There is a dislike of God-of-the-gaps arguments in the theological and philosophical community. A God-of-the-gaps argument is an argument for a theistic explanation of a feature that does not have a scientific explanation but where, as far as we can tell, a scientific explanation would in principle be possible.
But suppose that there is currently no good scientific explanation of a puzzling natural feature F of the universe, although we can give a good theistic explanation of F. This is likely to be evidence for theism. After all, if F is puzzling, then our best theories presumably do not predict it. Thus, given what we know, F is very unlikely on the naturalistic hypothesis. But if we can give a good theistic explanation, F is not going to be very unlikely on the theistic hypothesis. A good theistic explanation of F in this context will have to give reasons that God would have for actualizing F, and will have to give an argument for why God is not unlikely to act on these reasons, given the relevant background assumptions. This is not an impossible task--e.g., God would have strong reasons for creating persons, because the existence of persons has great worth, and a perfectly good being would be not very unlikely to act on such strong reasons.
Now, if F is evidence for theism, then surely a rational person cannot fail to consider F just because the evidence falls is of the "God-of-the-gaps" type. What matters evidentially is not whether the evidence is of the "God-of-the-gaps" type, but how much confirmation that evidence gives to the existence of God.
It may, of course, be that some evidence needs to be excluded for certain purposes on non-evidential grounds. Thus, a priest needs to ensure that his public interaction with a penitent is not affected by what he learned about the penitent in the confessional, lest inferences about what the penitent confessed be made from the priest's behavior, and it may be that the best means to that end is for the priest to exclude information about particular persons gained in the confessional from his general stock of information for making inferences. But it seems unlikely that the class of "God-of-the-gaps" evidence is at all analogous to cases like these.
Objection 1: God-of-the-gaps arguments are subject to revision in the light of changing scientific information, and hence will not yield the kind of stability that religious belief requires. One will be anxiously awaiting the next issue of Nature to see if one needs to revise one's religious views. (I've made this argument before, and I think in print.)
Response: Surely, the following line of argument is silly: "F is evidence for p. If I believed p on the grounds of F, my belief that p would be reviseable. Hence, I should not believe p on the grounds of F." The argument only shows that an argument that is not God-of-the-gaps is preferable in some way. But we take what we get, evidentially speaking.
Objection 2: If F is something that in principle can be explained by science, we have good reason to think that it will eventually be scientifically explained.
Response: What good reason do we have for this? Is there some inductively justified claim like: "For every feature that can be explained by science that we puzzle over, that feature finds a good explanation within 200 years of the first time we are puzzled over it"? Surely not--I think any such claim will either be gerrymandered or have too many counterexamples. Or is there some a priori argument for this? The only one that has any hope is, I think, a theistic one, along Descartes' lines--God would want us to be able to get at scientific truth. Besides, even if F will eventually be given a scientific explanation, that explanation may involve a theory that gives F a very low probability, and the theistic explanation may still be superior. Anyway, when we form our beliefs on evidence, we should form them on the evidence available.
But suppose that there is currently no good scientific explanation of a puzzling natural feature F of the universe, although we can give a good theistic explanation of F. This is likely to be evidence for theism. After all, if F is puzzling, then our best theories presumably do not predict it. Thus, given what we know, F is very unlikely on the naturalistic hypothesis.
Just curious, but doesn't this confuse:
(a) Given our present knowledge, we can't assign a high degree of confidence to F on the assumption of naturalism;
(b) Given our present knowledge, we can assign a low probability to F on the assumption of naturalism.
Maybe it doesn't matter for your argument whether (a) or (b) is true, but I found this passage sort of confusing.
Dear Alexander,
"For every feature that can be explained by science that we puzzle over, that feature finds a good explanation within 200 years of the first time we are puzzled over it"
At a meeting of the Joseph Butler society, a claim more or less similar to this was referred to as 'the optimistic induction': past success of science justifies us in expecting future success. I do think that 'God of the gaps' arguments fall prey to some form of optimistic induction, but it is indeed a problem trying to find the best way to formulate this induction.
Let me try to make the terminology more precise.
An Optimistic Induction, O, is any induction from past scientific success to future scientific success.
O1 is the Optimistic Induction introduced above by Alexander: "For every feature that can be explained by science that we puzzle over, that feature finds a good explanation within 200 years of the first time we are puzzled over it."
Of course, there is no evidence for O1. I take it that Alexander chose 200 years as an arbitrary number, but I doubt that there is any number of years we could substitute that would yield a plausible statement.So far as I know, nobody has ever tried to measure the average gap between our first puzzling about some feature, and our finding an explanation. Such a project would be extremely difficult: at what point, for example, did people begin to puzzle over lightning? Have people always puzzled over it (as opposed to being mystified by it), or did the puzzling only begin when lightning became an object of scientific interest? The kind of research that would be necessary to construct a plausible version of O1 is very tough: I cannot see the evidence that would justify any optimism about the vindication of O1, whatever period of time we substitute for 200 years.
However, there is an alternative approach. Any God of the gaps argument will always involve some particular feature, F, which is currently unexplained. Rather than waiting for some cover-all Optimistic Induction, the sceptic could point to analogies between F, and some other feature, F' that has been explained.
O2: Our current state of knowledge about F is analogous to our state of knowledge about F' in the year xxxx. Although we had no explanation for F' at xxxx, by yyyy, we did have such an explanation.
The plausibility of O2 would depend upon the strength of the analogy that is drawn in each particular case. Thus, rather than a single 'God of the Gaps' argument that stands or falls on a single Optimistic Induction such as O1, we should think of a 'God of the Gaps' argumentative strategy, that depends upon the possibility of filling in the blanks in O2 in a way that fits a particular case.
Thanks for posing the question.
Clayton:
I agree that the distinction needs to be made. I am thinking of things where it seems very unlikely that the feature would result given the laws as we know them. For instance, sexual reproduction might be such a case.
Ben:
That is a quite interesting approach. One difficulty with it is that it can be parodied. Suppose Dr. Jones proposes a new theory T as an explanation of a previously unexplained phenomenon F. But we give him the following argument: "Our current state of knowledge of F is like our knowledge of F' in year x. But an explanation of F' not involving T was eventually found. Therefore, an explanation of x not involving T will probably be eventually found." There is probably a way out of this difficulty.
A second difficulty is that there will also be items F'', F''', and so on, knowledge about which was also parallel to our knowledge about F, but where no naturalistic explanation has yet been found.
Moreover, the analogue F' would have to be a case where not only did we find an explanation for F', but where that explanation conferred a not very low probability on F' given the laws of nature. There will be many things that don't satisfy that standard. Consider, for instance, Newton's explanation of the motions of the planets. While the explanation conferred a high probability on the planets (in the short term at least) continuing in their orbits, given that they started in them, it did not confer a high probability of the planets having elliptical orbits given the laws. Most of the possible initial conditions involving planets and a star are going to have the planets having velocities too high to form an elliptical orbit around the star. Of course, in this case, we have another explanation right now, involving the Big Bang, and models of solar system formation out of accretion discs. However, this explanation will not confer high probabilities on the explanandum if it depends on low probability assumptions about the initial conditions.
Surely, the following line of argument is silly: "F is evidence for p. If I believed p on the grounds of F, my belief that p would be reviseable. Hence, I should not believe p on the grounds of F
Alex, I'm pretty sure that Kierkergaard gives an argument very much like that in defense of fideism. It might depend on what you take genuine faith to entail. Can someone say that they have (genuine) faith that God exists and also claim that they are not certain? If not, then if certainty is not responsive to new evidence (as in Bayesian epistemology) then the argument is not silly, since revisablity entails uncertainty.
A paper on a similar subject can be found here:
http://www.epsociety.org/blog/2008/11/2008-eps-papers-ganssle.asp
Mike:
Let's see how I can make the Kierkegaardian argument work. For the argument to work, it has to be the case that believing revisably that God exists is worse than not believing that God exists. It's certainly not intrinsically worse. But Kierkegaard might think it's extrinsically worse, because when I believe revisably that God exists, I may mistake my revisable belief for faith. Hence, paradoxically, I am more likely to get to faith when I don't believe that God exists than when I believe revisably, because when I don't believe that God exists, I am unlikely to think that I have faith.
Note that this objection is not just to a God-of-the-gaps argument, though. It applies to every argument that is less than apodeictic.
But I would disagree with this Kierkegaardian argument. If one believes in God, even revisably, this will help one to take seriously the possibility of divine revelation. And if one takes that possibility seriously, and maybe even think that possibility is very likely, that is something that grace can build faith on. After one gets faith, one commits irrevisably to what might previously have been revisable.
I also think it is a mistake to think of the belief that God exists as a part of faith. It is a preamble of faith, something presupposed and entailed by the doctrines of faith. There are other such preambles, such as the claim that there are other persons (else the command to love neighbor is empty).
I think you have to be careful what you consider a 'god-of-the-gaps' argument. Consider this argument:
1) We do not know how observed feature F could be produced on the naturalistic hypothesis.
2) On the theistic hypothesis, F could easily be produced by exercise of God's omnipotent power.
3) If the theistic hypothesis can explain the production of F and the naturalistic hypothesis cannot, F constitutes evidence for the theistic hypothesis.
:. 4) F constitutes evidence for the theistic hypothesis.
The problem here should be obvious: any possible feature could be produced by the exercise of omnipotent power (and, since F is actual, F is possible).
You seem to me to be assuming that the argument is not like this: you assume that the is that there is good reason to suppose God, if he existed, would produce F by non-naturalistic means. It seems to me that it is really hard to show this, for any F.
But, of course, this is very similar to 'fine-tuning' arguments, which don't seem to me to be objectionable 'God-of-the-gaps' arguments. If we say, rather than that God is likely to produce, e.g., minds by non-naturallistic means, that God is simply likely to produce minds by some means, then we have an explanation for why the universe would be such as to be hospitable to minds.
"there is good reason to suppose God, if he existed, would produce F by non-naturalistic means": Actually, there is no need to include "by non-naturalistic means" here.
Now that I think about it, what I have departs a little from standard God-of-the-gaps arguments. For standard God-of-the-gaps arguments may have the "by non-naturalistic means" qualifier in the proposed theistic explanation. But that qualifier only weakens the inference.
Kenny you write,
The problem here should be obvious: any possible feature could be produced by the exercise of omnipotent power (and, since F is actual, F is possible).
I'm not seeing the problem. If, as far as we know, E (some theistic explanation) is a possible explanation for F and E' (some naturalistic one) is not possible, then F confirms E. We have reason to believe that some (of the many possible) theistic explanation is what explains F. What does it matter that there are theistic explanations for lots of features? Most of those will be less plausible than competing naturalistic ones. I don't think, incidentally, that, taken non-trivially, any possible feature could be produced by the exercise of omnipotent power. But that's not so relevant here.
Why are naturalistic explanations ever better? Since the theistic hypothesis can explain anything, isn't it the case that once you accept it, the game's over, so to speak?
I don't think so, but I think this is an important part of the typical objection to 'God-of-the-gaps', and I think it is worth taking seriously.
Kenny: The problem here should be obvious: any possible feature could be produced by the exercise of omnipotent power (and, since F is actual, F is possible).
This is an interesting objection, but I think is solved by the fact that the God hypothesis is not limited to the content of "a being that could do anything." The additional content adds qualifiers to the hypothesis, for instance the absolute goodness of God. So even though God *could* do anything, the question is whether God *would* do anything. Atheists say no. They say he would not be so hidden, would not have made a world with so much suffering, would not have used evolution as a method of creation, etc.
So the mere actuality of something is not enough to be confirmational of the God hypothesis, rather, the feature under consideration must make sense given the existence of God for it to be evidence of His existence.
Alex,
Suppose the Descartes style argument you propose in your response to objection 2 works. What then? Would it follow that F is not, after all, evidence for theism?
If so, then wouldn't a theist need to decide whether she thinks such an argument is successful before she appeals to God-of-the-gaps arguments? She wouldn't want to offer as reason to accept P an argument that she thinks the atheist ought to reject upon accepting P. Or would the theist's appeal to God-of-the-gaps arguments be rather like a reductio? (I.e., given what the atheist believes, she should think that F is evidence for theism. For ...)
Why are naturalistic explanations ever better? Since the theistic hypothesis can explain anything, isn't it the case that once you accept it, the game's over, so to speak?
You have to consider the prior probability of the explanans or the hypothesis doing the explaining. If we can explain F on the hypothesis h that involves God acting in ways inconsistent with a regularity very highy confirmed as a law, (conservation of energy, say), or alternatively explain F on the naturalistic hypothesis h' that involves no such inconsistency, the priors for h' will be much higher than those for h, and h' will be the more probable explanation. In general, unless you're an occasionalist of some sort, God's intervention in the natural course of events is going to have small priors. These are all, I think, at least minor miracles.
Mike - that's right. But when you include that claim - the claim that, as I said earlier "there is good reason to suppose God, if he existed, would produce F by non-naturalistic means" - then I don't think you are doing the typical God-of-the-gaps line any more. At least, that kind of argument doesn't seem to me to be objectionable in the same way that God-of-the-gaps is usually thought to be objectionable.
Can anyone give me an example of an "F"? I would be curious to have an example of some "feature" (I assume by this you mean some physical phenomenon) that is given a theistic explanation because there is no scietific explanation.
Consider your response to the first objection you have identified:
"F is evidence for p. If I believed p on the grounds of F, my belief that p would be reviseable. Hence, I should not believe p on the grounds of F."
Why would you continue to believe a theistic explanation of F, if a scientific one become available? If you are going to walk down that road, then it seems all scientific explanations become expendable, whether or not they came subsequent to a "God of the gaps argument." Indeed, it seems that this is just what Kenny recognizes ain his comment above.
-js
Here's the kind of example I had in mind: sexual reproduction. We do not currently have a scientific explanation (but neither do we have any principled argument that a scientific explanation will never be found). Several theories have been proposed, but they each face problems, or so I understand. On the other hand, a theistic explanation in terms of creation imaging a triune God is possible.
As for why I would believe a theistic explanation if a scientific one became available, well I might have independent reasons (e.g., philosophical and/or theological ones) to think that God exists and that if God exists, he would be likely to produce F. In that case, I might have good reason to believe the theistic explanation, either instead of the scientific one (if the two are incompatible) or in addition to the scientific one (if the two are compatible).
Thank you for your prompt response. My background is in chemistry and the philosohy of science, but I have always been very interested in the philosophy of religion. I stumbled across this blog only today.
A scientifc explanation of what aspect of sexual reproduction? How it works? WHy it works? Why humans are inclined towards sexual reprodution? The theistic explanation you sight: "a theistic explanation in terms of creation imaging a triune God is possible", seems, at least to me, to suffer from the following problem: All living things reproduce (granted, not every living thing reproduces sexually). I guess I am wondering how, if at all, your theisitic explantion would apply to the sexual reproduction of lower animals. If human sexual reproduction lacks a scientific explanation, then is it fair to say that this lack of an explanation is present across the wider animal kingdom? If so, then can the sexual reproduction of a snake, for example, also be explained "in terms of creation imaging a triune God." On the other hand, if the lack of a scientific explanation is not present across the broader animal kingdom, then it seems that whatever scientific explanation is accepted with regards to the lower animals will serve as the model by which to frame a theory for human sexual reproduction.
-js
I was thinking that, as far as we know, we don't have a very good evolutionary explanation for how sexual reproduction got started. The matter is puzzling in light of the costs of sexual as opposed to asexual reproduction (e.g., if you reproduce sexually, you need to find a mate, while if you reproduce asexually, you just need yourself). There may be benefits, but they are not so easy to find.
As for the snake question, well I do think all life in some way images God.