Hostname: page-component-74656f9884-2dmjx Total loading time: 0 Render date: 2026-11-07T09:50:19.249Z Has data issue: false hasContentIssue false
  • English
  • Français

Moral motivation and the evil-god challenge

Published online by Cambridge University Press:  03 February 2020

LUKE WILSON Open the ORCID record for LUKE WILSON [Opens in a new window]
LUKE WILSON*
Affiliation:
Department of Philosophy, Purdue University, West Lafayette IN, 47907, USA
*
e-mail: wilso761@purdue.edu
Rights & Permissions [Opens in a new window]

Abstract

The evil-god challenge holds that theism is highly symmetrical to the evil-god hypothesis and thus it is not more reasonable to accept one rather than the other. But, since it is not reasonable to accept the evil-god hypothesis, it is not reasonable to accept theism. This article will primarily focus on defending the challenge from two recent objections which hold that it follows from the nature of moral motivation that theism is intrinsically much more likely to be true than the evil-god hypothesis. However, I will also argue that there is a different intrinsic asymmetry between the hypotheses which favours theism.

Information

Type
Original Article
Information
Religious Studies , Volume 57 , Issue 4 , December 2021 , pp. 703 - 716
Check for updates
Copyright
Copyright © The Author(s), 2020. Published by Cambridge University Press

Introduction

Stephen Law's (Reference Stephen2010) ‘evil-god challenge’ presents an argument that belief in the existence of a good god is not more reasonable than belief in the existence of an evil god. My primary aim is to defend the challenge against one type of objection from the nature of moral motivation given by Forrest (Reference Forrest2012) and Weaver (Reference Weaver2015), but in the final section I suggest a potential asymmetry between the good god and evil god hypotheses which favours the good god hypothesis. The objection to the challenge from the nature of moral motivation holds that if moral judgements are intrinsically motivating, so that an agent's judgement that an action is right would, ceteris paribus, motivate the agent to perform that action, then an omniscient being would be more likely to be motivated by the good than by evil, since an omniscient being would always make accurate moral judgements. After briefly presenting the evil-god challenge in a Bayesian form, I discuss Forrest's objection to the challenge and show that it requires the rejection of the Humean theory of motivation. I think Humeanism is a plausible theory and it is far from uncontroversial to accept a picture of moral psychology which presupposes its falsehood. I distinguish two versions of Humeanism, a standard version and a permissive version, and argue that Humeans, particularly permissive Humeans, are able to accommodate the anti-Humean intuitions to which Forrest appeals. Next, I show how Weaver's argument for the impossibility of an evil god presupposes anti-Humeanism and consider two further responses the challenger may give to Weaver. While I think that the arguments of Forrest and Weaver should not persuade those who take Humeanism to be a live option, in the final section I suggest that there is an intrinsic asymmetry in the content of the motivations of the good god and the evil god which favours the good god hypothesis even if Humeanism is correct.

The evil-god challenge

According to Law's evil-god challenge, classical theism (hereafter ‘theism’), the hypothesis that there is an omnipotent, omniscient, and perfectly good being, is highly symmetrical to evilism, the hypothesis that there is an omnipotent, omniscient, and perfectly evil being.Footnote 1 Placing the challenge in a Bayesian framework, we can distinguish the intrinsic symmetry and the evidential symmetry of the two hypotheses. Theism and evilism postulate the existence of a supernatural, personal being which is omnipotent and omniscient, and differ only in the moral character ascribed to that being. But, the properties of being perfectly good and being perfectly evil seem equally simple (compared to, say, having a mixed moral character). If this is correct, then theism and evilism are intrinsically symmetrical, and so we should hold them to be equally intrinsically probable.Footnote 2

Of course, even if theism and evilism are equally intrinsically probable, they seem to make very different predictions about the kind of world we would observe if one rather than the other were true and so they appear to be evidentially asymmetrical. However, according to the challenger, the evidence of natural theology taken to support theism may be parodied to equal effect in support of evilism.Footnote 3 For instance, teleological arguments hold that the existence of conscious embodied beings is much more likely on theism than on competitors such as naturalism. But, the existence of conscious embodied beings does not seem more likely on theism than on evilism. While a perfectly good god may want to create conscious embodied beings so that it can benefit or have positive relationships with them, a perfectly evil being may want to create conscious embodied beings so that it can harm or have negative relationships with them. So, challengers may argue that fine-tuning or apparent design favour theism and evilism equally well. Furthermore, theism and evilism face symmetrical evidential problems – theism faces the problem of evil, while evilism faces the problem of good. The challenger argues that theistic responses to the problem of evil can be parodied to provide equally plausible evilist responses to the problem of good.Footnote 4

So, according to the challenger, theism and evilism are equally intrinsically probable and explain the data equally (or nearly equally) well, and so it is no more rational to accept one than the other. But, the argument goes, since it is not rational to accept evilism, it is also not rational to accept theism.Footnote 5 As a challenge to the reasonableness of theistic belief, the argument is strengthened by the fact that while theism and evilism are mutually exclusive, they are not exhaustive; other versions of supernaturalism (such as morally mixed or indifferent gods, polytheism, and non-theistic versions of supernaturalism) as well as naturalism should take up some significant portion of one's epistemic probability space.Footnote 6

The evil-god challenge has recently received a number of objections and defences.Footnote 7 In the next few sections I focus on those who deny the intrinsic symmetry of theism and evilism by arguing that it is impossible (Weaver (Reference Weaver2015) ) or, compared to theism, highly improbable (Forrest (Reference Forrest2012) ) that an omnipotent and omniscient being would be perfectly evil. I argue that their claims about the nature of divine motivation require controversial metaethical assumptions which are not adequately supported; namely, that they require the falsehood of the Humean theory of motivation in both its standard form and a more permissive form. At this preliminary stage we may state Humeanism roughly as the view that desires are necessary for any agent to be motivated, and these desires must be paired with appropriately related means–end beliefs. So, beliefs, including moral beliefs, do not motivate on their own. I begin by discussing Forrest's argument that theism is much more intrinsically probable than evilism. I present his theory of moral motivation (axiarchism), show that axiarchism requires rejecting Humeanism, and describe his argument that axiarchism is supported by the phenomenology of resisting temptation.

Forrest's response to the evil-god challenge

Peter Forrest (Reference Forrest2012) argues that the most intrinsically probable, because the most simple, account of God's fundamental nature is that God does not have a moral character, by which he means that God does not possess virtues or desires to do good (nor does God have vices or desires to do evil).Footnote 8 However, it follows from axiarchism – the combination of objectivism and motivation internalism – that a god without a fundamental moral character will nevertheless be motivated to act in a morally good way.Footnote 9 This is because, if axiarchism is true, knowledge of the good generates morally good motivation in what would be, were axiarchism false, a morally indifferent being. Objectivism holds that ‘comparative evaluation and hence the judgement of which act has the best consequences is objective in the sense of being true or false independently of any mind, even God's. (Presumably it is therefore non-contingent.)’ (ibid., 40–41). According to motivation internalism, ‘that the consequences of act X appear to be better than the consequences of any other act being considered directly motivates the agent to perform X unless the agent suffers from akrasia. (The motivation would be indirect if it was due to a desire to do whatever appears to be the best.)’ (ibid., 41).Footnote 10 Add to this that God does not suffer from akrasia (since, presumably, akrasia can only affect beings with desires opposed to what they judge to be best) and it follows that, since what is best will always appear so to an omniscient being, an omnipotent and omniscient being without moral character will always be motivated to do what is best.Footnote 11 Thus all of the intrinsic probability space taken up by the god with no character if axiarchism were false is taken up by theism if axiarchism is true. So, if we have good reason to accept axiarchism, then we have good reason to think theism is more intrinsically probable than evilism.

A reason one might be sceptical of axiarchism is that there seems to be a tension between objectivism and motivation internalism; they appear to be incompatible with the Humean theory of motivation.Footnote 12 If moral judgements are just beliefs and moral judgements are sufficient for motivation, then some beliefs are sufficient for motivation without requiring the presence of an appropriately related desire. But, Humeanism denies that beliefs alone can be sufficient for motivation. Forrest argues that the phenomenon of overcoming temptation is strong evidence for axiarchism: ‘We experience various desires but decide to do what we believe is for the best. A purely naturalistic account would consider this merely a case of competing desires, in which either the strongest wins or the result is random. But this is not how we experience temptation’ (ibid., 41). (Although he states this as an argument against naturalism, it is more directly an argument against Humeanism, which is what he needs to defend axiarchism.)Footnote 13 We sometimes feel that what we take to be a moral requirement pulls against our desires, and yet we can choose to do what we think is morally right. If Humeanism were true, then these cases of overcoming temptation would have the phenomenology of a competition between two desires: the tempting desire and the desire to do what is right. But, since this is not the case, Humeanism is false.Footnote 14 In the next section I present two versions of Humeanism and describe how these views can account for our experiences of resisting temptation.

Humeanism about motivation

We can distinguish two versions of Humeanism, Standard Humeanism (SH) and Permissive Humeanism (PH), through two claims:

  • Standard Humean claim: Non-derivative desires are necessary for motivation and must be paired with appropriate means–end beliefs.Footnote 15

  • Permissive Humean claim: Pure beliefs are never sufficient for motivation.Footnote 16

I will label ‘impure’ those beliefs which one cannot have unless one has some appropriately related, independently intelligible affective state, where the affective state is explanatorily prior to the belief. All other beliefs are pure beliefs. So, permissive Humeans may allow that some beliefs, impure beliefs, may directly motivate but pure beliefs cannot directly motivate. Anti-Humeans would deny that pure beliefs cannot motivate. A derivative desire is a desire that is either explained by or brought about by a pure belief or ascribed to an agent just in virtue of their acting or being motivated. SH requires motivating desires to be non-derivative because some anti-Humeans take having a desire to be necessary for motivation but accept reductive accounts of desire (see below) or hold that desires can be caused by pure beliefs. (Similarly, PH requires affective states that are non-derivative.) Arguments for anti-Humeanism must be strong enough to reject PH as well as SH, yet SH is the typical target of anti-Humean arguments.Footnote 17 PH allows for two options for which SH does not: (1) non-desire affective states are intrinsically motivating and (2) some impure beliefs may be directly motivating.

Motivation by non-desire affective states

We can distinguish between reductive and non-reductive accounts of desire, where reductive accounts would include those anti-Humean views which ascribe a desire to Φ to an agent just in virtue of the fact that the agent Φs or is motivated to Φ (see Nagel (Reference Nagel1970) and Schueler (Reference Schueler1995), and discussion in Dancy (Reference Dancy1993), ch. 1 and Arruda (Reference Arruda2017) ) and those accounts which treat desires as beliefs of a certain type (see Oddie (Reference Oddie2005) and (Reference Oddie, Lauria and Deonna2017), Tenenbaum (Reference Tenenbaum2007), and Swartzer (Reference Swartzer2013) ).

Among non-reductive accounts, it is important to distinguish two views on the scope of desire. First, we might locate desire within a traditional belief/desire psychology, where belief and desire are the only categories relevant to motivation. There would be no room on this view for non-desire affective states to play a role in motivation, although proponents of SH may hold that certain non-desire mental states, such as emotions, may have motivational force in virtue of having desires as components (see Sinhababu (Reference Sinhababu2017), 63–66). Second, we may accept a narrower view of desire as one among many types of affective state. This is more in line with a phenomenological approach to desire, where we might treat such attitudes as caring or loving as distinct from desires.Footnote 18 On this view, care and love may entail certain desires (if you love someone, you desire their well-being) but are not reducible to or constituted by desires. These attitudes have a phenomenology different from those desires which are typically responsible for temptation – immediately satisfiable appetitive desires. An argument that there are cases of motivation in which (non-derivative) desires are not necessary, or in which we act contrary to our desires, would contradict only SH on this account of desire. On PH, it may be that in overcoming temptation we are not motivated by moral pure beliefs, but rather by non-desire affective states with which tempting desires may compete.

Two claims are central to the anti-Humean reconstruction of Forrest's argument: (1) resisting temptation need not involve two desires competing with one another and (2) if Humeanism were true, then resisting temptation must involve two desires competing with one another. However, the intuitive plausibility of (1) relies on accepting a narrow account of desire, where appetitive desires are the paradigm, and (2) is true only if Humeans must accept SH and the broader account of desire, according to which only belief and desire are relevant to motivation.Footnote 19 Standard Humeans may follow Sinhababu by accepting (2) and denying (1), but defuse intuitions against (1) by providing a richer account of desire on which there are different types of desires which have different phenomenological feelings.Footnote 20 Meanwhile, permissive Humeans may accept (1) but reject (2), holding that desires may not be responsible for motivation when we resist temptation but it does not follow that pure beliefs, rather than non-desire affective states, are responsible.

Motivation by impure beliefs

In addition to allowing that non-desire affective states may motivate, permissive Humeans may grant that moral beliefs directly motivate, as long as they are not pure beliefs. Consider the view that having genuine moral knowledge requires seeing the world with a sensitivity that is only possessed by someone with a virtuous character.Footnote 21 If possessing the proper sensitivity requires having certain non-derivative desires or affective states, then such beliefs would count as impure. Little, for instance, argues that ‘possession of various emotions and desires . . . is not just immensely useful to seeing the moral landscape, it is a necessary condition of doing so’ (Little (Reference Little1995), 118). Similarly, Scrutton holds that one may ‘assent intellectually’ to a proposition or believe that proposition in a way that is ‘affectively toned or emotionally enriched’ so that ‘there is a certain way of knowing or perceiving or believing that can be experienced through affective experience – and that cannot be experienced in any way except through affective experience’ (Scrutton (Reference Scrutton2011), 69).Footnote 22 This view differs from the standard Humean view that only means–end beliefs factor in motivation by pairing with desires for the ends. On PH, an explanation of resisting temptation need not appeal to two competing desires – a tempting desire competing with a desire to do what is morally right. Rather, there need only be a tempting desire competing with a moral belief, even though the moral belief gets its motivational force from some prior affective state. Applying this account to God, it follows that ‘there is a certain type of knowledge God can only have if he experiences the world emotionally’ (Scrutton (Reference Scrutton2011), 73). An individual's being omniscient would entail its having certain affective states and these affective states would be explanatorily prior to its moral judgements.Footnote 23

Interestingly, this view is compatible with axiarchism and so may be able to justify a reply to the evil-god challenge similar to Forrest's since it holds that no being could be omniscient without having certain prior virtuous affective states, and thus no being could be both omniscient and perfectly evil. The challenger might give a parody response that there are certain facts that cannot be known unless one has certain prior vicious affective states. This move would imply that no being can be omniscient without having conflicting affective states, but since to be a virtuous person is not to be a vicious person, it would follow that omniscience, understood as knowing every true proposition or as knowing every knowable proposition, cannot be exemplified. (This result could be avoided through a restricted account of omniscience – as knowing everything that a being with a given character could know – but this would have the unhappy consequence that two omniscient beings could have different beliefs and would not threaten the symmetry of theism and evilism.) Alternatively, the challenger may interpret this version of PH as requiring that one must have some prior affective states or other in order to see the moral facts but that these may be either virtuous or vicious.Footnote 24 This interpretation would re-establish the symmetry of theism and evilism. I think there is some force to these potential responses from the challenger, but even if they fail, accepting axiarchism and PH would lead to problems elsewhere in Forrest's defence of theism.Footnote 25 A view on which an omniscient being must have certain prior affective states to be omniscient is incompatible with Forrest's hypothesis that an omniscient being could have no moral character (apart from that derived from its knowledge of the good). Also, such a view would cost theism in simplicity, since theism would have to posit affective states as part of God's fundamental nature, rather than deriving them from more fundamental attributes.Footnote 26

Weaver's response to the evil-god challenge

In this section I discuss Christopher Gregory Weaver's (Reference Weaver2015) argument that evilism is impossible. I argue that, like Forrest's argument, Weaver's requires rejecting even a permissive version of Humeanism and suggest two further responses that a challenger might give to Weaver.

Weaver argues that the impossibility of evilism follows from a Kantian metaethical picture, which includes the conjunction of moral rationalism and what Weaver calls modest (reasons) internalism, joined with essentialism about the moral character of the good god and the evil god. Moral rationalism holds that ‘duties and/or moral obligations either strictly imply practical reasons for action, or are identical to such reasons’ (ibid., 10). Weaver's modest (reasons) internalism holds that ‘for any moral agent a, necessarily, if a has a good reason to Φ in circumstance C*, then possibly a will desire to Φ in C*’ (ibid., 12).

Consider the amoralist objection to motivation internalism which claims that it is possible for there to be an amoralist, someone who judges that they ought to Φ without having any motivation to Φ, and that this is incompatible with motivation internalism. Weaver thinks that, given moral rationalism, this objection is not plausible as an objection to his modest version of internalism.

In order for the amoralist objection to be applicable with respect to [modest (reasons) internalism], the amoralist would need to be essentially such that he or she is amoralist. I think that such an essentialist thesis is patently false, for the following claim is meek and mild: (4) For any amoralist x, possibly (x has a good reason to Φ in circumstances C*, and x is thereby motivated to Φ in C*). (ibid.)

So, there cannot be a perfectly evil being because, by essentialism, such a being would have to be essentially evil, i.e. completely evil at every world at which it exists, and yet, for any moral agent, there is some possible world in which it is motivated to do what it morally ought to do.

Weaver's argument requires moral rationalism, modest (reasons) internalism, and (4), but it also requires reading these theses in an anti-Humean way. One reading of (4) consistent with moral rationalism and modest (reasons) internalism is that (4) is true simply because for any being which does not desire to act morally, there is some (perhaps distant) possible world at which it has a different set of non-derivative intrinsic desires, including one to do what it ought. But it is equally plausible that a being which desires to do what it ought has, at some (perhaps distant) possible world, a set of non-derivative intrinsic desires which does not include that desire. So, this reading of (4) would disprove both the existence of an essentially evil being and the existence of an essentially good being. Of course, this interpretation would make reasons internalism almost trivially true and of little philosophical significance. Reasons internalists intend the more controversial claim that to have a reason is to be able to be motivated from one's actual beliefs, desires, affective states, and so on (i.e. one's mental set).Footnote 27 So, Weaver's argument requires reading (4) as claiming that an amoralist at the actual world is motivated in a different possible world not by different non-derivative intrinsic desires but by moral beliefs alone – i.e., that there is a possible world at which the intrinsic motivational force of the moral beliefs the amoralist has at the actual world is not blocked by some competing desire.Footnote 28 So, Weaver's argument must be understood in an anti-Humean way, and this is a contentious result which reasons internalists and moral rationalists need not accept. Thus, adding (4) to moral rationalism and modest (reasons) internalism gives a combination that is neither meek nor mild. Nevertheless, Weaver's demonstration of the impossibility of evilism on his Kantian, anti-Humean metaethical assumptions is important for showing how the plausibility of the evil-god challenge depends on the answers to significant metaethical debates about moral reasons and motivation.

However, the challenger may go further than pointing out the incompatibility of Weaver's argument with Humeanism and give more direct responses to his argument. I will briefly sketch two such replies. First, I am sympathetic to Collins's (Reference Collins2019) rejection of essentialism on the grounds that perfect goodness (or evilness) requires morally significant freedom, and this is inconsistent with essentialism.Footnote 29 If we think that being perfectly good/evil at a world is a matter of a being's having the right sort of fundamental motivations at that world, rather than having the right motivations essentially, then we can give an account that allows for significant freedom while avoiding the influential worries of Findlay (Reference Findlay1948) and Plantinga (Reference Plantinga1974) that a good god cannot just happen to be good.Footnote 30 Because an agent's fundamental motivations are modally stable even if the agent has different fundamental motivations at distant possible worlds, a perfectly good (or evil) being would not just happen to be good (or evil) despite not being good (or evil) essentially.

Second, a challenger might question the applicability of Weaver's metaethical assumptions to God. Moral rationalism entails that one who is not motivated to do what they morally ought is practically irrational. But, Weaver claims, an omniscient being cannot be practically irrational (see Weaver (Reference Weaver2015), 14–15, 22).Footnote 31 Evilism pictures an omnipotent, omniscient being which has evil desires and is aware of, but is not motivated by, its moral duties and acts solely to fulfil its malevolent desires. The challenger may accept that it is irrational for beings like humans to fail to be motivated by their moral knowledge yet hold that it is less clear that this is true of an evil god. Humans are highly dependent social beings whose flourishing is largely found in our relationships with one another and the health of those relationships requires us to act morally. But an evil god's flourishing might be very different from human flourishing and different standards of practical rationality might apply to a being with omnipotence, omniscience, aseity, and so on.Footnote 32

A potential asymmetry

In this final section I suggest a theism-favouring intrinsic asymmetry regarding the contents of the desires of the good god and the evil god. One puzzle in formulating the evil-god challenge is how to think about the evil god's relation to itself. There are a number of ways we might spell out the content of the evil god's desire, but it would seem most plausibly to be (a) a desire to do as much evil as possible or (b) a desire to bring about as much pain and suffering as possible.Footnote 33 If the content of a perfectly evil being's desire is captured by (a) or (b), then it must desire to self-harm as well as to harm its creatures.Footnote 34 However, the evil god's desire to harm itself must always be partially frustrated regardless of how it acts. In harming its creatures the evil god would be satisfying one desire but thereby frustrating its desire to self-harm, whereas in benefiting its creatures it would fail to satisfy its desire to harm them but would thereby satisfy its desire to harm itself. But this seems like a stranger picture of motivation than that on theism, particularly if we think that desire satisfaction is partly constitutive of one's flourishing or that desire satisfaction entails experiences of pleasure. A set of desires that includes a second-order instrumental desire that one's desires be frustrated is a less coherent set than one which does not include such a desire. Theism involves no such internal tension. The good god may desire the good of both itself and its creatures without conflict.

On this reading of the evil god's desires, while this set of conflicting desires makes evilism less intrinsically probable than theism, it may benefit the challenger by making it easier to respond to the problem of good than to the problem of evil. Theodicists may struggle to find some goods which justify particular evils suffered by creatures, but evilodicists will always be able to appeal to the evil suffered by the evil god which could not occur without its creatures enjoying some particular goods. However, this reading of the evil god's desires also makes evilism less explanatorily powerful than theism. The good god's unconflicted desire for the good makes a good creation probable on theism, but evilism does not straightforwardly make an evil creation probable due to the evil god's conflicted desires.

If we take the content of the evil god's desire not to be to bring about as much evil, or pain and suffering, as possible, but rather to bring about as much evil, or pain and suffering, as possible for its creatures, then the challenger can avoid the puzzle of the evil god's relation to itself. But this qualification would lead to an intrinsic asymmetry between evilism and theism by ascribing to the evil god a set of desires which is less simple and less coherent than the set of desires theism ascribes to the good god. The evil god's desires would not fit together as well as the good god's desires since the evil god must take one attitude towards itself and the opposite attitude towards its creatures, while the good god takes the same attitude towards both itself and its creatures.

Conclusion

The evil-god challenge presents an intriguing problem for theists and is particularly forceful in a Bayesian form. I have sought to defend the challenge against arguments that we can conclude from the nature of moral motivation that theism is significantly more intrinsically probable than evilism. These arguments presuppose the falsehood of a permissive (and, I think, highly plausible) version of Humeanism which entails that the motivation of either a good god or an evil god must originate from affective states which are not derived from its pure beliefs. I do not claim that Humeanism is correct, but rather that arguments that the nature of moral motivation favours theism over evilism must take seriously this Humean challenge. However, I have also argued that there appears to be a theism-favouring asymmetry in the way we understand the content of the desires of the good and evil gods. Nevertheless, if my Humean argument is correct, then plausibly both theism and evilism will take up a smaller portion of the overall probability space than if some form of axiarchism or antaxiarchism were true, since it opens up probability space for morally indifferent gods and gods with mixed moral character. So, the evil-god challenge deserves more attention for the role it can play in cumulative case arguments in natural theology.Footnote 35

Footnotes

1. I follow Weaver (Reference Weaver2015) in using the term ‘evilism’. See Lancaster-Thomas (Reference Lancaster-Thomas2018a) and (Reference Lancaster-Thomas2018b) for a good overview of the challenge.

2. For an in-depth discussion of intrinsic probability, see Draper (Reference Draper, Bergmann and Brower2016) and Swinburne (Reference Swinburne1997) and (Reference Swinburne2004).

3. By ‘natural theology’ I mean attempts to give reasons for or against belief in the existence of God from public evidence alone, rather than from purported revelation. (Prominent examples include cosmological arguments, teleological arguments, and the argument from evil.)

4. See New (Reference New1993), Murphree (Reference Murphree1997), Morriston (Reference Morriston2004), Law (Reference Stephen2010), Lancaster-Thomas (Reference Lancaster-Thomas2018a), and Collins (Reference Collins2019) for parody arguments.

5. See Lancaster-Thomas (Reference Lancaster-Thomas2018a, 2) for different ways of stating the challenge.

6. Stating the argument in Bayesian terms is, I think, the best way to motivate the view that theism and evilism cannot both be reasonable to accept since, being mutually exclusive, neither can have a probability above .5, and, being non-exhaustive, each plausibly has a probability well below .5. Also, on the Bayesian version, the challenge need not conclude that theism is unreasonable, but may instead make the more cautious claim that it provides one part of a cumulative case against theism. (Cf. Draper's (Reference Draper, Meister and Moser2017) argument from the intrinsic symmetry, and thus equal intrinsic probability, of source physicalism and source idealism to the low prior probability of theism.)

7. Recent objections have come from Forrest (Reference Forrest2012), Weaver (Reference Weaver2015), Scrutton (Reference Scrutton2016), Hendricks (Reference Hendricks2018), and Miller (Reference Calumforthcoming) with defences by Lancaster-Thomas (Reference Lancaster-Thomas2018b) and (Reference Lancaster-Thomasforthcoming), and Collins (Reference Collins2019).

8. Forrest (Reference Forrest2012), 38–40. More precisely, Forrest states that, given that there is a god, the intrinsic probability of a god with a malevolent character is at most .25, a god with a benevolent character is at most .25, and a god without moral character is at least .5 (ibid., 39). (This is on the simplifying assumption that gods with other fundamental motivations (such as those he mentions on p. 37) have been ruled out, leaving only the benevolent, malevolent, and no character options.) It is worth noting that by ‘god’ Forrest means ‘a first cause who is an all-powerful and all-knowing agent’ (36). This definition is neutral between theism and evilism (and gods with mixed moral character).

9. Forrest's use of ‘axiarchism’ is non-standard. Compare with the descriptions of axiarchism in Leslie (Reference Leslie1979) and Forrest (Reference Forrest2007).

10. Forrest states internalism in terms of appearances rather than judgements. It is not clear to me whether this is merely meant to emphasize that our moral judgements may be mistaken and yet motivating, or whether he accepts an internalism focused on appearances of value rather than judgements about value (see Oddie (Reference Oddie2005) ). I don't think the distinction between these versions of internalism will make much difference to my argument.

11. Forrest holds that God is literally neither good nor bad but is good in an analogical sense. I don't think this has much impact on the argument and I have stated the argument as holding simply that axiarchism plus God's omniscience (and lack of akrasia) are sufficient for God's literally being perfectly good. Forrest's argument is very similar to Swinburne's argument that omniscience and perfect freedom entail perfect goodness, where a being is perfectly free just in case there are no external forces, including non-rational desires, affecting it (Swinburne (Reference Swinburne2004), 94). I take it that God's lack of akrasia and lack of character traits at the fundamental level give the same result for Forrest as God's perfect freedom does for Swinburne.

12. See Smith (Reference Smith1994) and Sinhababu (Reference Sinhababu2017, chs 1 & 11) for clear statements of the tension between these views.

13. Forrest suggests that the intuition that objectivism and motivation internalism are in tension is due to the presumption of naturalism but, since naturalism is incompatible with evilism and theism, assuming it to reject axiarchism is not an option for the challenger (Forrest (Reference Forrest2012), 41–42). However, Humeans need not be naturalists, although these views fit together well. Naturalists may hold that some beliefs may be sufficient for motivation and Humeans may deny that all facts are reducible to scientific facts. So, Forrest's argument for axiarchism requires the rejection of Humeanism, not just the rejection of naturalism.

14. Forrest's argument may support the weaker claim that axiarchism is more plausible than antaxiarchism (the view that the bad is intrinsically motivating), given that one is true, but this does little to help the theist if we do not also have good reasons for rejecting Humeanism, since axiarchism and antaxiarchism assume anti-Humeanism. Nevertheless, the mere fact that axiarchism, unlike antaxiarchism, is a live option in moral psychology is a minor asymmetry supporting theism over evilism.

15. See e.g. Smith (Reference Smith1994) and Sinhababu (Reference Sinhababu2017). An appropriate means–end belief may be a belief about how to raise the probability that the desired state will obtain. (See Sinhababu (Reference Sinhababu2017), 2–3.) Note that Sinhababu does not take his account to give metaphysically necessary conditions for motivation (ibid., ch. 1).

16. This version is weaker than Sinhababu's (Reference Sinhababu2017) and Smith's (Reference Smith1994) in that it allows for motivation by non-desire affective states and impure beliefs, but it is stronger than Smith's in that it denies that pure beliefs alone can produce desires. (See Sinhababu's criticism of Smith's version of Humeanism (Sinhababu (Reference Sinhababu2017), 9–12).) However, Sinhababu's account of desire may incorporate much of what I say about motivation by non-desire affective states (see ibid., ch. 2).

17. A clear example is Swartzer's claim that Humeans hold that a person's beliefs influence her actions ‘only by laying out the paths by which her desires can be satisfied’ (Swartzer (Reference Swartzer2013), 975–976). Similarly, Döring characterizes Humeanism as holding that ‘all motivation is ultimately due to the desires an agent actually has,’ but defends a theory of motivation by emotions which amounts to a version of PH (Döring (Reference Döring2003), 220).

18. Cf. Radcliffe's interpretation of Hume: ‘Hume's view is that certain passions or affections, which may or may not be desires, are the necessary conative states [to motivate action]. Among passions he sees as motives are desire, aversion, grief, joy, hope, fear, despair, and security’ (Radcliffe (Reference Radcliffe2006), 355) See also Döring (Reference Döring2003). One motivation for accepting this account might be that such attitudes are directed towards objects whereas desires are directed toward states of affairs.

19. Forrest appeals to Campbell (Reference Campbell1938/1967), whose argument that we resist temptation by resisting our desires seems to accept a narrow account where the strength of a desire is a matter of how strongly it is felt. In a situation in which A is the end believed to be morally right and B is the end towards which one is tempted, he says: ‘the urge of our desiring nature toward the right end, A, is felt to be relatively weak. We are sure that if our desiring nature is permitted to issue directly in action, it is end B that we shall choose’ (ibid., 41). However, he also appears to assume that desires are the only type of affective state which can move us, so any motivation that does not come from desire must come from our moral beliefs: ‘nothing can be an object of possible choice which is not suggested by either the agent's desires or his moral ideals’ (ibid., 46). (His argument is no challenge to the Humean unless ‘moral ideals’ are pure beliefs, thus entailing traditional belief/desire psychology.)

20. These may include aversions, calm passions (Sinahababu (Reference Sinhababu2017), 36–7, 59, 200), and desires which we fully expect ourselves to act on (ibid., 152–156).

21. See McDowell (Reference McDowell1978) and (Reference McDowell1979), Little (Reference Little1995) and (Reference Little1997), Döring (Reference Döring2003), and Clarke (Reference Clarke and Snow2018). Radcliffe (Reference Radcliffe2006) and Tresan (Reference Tresan2006) argue in different ways for the similar view that a belief might count as a moral belief only if it is accompanied by or caused by an independently intelligible motivation or affective state. Cf. Kriegel (Reference Kriegel2012).

22. Emphasis in the original. Cf. Gibbard's (Reference Gibbard1990) account of norm-acceptance on which one may know that a certain action violates or conforms to a norm, but this would not move the person unless they accept the normative system in question.

23. See Scrutton (Reference Scrutton2013), 870–871. The argument that omniscience requires certain emotions is also made by Sarot (Reference Sarot1992) and Farmer (Reference Farmer2010).

24. Perhaps vicious people are better able than morally indifferent people to see certain moral facts due to their malevolent desires. (Think of Iago, for instance.) However, if moral vice is treated as a feature preventing one's acquisition of moral knowledge, analogous to intellectual vice, then this move is not open to the challenger.

25. Forrest (Reference Forrest2012, 36) qualifies the sense in which his response to the challenge is a defence of theism, and how it may not be useful to theists who reject merely analogical predication of goodness to God or who reject consequentialism.

26. Swinburne's (Reference Swinburne2004) argument would face analogous problems. On Humeanism generally, a being without any character would be simpler than one with a character, but such a being would not be motivated to act at all. One final comment on Forrest's argument: given the affect-infused nature of human moral education, it is not surprising, from a Humean perspective, that we do in fact come to care about acting in accordance with what we judge to be best, where the origin of this concern is affective. But, as Forrest (Reference Forrest2012, 39, 41) points out, we must be cautious in extrapolating from claims about human motivation to claims about motivation in general or about divine motivation. So, while there may be a tight connection between moral judgement and motivation for humans, it is dangerous to move from this to an analogous claim that this connection holds for all agents (cf. Collins (Reference Collins2019), 20).

27. Reasons internalists may follow two paths diverging from Weaver's Kantianism. The first holds that one may have objective moral obligations but that these need not generate reasons, thus rejecting moral rationalism. The other, which accepts moral rationalism, holds a type of moral subjectivism where one's moral reasons are dependent on one's affective states.

28. Gaining new desires by reasoning from the desires one actually has (perhaps on the basis of new non-normative information) is not an option for the good god or the evil god since each is omniscient and practically rational (in at least an instrumental sense).

29. See Collins (Reference Collins2019, 105–106) and also Reichenbach (Reference Reichenbach1980), Morriston (Reference Morriston1985), Howsepian (Reference Howsepian1991), Rowe (Reference Rowe2004), and Draper (Reference Draper, Buchak, Zimmerman and Swenson2019).

30. If we take all of a being B's affective states, desires, pro-attitudes, virtues, vices, and so on to constitute B's motivational set at W, B's fundamental motivations at W will be those members of the set which are not explained or grounded by any others. (I intend this in a way that is neutral between Humeanism and anti-Humeanism.) We can then understand a perfectly good being at W as one which has the best possible fundamental motivational set at W. It, perhaps, has a de dicto desire to do what is best, cares about the well-being of its creatures, and so on. (If one's motivations involve explanatory circles, then there may be degrees of fundamentality, or we may instead think of fundamental motivation in terms of originative priority. If B has some degree of concern for the good and this explains her developing a certain virtue and having this virtue in turn bolsters her concern for the good, then the original concern for the good would be a more fundamental motivation for B than her having the virtue.)

31. Some Humeans will reject moral rationalism, holding instead that rationality is a matter of satisfying one's intrinsic desires (see Schroeder (Reference Schroeder2007) and Sinhababu (Reference Sinhababu2017) ) while permissive Humeans may accept moral rationalism, but hold that practical rationality requires having certain non-derivative affective states, and the same consequences as described in the previous section would follow from this version of PH.

32. Cf. Murphy (Reference Murphy2017) who argues that, unlike humans, God is not bound by norms of ‘familiar welfare-oriented moral goodness’. However, he also thinks that a form of moral rationalism applies to God (ibid., 25–26).

33. On ways for formulating the content of the evil god's intention see Morriston (Reference Morriston2004), Forrest (Reference Forrest2012), Weaver (Reference Weaver2015), 4–7, and Collins (Reference Collins2019), 91. See also Miller (Reference Calumforthcoming, 9–10) for a discussion of theism-favouring asymmetries in different ways of formulating evilism. Thanks to an anonymous reviewer for making me aware of Miller's paper.

34. This assumes that, on (a), one can have moral duties towards oneself and, on (b), that the evil god would not be impassible.

35. Thanks to Paul Draper, Mike Jacovides, and two anonymous reviewers for helpful comments on previous drafts of this article.

References

Arruda, C. T. (2017) ‘What kind of theory is the Humean theory of motivation?’, Ratio, 30, 322–342.CrossRefGoogle Scholar
Campbell, C. A. (1938/1967) In Defence of Free Will: With other Philosophical Essays (Abingdon: Routledge).Google Scholar
Clarke, Bridget (2018) ‘Virtue as a sensitivity’, in Snow, N. (ed.) The Oxford Handbook of Virtue (Oxford: Oxford University Press), 35–56.Google Scholar
Collins, J. M. (2019) ‘The evil-god challenge: extended and defended’, Religious Studies, 55, 85–109.CrossRefGoogle Scholar
Dancy, Jonathan (1993) Moral Reasons (Oxford: Blackwell).Google Scholar
Döring, Sabine (2003) ‘Explaining actions by emotions’, Philosophical Quarterly, 53, 214–230.CrossRefGoogle Scholar
Draper, Paul (2016) ‘Simplicity and natural theology’, in Bergmann, M. & Brower, J. (eds) Reason and Faith: Themes from Richard Swinburne (Oxford: Oxford University Press), 48–63.CrossRefGoogle Scholar
Draper, Paul (2017) ‘God, evil, and the nature of light’, in Meister, C. & Moser, P. (eds) Cambridge Companion to the Problem of Evil (Cambridge: Cambridge University Press), 65–84.CrossRefGoogle Scholar
Draper, Paul (2019) ‘What if God makes hard choices?’, in Buchak, L., Zimmerman, D. W., & Swenson, P. (eds) Oxford Studies in the Philosophy of Religion, IX (Oxford: Oxford University Press), 18–30.CrossRefGoogle Scholar
Farmer, D. D. (2010) ‘Defining omniscience: a feminist perspective’, Faith and Philosophy, 27, 306–320.CrossRefGoogle Scholar
Findlay, J. N. (1948) ‘Can God's existence be disproven?’, Mind, 57, 176–183.CrossRefGoogle Scholar
Forrest, Peter (2007) Developmental Theism: From Pure Will to Unbounded Love (Oxford: Oxford University Press).CrossRefGoogle Scholar
Forrest, Peter (2012) ‘Replying to the anti-God challenge: a God without moral character acts well’, Religious Studies, 48, 35–43.CrossRefGoogle Scholar
Gibbard, Allan (1990) Wise Choices, Apt Feelings (Cambridge MA: Harvard University Press).Google Scholar
Hendricks, Perry (2018) ‘Sceptical theism and the evil-god challenge’, Religious Studies, 54, 549–561.CrossRefGoogle Scholar
Howsepian, A. A. (1991) ‘Is God necessarily good?’, Religious Studies, 27, 473–484.CrossRefGoogle Scholar
Kriegel, Uriah (2012) ‘Moral motivation, moral phenomenology, and the alief/belief distinction’, Australasian Journal of Philosophy, 90, 469–486.CrossRefGoogle Scholar
Lancaster-Thomas, Asha (2018a) ‘The evil-god challenge part I: history and recent developments’, Philosophy Compass, 13.CrossRefGoogle Scholar
Lancaster-Thomas, Asha (2018b) ‘The evil-god challenge part II: objections and responses’, Philosophy Compass, 13.CrossRefGoogle Scholar
Lancaster-Thomas, Asha (forthcoming) ‘Truth, consequences, and the evil-god challenge: a response to Anastasia Scrutton’, Religious Studies.Google Scholar
Stephen, Law (2010) ‘The evil-god challenge’, Religious Studies, 46, 353–373.Google Scholar
Leslie, John (1979) Value and Existence (Oxford: Blackwell).Google Scholar
Little, M. O. (1995) ‘Seeing and caring: the role of affect in feminist moral epistemology’, Hypatia, 10, 117–137.CrossRefGoogle Scholar
Little, M. O. (1997) ‘Virtue as knowledge: objections from the philosophy of mind’, Nous, 31, 59–79.CrossRefGoogle Scholar
McDowell, John (1978) ‘Are moral requirements hypothetical imperatives?’, Proceedings of the Aristotelian Society: Supplementary Volumes, 52, 13–29.CrossRefGoogle Scholar
McDowell, John (1979) ‘Virtue and reason’, The Monist, 62, 331–350.CrossRefGoogle Scholar
Calum, Miller (forthcoming) ‘Why God is probably good: a response to the evil-god challenge’, Religious Studies.Google Scholar
Morriston, Wes (1985) ‘Is God significantly free?’, Faith and Philosophy, 2, 257–264.CrossRefGoogle Scholar
Morriston, Wes (2004) ‘The evidential argument from goodness’, The Southern Journal of Philosophy, 42, 87–101.CrossRefGoogle Scholar
Murphree, W. A. (1997) ‘Natural theology: theism or antitheism?’, Sophia, 36, 75–83.CrossRefGoogle Scholar
Murphy, Mark (2017) God's Own Ethics: Norms of Divine Agency and the Argument from Evil (Oxford: Oxford University Press).CrossRefGoogle Scholar
Nagel, Thomas (1970) The Possibility of Altruism (Princeton: Princeton University Press).Google Scholar
New, Christopher (1993) ‘Antitheism: a reflection’, Ratio, 6, 36–43.CrossRefGoogle Scholar
Oddie, Graham (2005) Value, Reality, and Desire (Oxford: Clarendon Press).CrossRefGoogle Scholar
Oddie, Graham (2017) ‘Desire and the good: in search of the right fit’, in Lauria, F. & Deonna, J. A. (eds) The Nature of Desire, (Oxford: Oxford University Press), 29–56.Google Scholar
Plantinga, Alvin (1974) The Nature of Necessity (Oxford: Oxford University Press).Google Scholar
Radcliffe, Elizabeth (2006) ‘Moral internalism and moral cognitivism in Hume's metaethics’, Synthese, 152, 353–370.CrossRefGoogle Scholar
Reichenbach, Bruce (1980) ‘Why is God good?’, Journal of Religion, 60, 51–66.CrossRefGoogle Scholar
Rowe, William (2004) Can God Be Free? (Oxford: Oxford University Press).Google Scholar
Sarot, Marcel (1992) God, Passibility, and Corporality (Kampen: Kok Pharos).Google Scholar
Schroeder, Mark (2007) Slaves of the Passions (Oxford: Oxford University Press).CrossRefGoogle Scholar
Schueler, G. F. (1995) Desire: Its Role in Practical Reason and the Explanation of Action (Cambridge MA: MIT Press).CrossRefGoogle Scholar
Scrutton, A. P. (2011) Thinking Through Feeling: God, Emotion and Passibility (New York: Continuum).Google Scholar
Scrutton, A. P. (2013) ‘Divine passibility: God and emotion’, Philosophy Compass, 8, 866–874.CrossRefGoogle Scholar
Scrutton, A. P. (2016) ‘Why not believe in an evil God? Pragmatic encroachment and some implications for philosophy of religion’, Religious Studies, 52, 345–360.CrossRefGoogle Scholar
Sinhababu, Neil (2017) Humean Nature: How Desire Explains Action, Thought, and Feeling (Oxford: Oxford University Press).CrossRefGoogle Scholar
Smith, Michael (1994) The Moral Problem (Oxford: Blackwell).Google Scholar
Swartzer, Steven (2013) ‘Appetitive besires and the fuss about fit’, Philosophical Studies, 165, 975–988.CrossRefGoogle Scholar
Swinburne, Richard (1997) Simplicity as Evidence of Truth (Marquette MI: Marquette University Press).Google Scholar
Swinburne, Richard (2004) The Existence of God, 2nd edn (Oxford: Oxford University Press).CrossRefGoogle Scholar
Tenenbaum, Sergio (2007) Appearances of the Good: An Essay on the Nature of Practical Reason (Cambridge: Cambridge University Press).CrossRefGoogle Scholar
Tresan, Jon (2006) ‘De dicto internalist cognitivism’, Nous, 40, 143–165.CrossRefGoogle Scholar
Weaver, C. G. (2015) ‘Evilism, moral rationalism, and reasons internalism’, International Journal for Philosophy of Religion, 77, 3–24.CrossRefGoogle Scholar