Draper's Argument from Pain and Pleasure

Every version of the problem of evil met so far has turned on a claim about justification: that some suffering has no adequate reason behind it, or that no reason could be adequate to some suffering. The theist's replies have all attacked that claim, either by supplying a candidate reason or by arguing that our failure to find one shows nothing.

Paul Draper's argument, set out in "Pain and Pleasure: An Evidential Problem for Theists" in Noûs in 1989, does not make that claim at all. It concedes, for the sake of the argument, that every instance of suffering might have a justification we cannot see. What it asks instead is a comparative question: which of two hypotheses would have led you to expect the pattern of pain and pleasure that we actually find? The answer, Draper argues, is not theism, and the gap is large.

This makes the argument unusually difficult to answer, because the standard defensive moves were all built against a different attack.

A different kind of argument

The distinction matters for what counts as a successful reply. Against a claim of incompatibility, producing any coherent story on which God and the evil coexist is enough, which is why the free will defence settled the logical problem. Against a claim about evidence, a coherent story is not enough. The story has to be one that theism gave some reason to expect in advance, because a hypothesis is not supported by data it merely tolerates.

That was already the shape of the evidential problem of evil in its earlier form, which infers from apparently pointless suffering to the probable existence of gratuitous evil. Draper's argument is a further step away from justification talk. It never asserts that any suffering is gratuitous, so it never has to defend an inference from what we cannot see to what is not there. The rung on Rowe and skeptical theism takes up that inference directly; this one bypasses it.

Where the argument comes from

The move is not new, though its formal statement is. Part eleven of Hume's Dialogues Concerning Natural Religion has Philo grant, for the sake of argument, that the world's evils are all consistent with a benevolent designer, and then observe that consistency is not what natural theology needs. Philo's list of the general circumstances that produce suffering, among them the mechanism of pain as a spur to action and the frugal way in which creatures are supplied with the powers they need, is a list of design features rather than of individual tragedies, and the complaint is that these features are not what one would have anticipated in advance from a benevolent designer, whatever story can be told about them afterwards.

The second source is biology. Writing to Asa Gray on 22 May 1860, Darwin said he could not persuade himself that a beneficent and omnipotent God would have deliberately created the ichneumonid wasps to feed inside the living bodies of caterpillars, and added that he saw no evidence of design and beneficence in nature as clearly as others claimed to. The same letter contains the reply that theistic evolutionists have relied on ever since: Darwin also said he was inclined to see everything as the result of designed laws, with the details left to what one might call chance. Both the argument of this rung and its most common answer are present in a single letter, which is a useful indication of how little the terrain has shifted.

What Draper adds is not new evidence but a framework in which the observation can be stated as a claim about probabilities rather than as an expression of dismay, and therefore assessed.

The rival hypothesis

The comparison needs a second hypothesis, and Draper's choice of one is the argument's most carefully engineered feature.

The weakness is the point. A stronger rival would have been easier to attack: a theist could respond to naturalism by listing everything naturalism has trouble absorbing, which the previous rung catalogued. Indifference asserts almost nothing, and so incurs almost none of those costs. It also does not require the rival to be atheism, which means the argument can be run by someone who holds that a God exists but is not the sort of agent who arranges the biology of pain.

There is a second reason for the choice. A likelihood comparison is only informative if the two hypotheses are not wildly unequal in prior probability, since a large enough prior advantage can absorb a large likelihood gap. Draper argues that indifference is not less probable in advance than theism, and it is easy to see why the claim is at least reasonable: indifference is a weaker claim than theism, and weaker claims are, other things equal, more probable.

What is being explained

The evidence to be compared is not suffering in general. It is a specific and richly detailed set of facts about how pain and pleasure are distributed among living things.

Draper divides the relevant observations into three groups, and the division is not decorative: each group is claimed to favour indifference for a different reason.

The first is pain and pleasure experienced by moral agents that we know to be biologically useful. This is the case that looks least troubling for theism, since a theist can say the sensation serves a purpose. But the purpose it serves is biological, not moral, and that is the difficulty. Indifference expects it, because a process selecting for reproductive success would produce exactly this. Theism has to say that a being concerned with moral goods produced a system whose signals are calibrated to fitness.

The second is pain and pleasure experienced by sentient beings that are not moral agents, and that we know to be biologically useful. Here the theistic responses that appeal to moral development and free choice have nothing to work with. An animal that is not a moral agent is not being built into anything by its suffering, is not exercising a will, and cannot consent. Indifference expects the same integration here as everywhere, since the biology is the same. Theism has to import a purpose that the creature is incapable of participating in.

The third is pain and pleasure that we do not know to be biologically useful. This is where the older evidential argument would say "and therefore some of it probably is not useful, and therefore is probably gratuitous". Draper does not need that step. He needs only that indifference makes such cases entirely unremarkable, since a process not aimed at anything will produce plenty of misfires, whereas theism has to treat each as a case where a reason exists but is hidden.

The argument

Set out as a likelihood comparison, the structure is compact.

The second premise is where the argument lives, and it is where almost all the subsequent literature is directed. Everything else is either machinery introduced in the intermediate tier or a claim about priors that both sides have found difficult to make progress on.

Notice what the conclusion does not say. It does not say theism is improbable. A Bayes factor tells you how far to move, not where you end up, and where you end up depends on the prior and on all the other evidence. Draper's own statement of the conclusion is conditional in exactly this way: the evidence counts against theism unless there are countervailing considerations, and identifying those is the theist's job rather than his.

Why biology is the pressure point

It is worth being clear about why the biological detail carries the weight, since the argument is sometimes misread as just another appeal to the amount of suffering.

Quantity is a weak basis for an argument of this kind. Theism can say that any finite quantity might be necessary for goods we cannot assess, and skeptical theism turns that into a general policy. Quantity gives the theist a place to stand.

Pattern is different. The claim is not that there is a lot of pain but that pain and pleasure are wired into organisms in a highly specific way, and that the specification tracks reproductive fitness with considerable precision. A creature's suffering is calibrated to the biological stakes of its situation, not to its moral situation, its innocence, or its capacity to learn anything. And we have an independent account of why the calibration would take that form, since natural selection produces traits by exactly this route, and the account was not constructed to embarrass theism.

That last point is what makes the argument an inference to the best explanation rather than a bare appeal to intuition. Indifference is not merely compatible with the pattern. Combined with what we know about how organisms develop, it predicts it.

The other half of the title

The argument is almost always discussed as an argument from suffering, and the title of the paper names pleasure as well. The omission matters, because the pleasure half is what shows the argument is about pattern rather than about quantity.

Consider what indifference expects of pleasure. It expects pleasure to be attached to whatever tended to promote reproductive success in ancestral conditions: food, sex, status, warmth, the absence of threat. It expects the attachment to be approximate rather than exact, since selection works on ancestral conditions rather than current ones, which is why substances and foods that were rare in those conditions are now capable of hijacking the system. And it expects no correlation at all between how pleasant a life is and how the person deserves to have it go, because desert is not a variable selection can see.

That is what we find, and it is not obviously what theism would have led anyone to expect. A being concerned with moral goods might have been expected to arrange some connection, however loose, between how a creature acts and how its life feels, or to make the deepest available pleasures the ones connected to what the theology says matters most. Instead the strongest and most reliable pleasures are the ones tied to biological function, and the connection to moral standing is absent.

The theist has answers here too, and the best of them is that a world in which virtue were reliably pleasant would corrupt virtue, since acting well for the reward is not acting well. That is a serious reply, and it is the shape of much of the soul-making tradition. What it does not address is why pleasure tracks fitness so precisely, as opposed to merely failing to track desert. Random distribution would also protect virtue from corruption, and random distribution is not what we observe.

Including pleasure also blocks a natural misreading. If the argument were only about pain, a theist could respond by pointing to the enormous quantity of good in the world and treating the two as items to be weighed. Draper's argument is not a weighing of totals. Adding more pleasure to the world does not weaken it, because the pleasure is distributed on the same principle as the pain, and it is the principle that is at issue.

What theism leads one to expect

The complement of that claim is a claim about theism, and it is the one theists have most often contested: that theism gives some reason to expect a different pattern.

The reason offered is that a being with the attributes of rung 1 has moral concerns, and that a world arranged by such a being should show some sign of them in this domain. One would expect pain to bear some relation to what a creature has done, or to what it might become, or at least to be distributed in a way that a moral purpose could be discerned in. Instead the relation it bears is to what preserves a gene line.

The theist's first response is to deny that theism gives any such expectation, on the grounds that we cannot say what a perfect being would do. That reply is available, but it is expensive, and rung 15 is about how expensive. A theism that generates no expectations about how the world will look cannot be confirmed by how the world looks either, and the arguments of rung 11 all depended on its generating some.

The second response is to grant the expectation and satisfy it: to argue that the biological arrangement does serve a moral purpose. A world of stable natural laws is a precondition of agency, and a world of stable laws governing organisms will produce a nervous system that responds to damage. This is a real answer and not a trivial one.

That exchange is unresolved, and both sides are aware that what is really being disputed is how much a hypothesis of theism's generality is entitled to leave undetermined.

Conjoining a theodicy, and what it costs

The most instructive theistic response is the one that shows most clearly why likelihood arguments are hard to defeat by adding stories.

Suppose the theist has a theodicy: a specific account on which the biological facts do serve some moral good. Conjoin it with theism. The conjunction makes the observations much more probable, since the theodicy was designed to. The likelihood gap narrows or closes.

The cost is on the other side of the ledger. The conjunction of theism with a specific theodicy is a more detailed hypothesis than theism alone, and a more detailed hypothesis has a lower prior probability, because it can fail in more ways: it is false if theism is false, and also if theism is true but that particular theodicy is not the reason. What is gained in likelihood is paid for in prior, and the two effects run in opposite directions on the same calculation.

This is why the response cannot be run indefinitely. Adding enough detail to make the observations expected produces a hypothesis specific enough that its prior collapses. The Bayes factor for the conjunction is only an improvement if the theodicy is independently plausible, which returns the debate to whether the theodicy is any good, which is where it always was.

The point generalises well beyond this argument, and it is the reason accommodation is cheap and prediction is expensive. Any hypothesis can be made to fit any data by conjunction; what cannot be manufactured is a hypothesis that fitted the data before the data were known.

Whether skeptical theism helps here

The dominant theistic response to evidential arguments over the last thirty years has been skeptical theism, and it is worth asking in advance whether it engages this one, since rung 15 examines its costs in detail.

Against Rowe's argument the strategy is well aimed, because Rowe's inference runs from our inability to identify a justifying good to the probable absence of one, and the skeptical theist attacks exactly that step. Draper never takes that step, so the strategy has to be redeployed rather than reused.

The redeployment is available. If our grasp of the space of possible goods is as poor as the skeptical theist says, then we are also in no position to judge that the observations are much less probable on theism than on indifference, since that judgement requires an estimate of what a perfect being would be likely to arrange. The second premise of the argument is a judgement of exactly that kind, and the skeptical thesis undercuts it.

What the redeployment gives up is any claim that the observations are what theism would expect. It converts theism into a hypothesis about which no expectations can be formed in this domain, and a hypothesis that makes no predictions cannot be confirmed by observations either. That is the general cost, and rung 15 traces how far it spreads.

What the argument does not establish

Three limits are worth stating plainly, and Draper states them himself.

The conclusion is evidential and defeasible. The argument establishes a direction and a rough magnitude of evidential force. It says nothing about where the balance settles, and a theist who thinks the arguments of the earlier rungs are strong can accept the whole argument and remain a theist on the total evidence.

The conclusion does not concern gratuitous evil. Nothing in the argument requires that any particular instance of suffering lacks a justification, so replies built around the impossibility of establishing gratuitousness do not engage it. This is the argument's main structural advantage, and it is why it survived the skeptical theist turn better than Rowe's version did.

And the rival is indifference, not naturalism. Establishing that indifference is favoured is not establishing that a naturalist account of everything is favoured, and the costs the previous rung listed on the naturalist side are untouched. Someone could accept this argument and hold a form of deism, or a theology on which the divine is not an agent making decisions about vertebrate nervous systems.

Draper's own later work has continued along these lines, comparing theism with successively refined rivals rather than seeking a knockdown result, and the reasonableness of that ambition is part of why the argument has held its place.

Next

The next rung reverses the direction of attack. Plantinga argues that naturalism, combined with the claim that our faculties arose by unguided evolution, gives us no reason to trust those faculties, and that a naturalist who sees this acquires a defeater for everything they believe, including naturalism. The argument uses the same materials as this one, evolution and the shaping of organisms by selection, and turns them against the framework that most often deploys them.