Natural Selection Does Care about
Truth
Maarten Boudry and Michael Vlerick

True beliefs are better guides to the world than false ones. This is the common-sense

assumption that undergirds theorizing in evolutionary epistemology. According to Alvin

Plantinga, however, evolution by natural selection does not care about truth: it cares

only about fitness. If our cognitive faculties are the products of blind evolution, we have

no reason to trust them, anytime or anywhere. Evolutionary naturalism, consequently,

is a self-defeating position. Following up on earlier objections, we uncover three additional

flaws in Plantinga’s latest formulation of his argument: a failure to appreciate adaptive

path dependency, an incoherent conception of content ascription, and a conflation of

common-sense and scientific beliefs, which we diagnose as the ‘foundationalist fallacy’.

More fundamentally, Plantinga’s reductive formalism with respect to the issue of cognitive

reliability is inadequate to deal with relevant empirical details.

1. Introduction

True beliefs are better guides to the world than false ones. This is the common-sense

assumption that undergirds theorizing in evolutionary epistemology. An early

hominid who believes that tigers are dangerous predators, and acts upon that belief,

will fare better (in terms of survival and reproduction) than one who believes that

they are docile pets. Given that our brain has evolved to aid our survival and repro-

duction (as any biological organ), we can expect it to be geared towards churning

out true beliefs. False beliefs, or cognitive mechanisms that tend to produce such

beliefs, are mostly harmful and will be weeded out by natural selection.

CISP915651 Techset Composition India (P) Ltd., Bangalore and Chennai, India 5/8/2014

C
E

:
K

R
R

Q
A

:
C

o
ll

:

Maarten Boudry is at the Department of Philosophy and Moral Sciences, Ghent University. Correspondence to:

Vakgroep Wijsbegeerte en Moraalwetenschap, Universiteit Gent, Sint-Pietersnieuwstraat 49, 9000 Gent, Belgium.

E-mail: maartenboudry@gmail.com. Michael Vlerick is at the Department of Philosophy, University of Johannes-

burg. Correspondence to: Department of Philosophy, University of Johannesburg, PO Box 524, Auckland Park

2006, South Africa. E-mail: michaelvlerick@gmail.com

International Studies in the Philosophy of Science, 2014
Vol. 28, No. 1, 65 – 77, http://dx.doi.org/10.1080/02698595.2014.915651

© 2014 Open Society Foundation

5

10

15

20

25

30

35

40



According to a controversial argument developed by Alvin Plantinga (1993, 2000),

this assumption is false. If our cognitive faculties evolved by natural selection, he

argues, we have no reason to trust their deliverances. Natural selection does

not care about truth, but only about fitness. If evolution is blind and unguided,

then there is no guarantee at all that it will produce organisms with reliable, truth-

conducive cognitive faculties.

Naturalism and evolutionary theory, Plantinga contends, form a dangerous package

deal. Roughly, naturalism can be construed as the claim that no gods or supernatural

beings exist, and evolutionary naturalism as the standard scientific view of unguided

evolution through purely natural mechanisms. If both are true, we cannot expect our

cognitive faculties to be reliable and (the bulk of ) our beliefs to be true. In the end, the

evolutionary naturalist must fall into an abyss of scepticism, acquiring a defeater for all

of his beliefs, including his very belief in evolutionary naturalism.

Earlier critics (Fales 1996, 2002) have defended the common-sense view, against

Plantinga, that true beliefs are generally better guides in the world than false ones,

barring contrived exceptions, and that our expensive cognitive machinery is unlikely

to have evolved without conferring some adaptive benefits. Stephen Law (2012) has

argued that there is a strong conceptual link between beliefs and behaviour, which

makes beliefs ‘visible’ to natural selection. Natural selection weeds out beliefs that

are not conducive to an organism’s survival and reproduction, among which (often

enough) false beliefs.

Plantinga’s latest formulation of his evolutionary argument against naturalism

(EAAN) does not differ much from earlier versions, but suffers from three additional

and important flaws, which are relevant to the constructive project of evolutionary

epistemology: a failure to appreciate adaptive path dependency, an incoherent

conception of content ascription, and a conflation of common sense and scientific

beliefs. More fundamentally, Plantinga’s rigid formalist approach to the issue of

cognitive reliability fails to deal with relevant empirical findings.

2. Stark Choices

Before documenting the three flaws in Plantinga’s argument, we should say a few

words about the way he frames the issue of cognitive reliability. Plantinga forces us

to make a stark choice between two mutually exclusive propositions, to wit R (‘our

cognitive faculties are reliable’) or its negation, not-R (‘our cognitive faculties are

not reliable’). Having framed the discussion in those dichotomous terms, he then

proceeds to determine the value of P (R | N & E), where N is naturalism and E is
evolutionary theory. Serious evolutionary epistemology, however, does not frame

the reliability of human cognition in such stark, black-or-white terms. If only it

were that simple. Relevant empirical complications include the structure of the

ancestral environment, various design trade-offs involving production costs and

error rates, the specific adaptive problem at hand, the specific selection pressures

and developmental constraints at work, etc. For example, evolution may endow us

with ‘fast and frugal’ heuristics (Gigerenzer 2008) that sacrifice some accuracy for

66 M. Boudry and M. Vlerick

45

50

55

60

65

70

75

80

85



efficiency and speed; evolved cognitive mechanisms may produce occasional false

belief as a result of asymmetries in the costs associated with different errors (Haselton

and Buss 2000); cognitive mechanisms may be accurate or truth-conducive in adap-

tively familiar environments but lead us astray in new or modern environments; adap-

tive beliefs may be roughly accurate approximations of the world though limited in

other respects. Evolution may select true belief in some domains, but not necessarily

in others (De Cruz et al. 2011; Wilkins and Griffiths 2012). However, misbelief that is

directly adaptive turns out to be hard to find (McKay and Dennett 2009). In general,

the standard view of evolutionary epistemology was neatly summarized in a seminal

paper by Ryan McKay and Daniel Denett on the possibility of adaptive misbelief:

‘Although survival is the only hard currency of natural selection, the exchange rate

with truth is likely to be fair in most circumstances’ (McKay and Dennett 2009, 509).

The strand of reductive formalism that Plantinga promotes is—to put it mildly—

unlikely to yield a fruitful philosophical discussion, because it sweeps many empirical

details and qualifications under the carpet. Many philosophical commentators, defen-

ders of Plantinga and critics alike, have indulged in his reductionist framework,

instead of paying attention to the actual empirical work that is being carried out at

the moment. In James Beilby’s volume devoted to the EAAN (Beilby 2002), most con-

tributors (see Talbott, O’Connor, Bergmann, Sosa, and others) accept Plantinga’s

black-or-white distinction between R and not-R. They even grant the low value of P

(R | N & E), for the sake of the argument, and prefer to quarrel over various probabil-
ities, the exact formalization of the argument, and a whole network of defeaters and

defeater-deflecting defeaters (Evan Fales’s chapter is a notable exception). But that is

to give the game away right from the start. The only apposite response to Plantinga’s

formal dichotomy is the one voiced by Geoff Childers, in another critical paper on the

EAAN: ‘Such cartoonish skepticism arises only as a byproduct of forcing ourselves to

think in artificially rigid black and white categories, while ignoring our best current

naturalistic models of the biological and social evolution of cognition’ (Childers

2011, 201).

3. Evolutionary Scepticism

In this article, we will not accept Plantinga’s ‘cartoonish’ framing, but instead focus on

a number of specific and important flaws underlying his EAAN, which have not been

covered yet in the literature and which are relevant for the project of evolutionary

epistemology. Helen De Cruz et al. (2011) have distinguished between evolutionary

supportive arguments (ESA), which ground the truth-tracking abilities of our cogni-

tive faculties in their evolutionary origins, and evolutionary debunking arguments

(EDA), which try to sever or block at least one of the steps in the supportive ESA.

The general framework of the ESA runs roughly like this:

(1) Organisms need to interact with the environment to survive and reproduce

(premise)

(2) An organism’s beliefs guide its interaction with the world (premise)

International Studies in the Philosophy of Science 67

90

95

100

105

110

115

120

125

130



(3) There is a correlation between an organism’s fitness (likelihood of survival and

reproduction) and its beliefs (from 1 and 2)

(4) True beliefs are overall better guides to action in the world than false beliefs

(premise, see above for subtle empirical qualifications)

(5) Beliefs have a neural substrate (premise)

(6) Natural selection will weed out neural structures that give rise to false beliefs

(from 3, 4, and 5)

(7) Natural selection is the main driving force of evolution (premise)

(8) Evolution will produce organisms with cognitive systems that produce (overall)

true belief (from 6 and 7)

(9) Our belief in evolutionary naturalism is justified

One can distinguish different types of EDA depending on which of these steps in the

ESA are being blocked or challenged. Some philosophers have attacked step 7 in par-

ticular (Stich 1990). Plantinga’s case is more focused around step 4, although he has

taken issue with other steps as well. According to the EAAN, evolution by natural

selection does not care about truth; it cares only about taking adaptive courses of

action. There is no reason to think that we need true belief to achieve this. Maybe

our beliefs are causally inert, or they are mere by-products of adaptive behaviour,

or maybe false beliefs can be adaptive.

In the latest formulation of his argument, Plantinga commits a number of flaws that

are relevant for the project of evolutionary epistemology. The first reasoning error dis-

cussed in this article consists of an attempt to block step 4, whereas the second targets

step 5. The third flaw concerns step 9. Even if the ESA is not airtight up to step 8,

meaning that our cognitive systems are not completely reliable, we can still have con-

fidence in our belief in evolutionary naturalism. Plantinga’s conflation of different

sorts of belief is described as the ‘foundationalist fallacy’.

4. Adaptive Path Dependency

Does survival really require true or accurate belief? Many organisms, Plantinga points

out, seem to do perfectly fine without it. For example, most primitive organisms have

simple neural structures processing input from sense organs. These ‘indicators’ may be

adaptive, if they are correlated with specific courses of action (e.g. fleeing in the pres-

ence of a predator), but they need not involve belief states: ‘indication of this sort does

not require belief. In particular, it does not require belief having to do with the state of

affairs indicated’ (Plantinga 2011b, 328). It is true, of course, that survival does not

require belief: bacteria and other lower organisms thrive well without it. Even if organ-

isms have evolved neural architecture, their cognitive mechanisms may be relatively

simple and not require full-blown belief. But Plantinga’s turn of phrase reveals a

subtle error: ‘Fleeing predators, finding food and mates—these things require cogni-

tive devices that in some way track crucial features of the environment, and are appro-

priately connected with muscles; but they do not require true belief, or even belief at

all’ (Plantinga 2011b, 329; our italics).

68 M. Boudry and M. Vlerick

135

140

145

150

155

160

165

170

175



Plantinga equivocates between two different claims: (1) biological fitness does not

require beliefs (and hence a fortiori no true beliefs); (2) given that an organism has

action-directing beliefs, those beliefs need not be true. The second claim, however,

does not follow from the first, because it ignores adaptive path dependency (Boudry

2013). Though in general survival and reproduction need not require (full-blown)

belief, once evolution has gone down the road of belief generation as a means of

directing action, it better be producing beliefs—or so the evolutionary naturalist

claims—that are at least approximately true in ecologically relevant situations (and

this is where the empirical details come in).
1

The existence of creatures who thrive

without the help of beliefs at all is irrelevant to the claim, defended by the evolutionary

naturalists, that action-directing beliefs, once adopted, are adaptive if true and harmful

if false.

Adaptive path dependency is rampant in the living world. For example, most organ-

isms do not rely on powered flight to survive, so they can do without wings. Most birds

are dependent on wings, however, because their ancestors have ‘invested’ in the strat-

egy of flight. The adaptive value of wings is not diminished by the fact that most other

organisms make a good living without them. Early choices limit later options.

5. Arbitrary Content Labelling

If an organism has invested in neural machinery for action-directing beliefs, those

should better be approximately true beliefs (Fales 2002). Is the link between true

belief and fitness then secured? Not according to Plantinga, who argues that we still

have to bridge the gap between the neurophysiological (NP) properties of beliefs

and their contents: ‘The neurology causes adaptive behavior and also causes or deter-

mines belief content: but there is no reason to suppose that the belief content thus

determined is true’ (Plantinga 2011b, 327). For the evolutionary naturalist,

however, the content of a belief is associated with certain NP properties.
2

A belief

with a different content is linked with different physical properties (i.e. a different

physical realization in the brain), and will thus elicit different kinds of actions,

some of which will be fitness enhancing. Natural selection selects those beliefs that

result in adaptive courses of action.

There need not even be a causal link between belief content and behaviour. As Law

(2012) has shown, constraints on belief content obtain even if (semantic) epipheno-

menalism or substance dualism were true. The link obtaining between belief

content and behaviour is entirely conceptual, not dependent on any specific reduc-

tionist account of meaning. For example, for a dualist

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. (Law
2012, 46)

3

As long as NP states are associated with specific belief contents, either causally or con-

ceptually, belief content is within the reach of natural selection, and Plantinga’s EAAN

International Studies in the Philosophy of Science 69

180

185

190

195

200

205

210

215

220



fails. The precise relation between belief content and NP properties is irrelevant to

Plantinga’s argument.

Plantinga objects to this conceptual link between NP properties and content, using

an intriguing counterfactual argument that involves a second flaw:

It is exceedingly difficult to see, however, how they [beliefs] 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.
(Plantinga 2011a, 436; see also Plantinga 2011b, 335 – 337)

If there is a conceptual link between NP properties and belief content, however, the

question in virtue of which property beliefs cause certain courses of action becomes

moot, as far as the visibility for natural selection is concerned. Because of the concep-

tual links, evolution will select for belief contents even if they are causally inert. Plan-

tinga implicitly denies this conceptual link, however, as he asks us to imagine a NP

state with identical causal powers but ‘quite different content’ (Plantinga 2002,

214). If one accepts that beliefs supervene on NP properties, however, this will be

ruled out. For the physicalist, it is simply impossible that there can be difference in

content without difference in NP properties.

In a telling passage, Plantinga (2002, 337) claims that the NP properties of a con-

crete belief stand to its content as a ball shattering a window stands to its being a birth-

day present. The content is thus completely irrelevant. Does a ball break a window ‘by

virtue of being a birthday present’, asks Plantinga rhetorically? Of course not: some

balls are not birthday presents, and some birthday presents are not balls but rather,

say, stuffed animals, which usually do not shatter windows. Unlike belief contents

and NP properties, the property of being a birthday present is not correlated to its

breaking force and is thus irrelevant.

Plantinga’s counterfactual argument assumes that any belief content can be associ-

ated with any NP state, as if it were an arbitrary label: ‘would it not produce the same

physical effect, the same effect on nerves and muscles and glands, even if it had quite

different content?’ (Plantinga 2002, 214). Elsewhere, he writes that ‘those beliefs need

not be so much as about the objects involved in the state of affairs causing the subven-

ing properties. They could be anything’ (Plantinga and Tooley 2008, 232).

This view—let’s call it arbitrary content labelling—is much stronger (and stranger)

than semantic epiphenomenalism. Not only does it deprive belief contents of any

causal powers, but also it destroys any conceptual link with NP properties. Just as

any object can assume the role of a birthday present (though not necessarily a good

one), any NP structure can have the content ‘there is water five miles north’. To see

what Plantinga’s view would amounts to, consider an old-fashioned calculator.

When I press some keys, numbers appear on the screen. Now suppose that Plantinga

would come along and say that ‘the calculator would produce the same result if the

keys had different meaning’. By way of demonstration, he removes the labels from

keys 3 and 4 and switches them: ‘see, the calculator still produces the correct

results’. But it should be obvious that the inscription or label on the key is not

where meaning resides. The pressure applied to the key relays a signal on an electronic

70 M. Boudry and M. Vlerick

225

230

235

240

245

250

255

260



circuit, regardless of the inscriptions on the key. The key represents number 3 in virtue

of the fact that it is connected with the device in a certain manner, such that it pro-

duces the expected results on the screen when pressed in combination with other

keys. The manufacturer has just conveniently arranged my keypad in such a way

that the inscriptions correspond to the internal representations, in this case

numbers and mathematical concepts. If somebody secretly swapped the key labels,

this will soon be discovered (‘this key is not “3” at all!’).

Believing X means being prepared to take certain courses of action, when conjoined

with a set of desires and other beliefs. As William Clifford already wrote: ‘no belief is

real unless it guides our actions, and those very actions supply a test of its truth’ (Clif-

ford [1877] 1999, 82). If some NP property does not dispose us, ceteris paribus, to the

behaviour that we would expect from some belief state X, then it cannot be that belief

state X. For instance, if the belief ‘this snake is dangerous’ does not produce flight be-

haviour in the presence of a snake, when coupled with the desire ‘I don’t want to get

bitten by a poisonous snake’, and with other background beliefs, such as ‘if I run away

from X, then X is less likely to hurt me’, then something must have gone wrong in our

belief ascription.
4

Belief contents are not arbitrary labels that one can tag to NP prop-

erties at will, as Plantinga supposes. The counterfactual claim that some NP property

would have the same effect ‘even if it had quite different content’, therefore, is

incoherent.
5

Is that all there is to be said about Plantinga’s severing of behaviour and semantics?

No, because on other occasions, Plantinga slides from a counterfactual argument

about content, which suffers from incoherence, into a superficially similar counterfac-

tual argument centring on truth-values instead of belief contents. In this version, he

accepts a form of reductive materialism, which simply identifies certain NP properties

with belief contents. This ensures that evolution can track adaptive beliefs. Fair

enough, argues Plantinga, but does this give us any confidence that such beliefs are

also true?

Take a neurophysiological property P that is in fact adaptive, and is also identical
with the property of having content p: so far as adaptivity goes, it doesn’t in
general matter whether p is true or false. P would have the same adaptive effects,
one thinks, if p were false. How P contributes to motor output doesn’t depend
(except in special cases) upon the truth-value of p. (Plantinga 2002, 218)

In this version, Plantinga accepts the disputed conceptual link between content and

NP properties: ‘The neurology causes adaptive behavior and also causes or determines

belief content: but there is no reason to suppose that the belief content thus deter-

mined is true’ (Plantinga 2011b, 327). It is important to see that this is a different argu-

ment from the one above about belief content. Indeed, this argument is hardly

incoherent: a given NP property may instantiate a belief that is either true or false,

depending on the state of the world. The belief of an early hominid that ‘this mush-

room is not poisonous’ has that particular content in virtue of his neurological state

(within a particular environment). The NP properties of p will result in the same type

of behaviour (eating that tasty mushroom), regardless of the truth-value of p. By

themselves, the NP properties of p do not give us any information about the truth-

International Studies in the Philosophy of Science 71

265

270

275

280

285

290

295

300

305



value of p. If p is false, however, because the organism failed to identify the type of

mushroom, the NP properties linked to p will fail to be adaptive, pace Plantinga.
6

The link that the evolutionary naturalist seeks to secure is not between truth and

course of action, but between truth and adaptiveness. There are two connections at

play: the straightforward one between belief content and course of action, and the

higher-order connection between truth and adaptiveness. Plantinga muddies the

waters by mixing up these two levels, alternately asking what would happen if a

certain NP property had a different content, and what if it had different truth-value.

The two different links need to be kept apart: (1) The NP properties of my belief

‘there is a bear in the cave’ is linked with my disposition not to enter it, regardless

of whether that belief is true. (2) The adaptiveness of the NP properties that

dispose me not to enter the cave, under those circumstances, depends on the state

of the world (the bear’s presence), regardless of the semantic content of those NP prop-

erties. By taking either of those links in isolation, and switching back and forth

between them, Plantinga seems to undermine the confidence of the evolutionary nat-

uralist. There always seems to be a ‘missing link’. But appearances are misleading.

Taken together, however, those claims provide all the necessary chains to establish a

link between fitness and truth.

6. The Foundationalist Fallacy

Remember that Plantinga ends up rejecting step 8 of the ESA, by blocking one of the

earlier steps:

(8∗) Evolution will NOT produce organisms with cognitive systems that produce
(overall) true belief

From this assessment of our ‘naked’ and unassisted cognitive faculties, given evol-

utionary naturalism, Plantinga immediately jumps to the conclusion that scientific

beliefs—notably, belief in naturalistic evolution—must also be unreliable. In its

basic form, therefore, the punch line of the EAAN goes as follows:

(1) Given evolutionary naturalism, our belief producing cognitive faculties are

unreliable

(2) Evolutionary naturalism is itself a product of our cognitive faculties

(3) Conclusion: evolutionary naturalism is self-defeating

Indeed, Plantinga argues,

science, clearly, is an extension of our ordinary ways of learning about the world. As
such, it obviously involves the faculties and processes by which we ordinarily do
achieve knowledge. . . . For science to be successful, therefore, there must be a
match between our cognitive faculties and the world. (Plantinga 2011b, 269 – 270).

Once again, Plantinga sets up a stark dichotomy: either all our beliefs are reliable or

none of them is, including scientific beliefs. More generally, he conflates beliefs that

are the direct product or our naked, unassisted mind, and beliefs that arise out of

higher-order cognitive faculties (systematic theorizing and evidence-gathering).

72 M. Boudry and M. Vlerick

310

315

320

325

330

335

340

345

350



Underlying this argument is the assumption, which we call the ‘foundationalist fallacy’,

that if the foundations of our cognitive endeavours are not completely secure, then the

whole edifice built on top of it must collapse. If the ground is even a little shaky, no

amount of scaffolding will be sufficient.

As John Clendinnen (1989) has argued, however, epistemic confidence can be

gradually raised as evidence accumulates. Consider the triplet of propositions ‘If P

then Q’, ‘If Q then R’, ‘If R then P’. In a deductive model of justification, it is often

incorrectly assumed that each step is either completely justified or not justified at

all. However, it is possible to have an interdependence of justifications (P, Q, R) in

which we start out by an initially very weak and provisional version of P, whereby

our confidence is gradually raised through Q and R. In this regard, by starting from

the deliverances of our bare and unassisted mind, including folk logic and basic per-

ceptual capacities, we may gradually increase the reliability of our beliefs, by accumu-

lating evidence and thus bootstrapping or ‘scaffolding’ (Sterelny 2010) our way to

more justified beliefs. Therefore, even if we grant that step 8 of the ESA is not valid

across the board (see the empirical complications above), the EAAN’s conclusion

does not follow. Indeed, in his rigidly foundationalist approach, Plantinga leaves no

room for such gradual bootstrapping operations.

Let us have a closer look at the belief in evolutionary naturalism. The theory of evol-

ution by natural selection is by no means the product of our bare, unassisted mind. It

is in every respect the result of a long process of critical thinking and of ‘scaffolding’

our naked mind, as Kim Sterelny (2010, 480) puts it. It is supported by a massive body

of knowledge accumulated over many generations (about species, fossils, genes, etc.),

aided by formal scientific systems of analysis (e.g. mathematical models in population

genetics), and technological innovations (e.g. microscopes in genetic research, spec-

trometers for the carbon dating of fossils, etc.). Quite a different kind of belief than

the one held by Plantinga’s beloved prehistoric hominid Paul about the nature of

the hungry tiger in front of him (Plantinga 1993, 225 – 226). Indeed, as Sterelny

(2010, 480) points out, scientific theories about the world are primarily the product

of our ability to extend our mind’s capacity.

First, the impressive body of scientific knowledge is not the product of a single

mind, but the result of cooperation, both horizontally (groups of cooperating scien-

tists rather than isolated individuals) and vertically (cumulative progress over many

generations). Michael Tomasello, in this regard, points to our ability to read the

minds of conspecifics as the key cognitive adaptation accounting for human culture

(and by extension science). This, according to Tomasello, allows human beings ‘to

pool their cognitive resources both contemporaneously and over time in ways that

are unique in the animal kingdom’ (Tomasello 2001, 135). In a similar vein, Robert

Boyd, Peter J. Richerson, and Joseph Henrich argue that, rather than to innate intelli-

gence, we owe our success to ‘our uniquely developed ability to learn from others’

(Boyd, Richerson, and Henrich 2011, 1).

Second, our best scientific theories are supported by a series of cognitive artefacts

(e.g. mathematics, providing us with a formal system to represent and process quan-

titative information) and various notation devices, enabling us to store huge amounts

International Studies in the Philosophy of Science 73

355

360

365

370

375

380

385

390

395



of information and work out complex mathematical equations (De Cruz 2008). The

leverage provided by these scaffolds can hardly be overestimated. They radically trans-

form our ability to retain, transmit and correct knowledge about the world. Finally,

and equally important, we have a myriad of technological instruments at our disposal,

enabling us to detect data we could never gather by means of our naked senses (e.g.

telescopes, microscopes, barometers, and the like), and process information in ways

we could never do with our naked mind (computers).

Even on the level of individual belief formation, apart from the social organization

of science, we should distinguish between different types of beliefs. Daniel Kahneman

(2011) and other researchers of dual process theory have shown that, while the fast,

intuitive and almost automatic belief outputs of what he dubbed ‘system 1′ can
occasionally—and in some cases even predictably—lead us astray, we do possess the

cognitive ability to override these beliefs (Vlerick 2012). ‘System 2′ does indeed
provide us with a slower, voluntary and more strenuous cognitive activity, which

avoids many of the pitfalls of system 1. How else could we identify sloppy and

biased reasoning arising from system 1?

In this perspective, while natural selection may—as pointed out (cf. section 2)—

provide us with the occasional false intuitive belief (e.g. as the by-product of fast

and frugal heuristics or, in Kahneman’s terms, the output of system 1), this obviously

does not apply to beliefs that are supported by elaborate cognitive scaffolds and inter-

connected strands of evidence, such as the belief in evolution by natural selection. This

prompts Michael Levin (1997, 98) to conclude that ‘there is no evidence that evolution

has fated us to theoretical error, particularly about evolution itself ’ (our italics).

Indeed, science can be seen as a system of safeguards against our intuitive biases

and misconceptions, or what Francis Bacon called the ‘idols of the mind’ in his

Novum Organum. As Levin (1997, 97) points out, even if our cognitive faculties

mislead us, ‘we can find this out and correct for it by comparing the predictions of

our theories to observation’.

As it turns out, the belief in evolution by natural selection is a textbook case in over-

coming our intuitive grasp of the world. Notably, Darwin’s theory violates our intui-

tive species essentialism (a useful heuristic with which evolution endowed us to make

‘quick and dirty’ inferences about groups of living organisms), which predisposes us to

view species as eternal and endowed with an immutable essence. Population thinking,

which came to replace essentialism in biology, views species as loose aggregates of indi-

viduals showing a certain distribution of characteristics rather than an immutable set

of properties. Moreover, the creative power of blind selection also conflicts with our

teleological intuitions (in the sense of goal-directedness) about functional complexity,

and our intuitive design stance (Kelemen and Rosset 2009; Reiss 2009).

Ironically enough, creationism in particular and theism in general are typical cases

of belief systems that seem to entrench our biased, intuitive ways of understanding the

world (McCauley 2011; Blancke et al. 2012). In this regard, such beliefs would be the

first to succumb to a sceptical attitude towards the outputs of adaptive cognitive

modules. Thus, if anything, Plantinga’s EAAN backfires against theism. As Childers

(2011, 199) notes: ‘the atheist has a license to wonder whether theistic belief is one

74 M. Boudry and M. Vlerick

400

405

410

415

420

425

430

435

440



of those false, yet adaptive cognitive dispositions which, as Plantinga notes, would tend

to thrive under natural selection’.

7. Conclusion

Plantinga’s latest formulation of the EAAN contains three major flaws. The first is

based on a failure to appreciate adaptive path dependency; the second derives from

his confused account of belief content and NP (neurophysiological) properties, and

the third from an unwarranted conflation of common sense and scientific beliefs.

Uncovering these reasoning errors, we have aimed to show how Plantinga’s reductive

formalism, by forcing a stark choice on the evolutionary epistemologist, leads to scho-

lastic discussions that are out of touch with the relevant empirical literature. It is the

philosophical equivalent of performing brain surgery with an axe. Natural selection is

not the Cartesian evil genius Plantinga takes it to be, nor has it condemned us to be

slaves to our intuitive cognitive modules. Evolution does care about truth. Besides pro-

viding us with overall reliable beliefs, it has endowed us with the ability to ‘scaffold’

our way up to systems of beliefs that are even more reliable, accurate, and empirically

successful. But is theism among them?

Acknowledgements

We would like to thank Stephen Law, Eric Schliesser, Herman Philipse, and Alex

Broadbent for commenting on an earlier version of this paper. Thanks also to two

anonymous reviewers, and to James McAllister, for their valuable comments.

Notes

[1] For the time being, we take for granted that belief formation directs behaviour in the first place
(premise 2). As Fales (1996) has argued, the cognitive machinery for belief formation is
energy-consuming for the organism, and is unlikely to have evolved without making a differ-
ence in behaviour. Beliefs are not on a permanent holiday.

[2] Indeed, the natural properties realizing content need not be (solely) neurophysiological. If one
takes an externalist point of view, for example, content would be determined by the physical
state of the brain-plus-environment. For the sake of brevity, however, we will keep on referring
to NP properties as instantiating meaning.

[3] There are always contrived ways of avoiding this conclusion, for example, by imagining an
agent with an internal compass that has the wind directions exactly in reverse.

[4] This assumes a (moderate) form of semantic holism: beliefs ‘come to life’ in relation to other
mental states, and cannot be interpreted in splendid isolation.

[5] Plantinga (2011b, 338) has a reply of sorts to this objection, but it forces him to call the whole
semantics of counterfactuals into question. His alternative proposal accepts reasoning about
counterpossibles (‘if 2 had been greater than 3’). This, for instance, would permit him to say
things such as ‘If 13 were divisible by a natural number apart from one and itself, it would not
be a prime number’. This is already quite strange. So is it ‘in virtue of ’ not having any positive
divisors, other than 1 and itself, that 13 is a prime number? Or the other way around? But this
is to talk nonsense. In any case, Plantinga writes that he will not open ‘that particular (and
large) can of worms’. So neither shall we.

International Studies in the Philosophy of Science 75

445

450

455

460

465

470

475

480



[6] For a critique of Plantinga’s far-fetched scenarios where corrupt beliefs happen to result in
felicitous behaviour, see Fales (2002).

References

Beilby, J. K. 2002. Naturalism Defeated? Essays on Plantinga’s Evolutionary Argument against
Naturalism. Ithaca, NY: Cornell University Press.

Blancke, S., J. De Smedt, H. De Cruz, M. Boudry, and J. Braeckman. 2012. “The Implications of the
Cognitive Sciences for the Relation Between Religion and Science Education: The Case of
Evolutionary Theory.” Science and Education 21: 1167 – 1184.

Boudry, M. 2013. “Alvin Plantinga: Where the Conflict Really Lies. Science, Religion and
Naturalism.” Science and Education 22: 1219 – 1227.

Boyd, R., P. J. Richerson, and J. Henrich. 2011. “The Cultural Niche: Why Social Learning Is Essential
for Human Adaptation.” Proceedings of the National Academy of Sciences 108: 10918 – 10925.

Childers, G. 2011. “What’s Wrong with the Evolutionary Argument Against Naturalism?” Inter-
national Journal for Philosophy of Religion 69: 193 – 204.

Clendinnen, F. J. 1989. “Evolutionary Explanation and the Justification of Belief.” In Issues in
Evolutionary Epistemology, edited by K. Hahlweg and C. A. Hooker, 458 – 474. Albany: State
University of New York Press.

Clifford, W. K. [1877] 1999. The Ethics of Belief and Other Essays, edited by T. Madigan. Amherst,
MA: Prometheus Books.

De Cruz, H. 2008. “An Extended Mind Perspective on Natural Number Representation.” Philosophi-
cal Psychology 21: 475 – 490.

De Cruz, H., M. Boudry, J. De Smedt, and S. Blancke. 2011. “Evolutionary Approaches to Epistemic
Justification.” Dialectica 65: 517 – 535.

Fales, E. 1996. “Plantinga’s Case Against Naturalistic Epistemology.” Philosophy of Science 63:
432 – 451.

Fales, E. 2002. “Darwin’s Doubt, Calvin’s Calvary.” In Naturalism Defeated? Essays on Plantinga’s
Evolutionary Argument Against Naturalism, edited by J. K. Beilby, 43 – 58. Ithaca, NY:
Cornell University Press.

Gigerenzer, G. 2008. Rationality for Mortals: How People Cope with Uncertainty. Oxford: Oxford
University Press.

Haselton, M. G., and D. M. Buss. 2000. “Error Management Theory: A New Perspective on Biases in
Cross-sex Mind Reading.” Journal of Personality and Social Psychology 78: 81 – 91.

Kahneman, D. 2011. Thinking, Fast and Slow. London: Penguin Books.
Kelemen, D., and E. Rosset. 2009. “The Human Function Compunction: Teleological Explanation in

Adults.” Cognition 111: 138 – 143.
Law, S. 2012. “Naturalism, Evolution and True Belief.” Analysis 72: 41 – 48.
Levin, M. 1997. “Plantinga on Functions and the Theory of Evolution.” Australasian Journal of

Philosophy 75: 83 – 98.
McCauley, R. N. 2011. Why Religion Is Natural and Science Is Not. New York: Oxford University Press.
McKay, R. T., and D. C. Dennett. 2009. “The Evolution of Misbelief.” Behavioral and Brain Sciences

32: 493 – 561.
Plantinga, A. 1993. Warrant and Proper Function. New York: Oxford University Press.
Plantinga, A. 2000. Warranted Christian Belief. New York: Oxford University Press.
Plantinga, A. 2002. “Reply to Beilby’s Cohorts.” In Naturalism Defeated? Essays on Plantinga’s

Evolutionary Argument Against Naturalism, edited by J. K. Beilby, 204 – 275. Ithaca, NY:
Cornell University Press.

Plantinga, A. 2011a. “Content and Natural Selection.” Philosophy and Phenomenological Research 83:
435 – 458.

76 M. Boudry and M. Vlerick

485

490

495

500

505

510

515

520

525



Plantinga, A. 2011b. Where the Conflict Really Lies: Science, Religion, and Naturalism. New York:
Oxford University Press.

Plantinga, A., and M. Tooley. 2008. Knowledge of God. Oxford: Blackwell.
Reiss, M. J. 2009. “The Relationship Between Evolutionary Biology and Religion.” Evolution 63:

1934 – 1941.
Sterelny, K. 2010. “Minds: Extended or Scaffolded?” Phenomenology and the Cognitive Sciences 9:

465 – 481.
Stich, S. P. 1990. The Fragmentation of Reason: Preface to a Pragmatic Theory of Cognitive Evaluation.

Cambridge, MA: MIT Press.
Tomasello, M. 2001. “Cultural Transmission: A View from Chimpanzees and Human Infants.”

Journal of Cross-cultural Psychology 32: 135 – 146.
Vlerick, M. 2012. “How Can Human Beings Transgress Their Biologically Based Views?” South

African Journal of Philosophy 31: 707 – 735.
Wilkins, J. S., and P. E. Griffiths. 2012. “Evolutionary Debunking Arguments in Three Domains.” In

A New Science of Religion, edited by G. Dawes and J. Maclaurin, 133 – 146. London: Routledge.

International Studies in the Philosophy of Science 77

530

535

540

545

550

555

560

565

570