Sophia (2007) 46:177–187 DOI 10.1007/s11841-007-0021-8
Why Even a Believer Should Not Believe That God Answers Prayers Michael Veber
Published online: 24 May 2007 # Springer Science + Business Media B.V. 2007
Abstract Recent studies provide some support for the idea that prayer has curative powers. It is argued that even if prayers are effective in these kinds of cases it cannot be because God is answering them. While many have challenged theological explanations for the efficacy of prayer on epistemic grounds, the argument presented here concludes that the theological explanation conflicts with the standard conception of God. In particular, if God answers prayers in these kinds of cases then God is immoral. Keywords God . Prayer . Efficacy of prayer . Problem of evil Recent studies provide some support for the idea that prayer has curative powers. Here I will argue that even if it is true that prayers in these kinds of cases are effective, it cannot be because God is answering them. While many have challenged theological explanations for the efficacy of prayer on epistemic grounds, the argument presented here concludes that this sort of explanation conflicts with the standard conception of God. If God answers prayers in these kinds of cases, then God is immoral. In the first section, I offer a brief overview of the empirical evidence for the healing power of prayer. In the second section, I discuss the counterfactual theory of causation and the possible worlds analysis of counterfactual conditionals. It will be argued that in attempting to demonstrate a causal connection between prayer and healing these studies attempt to demonstrate a certain counterfactual claim. If these patients had not been prayed for, they would have continued to suffer. Once these preliminaries are out of the way, the central argument can be given. The truth of this counterfactual entails that there is a nearby world where God needlessly allows suffering and this is immoral. I answer objections to this argument and contrast it with the traditional argument from evil. M. Veber (*) Department of Philosophy, East Carolina University, Greenville, North Carolina 27858, USA e-mail:
[email protected]
178
M. Veber
The Evidence Claims about the healing powers of prayer have long been made. The scientific community has typically rejected such claims as empirically unwarranted. That is, until recently. A surprisingly large number of recent studies seem to have convinced even some scientists of respectable credentials.1 Many people who claim that prayers can heal also hold that this provides a rational basis for belief in a benevolent Deity of some sort. Those who believe that the efficacy of prayer provides evidence for the existence of God do so on the basis of the assumption that if prayer is effective then it is through divine intervention. The initial response from those who are skeptical of claims about the efficacy of prayer or about the existence of God is to attack the empirical evidence. To do that, one needs to know more about the details of the studies. To show that prayers are therapeutically effective it is necessary that prayer increases the probability of physiological improvement to a degree greater than what can be accounted for by the placebo effect. To do this, the experiments must be designed in an appropriate double-blind manner. In a study conducted at Duke University,2 150 patients who were scheduled for invasive cardiac procedures were divided equally into five groups: those who received standard coronary stenting and those who received standard coronary stenting plus either guided imagery, stress relaxation, healing touch or intercessory prayer. The prayers were conducted off-site without either the patient or the physician’s knowledge. The purpose of conducting prayers in this manner is, of course, to eliminate the possibility of a placebo effect. It is well-known that if a patient believes that he is receiving certain kinds of treatment his health will often improve even if he is not. But in these studies, since neither the patient nor the physician has the belief that he is being prayed for, this belief cannot be playing a role. It turned out that patients who received some form of non-traditional therapy, including prayer, showed a 25 to 30% reduction in adverse outcomes (such as death, heart failure, post-procedural ischemia, repeat angioplasty or heart attack) as compared to standard forms of treatment. Perhaps even more surprising is a study conducted at Columbia University’s College of Physicians and Surgeons. This study was conducted on women who were participating in an in vitro fertilization program and it was found that intercessory
1 Byrd, Randolph C., ‘Positive Effects of Intercessory Prayer in a Coronary Care Unit Population,’ Southern Medical Journal 81 (1988), 826–82; Harris, William S., et.al., ‘A Randomized, Controlled Trial of the Effects of Remote, Intercessory Prayer on Outcomes in Patients Admitted to the Coronary Care Unit,’ Archives of Internal Medicine 159 (1999), 2273–2278; Leibovici, Leonard ‘Effects of Remote, Retroactive Intercessory Prayer on Outcomes in Patients with Bloodstream Infection: Randomised Controlled Trial,’ British Medical Journal 323 (2001), 1450–1451; Cha, Kwang, Wirth, Daniel P., & Lobo, Rogerio, ‘Does Prayer Influence the Success of In Vitro Fertilization-Embryo Transfer? Report of a Masked, Randomized, Trial,’ Journal of Reproductive Medicine 46 (2001), Krucoff et al, ‘Integrative noetic therapies as adjuncts to percutaneous intervention during unstable coronary syndromes’, American Heart Journal 142 (2001), 760–769. 2
Krucoff et al, loc. cit.
Why even a believer should not believe that God answers prayers
179
prayer under similar double-blind conditions resulted in a 50% pregnancy rate compared with a rate of only 26% for those who did not receive any prayer.3 Critics will point out that there are only a handful of studies like this and therefore we are in no position to draw any surprising conclusions until more data come in. This is not an unreasonable point. Nonetheless, my challenge to the claim that prayers are effective forms of treatment will not rest on challenging the empirical evidence. In what follows, I will assume that the empirical evidence meets whatever the standards for good empirical evidence happen to be. The second sort of criticism that ordinarily arises in this context is directed against those who would claim that the effectiveness of prayer provides evidence for belief in God. Many, even many who find the evidence sufficient for belief in the curative powers of prayer, will raise doubts about whether divine intervention is its best explanation. The effectiveness of prayer might not be evidence for the existence of a benevolent deity so much as it is evidence for ESP or some hitherto unknown power of the human mind. The only response would be to construct some sort of experiment such that if certain results were obtained they could not be reasonably explained by anything other than Divine intervention. It is hard to imagine how such an experiment would go. Even if it cannot be conclusively shown that these kinds of prayers are effective because of Divine intervention, people of faith will still find such evidence very encouraging. My main contention is that this is misguided. But my argument for this will neither turn on whether there is sufficient empirical evidence for the efficacy of prayer nor on whether empirical evidence for the efficacy of prayer provides evidence for theism. Instead, I will attempt to show that the claim that God answers these kinds of prayers in this sort of situation contradicts the very essence of the standard conception of God.
Some Preliminaries: Counterfactuals and Causation At least since Hume, philosophers have seen a close connection between causation and counterfactuals. The connection is most prominent in Lewis's counterfactual analysis of causation.4 On this view, an event E causally depends on another event C if and only if the following counterfactual conditional is true: if C had not occurred, E would not have occurred. Much of the theoretical import in this analysis of causation is derived from the possible worlds semantics for counterfactual conditionals that accompanies it and that Lewis goes on to defend.5 In this way of understanding things, the counterfactual ‘If C had not occurred then E would not have occurred’ is true if and only if a world where C does not occur and E does not
3
Cha, Wirth and Lobo, op. cit.
4
Lewis, David, ‘Causation’, Journal of Philosophy, 70, (1970) 556–567.
5
Lewis, David, Counterfactuals, Cambridge, MA: Harvard University Press, 1973.
180
M. Veber
occur is nearer (or more similar) to the actual world than any world where C does not occur and E occurs anyway.6 The intuitive appeal of this account of causation and its complementary account of counterfactual conditionals are easy to see. The extent to which we are inclined to agree that Mr. X_s drinking caused the accident is the extent to which we are also inclined to say that if X had not been drinking there would have been no accident. And when we consider the other ways the evening might have turned out, it seems that situations where X does not drink and the accident happens anyway are more “far off” than those where he does not drink and everyone gets home safely. On the other hand, there are many well-known objections to this account of causation. The one that concerns us most in this context concerns causal overdetermination. Some object that the existence of causal overdetermination shows that the truth of ‘C causes E’ is not sufficient for the truth of ‘If C had not occurred E would not have occurred.’ Horwich offers the following example. A man_s death may be causally overdetermined if he is shot in the head simultaneously by two people Smith and Bloggs, acting independently of one another. In such a case the effect is not counterfactually dependent on its causes. The man would have died even if Smith had not shot him. Nevertheless, I think we would say that Smith's shot was a cause of his death.7 Lewis denies that this kind of case presents any real problem. For one thing, he claims that overdetermination of this sort is rare. He also denies that this is a clear counterexample because, he claims, it is unclear how to apply causal terminology to this sort of case. Even if this sort of objection is not decisive against the counterfactual analysis of causation, we must admit that the existence of causal overdetermination, or even simply the threat of its existence, presents serious epistemic problems. Experimental methods can be seen as attempts to circumvent these epistemic difficulties. Once we understand this, the connection between counterfactuals and experimental methods used in science becomes clear. To see this connection, some more terminology is required. Lewis defines a ‘determinant’ of any fact as ‘a minimal set of conditions, jointly sufficient, given the laws of nature, for the fact in question’.8 A fact is overdetermined when there is more than one determinant. According to Lewis, the nature of the universe is such that the sort of causal overdetermination alluded to in Horwich_s example is rare. A particular effect does not typically have more than one cause. But, according to him, from any one cause there will be many effects. ‘We may reasonably expect overdetermination toward the past on an altogether different scale from the occasional case of mild overdetermination toward the future.’9
6 As is well-known, Lewis himself believed that other possible worlds had to be the same sort of thing as the actual world. Others, of course, disagree and offer more deflationary accounts of possible worlds. Fortunately, the argument to be presented here is neutral on the question of how we ought to understand the ontological status of possible worlds. 7
Horwich, Paul, Asymmetries in Time, Cambridge, MA: MIT Press, 1987, p. 169.
8
Lewis, David, ‘Counterfactual Dependence and Time’s Arrow’, Nous, 13, (1979), p. 474.
9
Lewis, David, ‘Counterfactual Dependence and Time’s Arrow’, Nous, 13, (1979), p. 474.
Why even a believer should not believe that God answers prayers
181
How are we to know whether a putative cause C is a determinant for some effect E? Observing that C is followed temporally by E obviously will not do. This is because any particular event or cause will occur in conjunction with many other events. The world is a noisy place. It may be that one of these other events or causes that occur in conjunction with C is the determinant of E and C is, so to speak, along for the ride. Or, as in the case of causal overdetermination, it may be that C is a determinant of E and one or more of the other events that occur in conjunction with C are determinants as well. Once these possibilities are in play, it becomes difficult to know, of any particular C, whether it is a determinant of E and, likewise, it becomes difficult to know whether a particular C is a cause of E. One can see the purpose of experimental methods in science to be one of accounting for this very problem. Artificial situations are constructed so that we may isolate the relevant causal factors. If we want to know whether C is a determinant for E we must minimize causal noise. We do this by isolating C from any other possible determinants that might be occurring in conjunction with it. If C is followed by E in a situation like that, then we have good reason to believe that C is a determinant of E. If an event has only one determinant, then, were it not for the determinant, the event would not be. In other words, experimental methods attempt to construct an artificial situation where causal overdetermination is eliminated. If experimental methods fulfill their purpose, the threat of causal overdetermination will be minimized or removed altogether. In this sort of situation, our epistemic position with respect to the causal claim will be as good as it is with respect to the corresponding counterfactual. When there is no causal overdetermination, the truth of ‘C causes E’ is a sufficient condition for the truth of ‘If C had not occurred, E would not have occurred.’ This is true whether or not Lewis's analysis of causation is generally correct. All of this connects to the issue at hand in the following way. A patient is treated and then his health improves. Did the patient_s health improve because of the treatment? Or did his illness just go away on its own? Or was his improvement just a result of the placebo effect? Double-blind controlled studies are used to demonstrate causal connections by manipulating the environment in a way that isolates the phenomena of interest and removes any potential overdetermining factors.10 If these sorts of studies get positive results, we have support not only for belief in a causal connection between C and E but also for belief in the corresponding counterfactual: If C had not occurred E would not have occurred. We also see a close connection between causation, counterfactuals, and explanation. If we ask why the patient's health improves, the answer is clear: because of the treatment. Given the experimental results, what else could it be? In what follows, I will assume that there is this sort of connection between causation and counterfactuals. When double-blind experiments with controls are used to demonstrate a causal connection, they also serve to demonstrate the truth of a corresponding counterfactual. It is important to note that this is not to endorse Lewis_ counterfactual theory of causation. It is not to say that the truth of the counterfactual, ‘If C had not occurred E would not have occurred’ is always 10 For an effort to build this sort of idea into the analysis of causation itself, see Woodward, James, Making Things Happen, New York: Oxford University Press, 2003.
182
M. Veber
sufficient or always necessary for the truth of ‘C causes E’. This is just to say that double-blind controlled experiments, insofar as they support belief in a causal connection between C and E, also support belief in a counterfactual dependence between C and E. This is because worries about overdetermination are undermined in these conditions and therefore the truth of the causal claim is sufficient for the truth of the counterfactual. If the studies referred to above demonstrate that prayer caused improvement in health, they also demonstrate that if these people had not been prayed for, they would have continued to suffer from their maladies. In the next section, I will argue that if this counterfactual holds because of Divine intervention, then God is immoral.
The Argument Let us assume that the aforementioned prayer studies strongly support that prayer caused some people to heal from their illnesses. This entails that we have good reason to believe that if some of the subjects in these studies had not been prayed for they would have continued to suffer. Given the possible worlds analysis of counterfactuals, this means that there is a relatively near world w in which these people are not prayed for and they suffer. Here is where the trouble begins. In w, God knows that these people suffer and could do something to stop it but does not. Ex hypothesi, the only relevant difference between w and the actual world is that the former contains unknown third parties who are praying on behalf of them. But this is not a morally relevant difference. The fact that God relieves this suffering in the actual world shows that it is in no way morally necessary. And if it is not necessary in the actual world it is not necessary in this nearby possible world either. If God allows needless suffering in w, then God is immoral in w. Before laying out the argument more rigorously, two assumptions should be made explicit. First, I assume the standard philosophical conception of God.11 God is, by definition, all knowing, all powerful and perfectly good. To say that God possesses these properties ‘by definition’ is to say that He holds them essentially and that it would be contradictory to deny that He holds any of them.12 The role of that assumption should already be clear. Another assumption, whose role will become clear in what follows, is what I will call the ‘Imminence of Theodicy’. The idea here is that any explanation that we offer for God's allowance of suffering in some world w, must be an explanation that applies to w. For example, if we say that God allows people to suffer in w because it serves some greater good, it must be that this greater good is served in w. It must be that these goods are actual relative to the world we are talking about. This way of putting things may sound overly sophisticated, but the underlying idea is simple and intuitive. If God allows suffering to produce some sort of benefit, this benefit must not be a merely possible one. I include the modifier Fphilosophical_ in acknowledgement that the God of the philosophers may not be the God of Abraham, Isaac, Jacob or even ordinary folk.
11
12 To avoid begging any questions against St. Anselm, I leave it open whether ascribing such properties to God entails that God exists.
Why even a believer should not believe that God answers prayers
183
The central argument can be summarized as follows. Let α be a patient who received remote intercessory prayer in one of the aforementioned empirical studies. Let C be the claim that if α had not been prayed for then α would have continued to suffer. Let w be the nearest non-prayer world.13 And assume that C holds because of Divine Intervention. Then we have: 1. 2. 3. 4.
If C then α suffers in w. If α suffers in w then God is immoral in w. If God is immoral in w then God is not essentially good. Therefore, if C then God is not essentially good.
The first premise is analytically true given the standard analysis of counterfactual conditionals. The second premise is supported by the following line of reasoning. Since God relieves α_s suffering in the actual world it is unnecessary in the actual world and if α_s suffering is unnecessary in the actual world then it is unnecessary in w. The third premise is also analytically true and the argument is deductively valid. Whether or not we are theists, acceptance of the standard philosophical definition of God forces us to reject the claim that prayer was effective because of Divine Intervention in these kinds of studies. Either the studies are flawed and we should not accept C or the prayers were effective for some other reason.
The Traditional Argument from Evil: Similarities and Differences Since 1 and 3 above are analytically true, and the argument is deductively valid objections to the argument will, no doubt, center on premise 2. Readers will notice similarities between this argument and the traditional argument from evil. Both assume a simple “tri-omni” conception of God and both turn on worries over why such a God would allow suffering. And indeed, the support for premise 2 above is just the usual sort of support one gives for the crucial premises of the argument from evil. One ought to do what one can to help others and if we fail to do so, we are not acting as well as we could. But there is also an obvious difference. In this case, the difficulty is not to explain why God allows suffering in the actual world. Instead, the task is to explain why God allows suffering in some other nearby possible world. This might motivate the following sort of reaction. Why should anyone care whether God allows suffering in some other possible world? As suggested in the previous section, it certainly does matter if something you believe entails that there are possible worlds in which God behaves immorally. While some think that God contingently exists, no one thinks that God is contingently good. Any being who behaves immorally just could not be God. God’s goodness is essential. But to say that God’s goodness is essential is to say that in every world in which God exists, God is good. Therefore, to say that there are possible worlds in which God does bad things is to contradict a fundamental component of our conception of what sort of being God is. So the fact that the argument concerns God's behavior in some other possible world in no way trivializes it. 13 As Lewis suggests, there may not be a “nearest” possible world (any more than there is a real number “closest” to zero). I set aside this worry for ease of exposition.
184
M. Veber
In recognition of the similarities between this argument and the traditional argument from evil, one might also offer a free will objection to premise 2 above. This familiar idea is just that human suffering is a consequence of the fact that we have free will. Could the same sort of response be given here? It is difficult to see how the existence of free will in w could explain the instances of suffering we are worried about. Suppose Joe is a subject of the Duke study and he is one of the patients who ended up, as a result of chance, in the group that received intercessory prayer. If the prayer prevented him from suffering postoperative trauma then the following counterfactual is true: if he had not been prayed for then he would have suffered. In other words, in the nearest world in which he is not prayed for, he suffers. Now, whatever the merits of the free will explanation for evil generally, it does not work well here. One relatively near non-prayer world is a world where Joe randomly ends up in the control group. The fact that Joe, in the nearest non-prayer world, is not being prayed for (and therefore, is suffering) has nothing to do with decisions made by him. It is difficult to say whether the nearest non-prayer world fits this description. But the point is only that the differences between the actual world and the nearest non-prayer world are minor and morally irrelevant. Of course, one could reply that Joe_s suffering is necessary as punishment for things others have done or failed to do. In this example, perhaps Joe suffers in w as punishment for the fact that the third parties did not pray. This kind of position, however, is a non-starter. Any God who punishes people for the sins of others, especially when the sins are this minor (if they are sins at all), is an unjust God.14 What of the objection that all suffering is necessary for some greater good? The most prominent version of this sort of theodicy can be found in the work of Hick.15 On this sort of view suffering is necessary for human development and this is why a perfectly good God allows it. However plausible this line of criticism is against the traditional argument from evil, it fails here. It is assumed ex hypothesi that, in the actual world, God answers the intercessory prayer. If God prevented Joe from suffering on the basis of this prayer then, as far as the actual world goes, Joe’s suffering was not necessary for any greater good, including anyone’s spiritual development. But what about the nearest non-prayer world where Joe is not prayed for and he therefore suffers? The only relevant difference between this world and the actual one is that there is no intercessory prayer for Joe. But why does this fact make Joe’s suffering in w morally necessary? It does not. If Joe’s suffering is not necessary in the actual world then it is not necessary in w. Yet, if the counterfactual claim is true, God is allowing this unnecessary suffering to take place in w. Therefore, God is not essentially good. In response, one might point out that that Joe’s suffering in w is necessary for the truth of the counterfactual C. Since C is entailed by the claim that the aforementioned studies provide good evidence for the healing powers of prayer, we can have good evidence for the healing powers of prayer only if Joe suffers in w. This point can be combined with greater good theodicy to explain why God allows Joe’s suffering in w. He does so because this is the only way to provide good 14 This point is just an echo of the very common complaint that free will theodicy fails to address natural evils. 15
Hick, John, Evil and the God of Love, San Francisco: Harper, 1978.
Why even a believer should not believe that God answers prayers
185
evidence for the healing powers of prayer and, by providing this evidence, God provides all of us with an incentive to pray. By praying, we improve our spiritual lives and our relationships with God. That we do so for good scientific reasons is all the better. Again, Hick’s idea that God allows suffering in service of human development is confirmed. This response, however, conflicts with the Immanence of Theodicy assumption defended in the first section. Whatever benefits accrue from God’s allowance of suffering in w, those benefits must be, from the perspective of w, actual benefits. That God would allow Joe to suffer in w for reasons that are, as far as w is concerned, merely possible is unacceptable. This point applies whatever the benefits of prayer are taken to be. Since the prayer does not occur in w, its benefits in the actual world are irrelevant to the question of why God would allow patients to suffer in w.
Other Objections and Replies The above argument might appear to assume a rigid and, some would say, hamfisted categorization of actions of the sort found in very primitive versions of utilitarianism. On this sort of view, every act is either obligatory or prohibited. So far, it may appear that the above discussion assumes that when God relieves a patient’s suffering, this act is necessary either for the greater good or as a consequence of our having free will and thus it is obligatory. If it is not necessary for either of these, then the act is prohibited. It is false, however, that every act is either obligatory or prohibited. It may be that God heals the sick and it is good for Him to do so, even though it is not morally obligatory. When God answers intercessory prayers of this sort, it may be that this is an act of supererogatory mercy. And if the act is supererogatory, then, by definition, there is nothing wrong with refraining form doing it. Thus, it could be argued, premise 2 is false. God is not doing anything wrong by failing to heal people in w even though his healing people in the actual world is a good thing. Supererogatory acts are not obligatory but they are distinct from merely permissible acts in that they are praiseworthy. If an agent's action A is supererogatory then the agent is better for having performed A than he would have been had he refrained from doing A. This is true even if the agent would not be morally blameworthy in so refraining. Now suppose that an agent contingently performs supererogatory act A. In that case, there are possible worlds where the agent does not do A and therefore there are possible worlds where the agent is morally inferior to his real world counterpart. Again, this is true even if the agent in question is doing nothing wrong in worlds where he fails to do A. This line of reasoning shows that the current objection fails not because it is wrong to think that God can perform a supererogatory act or because it is wrong to think of God’s answering an intercessory prayer as supererogatory, but because God cannot contingently perform a supererogatory act. Suppose that God_s relieving Joe_s suffering is supererogatory. Given the truth of the counterfactual C, there is a world w where the prayer is not made and God does not help Joe. Since God does help Joe in the actual world, this means that it is possible for God to be morally worse than he is. To put it another way, in world w, God could do better, i.e., God
186
M. Veber
could heal people as He does in the actual world. But if God is morally perfect then it is not possible for God to be any better. Furthermore, if God is essentially morally perfect, then, necessarily, it is not possible for God to be better. Thus, there is no possible world in which it is true that God could be better. But if God contingently performs supererogatory acts, then there are possible worlds in which it is true that God could be better. In this case, it is true in w that God could be better than he is. Therefore, insofar as God is essentially morally perfect, He does not contingently perform any supererogatory acts. To say that He does would, once again, conflict with our conception of what sort of being God is. Of course, this is not to say that he does not perform any supererogatory acts, but that if He does, He must. If God’s act of healing Joe is supererogatory, then there is no world in which he refrains from doing it. And therefore he does not refrain from doing it in the nearest non-prayer world. And if that is true, then we must reject C. Jim would have been healed whether or not anyone prayed. This confirms the central thesis of the paper. To say that prayers, in these sorts of experiments, are effective because God is answering them is inconsistent with our conception of God. It is assumed here that God exists in w. If it turns out, for whatever reason, that the nearby worlds in which prayers are not made are also worlds in which there is no God, then the above argument will not go through. God cannot be blamed for not intervening and relieving or preventing the patient’s suffering in the absence of prayer. Therefore, it must be assumed that the nearest non-prayer worlds are not also non-God worlds. Could the conclusion be blocked by rejecting this assumption? If it is true that the nearest non-prayer world is also a non-God world, then so is the following counterfactual conditional: If no one had prayed for Joe, then God would not have existed. If the counterfactual theory of causation is correct and the truth of the counterfactual conditional is a sufficient condition for the truth of the corresponding causal claim, this amounts to saying that the prayer causes God to exist. This again contradicts our conception of what kind of being God is. It is one thing to say that God’s existence is contingent, but no one who accepts the Divine Intervention explanation for the effectiveness of prayer should want to make God’s existence that contingent. But I do not want to assume the truth of this controversial account of causation. Even if the counterfactual theory of causation is incorrect, the above counterfactual is clearly false. If God exists in the actual world, then God exists in the nearest non-prayer world. Oftentimes, when confronted with the argument from evil, theists assert that there must be some reason why evil is necessary even if we cannot figure out what it is. In that context, this kind of response sounds more like a capitulation speech than a genuine solution to the problem. But could a similar move be more successful in this context? One might reject the assumption that with respect to a patient like Joe, there is no morally relevant difference between the actual world in which he is relieved from suffering and the nearest non-prayer world in which he suffers. One might argue for this point in the following way. Anyone who countenances possible worlds talk should recognize that as we consider counterfactual worlds, that is, worlds in which P is true although P is false in the actual world, we must not only change that proposition P, but also every proposition that is logically connected to P. If we construct other possible worlds by changing the truth values of propositions about the actual world, then our changes must preserve logical consistency. Therefore,
Why even a believer should not believe that God answers prayers
187
there is no world identical to the actual world in every respect except that ‘There are bricks’ is false. For if there are no bricks then there could be no brick houses, nor any brick masons, nor any bricks thrown through windshields, and so on. It is here that a defender of the divine intervention explanation for the efficacy of prayer might find refuge from the above argument. It could be maintained that in every case where we assign a value of ‘False’ to propositions of the form patient X was prayed for, we end up having to change other propositions and, as a result, patient X’s suffering becomes morally necessary. In the context of the traditional argument from evil, this sort of defense gets whatever plausibility it has from its own vagueness. ‘God has His reasons, whatever they may be’ is supposed to liberate the theist from any obligation to establish exactly what that reason is. Here we can see another important difference between the argument here and the traditional argument from evil. In this case, we know exactly what God’s reason for allowing Joe to suffer in this possible world is: the third party did not pray. Since there appears to be no logical connection between propositions about these distant people praying and propositions relevant to whether Joe deserves to suffer, the bald assertion that logical consistency requires rejecting the proposition Joe’s suffering is unnecessary once we accept the proposition no third party prayed, is merely ad hoc. The burden is on anyone who makes this sort of claim to demonstrate that there is this logical connection. I see no plausible way to bear this burden.
Summary and Conclusion If recent studies justify belief in a causal connection between prayer and healing the sick, they also justify belief in the following counterfactual C. If certain people had not been prayed for then they would have continued to suffer. If C holds because of Divine intervention, then there is a near possible world w in which certain people are not prayed for and they suffer. And this, I’ve argued, is inconsistent with our own conception of what sort of being God is. There are at least two places for a theist to go from here. First, she can reject C and say that these people would have been healed whether or not anyone prayed. Given that the counterfactual is entailed by the claim that these studies support belief in a causal connection between prayer and healing, this person would also have to say that they do not (and indeed cannot) epistemically support such a connection. Many will find this line unacceptable because it amounts to settling the straightforwardly empirical question of whether prayers have curative powers on purely religious grounds. A second response would be to allow that perhaps C is true. But if it is true, it could not be because God answered those prayers. This would mean that the sort of evidence gained for belief in a causal connection between prayer and healing does not and, given the argument above, could not provide any support for belief in the existence of God. Many believers will find that position also unsettling.16 16 This paper was presented at Flagles College in 2003. The author thanks E.R. Klein and all who attended. The author also thanks A.J. Kreider and three anonymous referees for their helpful critical comments on an earlier version of this paper.