Ever since his 1989 article ‘Pain and Pleasure: An Evidential Problem for Theists’, Paul Draper has been making important contributions to the philosophy of religion. Prior to the publication of this book, however, it would have been difficult to construct a general picture of his overall defence of atheism, since his contributions are largely scattered throughout the literature in the form of isolated articles and book chapters. Thankfully, this is no longer the case. As his first book-length defence of atheism (decades in the making), Atheism and the Problem of Evil is a sweeping and punchy presentation of why he thinks God does not exist, all things considered.
The central aim of the first chapter is to characterise the main target of the book: ‘Core Theism’. Core Theism is the view that there exists an omnipotent, omniscient, and loving agent who created the cosmos for good reasons. Core Theism is distinct from Classical Theism since it does not build perfect rationality or moral perfection into the characterisation of God, but Draper thinks this is preferable and charitable because Classical Theism significantly lowers the initial probability of the theistic hypothesis.
The chapter opens with a discussion of the meaning of ‘atheism’, ‘theism’, and ‘God’. It then turns to what is the most interesting part of the chapter, a consideration of Perfect Being Theism and ontological arguments. Draper distinguishes between Hard and Soft Perfect Being Theism. The former says a being is perfect just in case it has every absolute perfection. The latter says a being is a perfect just in case it has every absolute perfection it is possible to have. Impossibility arguments attempt to show that a being cannot have all absolute perfections; and so that Hard Perfect Being Theism is false. The Soft Perfect Being Theist can avoid such arguments for, on their view, a perfect being need not have all absolute perfections, only as many as it is possible to have. There have been moves towards Soft Perfect Being Theism in recent years, but Draper offers some persuasive criticisms of this recent project. He also contends that Perfect Being Theists (of either variety) need an ontological argument to be justified in holding their view, but that these all fail. Indeed, he provocatively claims that a consensus regarding all modal ontological arguments – namely, that they all fail – may be soon within our reach.
The second chapter considers inconsistency arguments. These are arguments that attempt to show that theism is either (1) internally inconsistent (impossibility arguments) or (2) externally inconsistent (logical arguments). The former claim that either some divine attribute is logically incoherent or that two or more attributes are logically incompatible. The latter claims that some known (or probable) fact about the world entails atheism. He is pessimistic about inconsistency arguments for Core Atheism. Instead, he prefers evidential arguments. In this chapter, he discusses numerous impossibility arguments (e.g., the paradox of the stone), and three logical arguments (i.e., the foreknowledge problem, the hiddenness argument, and logical problems of evil).
Chapter 3 involves some important stage-setting. Draper frames his argument in terms of epistemic probabilities, understood as credences. The epistemic probability of a proposition p for someone in epistemic situation k is the credence he rationally ought to give to p in k. What is interesting here is Draper’s inclusion of epistemic situations. He will attempt to show that Core Theism is false relative to his epistemic or evidential situation (or at least one that substantially overlaps with it). It is not entirely clear what an epistemic situation is (I wish Draper had spent more time detailing this), but he suggests it can involve propositional as well as non-propositional elements. He seems to suggest what one’s evidence is (new evidence, forgotten evidence) and what one finds plausible (e.g., Hume does not find revealed religion plausible) is a part of one’s epistemic situation.
However, there are two problems facing any evidential argument for atheism: (1) the total evidence problem and (2) sceptical atheism. The requirement of total evidence stipulates that to assess the probability of a hypothesis, one must consider all the relevant evidence, not just a part of it. Draper claims the total evidence problem undermines all evidential arguments from evil for atheism he knows of (except Hume’s), for one can always perform a Moorean Shift against a key premise. In turn, ‘sceptical atheism’ is the flip side of ‘sceptical theism’. Where sceptical theism is the view that the probability of the existence of evil (or certain evils) on theism is inscrutable, sceptical atheism is the view that the probability of the existence of evil (or certain evils) on atheism is inscrutable. But this is precisely what Draper needs to run his argument. The reason it is inscrutable is atheism is simply the denial of theism, and this is compatible with a whole host of views. He draws some lessons from Hume to solve these problems.
For instance, in the Dialogues, Hume compares four hypotheses and concludes that the Indifference Hypothesis better accounts for the data of design and the distribution of good and evil in the world. Theism and the Indifference Hypothesis equally account for design, but the latter better ‘predicts’ the distribution of good and evil. This shows that theism is very probably false (not necessarily that the Indifference Hypothesis is probably true). This avoids the issue of sceptical atheism, because it is not to conclude atheism tout court, but a specific atheistic hypothesis – the Indifference Hypothesis. In turn, he avoids the problem of total evidence because his conclusions are relative to his epistemic situation, which is that revealed religion (miracle reports) are not a live option for him; only natural religion will do. Draper seeks to do something similar with respect to Core Atheism. He will find a hypothesis that entails Core Atheism and will argue that it is much more probable than Core Theism relative to his epistemic situation. This will show that Core Theism is very probably false (and so Core Atheism very probably true), without being a direct defence of atheism itself.
Chapter 4 considers the nature of intrinsic probability (the epistemic probability of a hypothesis in virtue of its intrinsic features alone). Draper grants these probabilities are relative to one’s epistemic situation (e.g., subjective rational intuitions), but nonetheless thinks there are substantial objective rational constraints (beyond mere consistency with the rules of mathematical probability). He thus defends objective Bayesianism.
These objective constraints are two (and only two): modesty and coherence. Modesty is a matter of a hypothesis’s degree of content (whether scope or specificity): how much content it contains that could be false. The more modest a hypothesis is (the less it asserts), the more it is likely to be true. Coherence is a matter of the degree to which the parts (its known logical implications) of a hypothesis fit together. This is a matter of how well the claims entailed by a hypothesis deductively or inductively support one another. Intrinsic inductive support, moreover, is cashed out in terms of uniformity. More uniform hypotheses (all apples taste sweet) have less ‘probabilistic tension’ than less uniform hypotheses (all Asian apples are sweet and all non-Asian apples are sour). He also attempts to show that simplicity, oftentimes put forward as a third candidate, can be dispensed with once we embrace modesty and coherence. His view assumes (as he admits) that basic inductive support relations can hold a priori between the parts of a hypothesis.
He also argues that Core Atheism has high intrinsic probability in this chapter. Source Physicalism and Source Idealism, he argues, are equally modest and coherent, and so have the same intrinsic probability. But Source Idealism is much more probable intrinsically than Core Theism (since the latter is a very specific version of Source Idealism). Thus, Source Physicalism is much more probable intrinsically than Core Theism. Since Source Physicalism entails Core Atheism, this implies that Core Atheism is much more probable than Core Theism. Moreover, since Core Atheism is the denial of Core Theism, it is not merely much more probable than Core Theism but can also be said to have high intrinsic probability. However, it is unclear why this chapter is necessary for his overall project in the book. The fact that Core Atheism has high intrinsic probability is not needed to run his two main arguments from evil, which come in the next two chapters.
The book contains two main arguments from evil for Core Atheism: a decisive evidence and a decisive priors argument. A decisive evidence argument begins with a Core Atheism-entailing hypothesis with roughly the same intrinsic probability as Core Theism, but that better predicts the evidence by a decisive ratio. This is akin to two runners starting a race at the same point, but such that the non-theistic alternative runs much faster. In turn, a decisive priors argument is one in which Core Theism and a Core Atheism-entailing hypothesis predict the evidence equally well, but such that the non-theistic alternative benefits from a much higher intrinsic probability. This is akin to the Core Atheism-entailing hypothesis getting a huge head start in the race.
The decisive evidence argument occurs in Chapter 5. The Core Atheism-entailing hypothesis that Draper considers as an alternative to Core Theism is Indifference Deism: there exists an omnipotent and omniscient agent who created the cosmos for good reasons and who is indifferent to the well-being of the sentient creatures in that cosmos. Draper claims Core Theism and Indifference Deism are tied, in terms of not only their intrinsic probabilities but also how well they predict the background evidence: namely, the sorts of intrinsically valuable things we observe in the world (order, consciousness, moral agency, beauty). However, Indifference Deism much better predicts QUALITY, or the quality of lives of human and conscious animals. This is what decisively breaks the tie.
How does it do that? Draper argues Indifference Deism better predicts the way pain and pleasure are distributed in the world. Since the deity on Indifference Deism is not concerned with the welfare of sentient creatures on earth, one would expect pain and pleasure to serve only biological roles. To this extent, they resemble many other parts of organisms in contributing to survival and reproduction. In contrast, Core Theists have much more reason to be surprised by the fact that pain plays a major motivational role in conscious beings on earth, for a loving God would want to produce more pleasure than pain in the world. He cites other benefits and harms besides pain and pleasure. The loving God of Core Theism would have a prima facie desire that everyone (not just the genetically lucky) experience goods like love, autonomy, and achievement and avoid evils like hatred, heteronomy, and the failure to achieve. This is to be expected much more on Indifference Deism than Core Theism.
Chapter 6 contains the decisive priors argument. As mentioned, this is one in which Core Theism and a Core Atheism-entailing hypothesis predict the evidence equally well, but such that the non-theistic alternative benefits from a much higher intrinsic probability. Of course, Draper had already argued that Core Atheism has greater intrinsic probability (Chapter 4), but he needs a distinct hypothesis in order to avoid the pitfall of sceptical atheism.
The chosen hypothesis is Anaxiarchism. This is the hypothesis that either (1) some entities are causally fundamental and aimed at neither goodness nor badness, or (2) no entities are causally fundamental but, for some time t, no entity prior to t was aimed at goodness or badness. It is the flip side of Axiarchism: goodness and badness are not doing any fundamental causal work. Since Core Theism is ‘biased’ towards good things, whereas Anaxiarchism is not, some of the data more strongly favours Core Theism over Anaxiarchism. To offset this, Draper notes that Anaxiarchism better accounts for QUALITY and VIRTUE, where the latter consists in what we know about the pattern of moral goodness and moral badness in human beings. Accordingly, matters come down to the intrinsic probabilities of Core Theism and Anaxiarchism. To argue that Anaxiarchism is many times more probable intrinsically than Core Theism, Draper claims that Anaxiarchism is at least as probable intrinsically as Axiarchism, and that Axiarchism is many times more probable intrinsically than Core Theism.
The remaining four chapters round out his defence. Chapter 7 turns to cosmological and design arguments, often thought to be the strongest arguments for Core Theism. He discusses and rejects the contingency argument, the nomological argument from regular laws of nature, the Kalam cosmological argument, the traditional fine-tuning argument, and the fine-tuning argument from discoverability.
Chapter 8 is an extended discussion of sceptical theism. Draper argues that, while effective against Rowe-style arguments from evil, it cannot be used to refute Humean arguments from evil, such as the one defended in the book. Chapter 9 discusses theory versioning. Some theists think that versioning theism can help solve the problem of evil, by either (1) Christianising Core Theism (or adding sectarian religious claims), (2) offering theodicies, (3) putting forward sceptical theses about the likelihoods involved, or (4) raising the probability of evil on theism by adding sceptical hypotheses to it. Draper uses a theorem of the probability calculus, the ‘weighted average rule’ (WAR), to see the impact of theory versioning on his problem of evil. As one might expect, he does not think any of these succeed.
Finally, Chapter 10 addresses two ‘gap problems’: namely, whether he can show that Core Atheism is true and not merely very probably true (he thinks he can), and whether his defence of Core Atheism can be extended to Global Atheism (he does not think it can, at present). This last chapter considers some interesting discussion of naturalised versions of theism.
All things considered, and compared to other recent book-length defences of atheism, Atheism and the Problem of Evil is a short book. Nonetheless, it is wide-ranging and packs a punch. It is also clearly written, well argued and original and advances the discussion in considerable ways. Philosophers of religion have no excuse but to rigorously engage with it.
