The problem of evil, both in its logical and evidential forms, is well known to philosophy of religion. Just as well known are a number of theodicies: defences of God, or explanations of why a good and all powerful being might have reasons for permitting the evil that exists.
But are theodicies ad hoc?
Stephen Law thinks so. His saying so was a reply to me, which was in turn a reply to him, so here’s the very condensed version of how the discussion got to this point, then I’ll get into what I want to say here. Some time ago now I had the pleasure of discussing Stephen’s “evil god challenge” with him and our host Just Brierley on the Unbelievable radio show, which I thought it went very well. That challenge is (to summarise incredibly) the claim that Christians have every bit as much reason for thinking that God is evil as they do to think that God is good. The evidence is just as good either way, and since Christians say the reasons for believing in an evil God are poor, Christians should grant that there are likewise no good reasons for believing in a good God. Sure, Christians can answer the problem of evil by appealing to theodicies (more on these below), but then, believers in an evil God can address the question of why there is so much good in the world by appealing to reverse theodicies (i.e. by saying that an evil god would have good reasons for allowing some goodness to exist in his world). Both sides are evenly matched. In short, the reply that I offered on that show (just part of the overall response I offer) is that even if we grant that both sides are evenly matched so far, this isn’t a problem if there are further reasons for supposing that God is good. After all, Christian don’t believe that God is good simply because there’s good in the world. One way they may make an argument for God’s goodness, I proposed, is by appealing to a moral argument for theism. We recognise that there are moral duties. I claim (and have elsewhere defended the claim) that there couldn’t be moral duties unless something like God existed, and the nature of those duties, such as the duty to be kind and merciful, to help others etc, point strongly in the direction of a God who himself is good rather than malevolent.
Shortly afterwards, Stephen began commenting here at the blog. He didn’t (as far as I could tell) actually seek to rebut the moral argument. Instead, he stepped back to a previous point in the argument where we were discussing how Christians respond to the problem of evil. In effect, his point was (again, my summary): “Wait, hold on. I know we granted for argument’s sake that Christian theists (and evil god theists) could offer their theodicies and be done with those problems. But now let’s revisit the problem of evil and the theodicies. You can’t proceed to offer a positive argument for God being good and not evil until you deal with the problem of evil, which makes it look as though God is evil and not good.” Now, I don’t see any reason why we have to treat the arguments in that order at all. The requirement lacks rationale (and may even be called ad hoc!). Why not deal with the moral argument first, as well as any positive reasons for thinking that God is good, and then consider objections, such as the problem of evil? But I’ll set that aside. The way that I replied to this was to point out that we have already granted for argument’s sake that there are theodicies that can be used, so unless there’s something wrong with those theodicies, what more is there to discuss about them? And then came what I now take to be Stephen’s one remaining line of argument. Rebuttal of the moral argument? There wasn’t one. An argument that the theodicies are not true, or not really compatible with a good God? None at all. Nothing. In fact there was precious little to salvage the evil god challenge in the end. The evil god challenge was left in the ditch, and now all we had was the old problem of evil, and the claim that the theodicies aren’t good enough. After all the to-and-fro, there was one consistent claim being reiterated as the killer argument: Theodicies are ad hoc, so they’re no good. That claim is made several times in this discussion, but here are a couple of examples to illustrate what was said.
But the thing about the theodicies, Genn [sic], is that they are what Popper calls ad hoc. They lead to now [sic] new tests. Or, if they do, but the further test fails, there’s always another gerrymandered explanation for the failure that can be cooked up.
Ad hoc explanations lead to no new tests. The theodicies are ad hoc, by Popper’s definition (he coined the phrase). Look it up. Or, when the theodicies are not ad hoc, and the further test is failed, they are salvaged by yet another defensive manouevre [sic], just as in the case of YEC, thereby rendering the theory unfalsifiable (ar an appeal to mystery, of course). Nutters who believe dogs are spies are [sic] the planet Venus, etc. employ the exact same strategy.
Since that discussion, I have also done a podcast episode on the evil god challenge, where I look not just at the moral argument, but also arguments around the nature of God and goodness, as well as at historical arguments for Christianity, and I try to show why the evil god challenge is simply unpersuasive, interesting though it might be.
But here let’s look at this last appeal for the evil god challenge (or rather, the problem of evil, which is all that’s left). Setting aside the fact that there are good responses to the challenge (I spelled three out in the podcast), the thread by which the challenge now hangs is this one hope: that all theodicies are ad hoc and therefore to be rejected. Are they?
Perhaps the strongest claim I have seen to this effect is that of a teacher who tells first year philosophy students that “A theodicy is an ad hoc attempt to explain why God permits evil and pointless suffering.” Now of course, if we get to just define our terms that way, then yes theodicies are ad hoc. But philosophy is typically about deeper things than just choosing to call things whatever you like. Prior to Stephen’s claims that this was the death knell to any reply to the problem of evil, I had not even encountered the argument that theodicies fail or are suspect because they are ad hoc. The reason I hadn’t encountered it is that in the literature on God and evil, the objection just doesn’t appear – or at least not that I have seen in any of the peer-reviewed articles, of which there are many on the problem of evil and theodicies. I’m also not enthused about the objection’s prospects by the fact that nearly the only place I can now find it suggested (apart from Stephen’s personal blog) is on fairly outspoken atheist blogs, sometimes in the comments section.
However, as any philosopher worth his salt knows, the novelty of Stephen’s objection does not show that the objection is mistaken, nor does the company it keeps! So I wanted to dissect the claim for all to see, giving what I hope is a clear anatomy of its weakness – which I think are considerable. Given that it’s a novel criticism that I think fails, see this blog entry as something of a vaccination rather than a cure, trying to pre-empt any influence this objection to theodicies might have.
What is an ad hoc explanation?
Ad hoc explanations exist for no reason other than to save a theory from defeat when it looks like the theory is in trouble. For some reason, when I hear descriptions of what it is for an explanation to be ad hoc, I’m always taken back to a conversation that took place at primary school when I was about nine years old. My friend was trying to impress us by telling us that their family car had monster truck wheels. Talk about cool! Unfortunately this friend had a reputation for telling tall tales and nobody believed him, but he certainly enjoyed telling the tale. I said, “but I see your car all the time when I come over to your place, and it doesn’t have monster truck wheels. It just has normal car wheels.” Then it came, the explanation that has come to occupy the space in my memory as the archetypal ad hoc explanation: “Yeah, the thing about that is, when we see you coming, my parents rush out and change the wheels, so when you get to our place you never see them.”
Oh, of course! Because that’s exactly what we would all expect the owners of a car with monster truck wheels to do, right? Wrong, we have no reason at all to expect that, and precisely there is the point. Sorry Damian – if you’re reading this, I didn’t buy your story.
This example highlights what is the main distinguishing feature of ad hoc explanations. They are “unmotivated.” That’s not to say people aren’t motivated to use them – of course they are, they want to save their story! No, being “unmotivated” here means that they aren’t naturally motivated by the circumstances in question, and are just cobbled together on a sort of impromptu basis (like the story that my friend’s parents swapped the wheels on their car, to save the story from defeat when I pointed out that I saw the car all the time with regular wheels). Here are some examples of ad hoc theories being denoted primarily as unmotivated: Nigel Thomas describes a problem for Description Theory, noting that it really doesn’t deal well with “ ‘classic’ imagery effects” (and just now it doesn’t even matter what these are), saying: “Although it is possible to devise explanations for them in terms which are consistent with basic descriptionist tenets, these explanations generally require otherwise quite unmotivated, ad hoc assumptions. None of the effects would have been predicted from the theory’s core principles.”
Michael Glanzberg finds fault with some ways of resolving the famous “Liar Paradox,” because they are “ad hoc solutions, unmotivated by our prior understanding of the concepts involved.” [Glanzberg, “A Contextual-Hierarchical Approach to Truth and the Liar Paradox,” Journal of Philosophical Logic 33:1 (2004), 27.]
In passing while discussing the problem of demarcation in philosophy of science, Thomas Nickles compares a number of thinkers, saying, “Like Popper, Lakatos and successors
such as Worrall and Zahar attempt to purify science of ad hoc statements, roughly, theory modifications that are heuristically unmotivated and that lead to no new predictions.” [Nickels, “Demarcation, Problem of” in Sarkar and Pfiefer (eds), The Philosophy of Science: An Encyclopedia (London: Routledge, 2005), 196.]
Or at a more popular level, here’s one from an internet discussion board, where a commenter explains why defences of belief in Bigfoot would be ad hoc (and I have no idea whether these defences are used or not):
Cryptozoologists who believe in creatures like Big Foot have been tromping through the forests for years now. By every reasonable extrapolation from how animals (or sentient tribes) behave, how big a population there would have to be, the likelihood of all these hunters somehow missing Big Foot is astronomically low. What’s the excuses? He can turn invisible, he can teleport, he’s not a physical being but from a parallel dimension, he’s an alien and has access to technology we can’t imagine, he’s psychic and erases memories from people’s head. If you think I’m making up this I’m really not. Can I prove that Big Foot aren’t all of these things? No, though I think I can give some arguments that these excuses are highly ad hoc and unmotivated, and only introduced to save the proposition “Big Foot exists”.
Of course, such explanations are indeed ad hoc, being quite unmotivated by anything other than to save belief in Bigfoot. If a large ape like creature exists in the forest and people believe this on the basis of having seen it (assuming they have), then we would not at all expect from this general claim that the being in question is also invisible or psychic. These explanations are unmotivated by the general theory (of Bigfoot’s existence), and have simply been tacked on to save the claim from defeat.
Jim Pryor’s summary is helpful:
You call something ad hoc when it’s introduced for a particular purpose, instead of for some general, antecedently motivated reason. So, for instance, an ad hoc decision is a decision you make when there’s no general rule or precedent telling you what to do.
Philosophers sometimes accuse their opponents of making ad hoc hypotheses (or ad hoc stipulations, or ad hoc amendments to their analyses, etc.). These are hypotheses (or stipulations or amendments) adopted purely for the purpose of saving a theory from difficulty or refutation, without any independent motivation or rationale. They will usually strike the reader as artificial or “cheating.”
For instance, suppose you analyze “bird” as “any creature that can fly.” I then cite mosquitos as a counter-example. They can fly, but they aren’t birds. Now, you might fix up your analysis as follows:
A bird is any creature that can fly, and which is not a mosquito.
This would be an ad hoc response to my counter-example. Alternatively, you might fix up your analysis as follows:
A bird is any creature that can fly, and which has a backbone.
This would be an independently motivated, and more appropriate, response to my counter-example. (Of course, someone may discover counter-examples even to this revised analysis.)
The problem with ad hoc explanations is not that they would never exist except to save a theory from defeat. Sometimes we have to wheel in explanations to save theories from defeat because we’ve actually got decent reasons for thinking that a theory is true, and we might otherwise not have thought of the explanation of apparent problems with the theory. For example, astronomers noted irregularities in the orbit of Uranus. But rather than giving up their views on gravity, which they had good reasons for holding, they hypothesised an unobserved object nearby. They were correct – that object was the planet Neptune. This new hypothesis was put forward to make sense of a model and to avoid having to give it up. But the hypothesis was legitimate nonetheless, because it was properly motivated. The proposal of a large mass beyond Uranus, interfering with its orbit, arises quite naturally from what people already knew about orbits, gravity, mass and so forth. Had they, on the other hand, suggested that “all planets are subject to the Sun’s gravity in the same way – unless that planet happens to be Uranus,” then we’d have a pretty unmotivated and ad hoc proposal on our hands.
Karl Popper made it clear that, in his view at least, being ad hoc may or may not involve being testable, saying (for example) that the best theories have numerous features including explanatory power, simplicity, being the least ad hoc and also being testable [Popper, The Logic of Scientific Discovery (London: Routledge, 2005), 438]. Not being testable doesn’t, in itself, make a hypothesis ad hoc. The main point to make about an ad hoc hypothesis is that it is unmotivated for any reason other than to save a theory. That being said, Stephen Law had a point when he said that ad hoc hypotheses lead to no new tests. Although claims that can’t be tested might not be ad hoc, it’s true that genuinely ad hoc hypotheses don’t lend themselves to new tests, which leads to one way of showing that a hypothesis is not ad hoc: It does indeed invite further testing.
Reading through any in-depth work that discusses the term quickly reveals that not all philosophers and scientists agree on what is and is not an ad hoc explanation or hypothesis, or on whether or not being ad hoc is always a bad thing (or on when it is and when it isn’t). Popper’s own take on what is and isn’t good science and what counts as pseudoscience or inappropriate ad hocness has been rejected by a number of good philosophers of science, and I have no real interest in those controversies. For a particularly strong critic of Popper here, see Greg Bamford. See HERE also.
What I have described, then, is just the barest essentials of what ad hoc hypotheses are like.
Why theodicies are offered
Theodicies are offered because they are asked for. The problem of suffering is a challenge to Christian theists (or any other theists who maintain that God is personal, all-powerful and good), and a challenge is only meaningfully a “challenge” if the person who puts it forward does so in good faith, to see if it can be met (and not, say, to announce that it couldn’t possibly be met, not intending to seriously listen to responses).
Here the line of thought in offering a theodicy is: “I think that I have reasons to think that God is a personal, all-powerful and good being. However, some people claim that the existence of such a being is incompatible with the existence of the amount of suffering that we see in the world. In order for this to be true it must be the case that there are no circumstances under which a being like this could conceivably permit the amount of suffering that actually exists. All I need to do here is show that this universal claim is not true, and that there are some circumstances under which we could imagine such a being permitting the suffering that exists. And I can!”
Here the believer is presented with a potential defeater for his beliefs. The defeater by its very nature makes a very strong universal claim – that there are no explanations we could (realistically) think of that might explain why a personal, all-powerful and good being might permit the amount of suffering that exists. This claim is very strong. Bear in mind that the theodicies are offered to answer the problem of suffering simply because the problem of suffering asks for theodicies, so it makes little sense to complain that they are only put forward in the face of a challenge. But is there any reason to think that the theodicies are believable apart from the fact that, if successful, they defuse a problem with a particular view of God? In other words, are they properly motivated?
What are theodicies like?
As this is not a defence of the truth of any theodicies, I need not assess their strength here. However, in order to see whether or not there is something about them all that makes them all ad hoc, I’ll briefly outline what some of the theodicies are and then look at whether or not they are unmotivated.
Free Will Theodicies. Free will theodicies maintain that an all-powerful and loving being, one who knew that loving him was what is genuinely best for people, would want them to freely love him rather than love him out of necessity. This shows a greater regard for his creatures, and it also takes account of the fact that love freely given is a greater good than love that is not freely given. But free will also carries a price, namely the price that some creatures will not freely love God and will do evil rather than good. This theodicy is motivated by what we might expect an all-powerful and good being to allow if they want to bring about this good, as it is plausible to think that such a being might indeed want.
“Greater Good” Theodicies. “Greater Good” theodicies (for want of a better term) maintain that we should naturally expect that an all-powerful, loving and all-knowing being knows the future and hence can see the outcome of all situations much better than we can (perfectly, in fact). If such a being exists, then it would surprising if they did not have some over-arching plan that we can’t fully appreciate. Hence there may be events that take place in the world that are hard for us to make sense of and seem tragic and pointless to us, but which nonetheless progress history towards the goals that this being has in mind, goals that are in fact very good. This theodicy is motivated by what we understand by the concept of a being who knows all things and who has meta-goals that are beyond our ken.
“Soul Making” Theodicies. Soul making theodicies maintain that if we were created by an all-powerful and loving being, it makes sense to think that such a being would want us to develop and grow in positive ways, including the development of virtues. However some virtues are such that they are best cultivated in the presence of evil. Virtues like compassion, mercy and so on are like this. Thus in order for a being like this to allow for the kind of development that we would naturally expect such a being to desire, evil is permitted. Notice that this theodicy is motivated by what we would already expect based on the core concepts of the theory involved (that God is our maker, that he is all-powerful and that he is loving).
Natural Law Theodicies. Natural Law theodicies maintain that if God is all-powerful and good, God would want the universe we live in to be one that, all things considered, is naturally conducive to our flourishing. However, as the saying goes, if you want to make an omelette, you have to break a few eggs. A universe that is naturally conducive to our flourishing, all things considered, will not have one-off (i.e. ad hoc) laws that exist just for a limited class of circumstances. It will have generalisable laws such that phenomena in the Universe will be predictable. The net result of this is very good, allowing human life, all things considered, to flourish. But there are obvious down-sides here as well. In order for the earth to support life as it does, having a spinning molten core and plate tectonics is a good thing. The effects of this set-up are on the whole very good. This remains true in spite of the fact that it also causes devastating earthquakes, for example. Here the theodicy can be seen as motivated by a view of providence in general. It arises from considering the kind of world an all-powerful and loving being would need to construct if it is to achieve certain ends that we would naturally expect such a being would want to see achieved.
Are theodicies ad hoc?
And so we arrive at the question. In light of what it is for an explanation to be ad hoc, and in light of what theodicies are like, is it true that theodicies are ad hoc?
What of the features of ad hoc explanations that I mentioned earlier – that they’re not motivated and they don’t allow of any testing? Do theodicies have these features?
As it turns out, theodicies like the ones outlined here (and probably many others) are both motivated and testable. In outlining a number of possible theodicies I have already described how they are motivated. Compare them with the supremely ad hoc explanation my school friend offered for why I had never seen his parents’ care with monster truck wheels. Starting with the theory that his parents have a car with monster truck wheels, there’s nothing inherent to that theory that would lead anyone (i.e. that would motivate anyone) to expect them to hurriedly change the wheels every time they see me coming just to keep me from finding out about them. On the other hand, starting with the theory that a God like that of Christianity exists (who is all-powerful, personal, all-knowing, perfectly good and so on), there already exists in the theory itself reasons for supposing that – for example – such a being might want people to freely love him and so to allow them freedom in spite of the risks it carries. Or for supposing that such a being might want a self-sustaining universe that is capable of supporting intelligent life, in spite of the side effects of the required processes for it to do this.
It may certainly be possible to construct an unmotivated theodicy. However, the main theodicies that have been proposed are not unmotivated in the least. As such they are not ad hoc.
Are theodicies testable? If they are, then for this independent reason, they’re not ad hoc. Now of course a theodicy might not admit of empirical testing because theodicies are usually not empirical claims. Some theodicies might be empirical claims, such as: “Contrary to what you claim, in fact the Israelites did not, under God’s direction, carry out mass genocide of entire people groups.” But the more generalised theodicies, such as those described above, are not empirical claims.
There are of course plenty of meaningful theories that can be questioned and probed for weaknesses and strengths without being empirical theories. Take theories of numbers (or abstract objects in general), or metaphysical claims about objects, or larger theories like nominalism and realism for example. While all of these are the kinds of things that can be tested in the sense that they can be challenged and we can tell whether or not they are subject to certain weaknesses, nobody thinks that this means we would carry out empirical tests to evaluate the theories. If you want to test a theory that is not an empirical theory, then whatever testing you do will not be empirical.
As it turns out, in any relevant sense, theodicies can be tested. They are open to question. For example, take the free will theodicy. The claim is that if we think carefully about what it is for a being to be personal, all-powerful, all-knowing and loving, then we will see that such a being is likely to want people to freely love him. We can certainly do this sort of analysis once we have a sufficiently clear view of what it means to be personal, all-powerful, all-knowing and loving. And we can certainly challenge the theodicy with counter examples or denials of the implications of God’s characteristics. In fact some classical theists (those who maintain determinism) will deny that the claims of the theodicy are true.
It is true that a theodicy could be untestable. We could claim that “God told me that he has a reason for allowing X but he hasn’t explained himself to anyone else.” But it’s clear that in principle a theodicy need not be untestable, and the typical theodicies that are offered do in fact turn out to be testable.
- If an explanation is not unmotivated (pardon the double negative), then it isn’t ad hoc.
- Theodicies in general are not unmotivated.
- Therefore theodicies in general are not ad hoc.
- If an explanation is testable, then it is not ad hoc.
- Theodicies in general are testable.
- Therefore theodicies in general are not ad hoc.