Nomic Regularity versus Induction

» Downloadable RTF of this piece

Michael J. Murray discusses the wide range of implications of the role of evil in the world, particularly emphasizing one particular argument, what he terms “nomic regularity.” Nomic regularity is the general principle that the functioning of the world will tend to follow common rules or principles; we will generally find that things behave “regularly,” and hence that we can form certain expectations about them. Murray examines and discards a number of applications for the nomic regularity principle as it concerns arguments for God; however, I want to raise a more general point here regarding the overall way that this tool works, because it seems to suggest an additional problem with this theistic strategy.

The basic notion here is essentially nothing more or less than induction. By observing, say, that when we drop an apple, it falls to the ground, we suspect that perhaps other apples will do the same. Dropping more apples, our suspicions are reinforced, and we even begin to wonder whether this might not apply to non-apple objects. With enough of this evidential ballast, we form the belief that all things fall when dropped, and this lets us comport ourselves usefully with regards to apples, airplanes, etc. This same process applies to other phenomena and indeed, taken on the whole, we might compact them all into more general conjectures, like the fundamental inductive tools from which we build science.

What’s suggested by Murray et al. is that we really need to have this sort of thing available to us in the world, or else much would be lost. At the worst, life would hardly have any coherence at all; at best (and more germanely) we would at least not have any meaningful free will, because a free choice is not much of a choice if we cannot make it in the context of certain expected results. Hence, God seems to be limited in the extent to which he can intervene or interfere with the ongoing events in our world. He cannot heavy-handedly steer us from car crashes or avert a hoodlum’s revolver, because then we would find ourselves in a world where cars seem magically never to crash and guns can shoot in any direction except towards the innocent. It would, in short, lack regularity, and without regularity we have very little. Suffering is therefore an unfortunate but necessary implication of the sort of world we want to inhabit.

The point I want to draw attention to is the seemingly banal one I already raised—that all we’re really discussing here is the practice of induction. Just why is it that I can expect an apple to fall when dropped? Merely because I have a mental model of things which says that apples will tend to behave like apples, and apples tend to fall, all based on my previous apple experience. We are able to practice this vital sort of induction because of the regularity that obtains in apples, and indeed, in general.

Now, we are hardly philosophers at all if we cannot recall Hume’s problem of induction, and if we accept the basic Humean argument that induction is essentially an unsupported practice, we might start to ask what ramifications this has for nomic regularity. In particular, on what grounds do we base our induction, and could the two be separated?

One indeed wonders what a world might look like if we did not have induction. If God built the world from whole cloth, we might imagine him either checking the box next to Induction on his order-form, or not; assuming he did indeed check that box in our current world, what would things be like had he not?

There seem to be four logical (though perhaps not actual) possibilities. First, we might imagine a world without nomic regularity, and hence without induction. Second, we might imagine a world with nomic regularity, but where, for whatever reason, induction was disallowed. Third, we can imagine a world with induction but wihout nomic regularity. (The fourth possibility, a world with both nomic regularity and induction, is apparently the one we currently inhabit, and therefore uninteresting as far as counterfactuals go; it will serve as our control case.)

The first possibility is largely explored by Murray, and his point is compelling that such a world would be extremely unappealing. I think we can go further, and say even that such a world would be not so much a world as a chaos—depending anyway on just how much regularity we still have, versus how much has been compromised—but in any case it is not much of an option.

The second possibility is more interesting. The thing we’re really interested in is the ability to use induction; that’s what lets us make sense of the world, and nomic regularity is only important insofar as it seems like an essential prerequisite to allowing induction. Surely it is the very essence of regularity that we can assume things are fundamentally the same except where they differ, and therefore, so long as we take into account those differences, we can expect them to have the same qualities. But suppose that, by the machinations of God (we can probably grant that this would not be the work of evolution, since such an omission would be a colossal failure on the part of natural selection), we simply lacked the psychological ability to apply induction. We have the same sensory inputs, we can even recognize patterns, but we are unable or unwilling to make the logical leap inferring from known to unknown on inductive grounds. We are like philosophy students who really did take to heart their lesson on Hume, and somehow managed to explode all vestiges of their inductive habits and customs. In this case, it seems clear that we would lack all of the things Murray wants us to keep; being unable to think in this way, we would have no better reason for expecting milk to stir into our cereal than to spontaneously combust, and certainly no justification for thinking that saving a drowning child would make him any happier. Without induction we have given up the game; so it is clearly induction that is the essential quality here.

So what of the last permutation—a world without nomic regularity, but nevertheless with induction? On first blush this seems absurd, as we have already agreed that the whole functioning of induction is based on an assumption of regularity. But they are at least logically separable, and if nothing else we can play the omnipotence card—God could establish things such that we could make inductive inferences regardless of whether they adhered in the world. Murray argues strongly against this sort of thing, indicating that simply throwing God into the mix as a sort of grand deceiver is no solution at all—the more he has to trick us, the further we drift from anything remotely resembling the true, shared reality, and shortly we will all be lost in our own solipsistic virtual delusions, something that again seems undesirable.

But this might not be the case. We might instead be able to limit the necessary deception to one simple trick writ large—less than ideal if we are after no tricks at all, but really not bad when compared to immense suffering in the world. Consider this scenario:

An assailant points a gun at you and pulls the trigger. Based on your prior experiences, you expect inductively that a bullet will emerge and you may or may not die. In one counterfactual reality, you are in fact struck by the bullet, suffer in agony, and eventually die—but unbeknownst to you, in this world God intervenes, gently averts the barrel of the gun, and the bullet merely grazes your arm. You escape with hardly a scratch.

Now, in this story, you expected nomic regularity to obtain and consequently applied induction; in fact, nomic regularity did not obtain, but your conclusion was nevertheless true. Did you use induction? In a manner of speaking, yes; it was simply based on inaccurate premises. In short, your inductive reasoning worked, but for the wrong reasons.

In this scenario, the epistemic harm seems minimal. One might object that this is only because some element of probability was built into the event; it was possible, even in a regular world, for the bullet either to strike you or to miss, therefore it was no huge blow to your expectations when it missed. On the other hand, probability is always involved; even in a wholly deterministic world we must allow for uncertainty in any prediction (at least in a quantum sense, though more practically in a Bayesian one). The critic would go on to suggest that this sort of intervention could not be permissible when done on the larger scale or in more obvious ways—eventually we would start to catch on, to question the regularity in things, and our inductive expectations would diminish (just as we would probably start to see the finger of God on the scales, which is also undesirable).

But this need not be. All that is implied by our analysis here is that God should intervene in such a way that harm is reduced, yet our inductive expectations are still fulfilled. This is like rearranging the premises in an argument so that it still reaches the same conclusion, and there are countless ways it might be done. Could it be done wholly, such that suffering was completely eliminated? No. Undoubtedly that would prove impossible; that would fall outside of the range of plausible results, a world where bullets never hit their targets. But this isn’t necessary. All we need is to show that the world could have a little less evil than it currently has, with a corresponding decrease in nomic regularity, without negatively impacting anyone’s sense of induction. And it does seem that God could easily steer more bullets aside, aim storms into the sea rather than towards our cities, and help slippery tires stick to the road a bit more often, all while keeping well within the ranges allowed by chance.

If this is the case, then there is at least one more argument, albeit a limited one, why nomic regularity does not acquit an omnipotent and omnibenevolent God for the problem of evil. If he did exist, he might not eliminate evil altogether, but it is unclear why he might not diminish it a little more.