2. Preliminaries

Before developing our argument, it’s important to note that the logical problem of evil must be framed as an internal critique to be effective. To illustrate what this involves, consider a common (though flawed) response sometimes offered by theists when atheists raise the problem of evil:

‘To even speak of something being evil already implies the existence of God. After all, if God doesn’t exist, nothing can be either good or bad, so by pressing the problem of evil, you’ve inadvertently become a theist. Congratulations!’

This is, admittedly, a cartoonish version of the argument, but even in its less exaggerated forms, it remains problematic, primarily because they treat an internal critique from the atheist as an external critique in which moral realism is not just an assumption for the sake of the argument. Aside from the controversial assumption that moral realism depends on theism, this response misses the point: the problem of evil is meant as an internal critique. Even if the atheist is a moral anti-realist, they still have the right to point out an inconsistency within the theist’s commitments. What matters here is whether the theist’s commitments are internally consistent, not whether the atheist’s commitments align with their own worldview.

That said, it’s equally important to emphasize that an internal critique works both ways. The atheist cannot press the problem of evil by assuming things the theist does not accept – such as a different axiology. Running a problem of evil argument from a hedonic utilitarian perspective might work against some theists (perhaps?), but it won’t work against classical theists, who reject such a theory. The key point is this: just as the theist cannot dismiss the problem of evil by claiming the assumptions required to run it are inconsistent with atheism, the atheist cannot raise the problem of evil by ignoring the commitments of the theistic view they are critiquing.

To further illustrate, we might emphasize one assumption already highlighted by classical theists in response to the problem of evil: that God is not a moral agent in the same way humans are moral agents. (We won’t focus on defending this argument here, but it’s worth noting in evaluating an internal critique, as our strategy follows a similar structure in offering an in-principle reason – internal to classical theism – for why no logical argument from evil can succeed.)

In short, classical theists are committed to a theory that goodness is equivalent to being, under the aspect of desirability or perfection.Footnote ⁶ For example, we naturally consider a tree with deep roots, strong branches, and vibrant leaves as a good tree – better than one that fails to achieve these things. Badness (or evil, broadly) is understood by classical theists as a privation of goodness, and this concept extends to their moral theory.

In the context of morality, these ideas apply to rational agents capable of contingent self-determination. Such agents can consider reasons for action, including facts about human nature, and either will to promote their flourishing or thwart it. The former is a rational, morally good act; the latter is an irrational, morally bad act. Humans are perfected to the extent that they act rationally, since rationality is their species-specific potency. It is what separates humans from other animals and makes them fully human.

Now, here is the key point internal to classical theism: moral obligations – for humans, at least – stem from human nature, and God is not a human being. He does not possess a human nature, as God is an immaterial, purely actual being, with no potentials to fulfill or higher state to attain. For classical theists, God simply is, always and fully actual.

Fr Brian Davies argues that these points render the (logical) problem of evil ineffective.Footnote ⁷ He claims that the problem of evil arises from a fundamental confusion of categories – namely, between creature and creator. To apply moral requirements that pertain to human beings to God – even if classical theists agree with those requirements for humans – introduces an assumption that classical theists have every right to reject. Those moral requirements apply only to beings with human nature, which God essentially lacks. For this reason, Davies and others argue that the problem of evil, when directed against classical theism, is dead on arrival.

Again, we are not defending Davies’s approach here – several notable classical theists disagree with Davies – but illustrating what is meant by an internal critique. If – if, if, if! – Davies is correct in his ‘God is not a moral agent’ claim, then an internal critique of classical theism cannot succeed by imposing moral requirements applicable to human beings onto God. Insisting otherwise would beg the question. On the other hand, if one holds a non-classical theistic perspective – where God is seen as subject to similar moral obligations as humans – then the problem of evil might still be available to the skeptic. This could be because the non-classical theist doesn’t view God as fundamentally different from humans, or because they adhere to a different ethical theory, perhaps one where moral obligations are not just objective but absolute. In such cases, these theists would need to offer a theodicy or justification for why God is not morally defective in allowing negative states of affairs.

With the idea of an internal critique now (hopefully) clarified, we aim to further defend classical theism from logical problems of evil by highlighting another feature inherent to the classical theistic paradigm: layers of reality.

Our primary contention is this: the inferences we can legitimately make across layers of reality – i.e., what we can infer about one layer of reality based on what occurs in another – are extremely limited, particularly in the moral realm. In fact, we argue that no inference can be made about the moral status of God (at the fundamental level of reality) based on negative states of affairs at less fundamental levels of reality, even if one assumes (contrary to classical theists) that God is a moral agent like us. This blocks any possibility of running a successful logical argument from evil because, to make the argument work, the skeptic would have to assume either that God exists on the same level of reality as humans (an assumption classical theists emphatically reject) or introduce a linking premise in the argument that is question-begging, as we aim to show.

3. Initial statement of the layers of reality response

We mentioned that classical theists are committed to layers of reality. But what does this mean?

Simply put: Some layer of reality is less fundamental if entities in that layer could not exist unless some entity or collection of entities of a more fundamental layer exist and produce and sustain the existence of the less fundamental entities and whatever properties they have. This understanding quickly gives rise to the notion of metaphysical causation, or deep dependence, as grounding theorists sometimes call it, which is simply the relation the less fundamental reality bears to the more fundamental reality producing and conserving all its beings, which would include all its properties. For example, with Tolkien actively thinking up his story, the story and all its features exist, and really exist,Footnote ⁸ but only insofar as Tolkien is thinking it up. To clarify, we are not speaking of books that symbolically convey the story, but the story per se. Apart from the activity of Tolkien’s imagining mind, there is nothing imagined.Footnote ⁹

Now, we ask: what can be inferred across these layers of reality, that is, between less and more fundamental layers related as (metaphysical) effect to cause? For instance, what can we infer about Tolkien from the story he is currently imagining? (Here, we take Tolkien and his imagined story as examples of different layers of reality.) Let’s say, for familiarity’s sake, that the story is The Lord of the Rings. There are certainly some things we can infer. For example, we can infer that Tolkien exists. Clearly, the less fundamental reality – the imagined Lord of the Rings – could not exist without the existence of the imagining mind that grounds it.

What else can we infer? It’s harder to say. There are obviously many things we cannot infer. For instance, from the fact that there are orcs in Tolkien’s story, we cannot reasonably infer that Tolkien himself is an orc – at least not with any confidence. Why can we infer Tolkien’s existence but not his ‘orc-ness’, even though both features exist in the less fundamental reality? The reason is simple: we can infer Tolkien’s existence because it is impossible for the less fundamental layer of reality to exist without the more fundamental layer also existing. This makes the inference legitimate. However, there is no similar logical connection that allows us to infer that Tolkien is an orc simply because he imagines orcs. It’s entirely possible for him to imagine orcs without being one himself.

Thus, we can infer certain things (features, broadly) about the more fundamental reality from the less fundamental reality if it’s impossible for the latter to have those features without the former also having them. Otherwise, we would need inductive support – that is, independent examples where, in similar cases, the more fundamental reality shares a feature when the less fundamental reality has it. But we have no inductive evidence suggesting that stories featuring orcs are typically imagined by orcs. This line of reasoning does not support the ‘orc-thesis’.

With that informal statement out of the way, let us, following James Ross, introduce what we will call the rule of restricted inferences between layers of reality:

Conditions:

• • It is logically impossible for A to have some feature F if B does not have F.

• • Inductive evidence supports that feature F in A implies F in B based on other known instances of metaphysical dependence.

Now, let us ask a question that brings us closer to the problem of evil. The Lord of the Rings contains many negative states of affairs – many atrocious evils. Can we infer from these features anything about Tolkien, particularly that he is somehow defective for creating a world with such negative states or for failing to prevent them, especially when he could have easily done so at no cost to himself? The answer is confidently no. We cannot infer any such thing about Tolkien based on his imagining of The Lord of the Rings.

So, we can infer the existence of an imagining mind from the existence of an imagined story because the story could not exist unless the more fundamental reality – the imagining mind – existed to conceive it. However, beyond this, many other inferences – especially moral ones – are not so straightforward. For instance, it is not legitimate to infer moral failure in the more fundamental reality simply because moral failure (or defect of some kind) exists in the less fundamental reality that depends on it. Such inferences violate the Rule of Restricted Inferences, as we will demonstrate.

Thus, once it is clear that God does not relate to the world in the same way that things within the world often (but not always) relate to each other – that is, once we understand that God is the deep metaphysical cause of the world, producing and sustaining it in existence – and once the rule of limited inferences is understood, the logical argument from evil loses its force.

Let us now substantiate this.

4. The layers of reality response, developed

That there are layers of reality is, we contend, obvious, but more importantly it is a critical traditional component of the classical theistic worldview, since theists have long claimed that God transcends the physical world, occupying the absolute basement of reality as the necessary condition for any finite, contingent thing. As we emphasized earlier, if a logical argument is to succeed, it must show some inconsistency between the components internal to the classically theistic worldview themselves or between those commitments and some obvious fact of the world – like the fact of evil’s occurrence.Footnote ¹⁰ Contemporary work in grounding theory lends strong support the notion of a hierarchical, or layered, structure to reality, as does common sense when considering things like imaginers and what they imagine, dreamers and what they dream, or even just how a waltz is metaphysically dependent upon the people waltzing.Footnote ¹¹ So, the skeptic will not have available to them the ability to deny this commitment without begging the issue by simply asserting or assuming a flattened view of reality that theists (and many non-theists) would reject.

With that in mind, let us now return to our example of entities at different level of reality and ask some questions about their relationship. Consider once again Tolkien and his story – once more, say aspects of what would eventually become The Lord of the Rings. As Tolkien is imagining different scenes of his story, including characters suffering greatly and committing various heinous crimes, what might be inferred of Tolkien? From the story alone, very little and certainly not something that will help the proponent of the logical argument from evil. To clarify, it may be possible for us to infer something of the character of the author based on the sort of narrative he composes, for which it is worth mentioning that such compositions are on the same layer of reality as the author, but not based on the sort of world he imaginatively builds (this is a point we’ll return to later).

To clarify, we are not claiming we are characters in God’s story, or anything like that. We are not fictions. We are appealing to an example of different layers of reality showcasing the rule of restricted inferences, where it is obvious that we cannot infer the feature of failure or defect (with respect to moral character or otherwise) of the more fundamental reality from the observance of failure or defect from the less fundamental reality metaphysically dependent upon it.

At this point, we invite the reader to take any of the moral principles proposed by proponents of arguments from evil – Mackie’s or Sterba’s, etc. – and test them in the current case. The result is one of obvious breakdown.

For instance, let us briefly inspect James Sterba’s formulation with his attempted deployment of the Pauline Principle – namely, that one may not do evil that good may come of it.Footnote ¹² To be precise, Sterba does not think the Pauline Principle itself actually is inviolable (he takes it that one may intentionally lie to save a life, for example), but he believes certain components of it are inviolable, which he labels Moral Evil Prevention Requirements. They are:

1. 1) Prevent, rather than permit, significant and especially horrendous evil consequences of immoral actions without violating anyone’s rights (a good to which we have a right), as needed, when that can easily be done.

2. 2) Do not permit significant and especially horrendous evil consequences of immoral actions simply to provide other rational beings with goods they would morally prefer not to have.

3. 3) Do not permit, rather than prevent, significant and especially horrendous evil consequences of immoral actions on would-be victims (which would violate their rights) in order to provide them with goods to which they do not have a right, when there are countless morally unobjectionable ways of providing those goods.

But these requirements fail to produce the result Sterba is looking for once we consider entities relating at different orders of reality, since, to return to our example, we cannot infer anything about Tolkien’s moral character by his creating a world in which negative states of affairs occur (horrendous or otherwise) that he could easily have avoided among (admittedly imaginary – we’ll come back to this later) rational beings, regardless of whether he was intending certain goods to come out of it, such as a more exciting story or particulargoods (logically dependent or otherwise) to individual entities. All to say, even if we think the Pauline Principle is true, as something with binding force among beings on the same layer of reality, it is quite clear that it does not apply vertically, that is, when trying to relate beings at different layers of reality. It does not enable the sort of inference Sterba or other proponents of the logical argument from evil require to strongly refute the existence of God. Minimally, these cases of entities relating at different layers of reality should cause us to restrict our commitment to said principles, such that, even if they are objective (viz., have stance-independent force among entities of the same natural kind or the same layer of reality or both, say), they are not obviously absolute (applicable between all beings, including at different layers of reality or different kinds even).

We anticipate, of course, the obvious retort that we are currently containing an example comprised of imaginary as opposed to real beings like us. This objection will be dealt with later, because it importantly sets up for our primary contention of there being no way to issue a linking principle that applies vertically without begging the question and no way to apply the linking principle horizontally without flattening the God of Classical Theism into some anthropomorphic superman on our ontological plane. With that promissory note issued, we now continue with developing the previous line of thought.

Briefly, we might also wish to consider Mackie’s proposed linking premise in his logical argument from evil, namely, ‘that good is opposed to evil, in such a way that a good thing always eliminates evil as far as it can, and that there are no limits to what an omnipotent thing can do’.Footnote ¹³

Though temping, we must continue to set aside wider aspects of the debate, including the (non-insignificant) fact that many have challenged Mackie’s principle both with respect to the idea that a good thing always eliminates evil as far as it can (since there may be justifying goods that are logically dependent upon the permission of negative states of affairs) and that there are limits to what an omnipotent thing can do (an omnipotent being cannot bring about logically dependent goods without that which those goods are dependent upon, for example). None of that is our concern; rather, we are concerned with what sort of inferences this principle might legitimately enable.Footnote ¹⁴ Just ask: can we infer that Tolkien is not good because he did not eliminate all the evil in his story, though it would have been enormously easy for him to do so? Obviously not, and this is not merely because the evil is fictional, as we will discuss later.

With Sterba’s and Mackie’s proposals having been considered, and toward the end of continuing our discussion efficiently, we shall avoid independently testing other principles from the literature on this subject (though, again, we invite the reader to do this themselves; the results are returned rather quickly and decisively, we submit).Footnote ¹⁵ Instead, we simply offer what we consider the most plausible generalized linking premise for the logical problem of evil. While this principle will (spoiler alert!) breakdown as well, it will allow us to see how the modification necessary to get the logical argument from evil to run inevitably leads to begging the issue. Our principle, somewhat similar to one of Sterba’s Requirements, says this:

This principle is definitely plausible but only when considering entities and their relations on the same level of reality, since this principle, like the others, clearly breaks down when thinking of the relation between entities on different layers of reality, i.e., Tolkien and his story.Footnote ¹⁶

Now, consider that most would accept (as they should accept) that if some passerby can save a child from drowning, with no obvious cost such as failing to prevent something worse, they should do so. If they failed to act to prevent the child’s death in such a situation, we would consider them to be morally at fault. Same with many of the other principles offered by formulations of the logical argument – that is, we should accept said principles as binding upon us in relation to other (presumably rational) beings on the same plane of reality.

But again, this principle is not applicable when concerning entities at different layers of reality; the principle, in fact, obviously does not apply in the most relevant respect, that is, insofar as it allows one to make any justifiable moral inference about the more fundamental level from the less fundamental layer. For Tolkien, highly plausibly, could have precluded all the suffering and sin of his characters without cost and yet we do not find him morally defective or blameworthy for failing to do so.Footnote ¹⁷

We take it that the matter, as so far stated, is quite uncontroversial. If that is correct, then to salvage the logical argument from evil, the skeptic must modify the principle to avoid breakdown when attempting to apply it across different layers of reality.

Can this be done? Perhaps the following alteration will do the trick.

While this principle addresses the issue of different layers of reality, it is no more evidently true than the previous version. Simply adjusting the principle to account for relations between entities on different levels of reality does not salvage the logical argument from evil, as the same reasons for rejecting the previous iteration apply here as well. Specifically, Tolkien, as the metaphysical cause of the beings in his story, is not morally at fault for failing to prevent the suffering and wrongdoing imagined to occur within the world. This leaves one remaining option, which undoubtedly is the most obvious option. That is, to qualify the principle to cover only real (as opposed to fictional or imaginary) beings.

This would mean proposing something as follows:

The advance of this principle is that it does initially propose what might appear like a relevant difference to avoid the breakdown that occurred with the prior alteration. However, this modification will not – in fact, cannot – succeed either. Several points are in order to explain why.

First, restricting it misses the general point that moral principles governing entities on the same layer of reality do not obviously apply when relating entities at different layers. In fact, they seem clearly inapplicable in the best examples we have concerning different layers of reality, such as the mind-story or dreamer-dreamed analogy (Socrates and his singleton might be another example of different layers, and since there is no moral relationship between them, this may further corroborate our point). The skeptic’s burden is to justify why moral principles that apply to entities on the same layer of reality should automatically extend to relationships between different layers of reality. Without this justification, simply asserting that God is morally at fault for not preventing suffering is equivalent to the conclusion the skeptic hopes to establish.

Moreover, by closing off fictional or imaginary examples, the skeptic also closes off all possible inductive evidence that could support their case. If they cannot build an inductive case for the required inference, they must show it is logically impossible for a less fundamental reality to have some relevant feature (such as defectiveness) if the more fundamental reality does not. But this claim will not hold, as it merely asserts that a perfect being cannot create an imperfect world, which is logically equivalent to the conclusion the skeptic is trying to establish. This is question-begging in the most straightforward sense.

Ultimately, and most importantly, the restriction itself is question-begging, and flagrantly so.

Why is this? Simply because the only possible instance of the restricted principle is the situation of God and the world, since only God – within the classical theistic paradigm – can create real beings such as us.Footnote ¹⁸ Hence, the restricted principle is just the same as saying: If God, or the being that grounds reality and has the capacity to prevent some aspect of reality from coming to be, i.e. has will or agency, fails to prevent the suffering and sin of real beings metaphysically dependent upon him when He could have done so without cost, God is morally defective. This means the principle used to support this claim is just a restatement of it. To thus insist that the principle is the premise in an argument to establish this conclusion (when it is merely a restatement) is false.Footnote ¹⁹

This means the restricted principle fails to advance the problem of evil unless one can somehow show the restricted principle is true from experience. This, however, would be impossible unless, of course, God revealed it, because the only experience relevant to the principle is the one of God creating real beings. So, God would have to tell us the restricted principle is true, because, presumably, God is the only one in such a position to know, since all other examples concerning non-real beings imply the opposite. However, even if this were revealed by God, it would show, ironically, that God is defective and not someone to be trusted (i.e., that this someone just isn’t God, in other words, and so we would have grounds for being highly suspect that what we were dealing with was, in fact, divine revelation). So, we would have no grounds for accepting as true this purported revelation as it would be a self-defeating purported revelation.Footnote ²⁰ It thus seems there is no way to know whether this principle is true even if it were true. This, atop of the inductive grounds from other examples of more and less fundamental realities for supposing it to be false.

This shows that the stronger form problem of evil is not only not successful but cannot possibly be made successful.Footnote ²¹

Let us now consider some objections.

5. The wholesomeness objection

To finish, we defuse several concerns that we expect our argument to raise. Doing so should allow us to clarify (and hopefully strengthen) our overall response in several respects.

Perhaps someone will think to contrast our example of Tolkien with someone who imagines and pens a particularly graphic or pornographic narrative, suggesting that material of sufficient depravity does provide warrant for inferring the moral character of the one imagining the story. Presumably, as being themselves depraved. Call this the Wholesomeness Objection.

Let us grant, for the sake of this objection, that the moral character of, say, an author, can be inferred from the plotline of their narrative. For instance, some have suggested that George R.R. Martin’s A Song of Fire and Ice describes a morally ambiguous world in which gratuitous sex and evil are prevalent and described in some detail. Does this suggest that Martin is less saintly than Tolkien? Perhaps. Certainly, the works of E.L James or the Marquis de Sade make a better case for inferring the moral ineptitude of the author from their work.

We argue, however, that the evils inherent in Tolkien’s works are just as wicked as those in Martin’s or any others. Sauron is terrible. Mordor is hell on Middle-Earth! The issue is not the existence of evil in the world, but way in which it is attended to and the intent of the author to illicit passions in the reader. And that is a relationship between the author’s composed narrative and the reader, which is on the same layer of reality.

God’s mode of knowing is not attending to creation in the same way an author attends affectually to aspects of his story. Contemporary proposals that God empathetically experiences reality from the subjective vantage point of all or some of his creatures are simply incompatible with God’s impassibility and so not relevant to our analysis here.Footnote ²² This is relevant since it is only because of the mode of attending that one might assume a depraved mind is behind a depraved narrative, for the simple reason that they enjoy attending to depravity. But since the mode of attendance does not apply to God, the comparison would be quite ineffectual. Classical theists deny that God attends to the world in any manner like Tolkien or Georg R.R. Martin attend to their narratives, so this account will not suffice or will itself result in begging the question by once again anthropomorphizing God (particularly God’s mode of knowing). For, what is attention other than a kind of focused observation. While God knows particulars, lowly things, and even evils, God does not direct attention on evils and lowly things in a focused or disproportionate manner as compared to the whole.

Tolkien’s world has horrific elements, but the world, per se, is not attended to in a gratuitous manner. One could imagine a depraved Tolkien could have attended to his story through a narrative that focuses on the gore, violence, and sex which is left unstated but assumed in the background of the world he built. Were a character aware of such evils, and attending to them, perhaps the character would assume the author of that world was depraved, but that assumes the very perspective Tolkien did not have when imagining Middle-Earth. A fortiori, the God of classical theism cannot be attending to evil in a depraved manner, since that would require a change of attention in God in which some aspect of his infinite knowledge is attended in a higher degree than the rest. This is simply not possible in a god of pure actuality.

Now, one might say that we must attend to the depravity of this world, and so God subjects us to horror in the way Martin, James, or de Sade inflict depravity on their readers. However, this is where the analogy breaks down. These authors can force the perspective and attention of the reader through the narrative in a way that God does not. Indeed, one may, within this creation, attend to the wickedness of this world so as to seek a voyeuristic thrill, or to focus upon those evils as a hero who seeks to remedy the evil. In effect, God gives us the world, but the narrative about that world is of our own making.

6. Moral collapse

There is one final concern that we should briefly like to highlight and to make some preliminary suggestions about: the concern of moral collapse. If what we have said in this article is correct, then do not all our concepts of morality break down in relation to God? Do we lose all ability to form reasonable exceptions of God, particularly with respect to how God will treat us? Do we not lose our grip on the moral dimension entirely?

The answer to the latter question is, we think, obviously not. Even if our knowledge of good and bad, right and wrong, as it applies to us, does not univocally apply to God, that is of little consequence for our daily lives, because it still applies to us and we can still have knowledge of such things – in fact, we do. So, that concern is not so much a serious one and can be set aside.

The more interesting question is related to the character of God. We say God is good, perfectly good, but has this notion become meaningless considering our analysis? Well, no, at least not under a traditional analogical notion of goodness. To say, for example, that something is good is – again, traditionally speaking – just to say it is adequately or sufficiently actualized as the kind of thing it is (given the convertibility thesis of being and goodness, another commitment of classical theists). A good tree is something that sinks deep roots and grows branches and leaves, or something like that. A good human being is something that actualizes its species-specifying potency – namely, its reasoning ability – and acts rationally in relation to relevant facts concerning human nature, etc.

As certainly broadly Thomistic and Aristotelian philosophers have emphasized, ‘good’ functions primarily as an attributive, not predicative, adjective. This means that ‘good’ does not denote some specific property or quality in the same way that ‘blue’ might describe the color of my bathrobe. Rather, ‘good’ modifies a noun directly and thus cannot be understood in isolation from the thing it describes. For example, to say ‘there is a good one in my hand’ doesn’t tell us much, but it does tell us something: namely, that there must be something which is sufficiently actualized along the relevant dimensions of its kind.Footnote ²³ Now, that doesn’t tell us whether that something is sharp (which we expect a good knife to be), or rational (as we expect a good human to be), or four-legged (which we expect a good dog to be), since whether those are good-making features of something depends entirely on the sort of something that is in the box. So, our notion of goodness is not rendered meaningless in relation to God, even if it is not informative either, a hat-tip to the apophatic elements of classical theism. We can know God is perfectly good, insofar as God is purely actual and so is not the sort of thing that features privations.

The question, then, of whether we can have any expectations of God relates to the question of how well we can understand the sort of thing God is. Everything acts according to its nature, and God has a nature – is identical to his nature, as classical theists maintain. If we believe we can have certain insights into the nature of God – philosophically, we take it – then perhaps we can begin to generate (at least broad) expectations of the sort of world God might create and how God might govern that world, if one can argue those expectations must be the case to be consistent with the nature of God (this is not to say God has moral obligations; only that God, like everything else, has constraints, if only logical).

If one is more apophatic or agnostic about discerning insights about the divine nature, then perhaps this cannot be done, and we must have recourse toward presumed divine revelation to tell us something about the moral character of God.

We leave such matters open for future inquiry.