1. The General Strategy: The G. E. Moore Shift
My response to the problem of evil takes its original inspiration from a famous argument of G. E. Moore (1939). Moore thinks that if the skeptic is free to argue that:
P1.1 if I might be dreaming, then I do not know that I have hands;
P1.2 I might be dreaming;
C1.1 I do not know that I have hands;
then the non-skeptic is free to argue in response:
P2.1 if I might be dreaming, then I do not know that I have hands;
P2.2 I do know that I have hands;
C2.1 it’s not the case that I might be dreaming.
Unless we have a good reason for thinking that P1.2 is inherently more probable than P2.2, then Argument 2 is at least as compelling as Argument 1.
I wish to argue that the position is similar with respect to the problem of evil. Whereas the problem of evil might be framed as follows:
P3.1 if gratuitous evil exists, then there is no classical God;
P3.2 gratuitous evil exists;
C3.1 there is no classical God;
the rational theist may respond with the following:
P4.1 if gratuitous evil exists, then there is no classical God;
P4.2 there is a classical God;
C4.21 gratuitous evil does not exist.
Moreover, I maintain that Argument 4 is no worse than Argument 3.
I am not the first person to propose applying the “G. E. Moore shift” to the problem of evil. In his classic introduction to the philosophy of religion, Brian Davies proposes a similar move. He suggests accepting that evil exists and that God exists and concluding that evil and God can exist simultaneously (Davies 1993, 53-54). However, there is one crucial difference between my proposal and Davies’s. Davies seems to think that the justification for the G. E. Moore shift in the context of the problem of evil depends on the presence of genuine knowledge of the existence of God. I disagree; I think that Argument 4 is as good as Argument 3, but not because P4.2 has been established. Indeed, I think that we may even assume to have no evidence whatsoever for the existence of God. Nevertheless, I think that Argument 4 is as good as Argument 3. This, of course, means that I am committed to rejecting the claim that we have any particularly good evidence for P3.2. I shall spend most of the rest of this paper arguing for this claim.
2. The “Evidence” for Gratuitous Evil
Numerous examples of manifest evil have been proposed as evidence for the non-existence of God. Murder, rape, natural disasters, the inevitability of death—there is no lack of available evidence for the existence of evil. I accept that evidence for evil exists and is overwhelming. However, that is not enough for the problem of evil. The argument from evil, as I set it out in the previous section (i.e., Argument 3), involves the premise “gratuitous evil exists.” Without further argument, there is no justification for inferring “gratuitous evil exists” from “evil exists,” so the obvious availability of cases of evil are only relevant to the issue of theism if some reason can be provided for thinking that one or more such cases is gratuitous. I do not think such a reason is available.
It is true that, historically, many philosophers and non-philosophers have taken the mere existence of evil to be sufficient evidential support for atheism. This line of thought is sometimes called “The logical problem of evil,” and its most famous exponents were David Hume (1779) and J. L. Mackie (1955). However, an emerging consensus amongst professional philosophers of religion, regardless of their position on the existence of God, is that the logical problem of evil fails. The reason, simply put, is that it seems, as a matter of conceptual analysis, that the idea of a divine reason for allowing evil is not incoherent. It is possible that the classical God exists and that evil exists at the same time, because God may have good reasons for allowing evil. If this notion does represent a logical possibility, then we cannot deduce from the existence of evil the non-existence of God.
With the failure of the logical problem of evil, the focus has turned to the evidential problem of evil. This differs from the logical problem in that it focuses on trying to show that there exists gratuitous evil—that is, evil for which there is no divine reason. The general argumentative strategy is to introduce evidential support for the premise that such evil exists by pointing to apparent instances of such evil. The issue is complicated by the presence of two well-known reasons God might have for allowing suffering: the free will and soul-making theodicies.
According to the free will theodicy, God allows evil, because even an omnipotent, omniscient God could not eliminate evil while still giving humans genuine freedom. Moreover, genuine freedom is so good that it is worth the evil that comes with it, and therefore a benevolent God would allow this evil in order to give humans freedom.
According to the soul-making theodicy, God allows evil because some virtues of the soul (e.g., moral rectitude and perseverance) are such that even an omnipotent, omniscient God could not eliminate this evil while still giving humans the opportunity to acquire these virtues. Moreover, these virtues are so valuable that they are worth the evil that comes with them, and therefore a benevolent God would allow this evil in order to enable humans to acquire these virtues.
It is by no means universally agreed that free will and soul-making do provide sufficient reasons for God to permit the existence of evil. But even if they do, it seems that individual instances of evil might still provide an evidential basis for the conclusion that there is gratuitous evil, because some instances seem wholly unrelated to human freedom and soul-making. A particularly problematic case is the case of unseen animal suffering. If freedom is a trait particular to humans (a common view among orthodox theists), then human freedom cannot provide a reason for God to permit animal suffering, or so it seems (I shall subsequently have reason to add a caveat to this claim). If perseverance, moral rectitude, and other traits of the soul are built up by facing evil (as seems very plausible), then these cannot provide reasons for God to permit evil beyond the sensory fields of humans. So if there is evil which is not inflicted by or on humans and does not occur within their sensory fields, it is a particularly good candidate for gratuitous evil. Unseen (by which I mean, unseen, unheard, unfelt, and so on) animal suffering seems to fit this description as well as other things could. It seems to many people that animal suffering, seen or unseen, is an intrinsic evil. There is a common presupposition that unseen animal suffering is widespread, and there does not seem to be any good reason for God to permit it. Each of these three claims can be challenged, however. To be more precise, the claim that unseen animal suffering provides evidential support for the claim that there exists gratuitous suffering depends on the truth of each of the following three theses: (a) unseen animal suffering is evil, (b) unseen animal suffering occurs, and (c) there is (probably) no reason why a classical God would permit unseen animal suffering.
Of these, I will take the truth of (a) for granted, simply because I find its rejection too great a departure from my own ethical intuitions. Others may, of course, have very different ethical intuitions from mine, but I am unable to endorse a defense against the problem of evil which involves rejecting (a). I am interested, rather, in which of (b) or (c) might be rejected. In recent decades, it has become both the common wisdom and scientific orthodoxy that animals experience suffering. This might be taken to suggest that we are better off rejecting (c) than (b). Indeed, an important tradition does reject (c). According to skeptical theism (see, e.g., Wykstra 1984; Alston 1991; and van Inwagen 1995), if theism is true and God does have reasons for permitting evil, humans would lack the right sort of epistemic access to be able to know what these reasons are. Given that, there is no good inference from “we are not aware of anything that might count as a reason for God to allow evil ϕ” to “there is probably nothing that would count as a reason for God to allow evil ϕ.” This is a well-developed and popular line of response to the problem of evil, but it will not be my line. There are a number of important objections to skeptical theism, but a central worry is that it commits its defender to too much skepticism, either a general skepticism about modal claims or a skepticism about moral claims: how could we know what actions are good or evil if we have no probably true beliefs about what is an overall good or bad action from God’s point of view? The second of these objections particularly bothers me, because I think that a grasp of how God would or wouldn’t act plays a central role in the theist’s attribution to God of the property of omnibenevolence, but also sometimes plays a role in practical decision-making in the context of religious ethics; that is to say, religious folks sometimes decide how to act (and justifiably do so, in my view) by asking themselves, “What would God do?” Both of these roles look threatened by skeptical theism. I do not think these are knock-down objections by any means, but I think, together with other objections, they make it worthwhile to consider alternative responses to the problem of evil.
My response takes as its starting point that, in some contexts, we know exactly how God would act and involves a rejection of (b), according to which unseen animal suffering occurs. Or rather, strictly speaking, it combines (b) and (c); it suggests that unseen animal suffering occurs far less than we might expect and that when it does, it is because God has a reason to allow it. I will develop this response in the next section.
3. Do Animals Suffer When There Is No One Around to See Them?
I think we know exactly what God would do if a genuinely gratuitous evil were about to take place. An omnibenevolent God would intervene to prevent it. I think that this is an analytic truth. Omnibenevolence rules out genuinely gratuitous evil. This is the major premise of Arguments 3 and 4, and, in keeping with my Moorean approach, I take this analytic fact as the cornerstone of my response to the problem of evil.
A similar skepticism to skeptical theism is at play here, but there is also a major departure from that approach. Whereas the skeptical theist is skeptical about our knowledge of the sorts of divine reasons that might or might not exist and therefore rejects the inference from the existence of evil to the existence of gratuitous evil, I am skeptical of the existence of certain kinds of evil, and I reject certain inferences that conclude that such evils exist. I am not skeptical of the existence of evil generally. That would be a much stronger position than I intend, and indeed, I think, one that is incompatible with orthodox theism. Rather, I am skeptical of the existence of evils that would make good candidates for gratuitous evil. My position depends on the claim that we do not have good evidence for the existence of these sorts of evil.
As a standalone defense against the problem of evil, mine would be unsuccessful. I do not dispute that we have very good evidence for all sorts of evil: moral evil, natural evil, and even animal suffering. However, I take it that each of these, and all the others for which we do have good evidence, can be accounted for by one or other of the two classical theodicies: free will and soul-making.
I think also that some instances of unseen animal suffering can be accounted for by the existing theodicies. I do not dispute that non-gratuitous unseen animal suffering occurs. My position is only that the rational theist is perfectly within her rights to reason that God would not allow animals to suffer when there is nothing to be gained by that suffering; so if there is a God, there is no animal suffering of this kind. Any apparent suffering of this kind must be illusory if God exists. I maintain that, in the case of unseen animal suffering at any rate, this does not render the existence of God improbable.
This entails that we do not have strong evidence for the unseen suffering of animals. I wish to briefly defend this claim. It seems that I must reject one of two claims: either unseen animals do not act as if they were in pain or unseen animals act as if they are in pain even though they are not. Both of these options might seem problematic. The first, I think, is much more problematic and I wish to reject it. I believe that we do have sufficient evidence that animals act as if they are in pain even when they are beyond our immediate sensory access. For example, there is ample evidence that animals die unseen in the form of carrion, fossils, oil, and so on. In certain cases, these remains can tell us about the behavior of the animals before their death. Instead, I think that the theist is rationally justified in believing that unseen animals might act as if they are in pain but not be in pain. In particular, I think this occurs when God miraculously intervenes to prevent an instance of gratuitous suffering that was about to occur.
Now, my claim that the theist is rationally justified in believing that unseen animals act as if they are in pain without actually being in pain might seem to be in conflict with my earlier appeal to the emerging scientific consensus that animals do feel pain. However, I believe that it is not. I maintain that scientists, and anyone else for that matter, are in a relevant epistemic position for discovering the internal mental states of seen animals, but not those of unseen animals. When an animal is there with us, writhing in pain, or expressing its discomfort in sound, I believe that we are provided with ample evidence for its suffering. That evidence, however, does not transfer over to the suffering of unseen animals. We have immediate, perceptual evidence that animals sometimes suffer. To get from this to the more general conclusion that animals suffer regardless of whether or not there are people around to see them, we must draw an inference. Inferences from cases of which we have perceptual experience to cases of which we do not have such experience are not always good inferences. In this case, the inference is backed up by the following considerations: animals plausibly constitute a natural kind, and, again plausibly, one of the features of this kind might be the ability to suffer. If these plausible assumptions are right, then it might be an empirically discovered law of biology or some other special science that, for example, all animals suffer when they are burned. This law would hold just as much of the unseen animals as the seen ones. Laws are projectable; that is to say, they license inferences from instances that have been experienced to instances that have not. On the face of it, this is a problem for my view. However, it is highly doubtful that we have an exceptionless law in this area. Certainly, there are exceptions to the proposed law “All animals suffer when they are burned.” We may put counterexamples to the side by the simple device of adding a ceteris paribus clause to our law: “all animals suffer when they are burned, unless… .” Nancy Cartwright (1980) argues that all laws of nature are of this form. We might still have a law here, but our justification for drawing inferences about particular instances on the basis of this law is only as good as our justification for thinking that those instances are not the exceptions set aside by the ceteris paribus clause. In effect, my proposal amounts to this: the purported laws which license the inference from seen animal suffering to unseen animal suffering are subject to the following ceteris paribus clause: “… unless the burning occurs in a situation in which there exists a God and the animal’s suffering would serve no greater purpose.” I think that both theist and non-theist should accept that this clause does genuinely constrain the possible applications of the law. Where there might be reasonable disagreement is on whether or not any given purported instance of animal burning occurs in a situation in which there exists a God. However, I think the rational theist need not attach a lower than 0.5 probability to this eventuality.
In short, the inference from the agreed instances of animal suffering to the best candidates for gratuitous suffering (that is, instances of unseen animal suffering) depend on an application of laws which the rational theist is within her rights to rule out as a misapplication. In other words, the theist is not rationally compelled, on the basis of available evidence, to conclude that unseen animals probably suffer. To return to the Moorean shift and Arguments 3 and 4 from Section 1, this means that, so long as it is the case that the rational theist is entitled to weight the probability of the existence of God no lower than 0.5, then P4.2 might be at least as well motivated evidentially as P3.2 (at least in so far as that premise is based on the evidence of unseen animal suffering). If the two premises, and therefore the two conclusions, are equally well motivated evidentially, then the rational theist may well be able to point to non-evidential reasons for preferring Argument 4 over Argument 3.
My position, as sketched in the previous section, has something in common with that defended by Michael J Murray in his Nature Red in Tooth and Claw (2008). Murray also defends a skeptical stance towards the existence of animal suffering. However, I am keen to distinguish my view from Murray’s. Murray defends the plausibility of the Cartesian view of animals—that they lack all subjective experiences whatsoever. This view is very strong and brings with it significant ethical consequences. Quite apart from the consequences for the morality of meat eating and the fur and diary industries, Cartesianism seems to have the consequence that deliberately harming, even torturing (that is, acting in such a way that torture would occur if animals could feel pain), animals is morally permissible or at the very least much less morally reprehensible than intuition tells us.
For this reason, I do not wish to commit myself to Cartesianism, and therefore I wish to draw an important distinction between Murray’s position and my own. I believe that animals do have fully developed mental lives. They regularly experience a range of positive emotions, and they sometimes experience negative emotions. In particular, animals experience negative emotions when those emotions serve a particular purpose justifiable from God’s perspective. I have already committed myself to the existence of two such purposes: animals suffer when their suffering is required to provide genuine freedom to human beings; for example, in order to truly be free, humans must be able to choose to actually inflict harm on animals, whether it be for food, clothing, or pleasure. Animals, therefore, might suffer when being tortured, when being slaughtered, held by humans in uncomfortable accommodation, or caught in deliberately set forest fires. Furthermore, human freedom may even account for the suffering of some animals at a great distance from humans, such as those that suffer as a result of man-made climate change. All of these sufferings are real and, at least, if the free will theodicy works, none are gratuitous. The other purpose that justifies some occasions of animal suffering is brought out by consideration of the soul-making theodicy. Animals that die from natural causes can, in some cases, provide suitable opportunities for human soul-making. Witnessing the death of a much-loved pet provides an opportunity for moral growth. It seems to me that instances of animal suffering that are explicable for these reasons are reasonably common. In other words, Cartesianism simply does not follow from my view.
The significance of all of this is that we are left with an interesting symmetry between our evidence for the occurrence of evil and the scope of traditional theodicies. I believe that ancient theodicies can account for all apparent evil that occurs suitably close to humans. The more plausible cases of gratuitous evil are to be found further away, but this is just what opens up the possibility of adopting a skeptical attitude towards these cases, without in any way falling into the apparent general skepticisms that seem to infect skeptical theism. This symmetry is not a perfect symmetry, as the climate change example shows, but the result is the same: the rational theist is within her rights to reject unseen animal suffering as an instance of (probably) gratuitous evil. In the final section of this paper, I will consider a possible objection against my defense that may have occurred to some readers.
5. Objections: Inherent Probability
The objection that I will consider is that my attempted defense fails, since it is inherently unlikely that there exists a God that intervenes every time an unseen animal is about to experience gratuitous suffering. A version of this objection goes like this: according to the standard probability theory, if proposition P asserts the existence of something, x, and proposition Q attributes some non-trivial qualitative property to x that was not already attributed to x by P, then the probability of (P&Q) is lower than the probability of P. Let us apply this to the current case. If P states that “there is a God” and Q states that “God intervenes regularly to prevent the gratuitous sufferings of animals,” then my position commits me to (P&Q), which has a lower probability than the prior probability of theism. This is important because if the principle of indifference is true, then assuming that we have no good argument either for or against the existence of God, the probability of P is no higher than 0.5. If that is right, then the probability of (P&Q) is lower than 0.5. In other words, so the thought goes, my position is unlikely to be true.
I have two responses to this argument. First, I am inclined to reject the principle of indifference and, in doing so, reject that without a good argument for the existence of God, the probability of God is no higher than 0.5. On the view to which I subscribe, all that we can say in the absence of good arguments either way is that we do not know what the probability of there being or not being a God is. If this is right, then even if it is true that the probability that God exists and God intervenes regularly to prevent the gratuitous sufferings of animals is lower than the probability that God exists, this still does not entail that the former probability is lower than 0.5. My second response to this objection is that I do not think it is true that the probability that God exists and God intervenes regularly to prevent the gratuitous sufferings of animals is lower than the probability that God exists. As I have said above, I consider “If God exists, God intervenes to prevent gratuitous evil that would otherwise happen” to be an analytic truth. This truth is implicit in the notion of God. Some analytic truths might nevertheless prove unexpected to those, for example, with only a limited mastery of the relevant concept. Not so with this one, I suggest. Anyone who is familiar with the concept of a classical God understands that God would not allow gratuitous evil. So it seems to me that “God exists and would intervene to prevent any gratuitous evil that would otherwise happen” is just a trivial unpacking of “God exists” and has the same probability. From the probability of these beliefs to arrive at a probability for “God exists and does intervene to prevent any gratuitous evil that would otherwise happen,” we take the prior probability of “God exists” and factor in the probability that there is evil that would otherwise happen. But this will not unduly lower the posterior probability, because it is obvious that there is such evil, and more importantly the problem of evil depends on there being evil that would otherwise happen. So if factoring in this evidence reduces the probability of “God exists and does intervene to prevent any gratuitous evil that would otherwise happen,” it also reduces the probability of the existence of gratuitous evil and thereby undermines the atheist’s argument.
In short, if I am right that “God exists and would intervene to prevent any gratuitous evil that would otherwise happen” has the same probability as “God exists,” and if we attach ≥0.5 prior probability to these claims, then it is at least as likely that God intervenes to prevent any gratuitous evil that would otherwise happen as it is that there is gratuitous evil.
By way of summary, I have presented in this paper a defense against a particular form of the problem of evil. Specifically, I have been concerned with animal suffering which happens beyond human sensory experience. The significance of this particular sort of evil is that it, perhaps uniquely, seems to avoid any possible response which appeals to the free will and soul-making theodicies. In other words, the conjunction of the two traditional theodicies with the position expounded here may be seen as a complete response to the problem of evil. According to the defense outlined in this paper, the rational theist is entitled to employ the “G. E. Moore shift” in response to supposed instances of unseen animal suffering. God, as classically conceived, would not permit the gratuitous suffering of unseen animals, so such suffering does not occur. I claimed that this argument is perfectly good even if the rational theist has no good argument in favor of the existence of God. This last claim depends on there being no good evidence for the gratuitous suffering of unseen animals. That, I have argued above, is just how things stand. There is plenty of evidence for the suffering of seen animals (and this suffering may very well be explicable by the theodicies), but as soon as the suffering we are interested in is sufficiently removed from human experience, our evidence for it also evaporates. The only real argument in favor of unseen animal suffering depends on an inference from how things are when seen by us to how things are when unseen by us. But, if we are justified in believing that there is a God, as classically conceived, then we are justified in rejecting this particular instance of this inference pattern. I, therefore, conclude that rational theism is not undermined by the purported case of unseen animal suffering
1. It might be objected that if, as I have granted above, we do have evidence that animals act as if they are suffering even when we do not see them, then our evidential relation to unseen animal suffering is really no different from our evidential relation to seen animal suffering. I don’t think that this is right. I subscribe to the so-called “perceptual hypothesis,” according to which we literally see the mental states of others. I think the best argument against Cartesianism—the view that animals lack subjective mental states—is that we see those mental states. We do not do this in the case of unseen animals.