Murphy's Anselmian theism and the problem of evil

Abstract

Mark Murphy has recently defended a novel account of divine agency on which God would have very minimal requiring reasons and a wide range of merely justified reasons. This account grounds his response to the problem of evil. If God would not have requiring reasons to promote the well-being of creatures, Murphy argues, then the evil we observe would not count as evidence against theism. I argue that Murphy's conclusion, if successful in undermining the problem of evil, also undermines probabilistic arguments for theism. However, there is good reason to resist his conclusion. Even if God does not have requiring reasons, but merely has justifying reasons, to promote creaturely well-being, God may nevertheless have most motivating reason to do so, and this would be enough to predict divine action, at least given Murphy's further assumption that God is perfectly free. It does not follow from the rational permissibility of God's Φ-ing that it is possible for God to Φ.

Introduction

Mark Murphy has recently defended a novel account of divine agency, on which God has only justifying, rather than requiring, reasons to prevent evils, and he uses this account to respond to the problem of evil. I will begin by summarizing Murphy's account of God's ethics and its relevance to the problem of evil. Then, I will discuss a problematic consequence of Murphy's argument for theists – that, if successful in undermining probabilistic arguments from evil, it also undermines probabilistic arguments for theism. Finally, I will argue that Murphy's account of God's ethics fails to undermine the problem of evil, since it does not follow from the fact that God is not required to prevent evils that it is possible for God to refrain from preventing evils.

Murphy on divine reasons and the problem of evil

Divine reasons

Murphy distinguishes the concept of God from the concept of an Anselmian being. To be an Anselmian being is simply to be an absolutely perfect being, while being God requires possession of additional attributes, such as being fully worthy of allegiance and worship, which an Anselmian being would not necessarily possess. However, an Anselmian being could qualify as God by taking on a certain contingent ethics (Murphy (2017a), 147–178). On Murphy's Anselmian theism, it is necessary that the Anselmian being essentially possesses each pure perfection to its maximal degree, and one such perfection is perfect practical rationality.¹ (Henceforth, for simplicity I will use ‘God’ as a general term covering both concepts unless the context requires a more specific use.)

Since God is necessarily perfectly rational, does it follow that God is morally perfect? Here we must distinguish two senses in which God may be said to be morally perfect (ibid., 23–24). As a perfectly rational agent, God is morally perfect in the sense that God always responds appropriately to value, but it does not follow that an appropriate response to value for God is also what would count as an appropriate response to value for creatures. If we understand moral perfection in a broad sense, so that one who always acts perfectly on whatever reasons they have is morally perfect, then necessarily God would be morally perfect. However, on a more specific account of moral perfection, on which moral perfection requires one to promote the well-being or flourishing of creatures, if God (qua Anselmian being) is under no obligation to promote the well-being of creatures, then God may not be morally perfect in this second sense. On Murphy's view, God is necessarily morally perfect in the first sense – that God perfectly responds to whatever reasons God has – but God (qua Anselmian being) is not morally perfect in the second sense – that God seeks to promote the flourishing of creatures – although God could contingently adopt a policy of promoting the flourishing of creatures (ibid., 22–29).

Murphy's argument that God need not be morally perfect in the second sense employs a distinction between requiring and justifying reasons (ibid., 58–60; see Gert (2004)). An agent who fails to act on a requiring reason is guilty of practical irrationality unless there is some sufficiently countervailing, incompatible reason on which the agent acts. A decisive requiring reason, then, would be a requiring reason for which there is no sufficiently countervailing, incompatible reason (Murphy (2017a), 98). A reason that countervails a requiring reason could be another, incompatible, requiring reason, or it could be a justifying reason. An agent who acts on a justifying reason ‘acts practically rationally unless there are some incompatible reasons that render acting on it practically irrational’ (ibid., 59). Importantly, justifying reasons can be ‘purely justifying’, so that however strong the justifying reason for an action may be, it does not follow that one is required to so act: ‘[A]n agent that acts on [a purely justifying] reason acts practically rationally, though that agent may fail to act on that reason without irrationality even in the absence of some other reason that precludes acting on it’ (ibid.).

Murphy thinks that while the value of creatures and their flourishing gives God reasons to create them and to promote their flourishing, these are purely justifying, rather than requiring, reasons (ibid., 22–66). God is required merely to respect creatures by not intending evil toward them (see ibid., 85–102).² Since God lacks requiring reasons to promote the flourishing of creatures, it is not necessary that God do so. Thus, while God necessarily acts perfectly on God's reasons, and so is perfectly rational, God is not necessarily perfectly good in the second sense described above – God does not necessarily act to promote the flourishing of creatures (ibid., 67; Murphy (2019)).³

Murphy's argument from Anselmian methodology

Murphy's primary argument that God would not have requiring reasons to promote creaturely flourishing is that God would be greater for not having such reasons, and if God would be greater for having some feature, then it should be included in our concept of God.⁴ This follows from his ‘Anselmian’ approach to perfect being theology. Part of his support for the claim that God does not have requiring reasons to promote creaturely flourishing is his argument that creatures lack the kind of value that would require God to promote their flourishing. I will focus on the argument from Anselmian methodology here, which can be summarized as follows:

1. (1) If P is a pure perfection, then God would possess P to its intrinsic maximum.

2. (2) Sovereignty is a pure perfection.

3. (3) God would not possess sovereignty to its intrinsic maximum if God would necessarily have requiring reasons to create or to promote the well-being of creatures.

Therefore,

1. (4) God would not necessarily have requiring reasons to create or to promote the well-being of creatures.

Premise (1) follows from the ‘absolute-greatness constraint’ on Murphy's Anselmian methodology, which holds that God ‘must exhibit each pure perfection to its limit’ and its ‘perfections should be conceived absolutely, rather than defined in relation to logically prior limits of metaphysical possibility’ (Murphy (2017b), 546). This contrasts with a view like Nagasawa's ‘maximal-God theism’ on which a being may count as God if it is the greatest possible being, even if it fails to exemplify some greatness-making property to its maximum degree because maximal possession of that property is either impossible or incompatible with a greater overall combination of greatness-making properties (Nagasawa (2013); (2017)).

It is largely uncontroversial that sovereignty is part of the concept of God, so I will assume (2) is true, but it is controversial how divine sovereignty should be understood and whether it entails (3). Murphy's argument that divine sovereignty is incompatible with God's having requiring reasons to create or promote the well-being of creatures relies on a strong conception of divine sovereignty involving two conditions.⁵ First, God is fully and solely the source of creation and is fully in control of creation – ‘all facts about creation are ultimately to be explained by God and facts about God’ (Murphy (2017a), 71). Omnipotence alone might seem to be sufficient for very extensive divine control over creation, but Murphy holds that God's control must extend even to the normative realm, so that the goodness of God is at least partially (and immediately) explanatorily prior to the goodness of creatures. God's ‘sovereignty over the normative order would be more fully and valuably realized if facts about God contribute to fixing the value of creatures’ (ibid., 77; see also Murphy (2011)). Second, God must have discretion over that which is not metaphysically or practically necessary in order to be fully sovereign – ‘the Anselmian being is free to settle by that being's own choices the way that the world will be’ (Murphy (2017a), 72). Thus, if the range of God's requiring reasons is very narrow and the range of God's justifying reasons is very broad, then God would have significant discretion since there would be a wide range of merely justified possible options between which it is entirely up to God to choose, including, on Murphy's view, whether to create and whether to promote the flourishing of creatures if God does create (ibid., 81; cf. Murphy (2021), 85–146). So, denying (3) would undermine God's discretion, and thus would undermine God's sovereignty. Thus, Murphy argues from his Anselmian methodology to the claim that God would not have requiring reasons to create or promote the flourishing of creatures.

Now, even if we accept Murphy's analysis of sovereignty, we might wonder if it is possible for any being to possess sovereignty. Here Murphy may appeal to a further feature of his Anselmian methodology, the ‘pressing outward’ constraint. To say that perfections press outward is to say that if we think that having a certain feature (or having a feature to a certain degree) would make a being greater than if it did not have it (or had it to a lesser degree), then that is a reason to think that this feature is possibly exemplified (Murphy (2017a), 21). If we have a reason from general metaphysical considerations to think that the exemplification of such a feature is not possible, then we cannot simply take that reason as conclusive but must weigh the strength of these competing reasons (ibid.). If we deny that perfections press outward in this way and have a fixed space of metaphysical possibility into which the Anselmian being must fit, we run the risk of violating Murphy's absolute greatness constraint since the greatest possible being might not be great enough to count as an Anselmian being.⁶ Given Murphy's pressing outward constraint, we have good (but defeasible) reason to think God would possess the kind of sovereignty Murphy describes.

God's ethics and the problem of evil

Murphy's account of God's ethics provides a response to the problem of evil. Like Murphy, I will focus on a probabilistic problem of evil according to which (very briefly) the evil we observe in the world (O) (or alternatively, the distribution of pleasure and pain that we observe) is much less likely given theism (T) than given atheism (~T) (or given some viable specific competitor to theism, such as naturalism).⁷ Thus, the Bayes's factor P(O|T)/P(O|~T) is very low. It would follow that a defence of theism must show either that theism has a comparative intrinsic probability that is very high (i.e. P(T)>>P(~T)), which is implausible, or that there is additional evidence that strongly supports theism. Murphy responds that his account of God's ethics entails that P(O|T)/P(O|~T) is inscrutable, rather than very low.

O does not provide any evidence against Anselmian theism, Murphy argues, since it follows from God's having only justifying, rather than requiring, reasons to promote or prevent setbacks to creaturely well-being that P(O|T) is inscrutable. ‘As the features of the created world are a matter of divine discretion, there are no such reasons [i.e. reasons to prevent evils] that so much as dispose, however mildly, the Anselmian being to create one way or another’ (ibid., 109). If this is true, then we would not be able to say that theism makes certain epistemically possible worlds more likely than others (apart from those where God intends evil or fails to respect God's own holiness, which would be ruled out by Murphy's Anselmian theism).

Given that whether to act on [reasons to prevent evils] is, by hypothesis, not required by rationality, and nothing in the Anselmian being's nature disposes that being to act on that reason or not, we have no basis at all for making an assessment on the extent to which the Anselmian being will take the pleasures and pains of creatures as reasons to promote their obtaining or their not obtaining. (ibid.)

Since the evidential argument Murphy is addressing is entirely comparative, the inscrutability from one member of the Bayes's factor carries over to the whole equation, so we cannot say that P(O|~T)>P(O|T) (ibid.). We can summarize this stage of Murphy's argument as follows:

1. (5) If God would not necessarily have requiring reasons to create or to promote the well-being of creatures, then it is possible that God refrains from creating or promoting the well-being of creatures.

Therefore,

1. (6) It is possible that God refrains from creating or promoting the well-being of creatures. (from 4–5)

2. (7) If it is possible that God refrains from creating or promoting the well-being of creatures and God has significant discretion, then we are not justified in holding that the evil we observe is less likely on theism than on naturalism (because P(O|T)/P(O|~T) is inscrutable).

3. (8) God has significant discretion. (from Murphy's strong account of sovereignty)

Therefore,

1. (9) We are not justified in holding that the evil we observe is less likely on theism than naturalism (because P(O|T)/P(O|~T) is inscrutable). (from 6–8)

Note that there are two different claims which Murphy's account of God's ethics might be taken to support:

1. (A) God is not required to create or to promote the well-being of creatures.

2. (B) God can refrain from creating and from promoting the well-being of creatures.

Murphy's account of God's ethics is intended to establish (A), but (as I will argue) his response to the problem of evil requires (B). It might appear that (B) follows from (A), as Murphy seems to assume, and this grounds his response to the problem of evil:

Prior formulations of the problem of evil. . .assume that God must have requiring reasons to prevent evils to creatures, and use that assumption as the basis for claiming that the existence (or types, or amount, or distribution) of evils in this world is either incompatible with or gives strong prima facie evidence against the existence of God. But given that God's reasons with respect to preventing evils are justifying, not requiring, reasons, no such argument can get off the ground. (ibid., 103)⁸

Below, I will show that this is false, and so (5) is not justified. What it is possible for God to do does not follow simply from what it is permissible for God to do. Before turning to that argument, however, I will describe a broader implication of Murphy's response to the problem of evil.

Murphy's argument and natural theology

Undermining probabilistic arguments for theism

Let's assume that Murphy is right that his account of God's ethics undermines probabilistic versions of the problem of evil. There is, nevertheless, a danger that his account also undermines probabilistic arguments in favour of theism such as Swinburne's.⁹ For Swinburne, God's knowledge of the good is necessarily motivating. So, if it is good that S obtain, and there are no overriding goods that would exist by S's not obtaining, then God would necessarily bring about S. The existence of ‘humanly free agents’ (i.e. animate substances ‘with moral awareness and limited free will, power, and knowledge’) is a great good (Swinburne (2004), 118). So, Swinburne argues, we have (defeasible) reason to think that the existence of humanly free agents (H) is not very low on theism. Thus if P(H|~T) is very low, then the fact that humanly free agents exist is evidence for theism.

However, on Murphy's view, even if God is justified in creating a world suitable for the existence of humanly free agents (H), if this is a matter of strong divine discretion, as Murphy thinks, we would not be able to say that such a world is to be expected on theism, nor would such a world be more probable on theism than on competitors to theism if the probability of (H|T) is inscrutable.¹⁰ This would leave us with only a comparison of the prior probabilities of theism and atheism, where theism would seem to be at a major disadvantage,¹¹ or one must look elsewhere than the probabilistic approach for an argument for theism. So, Murphy's strategy for undermining the probabilistic problem of evil appears to come at the cost of undermining probabilistic arguments for theism. This is a significant result given the importance of the probabilistic approach in contemporary natural theology. Furthermore, given that Murphy's argument is a response to Draper's (1989) formulation of the problem of evil, and is thus a move in probabilistic natural theology, it is significant to the assessment of Draper's influential argument if a response to it like Murphy's has the consequence of undermining probabilistic arguments for theism.

This result could be avoided if we had a sense of what limits there might be on the range of what God's justifying reasons allow, but Murphy (2017a, 103–128) suggests that any world is permissible (and so possible) for God to bring about as long as that world does not involve God's intending evil. But the observations we make are not fine-grained enough to distinguish worlds with, say, great creaturely suffering that God does not intend from those with divinely intended evil (see ibid., 116–122). The extra qualification added in Murphy (2021), that God could not create a world in which (or, more generally, act so that) God fails to properly respect God's own holiness, would seem to not give theism much more predictive power.¹²

While, on Murphy's view, God (qua Anselmian being) is not rationally required to promote the well-being of creatures, God could choose to adopt a contingent ethics, including one of ‘familiar welfare-oriented moral goodness’ (see Murphy (2017a), 22–23). If God were to adopt a contingent ethics of welfare-oriented moral goodness, this would give theism more predictive power, but theism would face the problem of evil. However, Murphy holds that other possible divine contingent ethics (the adoption of one of which would be sufficient for an Anselmian being to count as God) do not similarly face a problem of evil (see ibid., 147–198). But even if the adoption of some such contingent ethics may have the predictive power needed to ground probabilistic arguments for theism, we are left with a question of how likely it would be that an Anselmian being would take on such a contingent ethics. Murphy says it would violate divine sovereignty if ‘there are any norms that apply contingently to the Anselmian being the application of which is not due to the exercise of the Anselmian being's own discretion’ (ibid., 171). But, since we are largely in the dark about what an Anselmian being would choose of its own discretion (apart from not intending evil or failing to respect its own holiness), it seems that the probability of an Anselmian being's adopting some particular contingent ethics would be inscrutable.¹³ (Thus for an explanatorily powerful contingent ethics C that an Anselmian being might adopt, P(C|T)/P(~C|T) would be inscrutable.)

I will consider two further proposals for understanding God's range of justified options in a way that might allow the arguments of probabilistic natural theology, including the evidential problem of evil, to get off the ground. First, God's justified options might be limited if God would only act for the sake of creatures. This move is suggested, but ultimately not endorsed, by Murphy. Second, God's justifying reasons may be structured in such a way that would make some divine motivations more likely than others. I will briefly consider the first option before giving a more detailed discussion of the second.

Acting for the sake of creatures

Wielenberg (2017, 554–555) describes an epistemically possible world which it seems an Anselmian being would be justified in bringing about on Murphy's theory, and yet bringing it about would be intuitively morally unacceptable. In Wielenberg's example, the Anselmian being intends to create ‘a beautiful and complex visual display followed by a beautiful symphony’, and, as a means to this end, the Anselmian being creates a few humans who live short, excruciatingly painful lives. Since this scenario does not involve the Anselmian being's intending evil, but rather merely allowing evil as a means to something it does intend, it appears that, on Murphy's view, the Anselmian being would be justified in bringing about this world. One reply Murphy suggests is that, if the Anselmian being creates at all, it would create for the sake of some creature and creating the world described by Wielenberg, but not for the sake of any creature, may not be possible for the Anselmian being since there would be no point to it (Murphy (2017b), 576–577).

What should we make of the suggestion that, if the Anselmian being creates at all, it would create for the sake of some creature? If the scope of possible creatures it is possible for one to act for the sake of is limited to rational beings, or to sentient beings, or to living beings, etc., then this would limit the range of possible worlds the Anselmian being could bring about to worlds containing such beings. On this view, either the Anselmian being would not create or it would create a world with, say, sentient beings (let's assume this is the scope of creatures which one can act for the sake of). So, taking the evidence of fine-tuning for sentient life F as an example, the Bayes's factor (F|T)/(F|~T) would be very top-heavy rather than inscrutable (assuming (F|~T) is very low). This would bring Murphy's account closer to Swinburne's in holding that ascribing perfect rationality to God would lead us to expect a world similar to ours (i.e. with humanly free agents) on theism. Furthermore, this picture appears compatible with God's having minimal requiring reasons regarding how to act for the sake of creatures, and so appears compatible with a response like Murphy's to the problem of evil, while ruling out the type of scenario described by Wielenberg.

Nevertheless, there are problems for this move. First, it is not clear that God could not act for God's own sake. Ekstrom (2021, 178–179) presents an example similar to Wielenberg's in which God's purpose is not simply to bring about a beautiful state of affairs. Rather, the purpose is God's own enjoyment of that beauty. Of course, God would not lack anything which creating would provide, but this does not seem inconsistent with saying that God could act for God's own enjoyment, and thus for God's own sake. Given that God has significant discretion over merely justified possible creations and that any creation would reflect God's goodness, God would take some enjoyment from any possible creation, and thus creation would be for the sake of someone. So, not only would Wielenberg's scenario would not be ruled out, but Murphy's theism would face the original problem that P(F|T) would be inscrutable.

Even if we stick with the condition that it must be a creature for whose sake God would act, rather than for God's own sake, Wielenberg's example could be amended so that there are two groups of people, one whose members suffer and the other whose members, perhaps unaware of the source, enjoy the visual display and symphony. The suffering would then be for the sake of some creatures (those in the latter group), but still intuitively unacceptable. Indeed, Murphy (2017b, 577, n. 11) anticipates this point and this is one reason for his hesitation in accepting the condition that creation must be for the sake of someone.¹⁴ His main response to Wielenberg's challenge, then, is to accept the morally counterintuitive result that it may be possible for the Anselmian being to bring about such a world, but he stresses that the Anselmian being would not be a proper object of allegiance for the suffering creatures if it created such a world (ibid., n. 12). Nevertheless, Murphy holds that if it is possible for the Anselmian being to create a world like that in Wielenberg's example, it must have ‘good reason’ to do so. The notion of God's having good reason to act is important for my argument in the following section.

Divine reasons and motivation

Positive reasons vs merely permissive reasons

Beyond the requiring–justifying distinction, there is another distinction which is relevant to Murphy's argument – that between positive reasons and merely permissive reasons. A merely permissive reason for an action would simply make it not irrational, ceteris paribus, to perform that action, but it would not make it good in any way to so act.¹⁵ Surely there are reasons that have such merely permissive force and actions that are merely permitted; I have merely permissive reasons to wave my arms in the air while sitting alone in my office (no-one would be harmed, I would not frustrate any of my goals, etc.) but it is not good in any way that I do so.¹⁶ By contrast, one would have a positive reason to perform some action if it is good in some way that the action be performed.¹⁷

This distinction between positive and merely permissive reasons differs from that between requiring and justifying reasons. A requiring reason would never be merely permissive, but justifying reasons can be either positive or merely permissive. Importantly, positive reasons can be requiring, but they need not be. For instance, if Murphy is right about the nature of divine reasons, it may be good in some way that God perform a certain action, and there may be no countervailing requiring or positive reasons, and yet God may refrain from that action without irrationality. God would then have a positive reason which lacks any requiring force.

I think Murphy's discussion of God's justifying reasons can be read in terms either of positive reasons or of merely permissive reasons. One reason to think we should understand Murphy to be dealing with positive reasons is that he says the following about divine intentions: ‘that some outcome is intended by a fully rational being entails that that being judges it to be good in some way – good in itself, or at least as a means to what is good, or bearing some other positive relationship to the good’ (ibid., 92, italics in original).¹⁸ For a fully rational being to intend to Φ, then, it must have positive reasons to Φ. However, Murphy claims elsewhere that ‘as the features of the created world are a matter of divine discretion, there are no such reasons [i.e. reasons to prevent evils] that so much as dispose, however mildly, the Anselmian being to create one way or another’ (ibid., 109). A merely permissive reason would seem to fit this description, but it is not clear that a positive reason would. A merely permissive reason would allow a perfectly rational being to act on some pre-existing motivation, but it would fail to provide any motivation in its own right. But, as I will argue below, it seems that a perfectly rational being would be motivated by positive reasons to a degree corresponding to their weights, even without any pre-existing motivation. So, if justifying reasons to prevent creatures from suffering evils were positive reasons, then they should at least provide some motivation for God. (However, I will consider the possibility that God's positive reasons are not intrinsically motivating in the next section.) In what follows I will use the distinction between positive and merely permissive reasons to reject premise (5) of my reconstruction of Murphy's argument and defend the following argument against Murphy's inscrutability thesis:

1. (I) God would have positive reasons to promote (or prevent setbacks to) the well-being of creatures.

2. (II) God would be, ceteris paribus, motivated to act in accordance with God's positive reasons.

Therefore,

1. (III) God would be, ceteris paribus, motivated to promote (or prevent setbacks to) the well-being of creatures.

2. (IV) If God would be, ceteris paribus, motivated to promote (or prevent setbacks to) the well-being of creatures, then P(O|T) is not inscrutable.

Therefore,

1. (V) P(O|T) is not inscrutable.

As we have seen, there is good reason to think the first premise of my argument follows from Murphy's description of the Anselmian being's justifying reasons, particularly in his discussion with Wielenberg. Also, another argument for (I) will emerge from the discussion of the next subsection.¹⁹ My primary focus will be on (II), which will be discussed in the next subsection. Then I will briefly discuss the remaining premise of the argument, (IV).

Divine motivation

The second premise of my argument states that God would be, ceteris paribus, motivated to act in accordance with God's positive reasons. From this, along with the premise that God would have positive reasons to promote the well-being of creatures, it follows that God would be, ceteris paribus, motivated to act to promote the well-being of creatures. The second premise is eminently plausible. It is difficult to see how God could recognize a positive reason to bring about S and yet remain completely cold towards bringing about S. This premise is particularly plausible on Murphy's account since his argument requires that God can be motivated by merely justifying reasons.

As far as I can see, there are plausibly only two ways that God could be motivated by merely justifying reasons: either justifying reasons are intrinsically motivating for God, just as requiring reasons would be, or they can motivate because of some non-rational feature of God, namely a brute preference or desire.²⁰ The brute preference option is ruled out on Murphy's view because he holds that God is necessarily perfectly free in the sense that God is free from all external casual influences and from any desires that are not solely a consequence of God's recognition of God's objective reasons. ‘If the Anselmian being's agency is limited by other than that being's assessment of the reasons to act one way or another, then the Anselmian being is not exhibiting the perfection of freedom’ (Murphy (2017a), 27).²¹ If God is perfectly free, then God does not have any brute preferences or desires, so this would seem to leave only the first option, on which God's recognition of positive justifying reasons is intrinsically motivating.

Suppose, however, that God's positive reasons are not intrinsically motivating. Is there a third option, on which God ‘just acts’, selecting an option through an act of libertarian freedom without any prior motivation?²² It seems to me that there are just two ways this could go. First, perhaps God surveys a set of options and, without any pre-existing motivation toward any particular option, is simply struck by a preference for one option over the other. This, however, would be nothing more than a way for God to have a brute preference. While it would not be a pre-existing brute preference, it would nevertheless be a preference that is not derived solely from God's recognition of the objective reasons (since the reasons would not favour the preferred option over competitors) and thus it would be incompatible with divine perfect freedom. Alternatively, we might interpret the claim that God ‘just acts’ as involving God's making an arbitrary choice. Such arbitrariness need not be rationally objectionable. For instance, in a Buridan's ass scenario, the agent has good reason to perform one of a set of incompatible actions, but no good reason to prefer one over any other. In such a case, it does not seem objectionable for an agent to choose arbitrarily between those options.²³ Nevertheless, there is an important difference between Buridan's ass scenarios and God's choice. The agent in the former case is motivated by each option equally, whereas God is not (on the assumption that God's positive reasons are not intrinsically motivating and that God has no brute preferences). But it is far from clear that God could go from being entirely unmotivated to making an arbitrary choice without appeal to a brute preference or an intrinsically motivating reason. We cannot avoid this by appealing to a more general divine motivation, a motivation to pick some option or other from the set of merely justified options, as we can in a Buridan's ass scenario, since on Murphy's view God would only have merely justifying reasons for that more general motivation. But, if God is perfectly free and merely justifying reasons are not intrinsically motivating, then this more general motivation would similarly be in need of explanation. So, we should hold that, if God can have any motivation to act on merely justified reasons, God's positive reasons are intrinsically motivating if God is perfectly free.²⁴

The second premise says more than just that God is motivated by God's positive reasons. It adds that God is motivated in proportion to the strength of those reasons. I assume that positive reasons have weights (some considerations give agents stronger positive reasons than others) and, once we've granted that these reasons are intrinsically motivating, it would seem groundless to suppose that, for a perfectly free agent, their comparative motivational powers differ from their comparative weights. Perhaps the comparative motivational power of God's reasons need not correspond to their comparative weights if God has some brute preference favouring one reason over the other, but, once again, this option is not available if God is perfectly free. If God has more reason to do A than B, and so greater motivation to do A than B, and if God has no brute preferences, could God nevertheless ‘just act’ and choose B over A? It strikes me as incoherent to suppose that a perfectly free, perfectly rational and omnipotent being could be most motivated to do A rather than B, and yet choose B. This is consistent with God's having the causal power to bring about B, and thus being free in an agent-causal sense with respect to this choice, yet it does not follow that it is possible for God to choose B over A.²⁵ So, if there is more positive reason for God to Φ rather than to Ψ, God would be, ceteris paribus, more motivated to Φ than to Ψ.²⁶ As Murphy states, ‘if a perfectly free agent is also perfectly rational, then when the reasons decisively favor one course of action, it is not possible that that agent do anything other than what the reasons decisively favor’ (Murphy (2017a), 28, n. 10). Since God's positive merely justifying reasons can decisively favour one course of action, it would not be possible for God to what the reasons decisively favour.

The second premise does not violate the assumption that the reasons under discussion are merely justifying, rather than requiring. God would not be required to act on or be motivated by these reasons, but this is compatible with God's necessarily being motivated by them in proportion to their strengths and acting accordingly unless there are countervailing requiring reasons. Of course, one way God may be necessarily motivated to Φ is if God has a requiring reason to Φ (and no reasons that justify refraining), but it does not follow that if God is necessarily motivated to Φ, then God has requiring reason to Φ. If God has most all-things-considered positive reason to Φ, then God would be necessarily motivated to Φ, even if refraining from Φ-ing is justified.

One might object that if God is necessarily motivated to Φ, then failing to Φ is incompatible with perfect rationality. But if perfect rationality requires that God Φ, then God has requiring reason to Φ after all.²⁷ However, my argument holds that God's failing to Φ would be incompatible with perfect freedom, not with perfect rationality. My argument is compatible with a picture on which God is not perfectly free but has a brute preference for refraining from Φ-ing and so does not Φ, and yet is perfectly rational, even though God has more positive reason to Φ.²⁸ So, it would not follow from the fact that a perfectly free God would necessarily be motivated to Φ that God would be required to Φ.

If God were not perfectly free (in Murphy's sense), then we would have grounds to reject the second premise. If God could have brute preferences or desires, then God would have a source of motivations that originates elsewhere than God's reasons. Thus, while the reasons would, taken by themselves, motivate God in proportion to their objective weights, God's motivations may also be affected by God's non-rational preferences, so the overall weight of God's motivations may differ from the strength of the reasons.²⁹ This could be the case if God has a brute preference favouring acting on some positive reasons over others or if God has a desire to act on some merely permissive reason over a positive reason.

Indeed, the only way God could be motivated by a merely permissive reason would be if God had some suitably related brute desire. If premise 1 of my argument were false, and there were only merely permissive reasons for a perfectly free God to promote creaturely well-being, then God would not be motivated to promote creaturely well-being. In fact, if God's justifying reasons are to be understood as merely permissive and God's only requiring reasons are reasons to refrain from acting in certain ways (or with certain intentions), then God would have no motivation to create at all, since merely permissive reasons do not motivate without an independent desire. Thus, there is good reason to reject the combination of views on which (i) God is perfectly free, (ii) God has requiring reasons only to refrain from certain types of action/intentions, and (iii) God's merely justifying reasons are merely permissive. Murphy accepts the first two, and so should reject the third. (This gives us another reason to accept the first premise of my argument.)

I think the inclusion of divine brute preferences, and so the rejection of (i), would fit nicely into Murphy's account of divine reasons. It would allow for the discretion God possesses by having a wide range of merely justified actions to be not just a matter of what it is rationally acceptable for God to do, but also of what it is possible for God to do. This kind of leeway seems more fitting for genuine discretion and thus for greater divine sovereignty.³⁰ Alas, Murphy calls perfect freedom an ‘uncontroversial’ divine attribute (Murphy (2019), 94), so this move is not open to him.³¹ But those attracted to Murphy's application of the requiring–justifying distinction to divine reasons would do well to reject divine perfect freedom. Nevertheless, on Murphy's assumption that God would be perfectly free, there is nothing to ground divine motivation that is out of proportion to the weights of the positive reasons. So, we have good reason to accept the second premise on the assumption that perfect freedom is part of the concept of God.

From divine motivation to prediction

From the first two premises of my argument it follows that God would be motivated, at least to some degree, to promote the well-being of creatures. This would be enough for theism to have some predictive power and thus enough to run a version of the evidential problem of evil (and perhaps other probabilistic arguments for or against theism). Since God would be motivated to promote the well-being of creatures (and thus would act accordingly unless there were other types of positive reasons that count strongly enough against doing so), we have reason to think that O, the distribution of pain and pleasure that we observe, would be unlikely on theism, or at least that P(O|T)<P(O|~T). So, P(O|T) is not inscrutable.

Perhaps this conclusion is a bit too quick since there may be other sources of inscrutability, such as incomparability between different types of goods or sceptical theistic worries about our epistemic limitations. So, it may be that we do not know what other kinds of positive reasons God would have, nor how these other reasons would weigh against the reasons to promote creaturely well-being. For instance, perhaps for all we know, aesthetic reasons or reasons related to divine holiness can outweigh, or are incomparable with, reasons to promote creaturely well-being. (See e.g. Draper (2017a), 570–571; Draper (2019); Murphy (2021).) Nevertheless, whatever one thinks of these potential sources of inscrutability, Murphy's theory of divine reasons does not ground a distinct response to the problem of evil, as he claims, if my argument is correct (Murphy (2017a), 110–116). Thus, we cannot conclude that the inequality P(O|T)/P(O|~T) would be inscrutable just from Murphy's thesis that God lacks requiring reasons to promote, or prevent setbacks to, creaturely well-being.³²

Conclusion

I have argued that Murphy's account of God's ethics, on which God is not required to prevent evils, but has merely justifying reasons to do so, does not ground a successful response to the problem of evil. This is because God would have positive justifying reasons to prevent evils, and these reasons would intrinsically motivate God if God is perfectly free. So, even if it is permissible for God to refrain from preventing evils, it does not follow that it is possible for God to refrain or that the probability of God's refraining is inscrutable.