Plantinga’s Evolutionary Argument Against Naturalism refuted

<!– /* Style Definitions */ p.MsoNormal, li.MsoNormal, div.MsoNormal {mso-style-unhide:no; mso-style-qformat:yes; mso-style-parent:””; margin:0cm; margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:12.0pt; mso-bidi-font-size:10.0pt; font-family:”Times New Roman”; mso-fareast-font-family:”Times New Roman”; mso-no-proof:yes;} .MsoChpDefault {mso-style-type:export-only; mso-default-props:yes; font-size:10.0pt; mso-ansi-font-size:10.0pt; mso-bidi-font-size:10.0pt;} @page WordSection1 {size:612.0pt 792.0pt; margin:72.0pt 90.0pt 72.0pt 90.0pt; mso-header-margin:36.0pt; mso-footer-margin:36.0pt; mso-paper-source:0;} div.WordSection1 {page:WordSection1;} –> <!– /* Font Definitions */ @font-face {font-family:”Cambria Math”; panose-1:2 4 5 3 5 4 6 3 2 4; mso-font-charset:0; mso-generic-font-family:auto; mso-font-pitch:variable; mso-font-signature:-536870145 1107305727 0 0 415 0;} /* Style Definitions */ p.MsoNormal, li.MsoNormal, div.MsoNormal {mso-style-unhide:no; mso-style-qformat:yes; mso-style-parent:””; margin:0cm; margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:12.0pt; font-family:”Times New Roman”; mso-fareast-font-family:”Times New Roman”;} p.MsoFootnoteText, li.MsoFootnoteText, div.MsoFootnoteText {mso-style-noshow:yes; mso-style-unhide:no; mso-style-link:”Footnote Text Char”; margin:0cm; margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:12.0pt; font-family:”Times New Roman”; mso-fareast-font-family:”Times New Roman”;} p.MsoFooter, li.MsoFooter, div.MsoFooter {mso-style-noshow:yes; mso-style-unhide:no; mso-style-link:”Footer Char”; margin:0cm; margin-bottom:.0001pt; mso-pagination:widow-orphan; tab-stops:center 216.0pt right 432.0pt; font-size:12.0pt; font-family:”Times New Roman”; mso-fareast-font-family:”Times New Roman”;} span.MsoFootnoteReference {mso-style-noshow:yes; mso-style-unhide:no; vertical-align:super;} a:link, span.MsoHyperlink {mso-style-unhide:no; color:blue; text-decoration:underline; text-underline:single;} a:visited, span.MsoHyperlinkFollowed {mso-style-noshow:yes; mso-style-priority:99; color:purple; mso-themecolor:followedhyperlink; text-decoration:underline; text-underline:single;} span.FooterChar {mso-style-name:”Footer Char”; mso-style-noshow:yes; mso-style-unhide:no; mso-style-locked:yes; mso-style-link:Footer; mso-ansi-font-size:12.0pt; mso-bidi-font-size:12.0pt;} span.FootnoteTextChar {mso-style-name:”Footnote Text Char”; mso-style-noshow:yes; mso-style-unhide:no; mso-style-locked:yes; mso-style-link:”Footnote Text”; mso-ansi-font-size:12.0pt; mso-bidi-font-size:12.0pt;} .MsoChpDefault {mso-style-type:export-only; mso-default-props:yes; font-size:10.0pt; mso-ansi-font-size:10.0pt; mso-bidi-font-size:10.0pt;} /* Page Definitions */ @page {mso-footnote-separator:url(“Macintosh HD:Users:stephenlaw 1:Library:Caches:TemporaryItems:msoclip:0:clip_header.htm”) fs; mso-footnote-continuation-separator:url(“Macintosh HD:Users:stephenlaw 1:Library:Caches:TemporaryItems:msoclip:0:clip_header.htm”) fcs; mso-endnote-separator:url(“Macintosh HD:Users:stephenlaw 1:Library:Caches:TemporaryItems:msoclip:0:clip_header.htm”) es; mso-endnote-continuation-separator:url(“Macintosh HD:Users:stephenlaw 1:Library:Caches:TemporaryItems:msoclip:0:clip_header.htm”) ecs;} @page WordSection1 {size:595.05pt 841.7pt; margin:89.85pt 72.0pt 89.85pt 72.0pt; mso-header-margin:36.0pt; mso-footer-margin:36.0pt; mso-paper-source:0;} div.WordSection1 {page:WordSection1;} –>

Here’s my central criticism of Plantinga’s Evolutionary Argument Against Naturalism (EAAN). It’s novel and was published in Analysis last year.
Here’s the gist. Plantinga argues that if naturalism and evolution are true, then semantic epiphenomenalism is very probably true – that’s to say, the content of our beliefs does not causally impinge on our behaviour. And if semantic properties such as having such-and-such content or being true cannot causally impinge on behaviour, then they cannot be selected for by unguided evolution. But this argument requires, crucially, that there be no conceptual links between belief content and behaviour of a sort that it’s actually very plausible to suppose exist (note that to suppose there are such conceptual links is not necessarily to suppose that content can be exhaustively captured in terms of behaviour or functional role, etc. in the way logical behaviourists or functionalists suppose). It turns out that if such conceptual links exist, then (rather surprisingly!) natural selection will favour true belief even if belief content is epiphenomenal. So, contrary to Plantinga, even if belief content has no causal impact on behaviour, natural selection can still select for true belief. The EAAN is therefore refuted. To resurrect the EAAN, Plantinga would need to show that there no conceptual links of the sort I envisage between content and behaviour, links of a sort that, as I say, do seem to exist.
NATURALISM, EVOLUTION AND TRUE BELIEF
Stephen Law
Plantinga’s evolutionary argument against naturalism (EAAN) is currently one of the most widely discussed arguments targeting philosophical naturalism (see, for example, Beilby 2002).  Plantinga aims to show that naturalism, in combination with evolutionary theory, is, as he puts it, ‘incoherent or self-defeating’. His argument turns crucially on the claim that, in the absence of any God-like being to guide the process, natural selection is unlikely to favour true belief. This, Plantinga supposes, is because natural selection selects only for adaptive behaviour. It is irrelevant, from the point of view of unguided evolution, whether the beliefs that happen to cause that adaptive behaviour are true.
I argue that, even in its most recent incarnation, the EAAN fails. In particular, Plantinga overlooks the fact that adherents of naturalism may hold, seemingly quite plausibly, that there exist certain conceptual links between belief content and behaviour. Given conceptual links of the sort I envisage, natural selection will indeed favour true belief.
I then point out a further interesting, and perhaps somewhat surprising, consequence of the existence of such conceptual links: that even if semantic properties such as being a true belief are epiphenomenal – even if such properties have no causal impact on behaviour – unguided evolution will still favour true belief.
The EAAN
For those unfamiliar with the EAAN, here is a brief outline.[1]Let Naturalism (N) be the view that there’s no such person as God or anything at all like God, and Evolution (E) be the view that our cognitive faculties have come to be by way of the processes postulated by contemporary evolutionary theory. Then, argues Plantinga, the combination N&E; is incoherent or self-defeating. This, he maintains, is because if N&E; is true, then the probability that R – that we have reliable cognitive faculties (that is to say, faculties that produce a preponderance of true over false beliefs in nearby possible worlds) – is low. But, concludes Plantinga, anyone who sees that P(R/N&E;) is low then has an undefeatable defeater both for R and for any belief produced by their cognitive faculties, including their belief that N&E;.
But why suppose P(R/N&E;) is low? Plantinga supports this premise by means of a further argument. He begins by asserting that
materialism or physicalism is de rigeur for naturalism… A belief, presuming there are such things, will be a physical structure of some sort, presumably a neurological structure. (Forthcoming: 2)
According to a proponent of naturalism, then, this structure will have both neurophysiological (NP) properties and semantic properties. However, it is, claims Plantinga, unlikely that the semantic properties of the neurological structure will have any causal effect on behaviour:
It is easy to see how beliefs thus considered can enter the causal chain leading to behavior; current science gives us a reasonably plausible account of the process whereby volleys of impulses propagated along the efferent nerves cause muscle contraction, motor output, and thus behavior. It is exceedingly difficult to see, however, how they can enter that chain by virtue of their content. A given belief, it seems, would have had the same causal impact on behavior if it had had the same NP properties, but different content. (Forthcoming: 2-3)
Plantinga concludes that N&E; makes semantic epiphenomenalism (SE) likely. But, says Plantinga, if semantic properties such as having such-and-such content or being true cannot causally impinge on behaviour, then they cannot be selected for by unguided evolution. Given SE, truth and falsehood will be, as Plantinga puts it, invisibleto natural selection. In which case, (on the modest assumptions that (i) 75% of beliefs produced must be true in order for a cognitive mechanism to be reliable and, (ii) that we have at least 100 such beliefs) P(R/N&E;&SE;) will be low.
So runs the EAAN. Recently, Plantinga has refined the argument by trying to tackle a certain sort of objection. The objection is that by also embracing, for example, reductive materialism (RM), adherents of naturalism may, after all, quite reasonably suppose that they have evolved reliable cognitive faculties. Why so? Well, on Plantinga’s understanding of RM, content properties just are NP properties. But then, because NP properties cause behaviour, and semantic properties just are NP properties, so semantic properties can cause behaviour. And if semantic properties can cause behaviour, then they can, after all, be selected for by unguided evolution.
Plantinga’s argument that P(R/N&E;&RM;) is low
In his most recent presentation of the EAAN, Plantinga attempts to deal with the above objection. He focuses his attention on one semantic property in particular – truth. Even supposing that semantic properties such as being true can causally affect behaviour, why, he asks, should we suppose, that unguided evolution favour beliefs that are true?
According to Plantinga, the combination N&E;&RM; gives us no reason to suppose that the content of belief/neural structures resulting in adaptive behaviour is likely to be true. Suppose the belief/neural structure resulting in a piece of adaptive behaviour has the content q. While the property of having q as content does now enter into the causal chain leading to that behaviour, it doesn’t matter whether q is true:
What matters is only that the NP property in question cause adaptive behaviour; whether the content it constitutes is also true is simply irrelevant. It can do its job of causing adaptive behaviour just as well if it is false as if it is true. It might be true, and it might be false; it doesn’t matter. (Forthcoming:10).
But if the NP property can do its job of causing adaptive behaviour just as well whether the content is true or false, true belief cannot be favoured by natural selection. In which case, concludes Plantinga, (PR/N&E;&RM;) remains low.
Conceptual constraints on likely semantic content
There is, it seems to me, a fatal flaw in even this latest incarnation of the EAAN.
Plantinga supposes that what unguided evolution favours, in the first instance, is adaptive behaviour. As to what causes that behaviour, evolution doesn’t care. True beliefs, false beliefs, something else – it’s all the same to evolution. It is only the result – adaptive behaviour – that is preferred.
But even if unguided evolution doesn’t care what causes adaptive behaviour, just so long as it is caused, it may not follow, given certain further facts about be

lief that natural selection won’t also favour true belief.

Consider the suggestion that there exist certain conceptual constraints on what content a given belief can, or is likely to, have given its causal relationships to, among other things, behaviour. My claim is that, given the existence of certain conceptual constraints, unguided evolution will then tend to favour true belief.
To begin, let me sketch out a simple illustration of how such constraints might operate.  Suppose we just stipulatively introduce certain terms/concepts. Let’s say that a subject’s belief state has content MC1 iff that state has properties achieving a threshold of at least 30 points, with points allocated thus:
Property A     +20 points
Property B      +15 points.
Property C      +20 points
Property D     -12 points
Notice there’s no one property possession of which is essential if a state is to qualify as having the content MC1. Suppose we similarly stipulate that a subject’s belief state has content MC2iff that state possesses properties achieving a threshold of at least 30 points, with points allocated thus:
Property D     +20 points
Property E      +15 points
Property F      +20 points
Property A     -12 points
Note that if a subject has a belief state with properties A and B, then, ceteris paribus, that state is rather more likely to have the content MC1 than it is the content MC2 (though it might yet turn out to lack content MC1 and possess content MC2 instead if it also possesses properties D, E and F while lacking C). Now suppose that while not all these properties involve causal links to behaviour, some do, namely A, C, D and F. Property A is that of causing behaviour B1 in situation S1, C that of causing behaviour B2 in situation S2, D that of causing behaviour B3 in situation S3, and F that of causing behaviour B4 in situation S4.
Having introduced these conceptual constraints on what it is to have beliefs with the contents MC1 and MC2, we can now see how natural selection might select not only for or against certain behaviours in certain situations, but also for or against these two belief contents. Suppose that exhibiting B1 in S1 and B2 in S2 is in each case adaptive, while exhibiting B3 in S3 or B4 in S4 is maladaptive. Then, other things being equal, natural selection will tend to favour subjects holding beliefs with content BC1 over those holding beliefs with content BC2. So, given conceptual constraints on belief content of the sort outlined above, natural selection need not be blind to belief content. It will select for some contents over others, depending on the kinds of behavioural output with which they are conceptually associated.
So now suppose that constraints of this sort exist on the content of beliefs of the sort with which we are already familiar – contents such as that there is water five miles south, that Paris is the capital of France, and so on. Suppose these constraints conceptually link content with behavioural output. No doubt these constraints will be more complex than in my illustration. But, supposing they exist, with what sort of behaviour is a given content likely to be conceptually linked?
Suppose that, solely in combination with a very strong desire for water, a certain belief/neural structure typically results in a subject walking five miles to the south. Surely, if there are such conceptual links between behaviour and content, then the property of causing that behaviour in that situation will be among those properties lending, as it were, a considerable number of points towards that belief/neural structure achieving the threshold for having the content that there’s water five miles south. Other things being equal, that belief/neural structure is much more likely to have the content that there’s water five miles south than it is, say, the content that there’s isn’t water five miles south, or that there’s water five miles north, or that there’s a mountain of dung five miles south, or that Paris is the capital of Bolivia. Perhaps the belief/neural structure in question might yet turn out to have one of these other contents. We can know a priori, solely on the basis of conceptual reflection, that, ceteris paribus, the fact that a belief/neural structure causes that behaviour in that situation significantly raises the probability that it has the content there’s water five miles south. Among the various candidates for being the semantic content of the belief/neural structure in question, the content that there’s water five miles south will rank fairly high on the list.
But now notice that, given such conceptual constraints exist, unguided evolution will indeed favour true belief. Consider our thirsty human. He has a strong desire for water. He’ll survive only if he walks five miles south to where the only reachable water is located. He does so and survives. Suppose this adaptive behaviour is caused by a certain belief/neural structure. If there are conceptual constraints on belief content of the sort I envisage, and if a belief/neural structure in that situation typically causes subjects to walk five miles south, then it is quite likely to have the content that there’s water five miles south – a true belief. Were our thirsty human to head off north, on the other hand, as a result of his having a belief/neural structure that, in that situation, typically causes subjects to walk five miles north, then it’s rather more likely that the belief in question is that there’s water five miles north. That’s a false belief. Because it is false, our human will die.
So if beliefs/neural structures cause behaviour,

and if there are conceptual constraints linking content with behavioural output of the sort I am suggesting, then natural selection won’t just favour adaptive behaviour. It will also favour true belief.

True, there are other candidates for being the content of the belief that causes our human to head off in the right direction. Perhaps some are more likely candidates. Suppose our human has no conception of miles or south. Then, instead of the belief that causes his behaviour having the content that there’s water five miles south being, perhaps it has instead the content that there’s reachable water thataway. However, notice that, either way, the content of the belief in question is still true.
To sum up: what Plantinga overlooks, it seems to me, is the possibility that there exist conceptual constraints on content of the sort outlined here. The suggestion is that if beliefs are neural structures, then it is at least partly by virtue of its having certain sorts of behavioural consequence that a given neural structure will have the content it does. If such constraints exist, then one cannot, as it were, plug any old belief content into any old neural structure, irrespective of that structure’s behavioural output. We run up against certain conceptual obstacles. If such conceptual constraints exist, it appears natural selection will favour not only adaptive behaviour, but also true belief.
Neither materialism nor functionalism not presupposed
Note that to suggest that such conceptual constraints on belief content exist is not, of course, to presuppose that beliefs are neural structures or that materialism is true. Let’s suppose, for the sake of argument, that substance dualism is true and that beliefs are not neural structures, but soul-stuff structures. Then my suggestion is that we may be able to know on the basis of a little conceptual reflection that if beliefs are soul-stuff structures, and if a given soul-stuff structure in combination with a strong desire for water typically results in subjects walking five miles south, then ceteris paribus that soul-stuff structure is quite likely to have the content that there’s water five miles south, and is rather unlikely to have the content that there’s water five miles north.
Also note that to suggest that there exist conceptual constraints on content given behavioural output is not to presuppose the truth of some reductionist, materialist-friendly theory of content of the sort that Plantinga has gone on to attack[2], such as Dretskian indicator semantics or functionalism. Perhaps belief contents cannot be exhaustively characterized in terms of their causal connections to input and output, as some functionalists claim. That’s not to say that there are no conceptual constraints at all on what the content of a given belief is likely to be, given the causal links that belief has to behaviour. Perhaps there are. Consider my illustration involving contents MC1 and MC2. I stipulated that not all of the weighted properties involved causal connections with behavioural output. Properties B and E involved no such connections. Indeed, B and E might even be properties presenting an insurmountable obstacle to any attempt to characterize the content of MC1 and MC2 in wholly functionalist terms. It wouldn’t follow that there are no conceptual constraints at all on beliefs having content MC1 and MC2 given their behavioural output. Clearly there are.
So, while the combination N&E;&RM; might be self-defeating, it seems that the addition of CC – the thought that there are conceptual constraints on content of the sort I envisage – produces a combination of beliefs that is not, after all, self-defeating. It appears there are ways of embracing naturalism that sidestep Plantinga’s charge of incoherence.
How natural selection can still favour true belief even if SE is true
In fact, it turns out that in order to sidestep Plantinga’s charge of incoherence our naturalist doesn’t even have to sign up to RM. The addition of CC to R&E; aloneis sufficient to rescue naturalism from self-defeat, as I’ll now explain.
As we saw above, Plantinga’s initial worry about naturalism is that it makes semantic epiphenomenalism (SE) likely. He supposes the naturalist will hold that beliefs will be neural structures possessing both neurophysiological (NP) properties and semantic properties. However, Plantinga thinks that only the NP properties of those structures will then have any causal effect behaviour. A given belief would have the same causal impact on behaviour if it had the same NP properties but different semantic properties (or indeed no semantic properties at all).
So now let’s suppose our naturalist actually bites the bullet and accepts SE – they actually accept that the semantic properties of a given neurological structure have no causal impact on behaviour. Plantinga supposes such a naturalist is then compelled to accept that, because natural selection can only select for adaptive behaviour and the properties that cause it, so natural selection cannot select for the semantic property of being true. However, it turns out that Plantinga’s assumption that natural selection favours only adaptive behaviour and the properties that cause it is unwarranted. It turns out, somewhat surprisingly, that, given CC, natural selection will still favour true belief even if the property of being a true belief has no causal impact on behaviour.
To see why, let’s return again to our thirsty human. He has a certain belief/neural structure that, in conjunction his strong desire for water, causes him to walk five miles south. Given the kind of conceptual constraints outlined above, a belief/neural structure that causes a subject to walk five miles south given a strong desire for water will quite probably have the content there’s water five miles south. Notice it really doesn’t matter whether or not that belief/neural structure causes that behaviour by virtue of its having that semantic property. It remains the case that, if that sort of neural structure for whatever reason has that behavioural consequence, then, given CC, it quite probably has the conten

t there’s water five miles south and probably doesn’t have the conceptual content there’s water five miles north. It matters not whether SE is true: the behavioural output of a belief/neural structure still places constraints on its likely content.

But then, given such conceptual constraints, natural selection is likely to favour true belief even if SE is true. Odd though it might seem, given CC, natural selection will favour true belief even if the property of being a true belief has no causal impact on behaviour. This is a rather significant discovery, even setting aside its relevance to Plantinga’s EAAN.
Conclusion
Of course, I am merely making a suggestion. Perhaps there exist no such conceptual constraints on belief content of the sort I envisage. Still, the view that there are such constraints on content is widespread (it is by no means restricted to those wedded to some form of logical behaviourism or functionalism, for example). It seems intuitively obvious to many of us that belief content is not entirely conceptually independent of behavioural output: that one cannot plug any old belief content into any old neural structure (or soul-stuff structure, or whatever) entirely independently of its behavioural output. That intuition would appear to be, philosophically speaking, largely pre-theoretical. It cannot easily be dismissed by Plantinga as a product of some prior theoretical bias towards naturalism and/or materialism.
My central conclusion, then, is this. Plantinga has not shown that naturalism in combination with the theory of evolution is unavoidably self-defeating. It appears that an adherent of N&E; who also supposes CC is true can, after all, quite reasonably suppose they have evolved reliable cognitive faculties.
In response, Plantinga might now try to show that if naturalism is true, there are unlikely to be conceptual constraints on semantic content of the sort I describe. Perhaps he can do this. If so, then the EAAN might be resurrected. But as things stand, it is not naturalism that is defeated, but the EAAN.[3]
Heythrop College, University of London
London W8 5HN
References
Beilby, J. (ed) 2002. Naturalism Defeated? Essays on Plantinga’s Evolutionary Argument Against Naturalism. Ithaca: Cornell University Press.
Plantinga, A. Forthcoming. Content and Natural Selection. Philosophy and Phenomenological Research. Currently available on-line at Plantinga’s departmental webpage: http://philosophy.nd.edu/people/all/profiles/plantinga-alvin/documents/CONTENTANDNATURALSELECTION.pdf
Page numbers refer to the on-line version.


[1] I here follow the most recent version of the EAAN as presented in Plantinga (Forthcoming).
[2] See the latter part of Plantinga (Forthcoming).
[3] My thanks to Alvin Plantinga for his generous comments on earlier drafts.