Friday, August 16, 2013

A variant on Plantinga's evolutionary self-defeat argument

If naturalism is true, we would not expect our metaphysics to be reliable beyond the natural realm, since dealing with the natural realm is all our reasoning ability evolved for. But naturalism is a metaphysical thesis that goes beyond the natural realm—namely, it claims that there is nothing beyond it. Thus, if naturalism is true, we would not expect to be reliable in getting to claims like naturalism.


Richard Davis said...

Can the argument also be cast as in an inference to the best explanation?

Q := "Our reasoning is reliable about matters beyond the natural world."

P1 := "We are designed to reason reliably about such matters."

P2 := "Our faculties of reasoning result from an unguided evolutionary process."

On a first blush reading, it strikes me that P1 explains Q much better than P2 does.

Philothumper said...

I've tried to run an argument like this in the past and for the most part, the responses I got boiled down to this: we would evolve to be reliable metaphysicians because we'd evolve to be good logicians, and logic is metaphysics. E.g. Being able to disjunctively and hypothetically reason well requires grasping metaphysical truths such as the laws of logic, and provide a decisive advantage over not being able to. Idk though.

Alexander R Pruss said...

But metaphysics isn't just logic. :-)

Alexander R Pruss said...


Yes, that's another way to do it. I'd consider revising "unguided evolutionary process" to "evolutionary process 'aimed' at survival and reproduction".

Richard Davis said...

Re: Alex's last comment, I'd be interested to see a list of what specific sorts of thinking (other than logic) are involved in doing good metaphysics. I'm especially interested in seeing what sorts of thinking need to be reliable in order for us to arrive at a justified belief in naturalism.

My proposed list:

Empirical observation
Inference to best explanation
Ability to discern simplicity and elegance
Theoretical preference for simplicity and elegance
Rational intuitions
Ability to correctly identify which intuitions are rational intuitions
Emotional commitments
Visceral judgments
Aesthetic preferences

I'm not sure the naturalist has any problem in accepting that we are reliable with respect to rational intuition, since rational intuitions may be such that they are veridical by metaphysical necessity. If so, then evolution *couldn't* have made our rational intuitions unreliable. So maybe a naturalist can rely on logic, rational intuition, and empirical observation without any worries about the evolutionary argument against naturalism.

I find it doubtful, though, that we can get to a justified belief in naturalism relying only on logic, rational intuitions, and empirical observations. It seems to me that we'll at least need inference to the best explanation, and/or preference for simplicity and elegance, to get naturalism as a conclusion. Maybe hunches or visceral judgments or emotional commitments or a certain aesthetic preference, too.

But I'm unsure how well a naturalist can account for the reliability of any of those additional sorts of thinking with regard to metaphysics. It's not clear to me that their reliability with regards to the realm of ordinary physical things is any sort of guarantee that they'll still get us to the truth when we start thinking about philosophical issues. So on the face of it, it seems to me that naturalism has a real problem here.

I've always been a bit skeptical about the evolutionary argument against naturalism, so I'm very curious to hear what others think about this.

Philothumper said...

Richard Davis: Interestingly, Graham Oppy's The Best Argument Against God crucially depends upon preference for simplicity.

Glenn said...

Maybe reasoning faculties that are good at dealing with metaphysical issues in the natural realm have a spin-off trait of just being fairly good in general, and hence passably good at doing metaphysics related to the non-natural realm. Maybe.

Alexander R Pruss said...


Maybe. But do we have good reason to think so, given naturalism? Suppose I design a natural language AI system designed to answer yes/no English questions about astronomy. It's quite reliable about astronomy. Now you ask it yes/no about microbiology. And it still gives yes/no answers. Do you have much reason to think these answers are true? What reason do we have to think that whatever methods there are for answering astronomy questions work for microbiology?


Rational intuition about the metaphysical realm also seem very different from ordinary intuitions about the physical world.

Alexander R Pruss said...


It is pretty clear to me that if one understands simplicity of a theory in the right way--namely, as the simplicity of the explanatorily fundamental posits--a theistic worldview is much simpler than any naturalistic one.

The explanatorily fundamental posits of a theistic worldview can be expressed very simply:
(*) (Ex)(P)(Perfection(P)->P(x))
I.e.: there is a being that has all perfections. (There are variant formulations, somewhat more complex.)

Everything else is explanatorily posterior.

On the other hand, a naturalistic theory will presumably have as one of its fundamental explanatory posits some law of nature. Any candidate we have for that law of nature is much, much more complex than (*) when properly explained.

Richard Davis said...

Suppose E, F, and G are the faculties we need to rely upon in order to get to naturalism. Moreover, suppose that if we do rely on E, F, and G, then they will get us to a theory according to which E, F, and G reliably generate true beliefs. Now suppose the naturalist holds the view that he doesn't need to know any *explanation* as to why it is proper for him to rely on E, F, and G, in order for it to in fact be proper for him to rely on them. It seems that if this view is correct, then he should be able to rely on E, F, and G in order to get to naturalism in a non-viciously-circular manner.

I think this sort of thing is pertinent to the case of preference for simplicity. If we form our beliefs about the universe in accordance with a preference with simplicity, we'll end up believing that the universe is a simple place. And if the universe is a simple place, then it is the sort of place in which a preference for simplicity tends to get one to the truth. So if we start by trusting our preference for simplicity, even without knowing why we should, then we'll straightforwardly arrive at a reasoned belief that a preference for simplicity is something we should trust.

So, again, why can't the naturalist do this with whatever sorts of reasoning he needs in order to get to naturalism?

Obviously this can't be right for just *any* sort of reasoning. If I consistently reason from the fact that x exists to the conclusion that necessarily x is a pink bunny, I will of course arrive at the result that necessarily everything that exists is a pink bunny. And a world in which that is true is the sort of world in which my proposed pink-bunny reasoning process will reliably get me to true beliefs. It hardly follows that it would be rational for me to accept that process.

Nevertheless, there are very plausibly *some* processes of belief formation --- say, at least a priori intuition and introspection, and probably assent-to-undefeated-seemings as well --- which it is proper for us to rely on even prior to understanding *why* it is proper for us to rely on them. The naturalist can hold that the sorts of reasoning he needs are among those cases, rather than being like the pink-bunny case.

I'm not saying naturalism is correct. (I don't think it is.) But I am suggesting that the evolutionary self-defeat argument doesn't work against naturalism. Rather, the trouble with naturalism is that the sorts of reasoning it ostensibly relies on (preference for simplicity and elegance, logic, intuitions, empirical observations, and so on) *don't* lead to naturalism when consistently applied.

This reply on behalf of the naturalist seems, perhaps, too simple. Thoughts?

Alexander R Pruss said...


It's because of inchoate thoughts like this that I stopped short of saying that if naturalism is true, we are not justified in believing naturalism. I was merely claiming that if naturalism is true, we should not expect to be reliable in getting to claims like this.

What one gathers from this lack of expectation of reliability is a question for the epistemologists.

Richard Davis said...

If I were reasoning solely from the information that my reasoning processes resulted from an evolutionary process aimed at survival and reproduction, I don't see how I could reasonably expect that those reasoning processes would be good at other sufficiently unrelated things, like metaphysics. But that's not all the information I have. I gain an additional piece of pertinent information by first trusting my reasoning processes and then using them to reason about the world. When I do this, I arrive at the surprising conclusion that the metaphysical truths of the world are such that the sorts of reasoning that I need to rely upon in order to survive and reproduce are the same sorts of reasoning such that, if I apply them to questions of metaphysics, are likely to lead me to true beliefs. Reasoning from this new information --- even combined with the assumption of non-theistic macroevolution --- I can then reasonably expect that my reasoning processes are reliable about metaphysics.

So why can't the naturalist do all that? (Other than the fact that the reasoning processes in question do not, in fact, support naturalism?)

Richard Davis said...

Is the argument this: That given all the processes of reasoning that evolution might have conceivably have given us, it seems unlikely (on the assumption of naturalism) that it would have given us either the processes of reasoning we actually have or else some other processes which are also good at arriving at truths about metaphysics? Since it could instead have given us some other reasoning processes that weren't much good at metaphysics?

Alexander R Pruss said...

The surprise is that the reasoning processes lead to metaphysical claims such that if they are true, the processes reliably lead to metaphysical truth?

If so, I think this doesn't work. Let's grant that:
(a) the reasoning processes lead to naturalism.

And of course, if naturalism is true, the reasoning processes that lead to naturalism are reliable insofar as they lead to naturalism. But by itself, that's quite unimpressive. One might as well be impressed by the fact that if I will marry the dark stranger the crystal ball says I will marry, then the crystal ball is reliable in its prediction of whom I will marry.

To be at all impressive, we would have to have an argument that metaphysical conclusions ABC (e.g., naturalism + functionalism), we would expect reliability in metaphysical reasoning going beyond the these conclusions.

But in fact naturalism+evolution gives us no reason to think our metaphysical reasoning is going to be right about anything other than naturalism. For instance, it gives us no reason to think our metaphysical reasoning is going to be right about the nature of time, the persistence of objects, the nature of laws of nature, etc. But if there is only one thing we have confidence of the reasoning getting right, we should be suspicious.

Richard Davis said...

I was going for a stronger claim. First, define

(Reliable) For some structure F,
(i) if the world has F, then the reasoning processes will be reliable with respect to the general class of metaphysical truths (as opposed to being reliable solely with respect to some narrow subset of metaphysical truths); and
(ii) the world does have F.

What sort of structure is F? Well, for instance, a world that has F is a world in which the simplest and most elegant theories are usually true; and in which it tends to be the case that when beings roughly like us who think about metaphysics come into being, they come into being possessing a set of metaphysical intuitions whose contents tend to match the metaphysical truths; and so on for other reasoning processes.

Now suppose

(Self-Support) The reasoning processes lead to Reliable.

(YayNaturalism) The reasoning processes lead to naturalism.

(AgreeableNaturalism) Naturalism is rationally compatible with Reliable, Self-Support, and YayNaturalism.

I think Reliable and Self-Support are, in fact, true. If YayNaturalism and AgreeableNaturalism are also true, then it follows (I think) that we can reasonably expect that even if naturalism is true, then our reasoning processes are reliable about a broad class of metaphysical truths. I personally dispute YayNaturalism, but I think what you're denying is not that but rather AgreeableNaturalism or something close to it.

I'm not sure I understand on what grounds you are denying AgreeableNaturalism. I do understand the analogies to the astronomy program and the crystal ball, but I think the analogies fail for different reasons. The first analogy fails because unlike general reliance upon our basic human processes for gaining truth beliefs, general reliance upon the dictums of the astronomy program has no plausible claim to being properly basic. The second analogy fails because unlike reliance upon the narrow (if accurate) predictions of the crystal ball, general reliance upon our basic human reasoning processes does get us to a broad, impressive claim like Reliable. So unlike in the crystal ball case, there's no narrowness in our result to give cause for suspicion. We could try to broaden the crystal ball case so that it issues dictums about all sorts of issues instead of just the matter of whom I will marry --- but then the crystal ball case just collapses into the case of the astronomy program and the same objection applies.

So again, I take you to be denying AgreeableNaturalism (or something nearby), but I'm not sure I understand how the argument against that thesis goes.

Richard Davis said...

Put another way:

The Evolutionary Self-Defeat Argument against Naturalism seems to go like this:

(1) Naturalism entails that our reasoning processes result from an evolutionary process aimed at survival and reproduction (and not aimed at our being able to reason reliably about metaphysics).
(2) It is probable that if our reasoning processes result from an evolutionary process aimed at survival and reproduction (and not aimed at our being able to reason reliably about metaphysics), then our reasoning processes are not reliable in getting to truths about metaphysics.
(3) So it is probable that if naturalism is true, then our reasoning processes are not reliable in getting to truths about metaphysics.
(4) Necessarily if naturalism is true, then it is a truth about metaphysics.
(5) So it is probable that if naturalism is true, then our reasoning processes are not reliable in getting to naturalism.

I'm saying that I don't understand why the naturalist should feel pressured to accept premise (2). The astronomy-program analogy does not seem to me to work. We can strengthen the argument somewhat by weakening premise (2) by replacing 'probable' with 'not improbable', and making corresponding changes to the other steps. But I'm still not sure the naturalist has much reason to accept (2').

The best reason I can come up why the naturalist should accept (2) or (2') goes like this: The naturalist must accept that reliance on certain reasoning processes is properly basic. But any plausible account of why those reasoning processes are properly basic entails that certain other reasoning processes which disfavor naturalism (say, belief in response to religious experience or experience of the numinous or certain dualistic and/or rationalisic intuitions) are also properly basic. And if those processes which disfavor naturalism are properly basic, then our properly basic reasoning processes do not, in fact, *reliably* get us to naturalism, because they don't get us to naturalism at all. Their ranks are too heavily infiltrated by anti-naturalistic reasoning processes for them to do so. A rational person relies on (or at least begins at the outset of his investigation by relying on) his properly basic reasoning processes. So then the naturalist must accept that the reasoning processes that a rational person relies on do not get us to naturalism reliably (or at all).

This line of reasoning may work against naturalism if it turns out that belief in naturalism is unjustified unless certain reasoning processes (belief in response to religious experience, etc.) are not properly basic. But this line of attack seems a very different sort of argument than the one I took the Self-Defeat argument to be.

Drew said...

Alex, do you think that naturalism is compatible with innate knowledge, or do you think it entails some sort of empiricism? If the latter, it seems hard to believe that any ideas about the world outside sense experience would be reliable in the first place, if intelligible at all.

Even a neo-Kantian view whereby we have a categorical structure of the mind still forces us to ask: Does the categorical structure reflect reality as it is? If not, then the mind's structure seems unreliable about getting to metaphysical truths. If the former, that would be bizarre on naturalism.

Growing in God's Word said...

Hi Dr. Pruss, interesting argument here.
So it seems that even if EAAN fails and P(R/N&E) is high, the reliable cognitive faculties is only reliable in determining truths related to survival and reproduction. The claim here is that the truths required for survival would then be within the natural realm. Hence we must doubt the reliability of our cognitive faculties regarding any metaphysical beliefs and hence we have a defeater for naturalism which is a metaphysical claim. There is no reason to expect our cognitive faculties aimed at truths related to survival and reproduction to also be good at other unrelated things like metaphysics. For example, so what if one does not understand the logic of EAAN or if one doesn't accept the truth of EAAN, how does it relate to survival and reproduction?
I think a possible counter argument would be to pick at the fact that the truths related to survival and reproduction are limited to truths about the natural realm, or that the truths related to survival and reproduction excludes metaphysical truths. I think more needs to be defended and developed here.
Will think about this and use it alongside Plantinga's EAAN in the future! Thanks!