9 If Everything Happens for a Reason, Then We Don’t Know What Reasons Are

I have argued elsewhere, in a secular metaethical context, that normative realism—the position that there are robustly mind-independent truths about how to live—faces the following epistemological problem. On the supposition that normative realism is true, we must conclude that in all likelihood we are hopeless at discovering how to live. This skeptical conclusion is so implausible that we are forced to reject the realist supposition that leads to it.¹ In this essay, I explore a structurally analogous argument according to which theism—the position that there is a God in the sense of an omniscient, omnipotent, and morally perfect being—also leads to normative skepticism, and therefore should likewise be rejected.

According to the argument I will suggest: (P1) if theism is true, then (as the saying goes) “everything happens for a reason”; (P2) if everything happens for a reason, then we are hopeless judges of what reasons there are; but (P3) we are not hopeless judges of what reasons there are (a thoroughgoing normative skepticism is (i) implausible, (ii) practically paralyzing, and (iii) undermining of theism itself); therefore, (C) theism is false. For reasons of space, I focus almost exclusively on defending (P1) and (P2). I take it, however, that if (P1) and (P2) can be upheld, then (P3) and (C) carry a great deal of plausibility.

Normative antirealism and atheism are widely misunderstood. In the context of secular metaethics, it is often suggested that if antirealism is true, then “anything goes.” Similarly, in the context of debates about the existence of God, it is often suggested that if atheism is true, then “everything is permitted.” These thoughts are incorrect. While the point requires a positive case too large to make here, normative antirealism and atheism are both perfectly compatible with morality, and with a healthy degree of normative objectivity more generally.² In my view, it is actually normative realism and theism that are unable to make sense of normative objectivity in the end. I argue elsewhere that if normative realism is true, then we have no idea “what goes.” I will argue here that if theism is true, then we have no idea what is permitted. Theism, far from being a precondition for moral conviction, is ultimately incompatible with it, leading to a normative skepticism so deep that, as I will try to show, if we are theists and some horrendous evil starts to unfold in front of us, then we should be in doubt as to whether there is any good reason for us to try to stop it from happening.

The concept God is extraordinarily vague, and it is not my position that there is nothing answering to some recognizable version or other of the concept. It’s worth jumping up and down about this, since it is too easy to hear an argument of the kind I will be offering as an attack on broader ideas of a kind that I am emphatically not attacking—for example, the idea that our lives have meaning and value, or the idea that most of us have good reason to live morally, or the idea that there might be something of transcendent meaning and value in which we all are participating, and with which we are all in some sense capable of connecting. While I won’t say any more about it here, I think these broader ideas are true. Indeed, my entire argument is premised on the assumption that we can make moral sense of our lives. The complaint about theism is that it does not allow us to do that.

So it is not my position that there is nothing answering to any version of the God concept. But it is my position that there is nothing answering to the concept of God in what I’ll call the standard monotheistic sense, namely the sense of an omniscient, omnipotent, and morally perfect being. I deny, in other words, the thought that Alvin Plantinga calls “central to the great monotheistic religions—Christianity, Judaism, Islam, ” namely “the thought that there is such a person as God: a personal agent who has created the world and is all-powerful, all-knowing, and perfectly good” (2011:ix). Throughout the essay, unless otherwise noted, whenever I say God I mean God in the standard monotheistic sense. By theism I mean the view that God in the standard monotheistic sense exists; by atheism I mean the view that God in the standard monotheistic sense does not exist.

The argument I will be making is just another version of the age-old argument from evil. The bottom-line thought is the same as always, namely that it is impossible rationally to reconcile the moral evils that we see in this world with the supposition that God in the standard monotheistic sense exists. I hope, though, to press the age-old worry in a slightly new form. The argument I will offer differs from deductive versions of the argument from evil because it is granted at the outset that it is a perfectly coherent logical possibility that God and evil coexist. The argument differs from typical inductive versions of the argument from evil insofar as instances of evil are not put forward as pieces of evidence that God doesn’t exist. The model is not one of collecting instances of evil as one might collect fingerprints for a criminal trial, aiming gradually to assemble enough evidence to tip the scale in the direction of proof beyond a reasonable doubt.

The form of the argument, instead, is more akin to a reductio. The strategy is to suppose, with the theist, that God exists, and then to argue that this supposition, when coupled with factual observations about the kinds of things that happen in this world, has implausible substantive normative implications concerning the kinds of moral reasons for action there are, indeed to such an extent that on the supposition that theism is true, we must come to distrust our faculty of moral judgment across the board. This skeptical conclusion is unacceptable for a variety of reasons. First, it is extremely implausible. In particular, it is significantly less plausible than the supposition that “God” in one very particular sense of that expression (the standard monotheistic sense) exists. Second, the conclusion entails practical paralysis if genuinely accepted. Third, the skeptical conclusion means that theism is self-undermining insofar as theism itself is (as I will argue in section 6) a substantive normative claim.

The argument bears closest resemblance to a line of objection that has been developed against the theistic position known as “skeptical theism.”³ According to this line of objection, the skeptical theist cannot successfully contain his or her putatively restricted area of skepticism; instead, that skepticism inevitably spills out beyond its intended domain and becomes crippling. Almeida and Oppy (2003), Jordan (2006), Maitzen (2007, 2009, Forthcoming, and Sehon (2010), for example, all argue that the skeptical theist’s moral skepticism, in particular, cannot be contained. One difference between those arguments and mine is that mine is offered as a free-standing argument; it is not embedded in the particular dialectic of the evidential argument from evil, the skeptical theist’s response to that argument, and one way of objecting to that response. Another difference is the emphasis I wish to place on the point that theism is a substantive normative claim with substantive normative implications. In ways that I will try to bring out, the entire argument is offered from within the moral point of view on the world—a point of view that takes for granted the existence and authority of moral reasons, and then discovers, within that point of view, reasons to reject the specific moral hypothesis of theism.

Given the way atheists are often stereotyped in the culture at large, it is worth drawing attention to the fact that arguments from evil against the existence of God start from a place of moral conviction and moral common sense. More importantly, as I will try to bring out, they refuse to leave that place in the face of skeptical challenges from the theist. In the exact reverse of what is often supposed, it is the atheist who insists on taking moral appearances at face value until given a strong reason to do otherwise, and the theist who pushes a deeply skeptical hypothesis according to which moral reality is very different from what it appears to us to be.

So let us begin with some moral common sense. The example I’m about to give is disturbing, but if the subject of discussion is evil and what, morally, to make of it, then it is essential that we have in mind real-life cases. Philosophical positions concerning evil that might sound plausible in the abstract need to be tested against reality, whose horrors outstrip on a routine basis anything that one might otherwise have imagined was possible.

One among endless possible examples of a horrific real-life evil is a drunk-driving accident that occurred in the early morning hours of July 2, 2005, in Long Island, New York. In this accident, a drunk driver traveling 70 miles an hour the wrong way on a highway struck a limousine that was carrying six family members home from a wedding that had taken place earlier that day. In the crash, a seven-year-old girl, who had been a flower girl at the wedding, was decapitated. The limousine driver was also killed on impact, and the flower girl’s five-year-old sister, father, and maternal grandparents were critically injured. In the minutes that followed, the flower girl’s mother, who had also been in the limousine, pulled herself from the wreckage and began searching for her family. She knew that her five-year-old daughter was alive because she could hear her moans, but as she searched the wreckage she found her seven-year-old daughter’s decapitated head. The mother picked it up and clung to it, screaming to her husband that “Katie is dead.” In spite of repeated requests by emergency personnel, the mother refused to give up her daughter’s head, holding onto it for nearly an hour as she watched the rest of her family being cut from wreckage.⁴

The idea that there was a good moral reason to permit this scene of unimaginable horror to take place defies every last shred of moral common sense. This is so in the sense that if there was such a reason, then the moral reality of the world is very different from what our everyday moral and factual capacities are capable of discerning. I assume that no one among the likely readership of this essay would seriously entertain the thought that any of the parties involved deserved this. What, then? When we examine the world as we might have thought we knew it, we can find no circumstance—moral, empirical, or otherwise—that would seem to supply any good reason to permit such an event to occur. Importantly, for our purposes, this is not to say that there couldn’t be a morally good reason to permit such an event to occur. Of course there could be. There could be a morally good reason to permit anything. But it is to suggest that cleaving to the view that there was a morally good reason to permit this crash to happen—which, as I will argue, belief in God entails—might come at a very high price. It might come, in particular, at the price of our ability to trust our own faculty of moral judgment going forward. If there was a morally good reason to permit this to happen, in other words, then we are hopeless judges of moral reasons. The rest of this essay consists in a more formal exploration and development of that idea.

I said a moment ago that there could be a morally good reason to permit anything. That point requires some explanation since, for many of us, the mind almost ceases to compute when asked to suppose that there could have been a good moral reason to permit the crash of July 2, 2005. However, while many of us find it nearly impossible to imagine it as a live possibility that there was good moral reason to permit this accident to occur, we all should agree that it is a perfectly good conceptual possibility that there was such a reason. Theodicies of the kind offered by theists may be seen as helping to establish the point, but we don’t need a theodicy to see it. The secular moral theorist can make the same point by noting that, in a bizarre enough set of circumstances, there could be good moral reason to permit a crash like the one that occurred on July 2, 2005. For example, if for some reason there was a forced choice between permitting this tragedy to happen and permitting an even more horrendous evil to occur (in which, for example, scenes of equal or worse horror were played out a thousand times over), then there could be good moral reason to permit the crash to happen.

All parties to the discussion, then—theist, atheist, and otherwise—can and should agree that as a matter of conceptual possibility, there could exist a good moral reason to permit the drunk-driving crash of July 2, 2005, to take place. Additionally, all parties to the discussion can and should agree on a whole range of moral platitudes concerning the case that we can state with the help of an “other things being equal” clause. For example, all parties to the discussion can and should agree that other things being equal—where this is to say, assuming, for the sake of argument, the absence of any other good reasons, moral or otherwise, bearing on the situation one way or the other—it would be morally impermissible to (1) know this drunk-driving crash was going to occur unless one did something, (2) be capable of preventing it, and yet (3) decline to do so. Indeed, that, presumably, is an understatement. Presumably we can all agree that in such circumstances it would be morally depraved not to prevent the accident. Imagine, for instance, that a stranger somehow knew for certain that unless he acted, this drunk-driving crash in all its horror was about to happen, and that all he needed to do to stop it was to flip a switch at his fingertips (thereby, say, holding the limousine at a red light for an extra few seconds and ensuring that everyone got home safely that night). Presumably we can all agree that in such a case, in which everything else was equal and the agent knew it to be so, it would be morally depraved not to flip the switch.

Notice the breadth and depth of moral agreement between the theist and atheist thus far. First, all parties to the discussion are starting from a place of non-skepticism about morality: all parties to the discussion agree that some things are morally permissible and other things aren’t; that some actions are morally depraved and some aren’t; and so on. Second, the theist and atheist are also largely agreed on substantive moral matters, agreeing on a wide range of moral platitudes that may be stated with the help of an “other things being equal” clause. They all agree, for example, that other things being equal, one is morally obligated to prevent, if one can, the decapitation of an innocent child, the violent death of a limousine driver just doing his job, the sudden disruption of a family’s happiness by unspeakable tragedy, and so on. The matter at issue between the theist and the atheist concerns none of these things—in other words, whether there are moral reasons for action, nor even, for the most part, what those moral reasons are. Instead, their disagreement concerns the much larger, holistic question of how to make best overall moral sense of the world, and in particular, whether it makes sense, given the other things we think we know about morality and the way the world works, to suppose the existence of an omnipotent, omniscient, and morally perfect being. While obviously this is a major point on which to disagree, it is important to see how large the moral common ground is, common ground in terms of which the debate may go forward. The question whether God exists is a moral question that is being posed from within a largely shared moral point of view.

When one first entertains the question whether God exists, it is natural to assume that one is asking a purely non-normative question about whether, as a matter of empirical, metaphysical, or perhaps logical fact, a certain kind of being does or doesn’t exist “out there, ” somewhere, in the universe. One might suppose one is asking something analogous to the question whether anyone is at home in the house across the street, just on a much grander scale. But this natural way of thinking about the question is a mistake. Whatever else it is, theism is a substantive normative claim, and the question whether God exists is a substantive normative question.

To see the point, consider an analogy. Suppose someone points to the house across the street and says, “The man who lives there is a good man.” On first blush, this claim might appear to say little or nothing about how anyone morally ought to live. But that appearance is misleading, for in picking out a particular agent and predicating moral goodness of him, the claim implicitly stakes out the position that the actions this man takes, and the character traits this man displays, are morally good actions to take and character traits to have. In this way, the claim says something of potentially action-guiding relevance to us all.

The full normative content that lies disguised within the claim becomes evident when we encounter reason to question it. Suppose that one day you see the man from across the street standing there and watching impassively while one of his children drowns in front of him in the family swimming pool. The natural response to this factual observation would be to revise one’s view that the man from across the street is a good man. If one did so, one would be revising one’s view on the basis of a further moral assumption along the lines of “A good man does not stand by and do nothing while his child drowns in front of him.” Notice, however, that logic alone does not dictate that response. Another logically available option is to hold fixed the moral idea that the man from across the street is a good man, and instead revise one’s view that “A good man does not stand by and watch while his child drowns.” If, for some reason, one was unshakably convinced that the man from across the street was a good man, then even if one had no idea the man’s reason for standing there impassively and watching while his child drowned, one might opt to revise one’s commitment to the general moral principle about what a good man does, and conclude that “There can sometimes be good moral reason for a man to stand by and watch his child drown, and this was one of those cases, even though I don’t know what the reason was.” Now of course in the case at hand, the first of these two possible ways of revising one’s belief system does far less violence to our overall moral and factual view of the world than the second. But we see from the example how it is a question of what is most plausible all things considered, and how a claim concerning the moral goodness a particular agent could, if one cleaves to it, have far-reaching implications concerning the kinds of behaviors that are morally permissible.

Return to the claim that God exists. Like the claim that the man who lives across the street is a good man, theism picks out a particular agent (in this case, positing the existence of an omniscient and omnipotent being) and predicates moral goodness of him. In so doing, theism implicitly stakes out the position that the actions this agent takes, and the character traits he displays, are morally good actions to take and character traits to have. Theism is therefore a substantive normative claim in the following sense: When combined with non-normative observations about what actually happens in the universe, it necessarily entails conclusions concerning what good moral reasons for action there are.

The interesting and difficult question, to which we turn next, is whether it has any implications concerning our reasons for action. The point for the moment, though, is this. If theism is a substantive normative claim, then we should think about it and assess it in the same way we would any other substantive normative claim, which is to say by examining its substantive normative implications and assessing them for plausibility. In the remaining sections of the paper, that is what I will do. On the analogy I wish to explore, in much the same way that we should reject the moral claim that One should always save the greater number on the grounds that it implies that one should sometimes cut up a healthy person to distribute his organs to five others in need of transplants,⁵ so too we should reject the claim that There is an omniscient, omnipotent, and morally perfect being on the grounds that it implies that in a case in which one could easily save children from an impending catastrophe by calling 911, one should be in doubt as to whether there is reason to do so. Atheism, on this way of thinking about things, turns out to be the most plausible moral theory.

Return to the horrific crash of July 2, 2005, and assume for the sake of argument that God exists. Since God is omniscient, he knew, in some morally relevant sense, that the crash was going to happen unless he intervened to stop it. Since God is omnipotent, he was capable, in some morally relevant sense, of preventing it. For example, presumably it was physically possible for him to prevent the crash from happening, whether by somehow “reaching into” this world and stopping it, or by electing at the outset to create a different total possible world in which the crash didn’t occur.⁶ Now, we also know that other things being equal, it would be morally impermissible to (1) know the drunk-driving crash of July 2, 2005 was going to happen unless one did something to prevent it, (2) be capable of preventing it, and yet (3) not do so. As we saw earlier, this is just a moral platitude upon which all parties to the discussion can agree. Since God is morally perfect, he never does anything morally impermissible. From all of this combined with the fact that the crash occurred, it follows that other things are not, or were not, equal. In other words, there must be, or must have been, some circumstance or other that constituted a morally good reason for God to allow something to happen that it otherwise would have been morally impermissible to allow to happen.⁷

Let us call this circumstance, or set of circumstances, whatever they might be, circumstances C. They might be moral circumstances of which we’re unaware; they might be factual circumstances of which we’re unaware; or they might be some combination of the two. We will assume nothing about their precise nature; we pick them out simply as those circumstances, whatever they might be, such that other things were not equal. Some theists will have views about the likely general nature of those circumstances. The circumstances might, for example, have something to do with the good of free will, or the value of experiences of intimacy with God, but we needn’t ourselves assume anything one way or another on this score. The point is just that on the supposition that God exists, these circumstances certainly exist, and whatever their exact nature, they supply the “reasons-why” or the “beyond-our-ken justification”⁸ for God’s permitting the horrific crash to happen. Since the same reasoning applies to every evil that has ever occurred, we may conclude the following, on the assumption that God exists: For every evil that has ever happened, there existed some circumstance or other C which constituted a morally good reason for God to permit it to happen. This is a substantive normative implication of theism. Our task now is to see what further consequences we may draw from this.

So take some arbitrary evil E and let’s begin thinking in a general way about circumstances C. The first thing to note is that we already have some non-negligible information about the character of those circumstances, simply in virtue of the fact that talk of moral goodness is not empty. For example, while we have no idea the exact character of circumstances C—indeed, we don’t even know whether they are factual, moral, or some combination of the two—we can, at the very least, rule out possibilities such as the following: that the circumstances in question are things like that God thought it would be funny; or that God enjoys seeing people suffer; or that God was going for drama and suspense rather than a happy ending; and so forth. Were we not to rule out such possibilities, predicating moral goodness of God would have no meaningful content. God would be “morally good” only in the sense of that expression that allows that he might be a sadist or an amoral aesthete, which is to say not morally good at all.

Are there further implications we can draw out? Here we reach a key fork in the road. Take any given evil that we know to have taken place, and think again about circumstances C. The question we need to focus on now is this: We know that circumstances C constituted a good moral reason for God to permit the evil. But what reason-giving status (if any) do those same circumstances have with respect to us? In particular, do circumstances C also count as a good reason for us to permit the evil? Or do they count as no such reason?

To state this question with sufficient clarity, we need to distinguish between reasons in the fact-relative sense and reasons in the evidence-relative sense.⁹ The distinction is illustrated by cases like the following. Suppose that unbeknownst to you or anyone else, if you deliberately stick out your foot and trip the pedestrian next to you on the sidewalk, you will, as a matter of fact, save her life by preventing her from being hit by a car a short time afterward. Suppose, however, that there is no evidence available to you or anyone else that this is the case. Regarding such a scenario, we may say that, unbeknownst to you, you have overwhelming reason in the fact-relative sense to trip the person (the reason in question being that doing so will save her life). At the same time, since all the evidence available to you suggests that tripping her would be nothing but an act of gratuitous harm, you have overwhelming reason in the evidence-relative sense not to trip the person. Returning to our question and formulating it with the aid of this distinction, we may ask: For any given evil E, do the circumstances C that supply God with a good moral reason for permitting E also supply us with a good moral reason, in the fact-relative sense, to permit the evil? Or do they supply us with no such reason, in the fact-relative sense?

In entertaining this question, there are two main possibilities to consider. One possibility is that circumstances C constitute a good moral reason (in the fact-relative sense) for any agent (whether God or human being¹⁰) to permit the evil to happen. On this scenario, God’s reason for permitting the evil is what we may call agent-neutral—in other words, supplying anyone with a good reason to permit the evil. The other possibility is that circumstances C constitute a good moral reason for God to permit the evil, while providing no such reason for us. On this scenario, God’s reason for permitting the evil is what we may call agent-relative—counting as a good reason for God to permit the evil, but supplying us with no similar reason.¹¹

If we now step back and consider the world as a whole with all its evils, then there are three large-scale moral scenarios to consider. On the first, God’s morally good reason(s) for permitting evils are always agent-neutral. On the second, those reasons are always agent-relative. On the third scenario, God’s reasons for permitting evils are sometimes agent-neutral, other times agent-relative (though presumably we don’t know which in any given case). There is no obvious reason to assume that one versus another of these three scenarios is the one that obtains, so let us examine them all.

On the first scenario, God’s reasons for permitting evils are always agent-neutral (in the fact-relative sense). At first glance, this appears to be a perfectly plausible possibility. After all, we know that circumstances C, whatever their exact nature, constitute morally good reasons, and that fact alone encourages the thought that whatever the circumstances in question are, they might easily count as good reasons for us too. Moreover, given our ignorance of the exact nature of the reasons, we might think there is no problem involved in assuming them to have normative weight for us too. In particular, since the circumstances in question are reasons for us only in the fact-relative sense—with no one but God having any idea what they actually consist in—it might seem that these reasons are practically innocuous as far as we’re concerned—having no bearing (none that we can take into account, anyway) on our own actions.

But that thought is a mistake. To see why, assume for the sake of argument, in accordance with Scenario One, that circumstances C, whatever they are, always provide agent-neutral reasons in the sense that they constitute good moral reasons (in the fact-relative sense) for anyone to permit the evil. The basic point to notice about this scenario is that, contrary to what one at first might have thought, on it we suddenly have at our disposal a great deal of information about the kinds of things that there is often morally good reason to permit to happen. Specifically, all of history, and each additional passing moment, is a source of such information. For example, on the supposition that God exists and that his reasons for permitting evils are always agent-neutral, we know that:

So far, this is hardly an argument with regard to Scenario One, though if claims 1–5 generate some moral unease, they should. So far, all I have done is to state a handful of direct logical consequences of the substantive normative claim implied by the thesis that God exists, phrasing them in language that highlights the possibility that we are now considering—namely that God’s reasons for permitting evils are agent-neutral.

The next thing to note is how bizarre our moral epistemological situation is on this scenario. Theism is usually associated with an insistence on a certain kind of moral ignorance on the part of human beings, in the sense that theists often emphasize that we don’t know God’s deeper reasons for doing or allowing many of the things he does. And that no doubt has to be part of the position; it would be arrogance and delusion of the highest order, presumably, to claim that one knows why there was good moral reason to permit a seven-year-old girl to be decapitated, or for millions of people one never met to die in natural disasters, war, mass exterminations, and so on. But it turns out that on Scenario One, we have a terrific amount of reliable information about moral matters. For on this scenario, we know a great deal about what there tends to be good moral reason to permit, even while simultaneously being in ignorance about why there is morally good reason to permit it. In other words, while we have no idea what those “circumstances C” are, in virtue of which everything that happens is something there was morally good reason to allow to happen, we always know (as soon as whatever happens happens) that there was a good moral reason for permitting it. And this implication is highly problematic.

Here is an example to illustrate the point. Imagine you are at a highway rest stop when you see a stranger, intoxicated to the point of being almost unable to stand, get behind the wheel and turn on the engine. There are two children in the back of his car and they look frightened. The car pulls away, speeding and weaving, and then you see the driver turn the wrong way up an exit ramp onto a busy highway. You have your cell phone with you and could call the police with the license plate number and a description. Should you? Few things could be more obvious, from a moral point of view, than that you should. On Scenario One, however, the theist is unable to accommodate this point.

Consider the following reasoning. On the assumption that God exists and permits evils for agent-neutral reasons, it follows, as we’ve seen, that there is often (every 53 minutes, on average in the United States) morally good reason to permit a fatal drunk-driving accident to occur (and let’s suppose you are in the United States). Suppose you know, moreover, that roughly every 90 seconds, on average, someone is injured in a drunk-driving accident.¹⁴ That’s a lot of drunk-driving accidents, and you know that whenever they happen, there is a morally good reason for allowing them to happen. So now, when faced with a potential drunk-driving accident, it seems reasonable to wonder: “Might this be one of the ‘good’ cases?” That is, might this be one of the cases in which there is morally good reason to permit the accident to go forward? You know, of course, that if it is one of those cases, then it will be in virtue of circumstances of which you’re completely ignorant. The thing is, you know that whatever the relevant circumstances are that are capable of providing good moral reason to permit a drunk-driving accident to happen, they occur on a regular basis—on average every 90 seconds. If the past is any guide, then, those circumstances—the ones that provide morally good reasons for permitting such accidents—could easily be in play here too. Of course your everyday moral intuitions would have suggested that there is no morally good reason to permit such horrors to happen, but if we assume Scenario One to be the case, then your everyday moral intuitions about such matters are demonstrably unreliable. After all, you would have thought that 1–5 above are all false, but on the assumption that God exists and always acts for agent-neutral reasons, they are certainly true.

More broadly, under Scenario One it might appear that inductive reasoning is your best bet for discovering what there is morally good reason to allow to happen. For example, suppose you happen to know that far fewer drunk-driving accidents have occurred this year than is statistically normal by this time. Does that count in favor of not calling 911 about this drunk driver? The question is absurd. Yet on the assumption that God’s reasons for permitting evils are agent-neutral, it at least seems like a reasonable question to ask. Indeed, one might wonder, since everything that happens is a guide to what there can be morally good reason to permit, perhaps one should live one’s life by investigating what is statistically normal and then calibrating one’s responses to unfolding potential evils to that. On reflection, though, even that way of proceeding would seem to be a mistake. After all, what actually happens can, in any given case, diverge from what is statistically normal, and in such cases what actually happens, and not what would have been statistically normal, will turn out to be the thing that we can be certain there was morally good reason to permit. Practically, then, under Scenario One, one finds oneself at an utter loss about how to proceed. This is because for any potential evil that one might see coming—whether murder, tsunami, or drunk-driving accident—there will always be a way of categorizing it such that one knows for a fact that there are, on a regular basis, morally good reasons to permit such things to happen, and so it will be only rational to take seriously the possibility that the case in point is one of those.

The upshot is that on the assumption of Scenario One, we should suspend all confidence in our commonsense views about what there is morally good reason to permit to happen. The one thing we know for certain, on Scenario One, is that our commonsense views about how there is morally good reason to respond to unfolding evils are completely unreliable, as claims 1 through 5 above make clear. Notice what has happened: On Scenario One, all of history is converted to a source of evidence about our fact-relative reasons with respect to evils. And as soon as one has evidence that one has a fact-relative reason of a certain kind, one has information of direct practical relevance; in other words, one now has an evidence-relative reason too. On the assumption of Scenario One, we have indisputable evidence that there is, on a regular basis, fact-relative reason for us to permit evils. The only rational response to this evidence is to increase one’s credence, in the case of any given unfolding potential evil, that there is good reason to permit the evil to occur, even though one won’t have any idea in virtue of what.

Scenario One has disastrous consequences with respect to our moral epistemological situation. So let us turn to Scenario Two, in which, by hypothesis, God’s morally good reasons are always agent-relative—constituting good reasons for him to permit the evil while providing no such reasons for us. Just as there were some prima facie reasons to think Scenario One might be a plausible possibility, so there are here too. After all, agent-relative reasons are a familiar feature of human life, and one of the most prominent contexts in which they show up is the context of special relationships, many of which are characterized by some of the same features—for example, asymmetries of knowledge, power, and authority—that characterize the relationship between God (on the assumption he exists) and human beings.

As a mundane example of an agent-relative reason, consider a case in which a child is about to trip over a toy and it is likely the child will be slightly hurt and the toy will be broken. Person A might have good moral reason to permit this accident to happen if A is the child’s parent, and has repeatedly warned the child about such scenarios and believes this is the only way for the child to learn an important lesson and grow in character. But that doesn’t mean that some other person B—for example, a sitter left in charge for a few hours—has good moral reason to permit the accident to happen. As such cases illustrate, the mere fact that one agent has morally good reason to permit a certain evil to happen is not necessarily any indication that some other agent has morally good reason to permit it. So the model of agent-relative reasons seems to hold out promise as a way of understanding the relationship between God’s reasons for permitting evils and our own reasons with respect to those same evils. In short, on Scenario Two, there is no relationship, and we may infer nothing at all about our own reasons from what we know, thanks to factual observations of what happens in the world, about God’s reasons.

What, then, is the problem with this scenario? The problem reveals itself when we start asking how, on this scenario, we are supposed to figure out what our moral reasons are.

One possibility is that there is a secular answer to this question. After all, nothing prevents the theist, when asked how we’re supposed to figure out what our moral reasons are, from relying on a secular moral epistemology. There are two sub-possibilities to consider, depending on whether one assumes a realist or antirealist metaethical view.¹⁵ On a realist account, there are truths about normative reasons that hold in a way that is robustly independent of our evaluative attitudes. If the theist assumes a realist account and proposes to couple it with a secular moral epistemology to explain how we are supposed to figure out what our moral reasons are, then here I wish to rely on arguments I’ve made elsewhere that realism about normative reasons has no tenable secular epistemology.¹⁶ What about the other sub-possibility, which is to assume an antirealist account of normative reasons and couple it with a secular moral epistemology? (The thought is that the theist might draw on these accounts to explain how we know what our moral reasons are, while simultaneously accepting theism.) I return to this possibility in the conclusion, arguing that here too theism leads to normative skepticism.

Assuming, then, that the theist has no acceptable secular moral epistemology, what are his or her remaining options for explaining how, on Scenario Two, we’re supposed to learn what our moral reasons are? We saw already that under Scenario Two, we can’t learn about our reasons by looking to what God permits. So it seems the only thing left is to look to what God says. It would appear, in other words, that our only option is to learn about our moral reasons by way of communications from God. We should not, of course, make any assumptions at the outset about the form those communications might take. For example, we shouldn’t assume that those communications will be in the form of spoken words; presumably they could also come in the form of a “voice of conscience, ” private religious experiences, or an innate moral sense with which we are born.

But this is where we hit epistemological trouble again. If we are to learn about our moral reasons by way of communications from God, then it is of course a prerequisite that we be able to identify those communications. As I’ll now argue, however, on the assumption that God exists, we have no way of doing that.

Begin with another piece of moral common sense upon which the theist and atheist may presumably agree. It is natural to think that other things being equal, in a situation in which there is a massive asymmetry of knowledge, power, and moral goodness, and in which the less powerful party is reliant on communications from the more powerful party for information about how there is most reason to conduct his or her life—unable to glean that information from independent sources, for example, or from the more powerful party’s observable behavior—the more powerful party has reason to communicate with the less powerful party in terms that are clear and unambiguous, such that given the known cognitive powers of the less powerful party, there is little chance of mistaking the message or its source. While that statement is slightly complicated, I take it that it articulates a very basic moral intuition.

Now for a factual observation: God does not communicate with human beings in this way. I won’t dwell on this point, which I take to be obvious, but here are a few considerations. First, to the extent there is anything remotely resembling a shared universal “voice of conscience”—an innate “moral sense” that one might otherwise perhaps reasonably interpret as the voice of God in all of us—there is a secular evolutionary explanation of this that makes it reasonable to question the source.¹⁷ The plausibility of this alternative explanation would seem to count as an ingenious attempt by God to cover his tracks and speak to us in a less than clear and unambiguous fashion. The same may be said of extant religious texts, accounts of miracles, and so forth. Given the compelling alternative explanations of such phenomena, and given what we know (and what presumably God knows) of human cognitive traits and powers, it is not reasonable to expect that all human beings will be persuaded by such evidence.

Since some people believe, based on their own experiences, that they have received clear and unambiguous communications from God, it is worth pointing out that at the very least, it would seem that God does not communicate in clear and unambiguous terms with huge portions of the human population. One might of course try to explain why he doesn’t, or else explain this appearance away. One suggestion might be that God is communicating with all human beings in clear and unambiguous terms, but in virtue of some cognitive or moral deficiency, many people do not recognize his communications as such. Not only is this implausible, but it puts the theist in the awkward position of charging everyone who in good faith claims not to have received any clear and unambiguous communications from God with a cognitive or moral deficiency. Another suggestion is that God is not communicating with many people in clear and unambiguous terms, but that is because they don’t morally deserve to be communicated with. This suggestion, however, leads to the conclusion that we are hopeless judges of moral desert, and therefore of how we ourselves should live, since at least based on what we can see in terms of moral worthiness, there is no clear pattern in God’s choices about whom to communicate with in clear and unambiguous terms. A third possibility is that it is impossible to communicate with humanity in clear and unambiguous terms—or at the very least in any clearer or less ambiguous terms than what we see now. This is completely implausible, however, essentially constituting an abandonment of the claim that God is omnipotent.

If we (1) take our “other things being equal” moral thought about how we might otherwise have thought it would make sense for God to communicate with us, (2) combine it with the factual observation that he does not communicate with us in this way, and (3) assume for the sake of argument that God exists, then here too, as elsewhere, we may infer that other things are not equal. In particular, we may conclude that God has some morally good reason not to communicate with humanity in a clear and unambiguous way. It is hard to see what such a reason could be, of course, but then again it is also hard to see what reason there could be to permit the drunk-driving accident of July 2, 2005.

On the assumption of Scenario Two, then, we may be certain that God has some morally good reason not to be communicating with us in clear and unambiguous terms that any human being who is not cognitively deficient could reasonably be expected to recognize. But this knowledge, in turn, gives us reason to doubt that we have properly identified his communications. One was inclined to think one should call 911 to prevent a drunk-driving accident, but given what we now know to be God’s reasons not to communicate with us in clear and unambiguous terms, can we be confident we’ve properly interpreted his communications on such matters? Recall that we have no particular reason to believe that we are even in Scenario Two, as opposed to Scenarios One or Three. Given that uncertainty, and given God’s known reason or reasons not to be clear with us, what reason do we really have to be confident that calling 911 is the best course? Actions speak louder than words, after all, one might reason. So maybe we should in fact be looking for guidance to what God himself permits.

Notice what has happened under Scenario Two as compared with Scenario One. Under Scenario One, information about God’s practical reasons translated directly into information about our own practical reasons. Under Scenario Two, the information about God’s practical reasons doesn’t translate directly into information about our own practical reasons, but it translates nonetheless. The mechanism of translation in the latter case is that knowledge of God’s practical reasons gives us reason to doubt that we have properly identified his communications regarding our practical reasons (which, by stipulation under Scenario Two, may not be gleaned by attending to his). And that in turn has implications for our actions, insofar as we gain epistemic reason to doubt that we know what our practical reasons are. The problem is compounded by the fact that we don’t even know which of the three scenarios we are in.

One might object to this as follows: “But wait, what about the moral common sense that you yourself keep appealing to? We do have pretty strong, shared intuitions about many moral matters, as you yourself keep pointing out. God wouldn’t make our moral common sense unreliable. So we should understand this moral common sense (about calling 911, etc.) as coming from God. These are his communications to us; we have identified them.”

The problem with this response is that it depends on an assumption about what God would or wouldn’t do. In particular, the thought is that God wouldn’t make our “moral common sense” unreliable, presumably because (we think) his doing so would be morally extremely puzzling. But in having that thought, we rely on moral common sense as a guide to what God would or wouldn’t do. From the point of view of moral common sense, it doesn’t seem to make sense that God would make what we’ve been calling “moral common sense” unreliable. But everything we have seen so far is that moral common sense is no guide whatsoever to what God would or wouldn’t do with regard to any matter. Moral common sense would have suggested that God would prevent a flower girl’s decapitation, that he would not permit tsunamis that kill tens of thousands of innocent people at a time, that he would communicate with us in clear and unambiguous terms concerning his existence and how we have reason to live, and so forth. But in every case without exception, moral common sense has turned out to be no guide at all to what God will or won’t do. So in seeking to identify God’s communications—in trying to separate out the signal from the noise, so to speak—we have no basis at all for making predictions as to how God might or might not communicate with us. In particular, we are not entitled to assume that our own common sense is a reliable guide to what God will or won’t do. That assumption is eviscerated on a daily basis by every horrendous evil that God permits to happen for reasons that are completely opaque to us.

One might now object that I am asking the theist to prove, without in any way relying on moral common sense, that he or she is entitled to rely on moral common sense. And this, one might continue, is no more interesting than pointing out that you can’t prove that you’re not a brain in a vat without making a question-begging assumption according to which you are not. But this objection rests on a misunderstanding of where we are in the dialectic. The problem that the theist is now facing—in being asked to give some account of how, on the assumption of Scenario Two, human beings are supposed to figure out what our moral reasons for action are—is a problem that has arisen entirely from within the moral point of view. The line of reasoning that has brought us to this point is one that began by assuming that there are such things as moral reasons for action and we have some idea what they are, and that we’re all entitled to trust moral common sense until we find some good reason to think otherwise. All we have been doing since is (1) noticing that theism itself involves a substantive normative claim, and then (2) exploring the substantive normative implications of that claim by examining what follows when we couple it with non-normative observations about the way the world is. This line of inquiry is no more a radical skeptical undertaking than evaluating the moral claim that One should always save the greater number by looking at what it implies about particular cases. We reject the latter principle because it leads to the conclusion that one should sometimes cut up a healthy person to give his organs to others. We should reject theism because it leads to a battery of conclusions concerning God’s reasons which in turn force us (whether by way of the agent-neutral or the agent-relative scenario) to conclude that we don’t know what our moral reasons are.

Scenario Three needs no separate treatment. We don’t know what the mix of agent-relative and agent-neutral reasons would be under Scenario Three, after all, and from our examination of Scenarios One and Two, we’ve seen that it doesn’t really matter. No matter what the nature of God’s reasons for permitting evils—agent-neutral, agent-relative, or some combination—we are at sea. Our common-sense views about normative reasons, when coupled with the substantive normative claim of theism, lead over and over again to the conclusion that our common-sense views about normative reasons cannot be trusted.

What about the possibility that I bracketed earlier—the possibility of theism conjoined with normative antirealism? I close with a few brief remarks about this possibility.

According to normative antirealism, our normative reasons are mind-dependent in the sense that they depend for their existence on our (mortal) evaluative point of view on the world. On the particular constructivist view that I favor, whether some fact X is a normative reason for a given agent A to Y ultimately depends on that agent’s evaluative point of view, and in particular on whether the conclusion that X is a normative reason for A to Y follows from A’s global set of evaluative attitudes in combination with the non-normative facts.¹⁸ There is plenty of room for error about normative reasons on such a view, and the view is, at least in principle, compatible with a very strong form of moral objectivity.¹⁹ Taken by itself, an antirealist view faces no epistemological problems, in my view. Interestingly, however, if one adopts an antirealist view and then couples it with theism, then normative skepticism is once again the result.

To see why, note first off that it is not impossible that, even on an antirealist account of normative reasons, everything happens for a reason. Return one last time to the crash of July 2, 2005. On an antirealist account, what would it be for this event to have happened for a good reason? Roughly speaking, it would be for there to be non-normative facts (of which we are presumably unaware), which, by our own evaluative lights, supply a good reason for permitting the accident to happen. And that in turn is not impossible. As we saw earlier, there are many conceivable examples of such facts—for example, a forced choice between permitting this tragedy to happen and allowing scenes of equal or worse horror to be played out a thousand times over. If such a forced choice did in fact exist, and some powerful being opted to permit the accident for that very reason, then the accident happened for a reason after all.

So the antirealist can agree that it’s possible that everything happens for a reason. But to take that possibility seriously, the antirealist is going to have to find it plausible that either we don’t really understand our own “evaluative lights”—in other words, our own deepest loves, values, hopes, and aspirations—or else that the non-normative facts of the universe are very different from what they appear to us to be. (A third possibility is that we are making some gross logical or instrumental error that so far we have been unable to detect.) But these in turn are just far-fetched skeptical scenarios. In other words, the antirealist is going to agree that it’s not impossible that the crash of July 2, 2005 happened for a good reason, any more than it is impossible that the earth is flat, in spite of all our evidence to the contrary, but this possibility is going to strike him or her as one that involves giving up on the thought that we have any real idea of what is going on in this world.

Moreover, it seems to me this is how the possibility should strike anyone, regardless of his or her metaethics. If we simply pay attention to how things appear to us—both morally and factually—then the accident of July 2, 2005 would appear to be an utterly unmitigated evil. It would appear that there is nothing redeeming about its having happened, that there is nothing in the world that makes it okay that it happened. These are appearances that I think we should take at face value until we find an extremely good reason to do otherwise. To go with theism is to deny these appearances. It is to claim that, contrary to how things look, such horrors are not unmitigated after all—that in spite of how it might seem, there is something redeeming about this thing having happened, that there is something that makes it okay that this happened. To my mind, this is not only a radical denial of the appearances, but also a moral disservice to the people who were involved. It furthermore seems to me a disservice to any force at work in the universe that is worthy of the name “God.” Nothing makes it okay that this accident happened. While we can’t be certain of this, we should be as sure of this as we are of pretty much anything. In a choice between all outward moral and factual appearances and one specific conception of God, it seems to me clear which one should go.

Recall that to abandon the idea that God in the standard monotheistic sense exists is not to abandon the idea that there might be something at work in the universe that is worthy of the name “God.” Nor is it to abandon the idea that there might be true and comforting things to say to the victims of moral catastrophes, and to each other, when we contemplate these catastrophes. It is to abandon the idea that one of those true and comforting things is that the catastrophe happened for a reason. That does not mean that there was nothing deserving of the name “God” present on the Meadowbrook Parkway on July 2, 2005, nor does it mean that there is nothing deserving of that name present now. What it does mean is that if there is such a presence at work in the universe, then it is not omnipotent in anything remotely resembling the ordinary sense of that word. We reach this conclusion based on the same old reasoning that has always driven the argument from evil, namely that if such a presence were omnipotent, then it would have stopped this horror from happening, just as the best part of every one of us would have done.²⁰