Science in Christian Perspective



The Inductive Problem of Evil
Department of Philosophy
Asbury College
Wilmore, Kentucky 40390

From: JASA 33 (June 1981): 82-87.

There is no doubt that evil constitutes one of the most severe challenges to Christian theism. From time to time, different formulations of the problem of evil gain the interest of philosophers and theologians. Currently, much attention is being given to what can be called the inductive argument from evil, while interest in the deductive argument seems to be declining. I carefully distinguish among various formulations of the argument and focus on the one that is most formidable-the inductive argument from gratuitous evil. Drawing on concepts from philosophy of science, such as probability, hypothesis confirmation, etc., I fashion a rebuttal to the atheistic attempt to construe evil as a kind of scientific evidence against orthodox theism.

It is widely agreed that evil constitutes a formidable problem for Christian belief. Serious, educated persons have typically thought that the agonizing presence of evil in the world demands explanation on the part of those who give allegiance to a sovereign and loving God. Yet scholars differ over the exact nature of the problem of evil, i.e., over what its exact structure and strategy should be. However, amid all this diversity of opinion, there are discernible patterns. The broadest division in the literature on God and evil is between theoretical (intellectual) problems of evil and existential (emotional-volitional) problems of evil.1 Within the category of theoretical problems there is a further division between those that are deductive in structure and those that are inductive.2 One could refine the point even more and distinguish between problems over moral evils (e.g., lying, murdering, etc.) and those over natural evils (e.g., pain, deformity, etc.).3

Twentieth-century philosophers have been largely preoccupied with the deductive problems. Yet, in the last few years, there has been a marked shift of interest toward the inductive ones. In this article, I examine and attempt to refute the increasingly popular and important inductive arguments from evil that arise at the interface of science and religion. These inductive-type arguments rest on the general presumption that the existing evils of our world can be marshalled as significant, and perhaps devastating, evidence against the truth of Christian theism, much as empirical data might be compiled to disconfirm a given hypothesis in any established branch of experimental science. But before entering upon this subject, the demise of the deductive problems must be briefly explained.

The Deductive Problem of Evil

Those who advance what I here call a deductive problem of evil (also called the logical problem4 and the a priori problem5 seek to show that the concept of evil generates contradictions within orthodox Christian theism. More specifically, the opponent of Christian theism (let us use the term "atheist" in its literal sense) advances an argument the structure of which is deductive and the strategy of which is to derive a logical inconsistency by using the theist's own beliefs against him.

Two centuries ago, David Hume gave the problem classic expression:

Is [God] willing to prevent evil, but not able? Then he is impotent. Is he able, but not willing? Then he is malevolent. Is he both able and willing? Whence then is evil?6

J. L. Mackie's presentation of this type of argument from evil is representative of recent treatments:

In its simplest form the problem is this: God is wholly good; yet evil exists. There seems to be some contradiction between these propositions so that if any two of them were true the third would be false. But at the same time all three are essential parts of most theological positions; the theologian, it seems at once must adhere and cannot consistently adhere to all three.7

Since a collection of beliefs containing a contradiction-explicitly or implicitly-cannot be rationally accepted, atheists taking this line reject Christian theism outright.

Since statements of the deductive problem sometimes differ, it is helpful for our purposes to build a model of each of its extant versions, trying not to oversimplify one of the most important and subtle items of the philosophy of religion. A model helps us to understand better both its attractiveness and its fatal flaw. Let us first encapsulate the essentail theistic position in the proposition:

(G) An omnipotent, omniscient, wholly good God exists,

which is, of course, an amalgamation of the set of theistic propositions:

(1) God exists; (2) God is omnipotent; (3) God is omniscient; (4) God is wholly good.

Obviously, any question about proposition (G) is equivalent to a question about one or more of the propositions in the set (1) - (4).

Now (G) does not exhaust the scope of theistic belief, and it is in specifying one further theistic proposition-one regarding evil-that the dreaded inconsistency is claimed to arise. In fact, the extensive literature on the topic contains three distinct versions of the inconsistency charge, each version being determined by what specific proposition about evil is cited.

The second phase in constructing our model is to delineate the three propositions about evil that distinguish the several permutations of the deductive problem. The atheistic challenger may claim that proposition (G) is inconsistent with either the proposition that

(E1) Evil exists or that

(E 2 ) Large amounts, extreme kinds, and perplexing
distributions of evil exist

or that

(E1) Gratuitous or pointless evil exists.

The following chart displays the three different versions. Anyone familiar with the vast and quite sophisticated literature on the deductive problem knows that formulation I has been by far the most widely considered, and is therefore the one I treat here. But Il and III have also held some degree of interest for professional philosophers.9

The Demise of the Deductive Problem

In argument 1, which is our prime example of the deductive problem, the atheist ascribes both (G) and (E,) to the theist, which is an entirely acceptable maneuver. The next step in the attack is for the atheist to argue that (G) implies

(-E1) Evil does not exist,

(where - means negation) and that the theist is logically committed to it. When the atheist believes himself to have established that the theist must hold (-E1), he then reminds the theist that qua theist he is officially committed to (E1) as well. Since (E1) is the straightforward contradictory of (-Ed, the atheist seems to have made good his charge of logical inconsistency. This approach, if it can really be made to work, is so conclusive that it is not surprising that atheists have long been fascinated with it.

But the cruicial Phase of the argument that proceeds from (G) to (-E1) is not above suspicion. (G) does not by itself imply (-E1 ), but does so only if it is supplemented by other propositions, such as

(1) God
has being or independent ontological status, (2) God has the power to perform any logically consistent act, (3) God knows all the ways to eliminate evil, (4) God wants to eliminate all evil; (5) Evil is not logically necessary.10

Now, from (G) together with (1) - (5) it does follow that (-E~) is true. But this brings us to the very reason why such atheistic challenges have regularly failed. In order to play fair, the atheist must use additional propositions that are also parts of theism. However, there are numerous theists

Deductive Arguments From Evil

    1             11                 111
(G)             (G)             (G)
is                 is                 is
inconsistent inconsistent inconsistent
with             with             with
(E1                 (E1)             (E1)

We need not deny that theism is capable of evidential evaluation, and thus evade the inductive argument from evil. To do that would unwittingly be to make theism incapable of much desired confirmation as well.

(e.g., Lewis, Plantinga, Ahern, etc.) who have pointed out that various propositions in the set (I ) - (5 ) are not or need not be included in theism, and hence cannot be used to produce the rumored contradiction.

Ironically, while trying to expose the fallacy of selfcontradiction on the part of the theist, the atheist inevitably commits one of two fallacies: either begging the question by selecting propositions to which the theist is not really committed, or by lifting out of context propositions to which the theist is committed but attributing new and convenient meanings to them." As this fundamental mistake in the deductive arguments from evil becomes more widely recognized, philosophers are beginning to put much less stock in it.

The Inductive Problem of Evil

Not to be completely thwarted by the failure of the deductive problem of evil, some atheistic critics have begun to develop arguments from evil which are inductive in structure and geared to establish that theism, or (G), has a low probability relative to the evidence of evil. Speaking for this group of thinkers, William Rowe charges that Christian theism is "an extraordinary, absurd idea, quite beyond our belief."12

Interestingly, there are three identifiable renditions of the inductive argument (also called the probabilistic13 problem and evidential problem14) which correspond exactly to the three renditions of the deductive problem.

Again, each of these inductive-type arguments finds representation in the scholarly literature.15 However, the significance and popularity of these inductive arguments seems almost to be the reverse of their deductive counterparts. Whereas I is the classic deductive problem, VI is clearly the most interesting and important inductive problem, the reason for which will soon become clear. Hence I focus on VI and simply state here that my critique of it applies mutatis mutandis to IV and V.

Admittedly, there are several different ways of calculating the probability value of (G) given some (E)-like proposition. The personalist, frequency, logical, and inductive theories of probability are well-defined studies of probability, any one of which might be employed to assess the probability of (G). However, I concentrate here only on the last theory, the inductive, both because it creates the most formidable challenge from evil, and because Alvin Plantinga has already provided an excellent refutation of the other three theories relative to the problem of evil.16

A Sample of Inductive Reasoning

A momentary digression at this point into the fundamentals of the inductive evaluation of hypotheses is helpful. (In this paper I do not distinguish further among specific inductive techniques, e.g., elimination, enumeration, etc.) According to philosophers of science, scientific induction consists in setting up proper test conditions for a given hypothesis and then checking to see whether and the extent to which they occur. Specification of the test conditions for almost any meaningful hypothesis can be accomplished only by adding some supplementary or auxiliary assumptions. After taking these propositions for granted, the proposition offered as an hypothesis can then be tested by methods appropriate to its claim.17

A simple example exhibits the inductive process perfectly. Suppose that we are in ancient times and want to check the hypothesis that

(S) The earth is spherical.

Christopher Columbus thought that (S) was partially confirmed (i.e., rendered probable) by the factual observation that

(D) The decks of receding ships always disappear from sight before their mastheads.

Now there is no obvious connection between (S) and (D) unless it is also assumed that

(L) Light always travels in straight lines.

Thus (L) would make the inductive reasoning cogent, reasoning which can be schernatized as follows:

Inductive Arguments From Evil

IV                     V                     VI
(G)                     (G)                (G)
is is is
improbable         improbable     improbable
on                     on                     on
(E1)                     (E2)                 (E3)

If (S) is true, then, assuming (L) is true, 
will be true. 
appears to be true. 
Therefore, (S) is probably true. 

On the other hand, for a case of disconfirmation we would get: 
If (S) is true, then, assuming (L) is true, 
(D) will
be true 
appears to be not true 
Therefore, (S) is probably not true.

Examining the Inductive Problem of Evil

Returning now to the atheistic argument from evil of diagram VI and mounting it on our schema of probabilistic disconfirmation with (A) as the auxiliary assumption, we get: 
If (G) is true, then assuming (A) is true, 
will not be true.
(E,) appears to be true. 
Therefore, (G) is probably not true. 

Unfortunately, this type of reasoning is often buried under complicated philosophical jargon and persuasive rhetoric. Consider the argument made by Cornman and Lehrer:

... we seem warranted in concluding that the existence of what surely seems to be unnecessary evil in this world provides inductive evidence for the belief that God does not exist, because it is probable that if he once existed he would have created a different world and that if he now exists he would control the course of nature so as to avoid many pernicious events that do occur.18

Or, ponder the challenge posed by Madden and Hare, who state the inductive argument in the form of an emgma:

If God is unlimited in power and goodness, why is there so much prima facie gratuitous evil in the world? If he is unlimited in power he should be able to remove unnecessary evil, and if he is unlimited in goodness, he should want to remove it, but he does not. Apparently he is limited in either power or goodness, or does not exist at all. 19

These examples show that we have captured the essential atheistic argument in our model; but what prospect is there for answering it?

Strategy for Rebutting the Inductive Problem

The way to begin to generate a proper rebuttal to inductive arguments from evil is to review the prominent features and weaknesses of induction in general. First, there is the matter of ascertaining the occurrence or non-occuff ence of the test condition. Were the instruments accurate, the reports reliable? Was the observer in the proper position to gather the facts? In regard to the atheist's argument VI, therefore, the enormous difficulty of establishing the truth of (E,) is not taken lightly. There are clear grounds for theistic resistance to (E~), and indeed this is the spirit of historical Christian theodicy: showing why what appears to be gratuitous evils are not really gratuitous. After a minimum of debate it becomes clear that whether one believes a given evil to be gratuitous depends on his already accepted values, the time dimension over which he investigates the evil, and other highly debatable factors. This is why the factual or evidential premise (E,), as with such premises in other inductions, possesses a degree of probability short of complete certainty. It should be noted, however, that the probabilistic character of the factual premise(s) in induction is not in itself a reason for discrediting it.

Second, the inductive test of any hypothesis is only as reliable as the background assumptions that are connected to it. It seems clear that assumptions to the effect that God would not allow any evil or would not allow certain amounts, kinds, and distributions are much harder for the atheist to establish, since theists could offer a number of reasons why God could justifiably allow not only the sheer existence of evil but a very great degree and variety as well, and hence why neither (EI) nor (E,) count heavily against (G). But such principles would have to be presupposed in order to make arguments IV and V go through. By contrast, VI turns on a principle which is initially much more plausible than the others. I choose to call this assumption the Principle of Meticulous Providence20 and express it as follows:

Michael Peterson received the Ph.D. in Philosophy of Science from the State University of New York at Buffalo in 1976. Presently he is Department Head and Associate Professor of Philosophy at Asbury College, Wilmore, Kentucky. Dr. Peterson is the author of a number of articles on the philosophy of religion, many of which pertain to the problem of evil. Under a recent grant from the Institute from Advanced Christian Studies, Dr. Peterson completed a book entitled Evil and the Christian God, which will be published in 1981 by Baker Book House.

Whether or not it is expressed in precisely this way, (MP) plays an important role in the preponderance of contemporary inductive arguments from evil, and functions as (A) in my schema.

Although a great many theists would be inclined to accept (MP), close scrutiny reveals several potential problems with it. For example, it is entirely imaginable that God might allow gratuitous evils to occur if eliminating them meant curtailing the scope of human freedom or abrogating the regular operation of natural laws. Or, God might allow gratuitous evils to exist in order to enhance the quality of faith. Moreover, it seems that God might allow numerous trivial evils to exist while directing his sovereign activity to more significant evils and the broad contours of history. (I will not here attempt a detailed distinction between trivial and significant evils.) Last, it could be maintained that biblical passages (e.g., Romans 8:28) and other Christian teachings (e.g., total sovereignty) can best be interpreted as entailing the principle, not that God must meticulously compensate for or redeem every particular evil event, but that God seeks to redeem every person affected by those evils. This interpretation of God's ways not only avoids the errors of, say, Karmic-theorists and Job's comforters, but also enlightens us more fully about what it is to live in a world which is lost.

A third feature of any induction is the probabilistic character of its final result, which stems from the overall logic of induction as well as from the modality of its premises. In the case of confirmation, there could be alternative theories which would be verified by the same test results; or, the assumptions taken for granted may be debatable. The process of inductive confirmation also automatically commits the deductive fallacy of affirming the consequent, which therefore falls short of certainty. In the case of disconfirmation, the hypothesis that is seemingly falsified by certain test conditions may be verified by other conditions. Or, the hypothesis may be logically related to other already confirmed and accepted hypotheses, and thus be retained even in the face of some adverse evidence. It is now quite understandable why many philosophers of science impose the ideal requirement of total evidence on the inductive evaluation of any hypothesis, even though the requirement can never be fully met. In light of these considerations, then, induction gains its distinctive probabilistic, tentative character.

The ramifications of all this for the inductive problem of evil are clear. The apparent disconfirmation of (G) by (E,) may be overridden for the several reasons suggested above. So, the final and complete appraisal of orthodox theism is extremely complex and is hardly exhausted by the inductive problem of evil.

The Importance of Facing the Inductive Problem

In the last analysis, we must admit that evil-particularly instances of what seems to be senseless, pointless evil-may constitute prima facie evidence against Christian theism. After all, orthodox theists do claim that a God of power and love superintends the events of this world, and must square that claim with the facts. And squaring theism with the facts at least means treating it in a way analogous to the way we treat any scientific hypothesis which is amenable to evidence. This is why we are justified in rigorously examining the mechanics of scientific induction in general and the inductive argument from evil in particular. We must know as much as we can about the type of reasoning process which, when applied to evil, supposedly renders theism improbable, implausible, or unlikely.

Knowing more about induction helps us focus on the weak points of the atheist's inductive argument from evil. To sum them up, first, we can contest the basis and accuracy of the factual premise about the actual gratuity of evil. Second, we can question the legitimacy of the assumption (MP) which stipulates how God should dispose evil in the world. Third, we can insist that more and different evidence-rational, historical, existential-actually tends to confirm, not disconfirm, Christian theism. So, we need not deny that theism is capable of evidential evaluation, and thus evade the inductive argument from evil. To do that would unwittingly be to make theism incapable of much desired confirmation as well. As I have urged here, what we must do in response to the increasingly popular inductive argument from evil is to meet it on its own ground and fashion a competent rebuttal.21


1I defended this distinction in "The Existential Dimension of the Problem of Evil," a paper read before the Evangelical Theological Society/ Southeast, March, 1979. Other authors who have proposed such a distinction include George Schlesinger, "The Problem of Evil and the Problem of Suffering," American Philosophical Quarterly, 1 (1964), pp. 244-247; Erazim Kohttk, "The Person in a Personal World: An Inquiry Into the Metaphysical Significance of the Tragic Sense of Life," Independent Journal of Philosophy, I (1977), pp. 51-64; John Bowker, Problems of Suffering in Religions of the World (London: Cambridge University Press, 1970), p. 2.

2Authors who recognize the deductive problem include: J. L. Mackie, H. J. McCloskey, R. R. LaCroix, M. B. Ahern, N. Pike, and C. S. Lewis. Authors who recognize the inductive problem include: Wm. Rowe, R. Pargetter, M. Martin, Madden and Hare, A. Plantinga, and D. Basinger.

3Regarding the problem of moral evil, see A. Plantinga, God, Freedom, and Evil (Grand Rapids: Eerdmans, 1977). Regarding the problem of physical evil, see B. Reichenbach, "Natural Evils and Natural Laws: A Theodicy for Natural Evils," International Philosophical Quarterly, 16 (1979), pp. 179-196. For a discussion on making the distinction at all, see G. Wallace, "The Problems of Moral and Physical Evil," Philosophy 46 (1971), pp. 349-351.

4e.g., Wm. Rowe, Philosophy of Religion (Encino: Dickenson, 1978), pp. 80-86.

5e.g., A. Plantinga, God and Other Minds (Ithaca: Cornell, 1967), p. 128.

6David Hume,
Dialogues Concerning Natural Religion (New York: Hafner, 1948), p. 66.

7J. L. Mackie, "Evil and Omnipotence," Mind, 64 (1955), p. 200.

8Although orthodox Christianity makes a host of other claims, (e.g., Christological), these prime theistic propositions form the necessary foundation for the other, more specific claims. The attack from evil actually has force against the whole of Christian belief, since it threatens its base. Hence, our need to reply.

9Formulation 11. received treatment by Plantinga in God, Freedom, and Evil, p. 55; formulation III. is analyzed by T. Pennelhum in "Divine Goodness and the Problem of Evil" in Brody, ed., Readings in the Philosophy ofReligion, (Englewood-Cliffs: Prentice-Hall, 1974), p. 226.

10I discuss the import of these additional propositions in my "Christian Theism and the Problem of Evil," Journal of the Evangelical Theological Society, 21 (1978), pp. 38-39. 

11Ibid., p. 39. 

12Win. Rowe, op. cit., p. 89.

13e.g., A. Plantinga, "The Probabilistic Problem of Evil," Philosophical Studies, 35 (1979), pp. 1-53.

14e.g., Win. Rowe, op. cit., p. 86.

15Formulation IV. receives attention from G. Schlesinger, Religion and Scientific Method (Dordrecht: Reidel, 1977), p. 13formulation V. receives attention in Gordon Kaufman, God: The Problem (Cambridge: Harvard, 1972), pp. 172-173. Those treating formulation VI. are examined in the body of the paper.

16A. Plantinga, "The Probabilistic Problem of Evil."

17For a more complete explanation of scientific reasoning, see C. Hempel, The Philosophy of Natural Science (Englewood-Cliffs: Prentice-Hall, 1966).

18J. Cornman and K. Lehrer, Philosophical Problems and Arguments (New York: MacMillan, 1969), pp. 340-341; italics mine.

19E. Madden and P. Hare, Evil and the Concept of God (Springfield: Charles C. Thomas, 1968), p. 3; some italics mine.

20See my "Evil and Inconsistency: A Reply," Sophia 18, No. 2 (1979), pp. 20-27.

21 gratefully acknowledge a summer grant from the National Endowment for the Humanities under which I completed this article. This study also contributed to my forthcoming book, Evil and the Christian God, Baker Book House (1981).