrains from performing certain actions in situations where it is morally wrong to do so, all things considered, and knows that he is doing so.
(10) A being that intentionally refrains from performing certain actions in situations where it is morally wrong to do so, all things considered, and knows that he is doing so, is not morally perfect.
Therefore:
(11) If there is an omnipotent and omniscient being, then that being is not morally perfect.
Therefore:
(12) There is no omnipotent, omniscient, and morally perfect being.
(13) God is, by definition, an omnipotent, omniscient, and morally perfect being.
Therefore:
(14) God does not exist.
When the argument from evil is formulated in this way, a non-deductive inference is explicitly present in the argument itself - namely, in the inference from (7) to (8). But what exactly is the form of this inductive inference, and is it sound? In section 12 of this paper, I shall address those questions.
Finally, though I believe that a deontological formulation of the argument from evil is to be preferred to an axiological formulation, this issue will not bear upon the crucial question of the correct account of the logical form of the basic inductive step that is involved in the argument. In discussing answers to that crucial question, accordingly, I shall put things in terms of the statement of the argument that has been advanced by the person whose proposal I am considering.
4.4 The Naturalness of the Argument from Evil
The evidential argument from evil, formulated in either of the two ways just set out, is a very natural argument. For, at bottom, the basic idea is simply that, given that judgments about the character of a human agent based upon his actions and/or intentional omissions are frequently justified, why shouldn't similar conclusion be justified in the case of more powerful agents, including deities?
Initially, then, it may appear that, whatever the exact account of the inductive inference involved, the evidential argument from evil must be sound, since one can use reasoning that is exactly parallel to the reasoning that one uses in the case of human persons, so that the argument from evil could only be unsound if it were impossible to draw any conclusions about the moral character of other human persons from information about the actions that those persons do and do not perform. But surely the latter conclusions cannot be unjustified, and so the argument from evil must be sound as well.
It turns out, however - as we shall see in the next section - that there is a type of reasoning which one can employ in the case of human persons, and which seems perfectly sound, but which is not sound if applied to the case of an omniscient person. This being the case, the question arises as to whether there is some other form of reasoning which is also justified, and which can be used at the crucial inductive step in the evidential argument from evil.
5. A Crucial Response: Human Knowledge Is Incomplete
Let us now see why the evidential argument from evil is less straightforward than it may initially seem. Consider the axiological formulation of the argument from evil set out in the preceding section. As noted there, the core argument is itself a deductive argument, but the argument as a whole is inductive or probabilistic, since it is being claimed only that the crucial premise is very likely to be true, given relevant evidence, and not that one can know that it is true. Thus, the advocate of the above argument will happily concede that it could be the case that the events in question are in fact such that their prevention would either have made possible the existence of an equal or greater evil, or have ruled out the existence of an equal or greater good. The claim is merely that it is very unlikely that either of these things is so.
But can even this modest claim be sustained? One way of arguing that it cannot, and which has recently been advanced fairly frequently, appeals to the proposition that, since human cognitive powers are finite, there may very well be relevant, morally significant states of affairs of which humans have either no knowledge, or else knowledge that is limited in a crucial respect. This could happen, moreover, in any of three ways. First, there may be non-moral facts about the world of which humans have no knowledge, and where those non-moral facts are morally important. Thus it may be a fact, for example, that humans survive death, and have the opportunity of enjoying a certain sort of afterlife, even if humans have no knowledge that this is so. Such a fact might, moreover, be a very important one. Secondly, it may be that there are non-moral states of affairs that humans know to be morally significant, but which are logically connected, in ways that humans do not appreciate, with the morally troubling events on which the argument from evil focuses. Thirdly, it may be that humans do not have complete moral knowledge, and thus that they are unable to assess correctly the overall moral value of non-moral facts of which they are aware.
The idea, then, is that if one of these things is so, then the prevention, by an omniscient, omnipotent, and morally perfect person, of events such as those involving terrible suffering on the part of animals or children, might very well either (1) make possible the existence of states of affairs that possess badmaking properties, or else (2) rule out states of affairs that involve goodmaking properties, and where, moreover, the following two things are true: first, humans are unacquainted either with the non-moral states of affairs in question, or with the logical connections between those states of affairs and the relevant evils, or with some of the relevant, morally significant properties of those states of affairs; secondly, the morally significant states of affairs in question are sufficiently weighty to make it the case that the prevention of the events in question by an omnipotent and omniscient and morally perfect person would be morally worse than allowing those events to happen.
The problem, in short, is that there is a type of reasoning which works in the case of a human agent, but which does not work in the case of an omniscient person - namely, reasoning which makes use of the following claim:
(*) The agent in question is unlikely to be aware of important factors that I am not aware of, and so if the agent's action is morally wrong as judged by my information and moral standards, the agent is unlikely to have any grounds for overthrowing my assessment. So it is likely that the agent's action is morally wrong.
In the case of other humans, this claim seems reasonable: it is very unlikely, for example, that Hitler possessed either superior knowledge, or a fuller appreciation of morality, which would provide him with grounds for concluding that the Final Solution was morally sound. An omniscient being, on the other hand, might well possess additional knowledge, both of a purely factual sort, and of a moral sort. The use of assumption (*) in the case of an omniscient person, accordingly, is surely unsound.
6. The Question of the Crucial Inductive Inference
The evidential argument from evil initially seemed both natural and relatively unproblematic, since it appeared to be just another instance of a type of reasoning that we use everyday in drawing conclusions about the moral character of other human persons. We have just seen, however, that there is a type of reasoning that one might well be using in the human case that would not be acceptable in the case of the evidential argument from evil, since there we are attempting to draw conclusions about the moral character of a being that is omniscient. So the question is whether the idea that the evidential argument from evil shows that it is unlikely that God exists simply rests upon a mistake, or whether, on the contrary, there is a satisfactory account of the reasoning that is involved at the crucial inductive step.
In the next four sections, I shall set out and discuss two important attempts - both advanced by William Rowe - to show that the inductive step can be justified. I shall argue that neither attempt is successful. Given the failure of those attempts, it seems that finding an account of the reasoning that renders it sound is not an easy matter, and one might well wonder whether the evidential argument from evil is not in fact unsound. I shall then argue, however, that that is not the case.
7. William Rowe's Discussion and Defense of the Crucial Inference
William Rowe has discussed the objection that appeals to human cognitive limitations in a number of articles. In this section, I shall focus upon the response that he offers in his article, "Ruminations about Evil".[3]
In that article, Rowe formulates the premise of the crucial inference as follows:
(P) No good state of affairs that we know of is such that an omnipotent, omniscient being's obtaining it would morally justify that being's permitting E1 or E2.
Commenting on (P), Rowe emphasizes that what proposition (P) says is not simply that we cannot see how various goods would justify an omnipotent, omniscient being's permitting E1 or E2, but rather that
"The good states of affairs I know of, when I reflect on them, meet one or both of the following conditions: either an omnipotent being could obtain them without having to permit either E1 or E2, or obtaining them wouldn't morally justify that being in permitting E1 or E2."[4]
Rowe then goes on to say that "if this is so, I have reason to conclude that:
(Q) No good state of affairs is such that an omnipotent, omniscient being's obtaining it would morally justify that being's permitting E1 or E2."[5]
Rowe uses the letter 'J' "to stand for the property a good has just in case obtaining that good would justify an omnipotent, omniscient being in permitting E1 or E2."[6] When this is done, the above inference can be compactly represented as follows:
(P) No good that we know of has J.
Therefore, probably:
(Q) No good has J.
Rowe next refers to Plantinga's criticism of this inference, and he argues that Plantinga's criticism now amounts to the claim that
"we are justified in inferring Q (No good has J) from P (No good we know of has J) only if we have a good reason to think that if there were a good that has J it would be a good that we are acquainted with and could see to have J. For the question can be raised: How can we have confidence in this inference unless we have a good reason to think that were a good to have J it would likely be a good within our ken."[7]
Rowe's response is then as follows:
"My answer is that we are justified in making this inference in the same way we are justified in making the many inferences we constantly make from the known to the unknown. All of us are constantly inferring from the A's we know to the A's we don't know. If we observe many A's and note that all of them are B's we are justified in believing that the A's we haven't observed are also B's. Of course these inferences may be defeated. We may find some independent reason to think that if an A were a B it would likely not be among the A's we have observed. But to claim that we cannot be justified in making such inferences unless we already know, or have good reason to believe, that were an A not to be a B it would likely be among the A's we've observed is simply to encourage radical skepticism concerning inductive reasoning in general."[8]
Finally, Rowe points out that "in considering the inference from P to Q it is very important to distinguish two criticisms:
A. One is entitled to infer Q from P only if she has a good reason to think that if some good had J it would be a good that she knows of.
B. One is entitled to infer Q from P only if she has no reason to think that if some good had J it would likely not be a good that she knows of.
Plantinga's criticism is of type A. For the reason given, it is not a cogent criticism. But a criticism of type B is entirely proper to advance against any inductive inference of the sort we are considering."[9]
In view of the last point, Rowe concludes that "one important route for the theist to explore is whether there is some reason to think that were a good to have J it either would not be a good within our ken or would be such that although we apprehend this good we are incapable of determining that it has J."[10]
8. Comments on Rowe's Discussion of the Crucial Inference
First, Rowe is right that a criticism of type A does involve "radical skepticism of inductive reasoning in general".
But, secondly, having granted that point, how satisfactory is Rowe's account of the reasoning involved? To answer that question, what one needs to notice is that Rowe's claim that "if we observe many A's and note that all of them are B's we are justified in believing that the A's we haven't observed are also B's" is somewhat ambiguous, since while the claim that "we are justified in believing that the A's we haven't observed are also B's" might naturally be interpreted as saying
(1) We are justified in believing that all the A's that we haven't observed are also B's
it is possible to construe it as making, instead, the following, much weaker claim
(2) We are justified in believing of each of the A's that we haven't observed that that A is also a B.
It is crucial - as we shall see below - to distinguish between these two claims.
Before turning to that point, however, let's ask the general question: What objections might be raised to the inference from P to Q? One point that might be thought to be relevant is that J is a complex property. For a property P has property J if and only if (1) the goodmakingness of property P is greater than the badness of E1 or of E2, and (2) there is an instance of P that could not exist were it not for the existence of either E1 or E2. So J is a conjunctive property, and thus to say that something lacks property P is to say that that thing possesses a certain disjunctive 'property'.
On the face of it, it's not clear why that should be relevant. If one has drawn 100 marbles from an urn, and all have them have been red, one is surely justified in believing, in the absence of other relevant evidence, that the next marble drawn will probably be red as well. Is the situation any different if all of the 100 marbles drawn were either red or green? Isn't one justified in believing in the absence of other relevant evidence, that the next marble drawn will probably be either red or green as well?
Let us now turn to the distinction between
(1) We are justified in believing that all the A's we haven't observed are also B's
(2) We are justified in believing that of each of the A's we haven't observed that that A is also a B.
Rowe is certainly right that any criticism that claims that one is not justified in inferring (2) unless one has additional information to the effect that unobserved A's are not likely to differ from observed A's with respect to the possession of property B entails inductive skepticism. But it is not true that this is so if one rejects, instead, the inference to (1).
This is important, moreover, because it is (1) that Rowe needs, since the conclusion that he is drawing does not concern simply the next morally relevant property that someone might consider: conclusion Q asserts, rather, that all further morally relevant properties will lack property J. Such a conclusion about all further cases is much stronger than a conclusion about the next case, and one might well think that in some circumstances a conclusion of the latter sort is justified, but that a conclusion of the former sort is not.
One way of supporting the latter claim is by arguing, as I have elsewhere, that when one is dealing with an accidental generalization, the probability that it will obtain gets closer and closer to zero, without limit, as the number of potential instances gets larger and larger, and that this is so regardless of how large one's evidence base is. Is it impossible, then, to justify universal generalizations? The answer is that if laws are more than mere regularities - and, in particular, if they are second-order relations between universals - then the obtaining of a law, and thus of the corresponding regularity, may have a very high probability upon even quite a small body of evidence. So universal generalizations can be justified, if they obtain in virtue of underlying laws.
The question then becomes whether (Q) expresses a law - or a consequence of a law. If - as seems to me plausible - it does not, then, although it is true that one in justified in holding, of any given, not yet observed morally relevant property, that it is unlikely to have property J, it may not be the case that it is probable that no goodmaking (or rightmaking) property has property J. It may, on the contrary, be probable that there is some morally relevant property that does have property J.
This objection could be overcome if one could argue that it is unlikely that there are many unknown goodmaking properties. For if the number is small, then the probability of (Q) may still be high even if (Q) does not express a law, or a consequence of a law.
I am inclined to think that it may be possible to argue that it is unlikely that there are many unknown, morally relevant properties. Such an argument, however, would involve, I think, some controversial claims. For this reason, I shall attempt to show below that one can set out a different account of the crucial inference, and one which avoids the need for this line of defense.
………..
11.2 Theodicies and the Skeptical Consequences Argument
The skeptical consequences argument can also be developed in a way that bears upon the possibility of supplementing the above line of thought with a theodicy. The key to doing so involves comparing the value of a state of affairs, S, with the value of an action, A, that produces such a state of affairs. The argument can be put this way:
(1) If the theist is to be able to reject the conjunction of (c) and (d), without rejecting the conjunction of (a) and (b) as well, a natural idea is to hold that the problem with an omnipotent and omniscient being's preventing certain states of affairs is that this will deprive finite beings of the opportunity of preventing them. But this presupposes that the prevention of them by finite beings is more valuable than their prevention by an omnipotent and omniscient being.
(2) Consider the elimination of cancer from the world. Given our ordinary moral knowledge, almost everyone believes that it would be a very good thing for a person who knew of a cure for cancer to make that information available to other people.
(3) But given our ordinary moral knowledge, almost everyone also believes that it would be a very good thing for cancer to be eliminated from the world by some sort of accident, rather than by anyone's action, human or otherwise.
(4) How does the value of cancer's being eliminated from the world compare with the value of cancer's being eliminated from the world by the action of some human? Consider these alternatives: (i) Cancer is eliminated from the world, right now, by some fluke - e.g., the evolution of an anti-cancer virus; (ii) Cancer is eliminated from the world in ten year's time by a discovery made by a human. Is the second alternative preferable to the first? The argument that I am now advancing claims that that is not so. Furthermore, I would suggest that the situation is not altered if one considers even much shorter periods of time than ten years. But if so, then the value of having cancer eliminated via human action is almost entirely composed of the value of having a world without cancer: the value lies almost entirely in the relevant states of affairs, rather than in the action of bringing about that state of affairs.
(5) Next, how does the value of cancer's being eliminated from the world by a fluke compare with the value of cancer's being eliminated from the world by the action of an omnipotent and omniscient being? In particular, can it be plausibly maintained that the value of cancer's being eliminated from the world by the action of an omnipotent and omniscient being is less than the value of cancer's being eliminated from the world by a fluke? I suggest that it is not at all easy to see why this would be so.
(6) If the last two steps are right, what we have is this:
(a) The value of cancer's being eliminated from the world by human action is at most slightly greater than the value of its being eliminated by a fluke;
(b) The value of cancer's being eliminated from the world by the action of an omnipotent and omniscient being is no less than the value of cancer's being eliminated from the world by a fluke.
But taken together, these propositions imply that it is better for an omnipotent and omniscient being to eliminate cancer from the world than, say, waiting ten years for a human to do so.
(7) The theist may, of course, propose some other account of how it is that the conjunction of (c) and (d) is to be rejected, but not the conjunction of (a) and (b). When such accounts are advanced, one can see whether they will withstand scrutiny. In the meantime, it would seem that one must hold, if one is to avoid a devastating skepticism, that the conjunction of (c) and (d) cannot plausibly be rejected.
12. An Argument Based on Pure Inductive Logic
The thrust of this second response is that one can provide a positive defense of the crucial inductive step in the evidential argument from evil, and that one can do this, moreover, without appealing to any substantive moral theory. For one can show, simply by appealing to sound principles of inductive logic, both that an appeal to the claim that the moral knowledge that humans have is incomplete cannot suffice to provide an answer to the argument from evil, and that the evidential argument from evil, properly formulated, is in fact sound.
12.1 One Version of the Pure Inductive Logic Response
The first version of what I am calling a pure inductive logic argument has two parts. The first part focuses upon what things it is reasonable to believe, based upon fundamental principles of inductive logic, about the existence of rightmaking and wrongmaking properties in themselves, as contrasted with beliefs about the instances of such properties. Then, in the second part, I consider whether information about the instances of those properties provides one with any ground for modifying the conclusions drawn in part one.
12.1.1 Part 1: The Existence of Rightmaking and Wrongmaking Properties
In setting out the argument that follows, it will be helpful to use the following sort of diagram:
In this diagram, the region to the right of the vertical axis represents actions that one ought to perform, all things considered, and the further to the right an action is located, the stronger the obligation to perform it is. Similarly, the region to the left of the vertical axis represents actions that are morally wrong, all things considered, and the further to the left an action is located, the more wrong it is.
We shall need to represent rightmaking and wrongmaking properties, both known and unknown. Rightmaking properties will be represented by arrows pointing to the right, and wrongmaking properties by arrows pointing to the left, with the length of the arrows representing the seriousness of the property involved.
Finally, to distinguish between known and unknown rightmaking and wrongmaking properties, morally significant properties that are known will be represented by arrows above the horizontal axis, and unknown morally significant properties by arrows below the horizontal axis.
The argument in part one may be put as follows. Consider some action A that, judged in terms of the rightmaking and wrongmaking properties that one is aware of, is an action that is morally very wrong. If action A does not have any unknown rightmaking or wrongmaking properties, the situation can be represented by the following diagram:
But what about the possibility that action A has some morally significant properties about which one has no knowledge? The right response is, I suggest, as follows:
(1) Judged from a purely a priori point of view, the mere existence of wrongmaking properties (as contrasted with their instantiation) is no less likely than the existence of rightmaking properties. Similarly, judged from a purely a priori point of view, the existence of badmaking properties is no less likely than the existence of goodmaking properties.
(2) Judged from a purely a priori point of view, the likelihood that a rightmaking property will have a certain moral weight is no greater than the likelihood that a wrongmaking property will have that moral weight. Similarly, judged from a purely a priori point of view, the likelihood that a goodmaking property will have a certain moral weight is no greater than the likelihood that a badmaking property will have that moral weight.
(3) Is the situation different if one takes into account the knowledge that we have about the existence of wrongmaking properties and rightmaking properties - whether properties that are instantiated or properties that are not - and their moral weights, and similarly, about the existence of badmaking properties and goodmaking properties, and their moral weights? Are there, for example, more rightmaking properties (not: instances of rightmaking properties) than wrongmaking properties? Or are rightmaking properties typically more weighty than wrongmaking properties? I do not believe that the moral knowledge that we have supports affirmative answers to either of these questions. So it seems to me that, judged from an a posteriori point of view, the existence of wrongmaking properties is no less likely than the existence of rightmaking properties, nor are wrongmaking properties likely to have less moral weight than rightmaking properties, and similarly with respect to badmaking properties versus goodmaking properties.
(4) Consider, now, some particular state of affairs, S, which is such that, judged only by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, it is reasonable to believe that an omnipotent and omniscient being's allowing S to exist would be prima facie morally very wrong. It is certainly possible that there are rightmaking properties that we are not aware of, and that make it the case that an omnipotent and omniscient being's allowing S to exist is not morally wrong, all things considered.
It is possible, then, both that action A - the action of allowing S to obtain - has some unknown rightmaking property, and that, as depicted in the following diagram, the unknown rightmaking property - UR - is weightier than the known wrongmaking property - KW - with the result that the overall moral status of action A is positive, so that action A is one that a person with the relevant knowledge should perform:
(5) Notice that in allowing that this may be so, I am rejecting the conclusion that Rowe attempted to establish by means of the argument discussed earlier in sections 7 and 8, since if things are as in the above diagram, then there is a God-justifying property - namely, UR. My goal, accordingly, is not to provide a sound argument that shows that it is unlikely that there is any unknown rightmaking property that is sufficiently serious to outweigh the known wrongmaking property (or properties). I am arguing, instead, for a number of other theses - one of which is that it is unlikely that the action in question is morally permissible, all things considered.
(6) If the claim advanced at step (3) is correct, then it is just as likely that any unknown rightmaking properties and wrongmaking properties that exist make it the case, instead, that an omnipotent and omniscient being's allowing S to obtain is morally much worse than it already is relative to known rightmaking and wrongmaking properties, than that they make it the case that an omnipotent and omniscient being's allowing the existence of S something that is morally permissible, all things considered.
In particular, it is no less likely that the situation, rather than being as in the diagram just set out, is instead as follows:
(7) What, then, is the likelihood that it is not morally wrong, all things considered, for an omnipotent and omniscient being to allow the existence of S? One cannot give any precise answer to this, as it will depend on how many unknown morally significant properties there are, and how weighty each one is. But one can say that it is more likely than not that an omnipotent and omniscient being's allowing the existence of S is morally wrong, all things considered, since a shift of a given amount in the direction of greater moral wrongness is, in view of (4), as likely as a shift of an equal amount in the direction of less moral wrongness, and we are starting with a state of affairs, S, which is such that an omnipotent and omniscient being's allowing S to exist is an action that, judged by known rightmaking properties and wrongmaking properties, is prima facie very seriously wrong. We have, therefore, the following conclusion:
(a) If S is a state of affairs which is such that an omnipotent and omniscient being's allowing S to exist is an action that, judged by known rightmaking properties and wrongmaking properties, is prima facie very seriously wrong, then the probability that, all things considered - including relevant unknown rightmaking and wrongmaking properties - the action by an omnipotent and omniscient being of allowing S to obtain is morally wrong is greater than one half.
(8) But the same is also true with respect to the probability, all things considered, that the action by an omnipotent and omniscient being of allowing S to obtain is morally very seriously wrong. For given that we are starting with a state of affairs which is such that an omnipotent and omniscient being's allowing that state of affairs to exist is prima facie very seriously wrong, very small shifts in the direction of rightness will still leave one with a state of affairs that it is very seriously wrong for an omnipotent and omniscient being to allow to exist So we also have this second conclusion:
(b) If S is a state of affairs which is such that an omnipotent and omniscient being's allowing S to exist is an action that, judged by known rightmaking properties and wrongmaking properties, is prima facie very seriously wrong, then the probability that, all things considered, the action by an omnipotent and omniscient being of allowing S to obtain is very seriously wrong must also be at least slightly greater than one half.
(9) Suppose that S is a state of affairs which is such that the prima facie wrongness of an omnipotent and omniscient being's allowing S to exist, as judged by known rightmaking properties and wrongmaking properties, is equal to V. What is the best estimate of the moral worth of that action, all things considered, including unknown rightmaking and wrongmaking properties? The answer is that if neither a priori nor a posteriori considerations provide any reason for thinking that rightmaking properties are either more numerous, or more weighty than wrongmaking properties, or vice versa, then the best estimate of the real moral worth of the action in question is that it is equal to V, since for every unknown rightmaking property that might exist, and that the action might possess, it is just as likely that there is some equally weighty unknown wrongmaking characteristic that the action possesses. The basic idea, then, is that these two, equally probable possibilities involving unknown rightmaking and wrongmaking properties cancel each other out. This is clearly so when, as depicted in the following diagram, both are present:
The general point, however, needs to be expressed probabilistically. Thus, let p be the probability that there is some unknown rightmaking property, UR, with a weight of +n. It is just as likely that there is an unknown wrongmaking property, UW, with a weight of -n. So given that action A has the known wrongmaking property, KW, of weight -k, the expected moral status of action A, taking into account just the possibilities of UR and UW, will be as follows:
Expected Moral Status of action A = -k + (p x weight of UR) + (p x weight of UW)
= -k + (p x n) + (p x -n)
= -k + pn -pn
= -k.
There may, of course, be any number of other unknown rightmaking properties, but the point is the same in all cases - namely, that it is just as likely that there is an equally weighty unknown wrongmaking property, with the result that when one calculates the expected moral status of the action in question, the positive possibilities with respect to possible unknown rightmaking properties will be exactly canceled by the negative possibilities with respect to possible unknown wrongmaking properties.
We have, accordingly, this third conclusion:
(c) If S is a state of affairs which is such that the prima facie wrongness of an omnipotent and omniscient being's allowing S to exist, as judged by known rightmaking properties and wrongmaking properties, is equal to V, then the best estimate of the real moral worth of the action in question is that it is equal to V.
(10) Next, notice that so far I have considered only a single state of affairs - S - whose existence it is prima facie very seriously wrong for an omnipotent and omniscient being to allow. It is, however, reasonable to believe that there are very many such states of affairs. But, then, if the probability, given complete moral knowledge, that it is not morally wrong for an omnipotent and omniscient being to allow any one such state of affairs to exist is less than one half, then, in the absence of any evidence that a favorable situation with regard to the possession of unknown moral properties in the case of one state of affairs, S, is likely to be very strongly correlated with a favorable situation with regard to the possession of unknown moral properties in the case of another state of affairs, T, the probability that no such states of affairs are such that it is morally wrong, all things considered, for an omnipotent and omniscient being to allow them to exist, will be extremely low. We have, in short, this fourth conclusion:
(d) If S1, S2, S3, S4, . . . Si, . . . Sn are states of affairs which are such that, for each Si, an omnipotent and omniscient being's allowing Si to exist is an action that, judged by known rightmaking properties and wrongmaking properties, is prima facie very seriously wrong, then the probability that, all things considered - including relevant unknown rightmaking and wrongmaking properties - the action by an omnipotent and omniscient being of allowing all of S1, S2, S3, S4, . . . Si, . . . Sn to obtain is not morally wrong is extremely low.
Since I believe that a deontological formulation of the argument from evil is to be preferred over an axiological one, I have put things in terms of rightmaking and wrongmaking properties. But if one preferred formulating things in terms of goodmaking and badmaking properties of states of affairs, the above arguments could easily be reformulated. The corresponding conclusions would then be as follows:
(a*) If S is a state of affairs that, judged by known goodmaking properties and badmaking properties, is prima facie very bad, then the probability that S is bad, all things considered, including relevant unknown goodmaking and badmaking properties, is greater than one half.
(b*) If S is a state of affairs that, judged by known goodmaking properties and badmaking properties, is prima facie very bad, then the probability that S is very bad, all things considered, including relevant unknown goodmaking and badmaking properties, must also be at least slightly greater than one half.
(c*) If S is a state of affairs that, judged by known goodmaking properties and badmaking properties, has an overall value V, then the best estimate of the real value of the state of affairs in question, all things considered, including unknown goodmaking and badmaking properties, is that it is equal to V.
(d*) If S1, S2, S3, S4, . . . Si, . . . Sn are states of affairs which are such that each Si, judged by known goodmaking properties and badmaking properties, is prima facie very bad, then the probability that, all things considered, including relevant unknown goodmaking and badmaking properties, none of S1, S2, S3, S4, . . . Si, . . . Sn is bad, is extremely low.
12.1.2 Part 2: The Existence of Instances of Rightmaking and Wrongmaking Properties
As formulated in part one, the method of arriving at the best estimate of the rightness or wrongness of an action appeals to the following two propositions:
(1) The mere existence of wrongmaking properties is a priori no more likely than the existence of wrongmaking properties;
(2) As regards rightmaking and wrongmaking properties that we know of, there are just as many wrongmaking properties as rightmaking properties.
One now needs to consider, however, whether the conclusions that were drawn in part one of the argument on the basis of propositions (1) and (2) might not be undermined by additional information of a certain sort - such as, for example, information not about the number of rightmaking versus wrongmaking properties that exist - whether instantiated or not - but, instead, about the number of instances of such properties.
Suppose, for example, that we knew of more instances of rightmaking properties than of wrongmaking properties. Why could not one argue that that fact would make it more likely that there are more instances of rightmaking properties that we are not aware of than of wrongmaking properties that we are not aware of? But if that is right, might we not be justified in concluding, then, that it may very well be likely that, all things considered, allowing Sue to be raped, brutally beaten, and murdered, was morally obligatory, all things considered - or, at least, morally permissible?
The first point to be made about this response is that it will lead, once again, to moral skepticism of the kind described above. My objective, however, is not merely to show that there is something deeply problematic, or unappealing, about the response to the argument from evil that appeals to human cognitive limitations. It is, rather, to show that that response is unsound.
The question, accordingly, is whether the pure inductive logic argument is undercut when one brings in information about the number of instances of rightmaking and wrongmaking properties. Here there are two issues. First, is it or is it not true either that there are more instances of rightmaking properties than wrongmaking properties, or, even if there are not, that rightmaking properties are more weighty on average than wrongmaking properties, or both? Secondly, if this were so, would that fact mean that the inductive inference was unsound?
As regards the first question, it seems to me that it is far from clear that instances of rightmaking properties are either more numerous, or more weighty, than instances of wrongmaking properties. For while it may seem plausible that people more often act in ways that benefit other people, than in ways that harm them, in determining how instances of rightmaking properties compare with instances of wrongmaking properties, one needs to consider not just the things that people do, but the things that they fail to do. Consider, then, the fact that many people in the world do not have enough to eat, do not have adequate health care, etc. If such failures on the part of those of us living in very affluent countries are, as many moral philosophers have argued, morally wrong, then I think that it is far from clear that the world contains more actions that are in accordance with one's duties than actions that are contrary to it. Moreover, given that those failures lead in many cases not only to great suffering, but to death, there may very well be good reasons for holding that instances of wrongmaking properties are just as numerous, and just as weighty, as instances of rightmaking properties.
Secondly, suppose, for the sake of argument, that this is not so. Where does that leave the pure inductive logic argument?
If we found that, with regard to the rightmaking and wrongmaking properties that we are aware of, instances of rightmaking properties are either more numerous or more weighty than wrongmaking properties, then, given that evidence alone, we would surely be justified in projecting from that sample unto the set of rightmaking and wrongmaking properties that we are not aware of, and concluding that it was likely that, as regards the latter set of properties, instances of rightmaking properties were either more numerous or more weighty than wrongmaking properties.
Would this provide one with grounds for concluding that even acts of allowing the occurrence of events that, judged only on the basis of what we know, are morally horrendous - such as the Holocaust - are likely to be morally justified, given all of the rightmaking and wrong-properties that exist, known and unknown? Perhaps. Perhaps not. The answer will depend upon whether we can know anything about how the number of unknown morally significant properties compares with the number of known morally significant properties. If, for example, we had any reason for holding that the former is unlikely to be larger than the latter, then we might well be able to argue that it was unlikely that allowing what were, given only our own knowledge, morally horrendous events would turn out to be morally justified.
But even if that is right, there does not seem to be any uncontroversial argument for the claim that the class of unknown morally significant properties is unlikely to be larger than the class of known morally significant properties, and so as long as this is so, I think that it is the case that the inductive inference is undercut.
Yet consider the following. Imagine two possible worlds, W1 and W2, that are as follows:
(1) The amount of (what appears to be ) natural evil is very similar in W1 and W2, and is very great in both worlds;
(2) In world W1, parents do an excellent job of raising their children, with the result that there are many more instances of rightmaking properties than of wrongmaking properties;
(3) In world W2, by contrast, parents do a very bad job of raising their children, either failing to inculcate sound moral behavior, or actually inculcating behavior that is deeply wrong, with the result that there re many more instances of wrongmaking properties than of rightmaking properties.
Consider now corresponding, serious natural evils in W1 and W2: fauns die agonizing deaths due to forest fires; children die slowly and painfully from cancer; thousands of people are killed in earthquakes. If the relative numbers of instances of rightmaking and wrongmaking properties is crucial for the inductive step in the evidential version of the argument from evil, then the argument from evil should be sound in world W2 but not in world W1. But how can this be? How can the difference in the way people behave in the two worlds make it the case that equally terrible natural evils make it unlikely that God exists in world W2 but not in world W1?
It may seem as if we have conflicting intuitions here. On the one hand, it is hard to see, in view of worlds W1 and W2, why the behavior of people should affect the argument. On the other hand, inductive logic seems to entail that the relative numbers of instances of rightmaking and wrongmaking properties must be relevant to the crucial inference. In fact, however, there is no conflict here. For we can affirm both that the relative numbers of instances of rightmaking and wrongmaking properties is, other things being equal, relevant to the crucial inference, and that the evidential argument from evil is no less forceful in world W1 than it is in world W2, and the reason that we can affirm both of these propositions is that the descriptions of worlds W1 and W2 contain additional information that is relevant to the inductive inference. In particular, we know that the reason that instances of rightmaking properties are much more numerous than wrongmaking properties in world W1 - and the opposite in world W2 - is that people in W1 are disposed, because of the way in which they have been raised, to perform actions that are morally right. This disposition, of course, cannot make any difference with regard to unknown rightmaking properties and wrongmaking properties, and since it is precisely the disposition of people in W1 to perform actions that are morally right, by their lights, that explains why instances of known rightmaking properties are much more numerous than instances of known wrongmaking properties, we are not justified in projecting that fact about the distribution of known rightmaking and wrongmaking properties from the class of known properties to the class of unknown properties.
What is relevant here, in short, is a principle mentioned earlier, in section 10:
The Total Evidence Requirement
The probability that one should assign to a given proposition's being true is the probability that that proposition has relative to one's total evidence.
Given only the information that instances of rightmaking properties are more numerous than instances of wrongmaking properties. the pure inductive logic defense of the crucial inductive step is undermined. But even if one assumes - what I have argued is not at all clear - that instances of rightmaking properties are more numerous than instances of wrongmaking properties, this is not the totality of the relevant evidence. For one also knows that if there are more instances of rightmaking properties than wrongmaking properties in this world, this is because human agents have, on the whole, received a reasonably sound moral education, and thus have acquired the disposition to do what is right, rather that what is wrong. But given this explanation of the existence of more instances of known rightmaking properties than of known wrongmaking properties, there is no reason at all for thinking that, as regards rightmaking and wrongmaking properties that humans are not aware of, there will be more instances of the former than of the latter, since the causal mechanism in question cannot operate in the case of rightmaking and wrongmaking properties that humans are not aware of.
The upshot is that the fact - if it is a fact - that our world contains more instances of rightmaking properties that humans are aware of than of wrongmaking properties that they are aware of provides no ground at all for concluding that it is likely that our world contains more instances of rightmaking properties that humans are not aware of than of wrongmaking properties that they are not aware of.
12.2 A Second Version of the Pure Inductive Logic Response
The above is one form of the pure inductive logic response to the objection based upon the proposition that it is unlikely that humans have complete moral knowledge. But a slightly different inductive logic argument can also be advanced that starts out from the claims advanced at steps (1), (2), and (3) above, and that then proceeds as follows:
(4*) As before, let S be some particular state of affairs such that, judged by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, an omnipotent and omniscient being's allowing that state of affairs to exist would be prima facie very wrong. It will often be the case that one can find some other state of affairs, G, such that, judged by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, an omnipotent and omniscient being's allowing G to exist would be prima facie obligatory to roughly the extent to which allowing S to exist would be prima facie wrong.
(5*) In view of the premises, first, that the existence of unknown wrongmaking properties is just as likely as the existence of unknown rightmaking properties, and, secondly, that unknown wrongmaking properties are just as likely to have a given moral weight as unknown rightmaking properties are, it is just as likely that an omnipotent and omniscient being's allowing G to obtain will turn out to be, all things considered, morally wrong, as it is that an omnipotent and omniscient being's allowing S to obtain will turn out to be, all things considered, morally right.
(6*) Let W be the set of states of affairs each of which is such that, judged by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, it would be morally wrong for an omnipotent and omniscient being to allow the state of affairs in question to obtain. Let R be the set of states of affairs each of which is such that, judged by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, it would be morally obligatory for an omnipotent and omniscient being to allow the state of affairs in question to obtain. Then, in view of (5*), it is reasonable to believe that if there are morally significant properties lying outside our ken which often make it the case that states of affairs belonging to set W are states that an omnipotent and omniscient ought to allow to exist, all thing considered, then those unknown moral properties will also make it the case, approximately as often, that states of affairs belonging to set R are states that an omnipotent and omniscient ought not to allow to exist, all thing considered.
(7*) The upshot is that it is reasonable to believe, first, that, if all morally significant properties - both known and unknown - are taken into account, it will still turn out that the world contains many states of affairs that an omnipotent and omniscient ought not to allow to exist, and secondly, that, in the absence of detailed information about differences with respect to the number of rightmaking versus wrongmaking properties, or with respect to the weightiness of those properties, there is no reason to expect that the number of actual states of affairs that it would be wrong for an omnipotent and omniscient being to allow to exist, all things considered, will be any less than the number of actual states of affairs that, as judged by our current knowledge of rightmaking and wrongmaking properties, together with our non-moral knowledge, it would be wrong for an omnipotent and omniscient being to allow to exist.
Conclusion
The argument from evil is a very natural argument, since in everyday life we constantly draw conclusions about the moral character of individuals based upon information both about what the individual in question does, and about what the individual intentionally refrains from doing. When one attempts to extent this type of reasoning to the case of an omnipotent and omniscient being, however, a crucial question arises concerning what exactly the form of the central inductive step is. Moreover, when this question is carefully considered, it turns out to be rather more difficult than one might have expected to provide a satisfactory account, and we saw, in particular, that two accounts that may initially have seemed promising do not work. I then went on to argue, however, that an approach based upon principles of pure inductive logic does succeed. The conclusion, accordingly, is that the evidential argument from evil does show - unfortunately - that it is unlikely - indeed, extremely unlikely - that God exists.
Michael Tooley
Department
of Philosophy
University
of Colorado at Boulder