© 1994 Alvin Plantinga

Reproduction on other websites is expressly prohibited.

Links to this site are permitted.

Naturalism Defeated

In the last chapter of Warrant and Proper Function[1] I proposed an "evolutionary argument against naturalism".

Take philosophical naturalism to be the belief that there aren't any supernatural beings--no such person as God, for

example, but also no other supernatural entities.[2] My claim was that naturalism and contemporary evolutionary

theory are at serious odds with one another--and this despite the fact that the latter is ordinarily thought to be one of the main supporting beams in the edifice of the former.[3] More particularly, I argued that the conjunction of naturalism with the belief that

human beings have evolved in conformity with current evolutionary doctrine--'evolution' for short--is in a certain interesting way

self-defeating or self-referentially incoherent. Still more particularly, I argued that naturalism and evolution--'N&E' for

short--furnishes one who accepts it with a defeater for the belief that our cognitive faculties are reliable--a defeater that can't

be defeated. But then this conjunction also furnishes a defeater for any belief produced by our cognitive faculties, including,

in the case of one who accepts it, N&E itself: hence its self-defeating character. Now oddly enough, not everyone who has

heard this argument has leapt to embrace it; there have been a number of fascinating objections, some published[4] and some unpublished. These objections for the most part revolve around the notion of a defeater--a notion crucial to contemporary epistemology, but so far largely unexplored. In this paper I want to examine and respond to those objections, in the process

hoping to learn something about defeaters.

I The Argument

Since you may not have a copy of WPF on your desk at the moment, I'll briefly outline the original argument here.

It begins from certain doubts about the reliability of our cognitive faculties, where, roughly,[5] a cognitive faculty--memory,

perception, reason--is reliable if the great bulk of its deliverances are true. These doubts are connected with the

origin of our cognitive faculties. According to current evolutionary theory, we human beings, like other forms of life, have

developed from aboriginal unicellular life by way of such mechanisms as natural selection and genetic drift working on

sources of genetic variation: the most popular is random genetic mutation. Natural selection discards most of these

mutations (they prove deleterious to the organism in which they appear), but some turn out to have survival value and to

enhance fitness; they spread through the population and persist. According to this story, it is by way of these

mechanisms, or mechanisms very much like them, that all the vast variety of contemporary organic life has developed;

and it is by way of these same mechanisms that our cognitive faculties have arisen.

Now according to traditional Christian (and Jewish and Muslim) thought, we human beings have been created in

the image of God. This means, among other things, that he created us with the capacity for achieving knowledge—

knowledge of our environment by way of perception, of other people by way of something like what Thomas Reid calls

sympathy, of the past by memory and testimony, of mathematics and logic by reason, of morality, our own mental life,

God himself, and much more.[6] And the above evolutionary account of our origins is compatible with the theistic view

that God has created us in his image.[7] So evolutionary theory taken by itself (without the patina of philosophical naturalism

that often accompanies expositions of it) is not as such in tension with the idea that God has created us and our cognitive

faculties in such a way that the latter are reliable, that (as the medievals like to say) there is an adequation of intellect to reality.

But if naturalism is true, there is no God, and hence no God (or anyone else) overseeing our development and

orchestrating the course of our evolution. And this leads directly to the question whether it is at all likely that our cognitive

faculties, given naturalism and given their evolutionary origin, would have developed in such a way as to be reliable, to

furnish us with mostly true beliefs. Darwin himself expressed this doubt: "With me," he said,

the horrid doubt always arises whether the convictions of man's mind, which has been developed from the mind

of the lower animals, are of any value or at all trustworthy. Would any one trust in the convictions of a monkey's

mind, if there are any convictions in such a mind?[8]

The same thought is put more explicitly by Patricia Churchland. She insists that the most important thing about the human

brain is that it has evolved; this means, she says, that its principal function is to enable the organism to move appropriately:

Boiled down to essentials, a nervous system enables the organism to succeed in the four F's: feeding, fleeing,

fighting and reproducing. The principle chore of nervous systems is to get the body parts where they should be in

order that the organism may survive. . . . . Improvements in sensorimotor control confer an evolutionary

advantage: a fancier style of representing is advantageous so long as it is geared to the organism's way of

life and enhances the organism's chances of survival [Churchland's emphasis]. Truth, whatever that is,

definitely takes the hindmost.[9]

What Churchland means, I think, is that evolution is interested (so to speak) only in adaptive behavior, not in

true belief. Natural selection doesn't care what you believe; it is interested only in how you behave. It selects for certain

kinds of behavior, those that enhance fitness, which is a measure of the chances that one's genes are widely represented

in the next and subsequent generations. It doesn't select for belief, except insofar as the latter is appropriately related to

behavior. But then the fact that we have evolved guarantees at most that we behave in certain ways--ways that contribute

to our (or our ancestors') surviving and reproducing in the environment in which we have developed. Churchland's claim,

I think, is best understood as the suggestion that the objective[10] probability that our cognitive faculties are reliable, given

naturalism and given that we have been cobbled together by the processes to which contemporary evolutionary

theory calls our attention, is low. Of course she doesn't explicitly mention naturalism, but it certainly seems that she is

taking it for granted. For if theism were true, God might be directing and orchestrating the variation in such a way as to

produce, in the long run, beings created in his image and thus capable of knowledge; but then it wouldn't be the case that

truth takes the hindmost.

We can put Churchland's claim as

P(R/N&E) is low,

where 'R' is the proposition that our cognitive faculties are reliable, 'N' the proposition that naturalism is true, and 'E' the

proposition that we have evolved according to the suggestions of contemporary evolutionary theory.[11] I believe this

thought--the thought that P(R/N&E) is low--is also what worries Darwin in the above quotation: I shall therefore call it 'Darwin's

Doubt'.

Are Darwin and Churchland right? Well, they are certainly right in thinking that natural selection is directly interested

only in behavior, not belief, and that it is interested in belief, if at all, only indirectly, by virtue of the relation between

behavior and belief. If adaptive behavior guarantees or makes probable reliable faculties, then P(R/N&E) will be rather high:

we (or rather our ancestors) engaged in at least reasonably adaptive behavior, so it must be that our cognitive faculties

are at least reasonably reliable, in which case it is likely that most of our beliefs are true. On the other hand, if our

having reliable faculties isn't guaranteed by or even particularly probable with respect to adaptive behavior, then presumably

P(R/N&E) will be rather low. If, for example, behavior isn't caused or governed by belief, the latter would be, so to speak,

invisible to natural selection; in that case it would be unlikely that most of our beliefs are true, and unlikely that our cognitive

faculties are for the most part reliable. So the question of the value of P(R/N&E) really turns on the relationship between

belief and behavior. Our having evolved and survived makes it likely that our cognitive faculties are reliable and our beliefs

are for the most part true, only if it would be impossible or unlikely that creatures more or less like us should behave

in fitness-enhancing ways but nonetheless hold mostly false beliefs.[12]

Is this impossible or unlikely? That depends upon the relation between belief and behavior. What would or could

that relation be? To try to guard against interspecific chauvinism, I suggested that we think, not about ourselves and our

behavior, but about a population of creatures a lot like us on a planet a lot like earth (Darwin suggested we think about

monkeys in this connection). These creatures are rational: that is, they form beliefs, reason, change beliefs, and the like.

We imagine furthermore that they and their cognitive systems have evolved by way of the mechanisms to which

contemporary evolutionary theory direct our attention, unguided by the hand of God or anyone else. Now what is

P(R/N&E), specified, not to us, but to them? To answer, we must think about the relationship between their beliefs

and their behavior? There are four mutually exclusive and jointly exhaustive possibilities.

(1) One possibility is epiphenomenalism:[13] their behavior is not caused by their beliefs. On this possibility, their

movement and behavior would be caused by something or other--perhaps neural impulses--which would be

caused by other organic conditions including sensory stimulation: but belief would not have a place in this

causal chain leading to behavior. This view of the relation between behavior and belief (and other mental

phenomena such as feeling, sensation, and desire) is currently rather popular, especially among those strongly

influenced by biological science. Time (December, 1992) reports that J. M. Smith, a well-known biologist,

wrote "that he had never understood why organisms have feelings. After all, orthodox biologists believe

that behavior, however complex, is governed entirely by biochemistry and that the attendant

sensations--fear, pain, wonder, love--are just shadows cast by that biochemistry, not themselves vital to

the organism's behavior . . . ." He could have added that (according to biological orthodoxy) the

same goes for beliefs--at least if beliefs are not themselves just biochemical phenomena. If this way of

thinking is right with respect to our hypothetical creatures, their beliefs would be invisible to evolution;

and then the fact that their belief-forming mechanisms arose during their evolutionary history would confer

little or no probability on the idea that their beliefs are mostly true, or mostly nearly true. Indeed, the

probability of those beliefs' being for the most part true would have to be rated fairly low.

On N&E and this first possibility, therefore, the probability of R will be rather low.

(2) A second possibility is semantic epiphenomenalism: it could be that their beliefs do indeed have causal

efficacy with respect to behavior, but not by virtue of their content. Put in currently fashionable jargon, this

would be the suggestion that beliefs are indeed causally efficacious, but by virtue of their syntax, not by

virtue of their semantics. On a naturalist or anyway a materialist way of thinking, a belief could perhaps be

something like a long-term pattern of neural activity, a long-term neuronal event. This event will have properties

of at least two different kinds. On the one hand, there are its electrochemical properties: the number of neurons

involved in the belief, the connections between them, their firing thresholds, the rate and strength at which they

fire, the way in which these change over time and in response to other neural activity, and so on. Call these

syntactical properties of the belief. On the other hand, however, if the belief is really a belief, it will be the

belief that p for some proposition p. Perhaps it is the belief that there once was a brewery where the Metropolitan

Opera House now stands. This proposition, we might say, is the content of the belief in question. So in

addition to its syntactical properties, a belief will also have semantical [14] properties--for example, the property

of being the belief that there once was a brewery where the Metropolitan Opera House now stands. (Other

semantical properties: being true or false, entailing that there has been at least one brewery, being

consistent with the proposition that all men are mortal and so on.) And the second possibility is that belief is

indeed causally efficacious with respect to behavior, but by virtue of the syntactic properties of a belief, not its

semantic properties. If the first possibility is widely popular among those influenced by biological science, this

possibility is widely popular among contemporary philosophers of mind; indeed, Robert Cummins goes so far

as to call it the "received view."[15]

On this view, as on the last, P(R/N&E) (specified to those creatures) will be low. The reason is that truth or

falsehood, of course, are among the semantic properties of a belief, not its syntactic properties. But if the former aren't

involved in the causal chain leading to belief, then once more beliefs--or rather, their semantic properties, including

truth and falsehood--will be invisible to natural selection.[16] But then it will be unlikely that their beliefs are mostly true

and hence unlikely that their cognitive faculties are reliable. The probability of R on N&E together with this possibility,

(as with the last), therefore, will be relatively low.

(3) It could be that beliefs are causally efficacious--'semantically' as well as 'syntactically'--with respect to

behavior, but maladaptive: from the point of view of fitness these creatures would be better off without them. The

probability of R on N&E together with this possibility, as with the last two, would also seem to be relatively low.

(4) Finally, it could be that the beliefs of our hypothetical creatures are indeed both causally connected with their

behavior and also adaptive. (I suppose this is the common sense view of the connection between behavior and belief in

our own case.) What is the probability (on this assumption together with N&E) that their cognitive faculties are reliable;

and what is the probability that a belief produced by those faculties will be true? I argued that this probability isn't nearly

as high as one is initially inclined to think. The reason is that if behavior is caused by belief, it is also caused by desire

(and other factors--suspicion, doubt, approval and disapproval, fear--that we can here ignore). For any given adaptive

action, there will be many belief-desire combinations that could produce that action; and very many of those belief-desire

combinations will be such that the belief involved is false.

So suppose Paul is a prehistoric hominid; a hungry tiger approaches. Fleeing is perhaps the most appropriate

behavior: I pointed out that this behavior could be produced by a large number of different belief-desire pairs. To quote

myself:

Perhaps Paul very much likes the idea of being eaten, but when he sees a tiger, always runs off looking

for a better prospect, because he thinks it unlikely that the tiger he sees will eat him. This will get his body parts

in the right place so far as survival is concerned, without involving much by way of true belief. . . . . Or perhaps

he thinks the tiger is a large, friendly, cuddly pussycat and wants to pet it; but he also believes that the best way

to pet it is to run away from it. . . . or perhaps he thinks the tiger is a regularly recurring illusion, and, hoping

to keep his weight down, has formed the resolution to run a mile at top speed whenever presented with such an

illusion; or perhaps he thinks he is about to take part in a 1600 meter race, wants to win, and believes the

appearance of the tiger is the starting signal; or perhaps . . . . Clearly there are any number of belief-cum-desire

systems that equally fit a given bit of behavior (WPF pp. 225-226).

Accordingly, there are many belief-desire combinations that will lead to the adaptive action; in many of these combinations,

the beliefs are false. Without further knowledge of these creatures, therefore, we could hardly estimate the probability of R

on N&E and this final possibility as high.

A problem with the argument as thus presented is this. It is easy to see, for just one of Paul's actions, that there

are many different belief-desire combinations that yield it; it is less easy to see how it could be that most of all of his beliefs

could be false but nonetheless adaptive or fitness enhancing. Could Paul's beliefs really be mainly false, but still lead to

adaptive action? Yes indeed; perhaps the simplest way to see how is by thinking of systematic ways in which his beliefs

could be false but still adaptive. Perhaps Paul is a sort of early Leibnizian and thinks everything is conscious (and suppose that

is false); furthermore, his ways of referring to things all involve definite descriptions that entail consciousness, so that all

of his beliefs are of the form That so-and-so conscious being is such-and-such. Perhaps he is an animist and thinks

everything is alive. Perhaps he thinks all the plants and animals in his vicinity are witches, and his ways of referring to them

all involve definite descriptions entailing witchhood. But this would be entirely compatible with his belief's being adaptive; so

it is clear, I think, that there would be many ways in which Paul's beliefs could be for the most part false, but adaptive

nonetheless.

What we have seen so far is that there are four mutually exclusive and jointly exhaustive possibilities with respect

to that hypothetical population: epiphenomenalism simpliciter, semantic epiphenomenalism, the possibility that their beliefs

are causally efficacious with respect to their behavior but maladaptive, and the possibility that their beliefs are both causally

efficacious with respect to behavior and adaptive. P(R/N&E) will be the weighted average of P(R/N&E&P_i) for each

of the four possibilities P_i --weighted by the probabilities, on N&E, of those possibilities. The probability calculus gives us

a formula here:

P(R/N&E) = (P(R/N&E&P₁) x P(P₁/N&E)) + (P(R/N&E&P₂) x P(P₂/N&E)) + (P(R/N&E&P₃) x

P(P₃/N&E)) + (P(R/N&E&P₄) x P(P₄/N&E)).

Of course the very idea of a calculation (suggesting, as it does, the assignment of specific real numbers to these various

probabilities) is laughable: the best we can do are vague estimates. But that is all we need for the argument. For consider

the left-hand multiplicand in each of the four terms on the right-hand side of the equation. In the first three, the sensible

estimate would put the value low, considerably less that 1/2; in the 4th, it isn't very clear what the value would be, but it

couldn't be much more than 1/2. But then (since the probabilities of P1 and of P2 (the two forms of epiphenomenalism)

would be fairly high, given naturalism, and since the right hand multiplicands in the four terms cannot sum to more than 1)

that means that the value of P(R/N&E) will be less than 1/2; and that is enough for the argument.

But the argument for a low estimate of P(R/N&E) is by no means irresistible; our estimates of the various probabilities

involved in estimating P(R/N&E) with respect to that hypothetical population were (naturally enough) both imprecise

and poorly grounded. You might reasonably hold, therefore, that the right course here is simple agnosticism: one just

doesn't know what that probability is. You doubt that it is very high; but you aren't prepared to say that it is low: you

have no definite opinion at all as to what that probability might be. Then this probability is inscrutable for you. This

too seems a sensible attitude to take. The sensible thing to think, then, is that P(R/N&E) is either low or inscrutable.

Now return to Darwin's doubt, and observe that if this is the sensible attitude to take to P(R/N&E) specified

to that hypothetical population, then it will also be the sensible attitude towards P(R/N&E) specified to us. We are

relevantly like them in that our cognitive faculties have the same kind of origin and provenance as theirs are hypothesized

to have. And the next step in the argument was to point out that each of these attitudes--the view that P(R/N&E) is low

and the view that this probability is inscrutable--gives the naturalist-evolutionist a defeater for R. It gives him a reason to

doubt it, a reason not to affirm it. I argued this by analogy. Among the crucially important facts, with respect to the

question of the reliability of a group of cognitive faculties, are facts about their origin. Suppose I believe that I have

been created by an evil Cartesian demon who takes delight in fashioning creatures who have mainly false beliefs

(but think of themselves as paradigms of cognitive excellence): then I have a defeater for my natural belief that my faculties

are reliable. Turn instead to the contemporary version of this scenario, and suppose I come to believe that I have been

captured by Alpha-Centaurian superscientists who have made me the subject of a cognitive experiment in which the

subject is given mostly false beliefs: then, again, I have a defeater for R. But to have a defeater for R it isn't necessary

that I believe that in fact I have been created by a Cartesian demon or been captured by those Alpha-Centaurian

superscientists. It suffices for me to have such a defeater if I have considered those scenarios, and the probability

that one of those scenarios is true, is inscrutable for me--if I can't make any estimate of it, do not have an opinion

as to what that probability is. It suffices if I have considered those scenarios, and for all I know or believe one of

them is true. In these cases too I have a reason for doubting, a reason for withholding[17] my natural belief that my

cognitive faculties are in fact reliable.

Now of course defeaters can be themselves defeated. For example, I know that you are a lifeguard and believe

on that ground that you are an excellent swimmer. But then I learn that 45% of Frisian lifeguards are poor swimmers, and

I know that you are Frisian: this gives me a defeater for the belief that you are a fine swimmer. But then I learn still further

that you graduated from the Department of Lifeguarding at the University of Leeuwarden and that one of the

requirements for graduation is being an excellent swimmer: that gives me a defeater for the defeater of my original belief:

a defeater-defeater as we might put it.[18] But (to return to our argument) can the defeater the naturalist has for R be in turn

defeated? I argued that it can't (WPF 233-234). It could be defeated only by something--an argument, for example,

that involves some other belief (perhaps as premise). But any such belief will be subject to the very same defeater as R is.

So this defeater can't be defeated.[19]

But if I have an undefeated defeater for R, then by the same token I have an undefeated defeater for any other

belief B my cognitive faculties produce, a reason to be doubtful of that belief, a reason to withhold it. For any such

belief will be produced by cognitive faculties that I cannot rationally believe to be reliable. But then clearly the same

will be true for any proposition they produce: the fact that I can't rationally believe that the faculties that produce

that belief are reliable, gives me a reason for rejecting the belief. So the devotee of N&E has a defeater for just any belief

he holds--a defeater, as I put it, that is ultimately undefeated But this means, then, that he has an ultimately undefeated

defeater for N&E itself. And that means that the conjunction of naturalism with evolution is self-defeating, such that

one can't rationally accept it.

I went on to add that if naturalism is true, then so, in all probability, is evolution; evolution is the only game in

town, for the naturalist, with respect to the question how all this variety of flora and fauna has arisen. If that is so,

finally, then naturalism simpliciter is self-defeating and cannot rationally be accepted--at any rate by someone who is

apprised of this argument and sees the connections between N&E and R.

II Objections

Now I believe this argument, while inevitably a bit sketchy, has a great deal to be said for it. My exalted

opinion of it, however, has not sufficed to protect it from a number of extremely interesting objections. Despite the

objections I continue to believe that the argument is a good one: I therefore want to examine and reply to the objections.

I have a further and ulterior motive. The objections have to do crucially with the notion of defeaters; although this notion is

absolutely central to contemporary epistemology, it has so far received little by way of concentrated attention.[20]

Exploring these objections will give us, as an added bonus, a good chance to learn something about defeaters. I shall first

briefly set out the objections, then make some suggestions as to how defeaters work, and then assess the objections

in the light of what (I hope) we will have learned about defeaters.

A. The Perspiration Objection (Michael DePaul, Frederick Suppe, Stephen Wykstra, others). This objection goes as

follows. "You claim that the naturalist has a defeater for R in the fact that the probability of R on N&E is either low or

inscrutable. But this can't be right. The probability that the function of perspiration is to cool the body, given (just) N&E, is

also low, as is the probability that Holland, Michigan is 30 miles from Grand Rapids, given N&E. But surely it would be

absurd to claim that these facts give the partisan of N&E a defeater for those beliefs.

B. Austere Theism a defeater for Theism Simpliciter? (Earl Conee, Richard Feldman, Theodore Sider, Stephen

Wykstra, others) This objection comes in three varieties.

1. "If you are a theist, then, unless your inferential powers are severely limited, you also accept austere theism, the view

that there exists an extremely powerful and knowledgeable being. But the probability of theism with respect to austere

theism, like that of R with respect to N&E, is low or inscrutable; hence (if the principles underlying your argument against

N&E are correct) austere theism furnishes the theist with a defeater for theism. But every theist is an austere theist: so every

theist has a defeater for theism. Furthermore, this defeater can't be defeated, as is shown by an argument exactly paralleling

the one you gave for supposing that the defeater for N&E can't be defeated. So if your argument is correct, the theist

has an ultimately undefeated defeater for theism."

2. "If N&E is self-defeating in the way you suggest, then so is austere theism. For relative to austere theism, the probability

of R is low or inscrutable; the austere theist therefore has an undefeatable defeater for R, but then also for any other belief

she holds, including austere theism itself. Austere theism, therefore, is self-referentially self-refuting (if N&E is) and hence

cannot rationally be accepted. But of course theism entails austere theism; if it is irrational to accept a proposition p, it is also

irrational to accept any proposition that entails it; hence if the argument defeats naturalism, it pays the same compliment to

theism."

3. "As in (b), the probability of R on austere theism is low or inscrutable; so the theist has a defeater for R, and hence for

anything else he believes; but then he has a defeater for theism, one he can't lose as long as he accepts theism."

C. Can't the Naturalist Just Add a Little Something? (Fred Dretske, Carl Ginet, Timothy O'Connor, Richard Otte,

John Perry, Ernest Sosa, Stephen Wykstra, others). An austere theist who wasn't also a theist would face the same

defeater as the partisan of N&E. Although the theist also accepts austere theism, she escapes defeat because she

accepts not just austere theism, but something additional, the difference, we might say, between austere theism and theism.

But if it is right and proper for the theist thus to elude defeat, why can't the naturalist do the same thing? Thus Ginet:

. . . if we delete this component (the difference between theism simpliciter and austere theism) and consider

just the hypothesis T- that there is a perfect being who creates everything else, then it looks as if we could argue

in just the same way Plantinga argues concerning P(R/N&E&C[21]) to the dismal conclusion that P(R/T-&E&C)

is low or unknown. Now how is it that the theist is allowed to build into her metaphysical hypothesis something

that entails R or a high probability of R but the naturalist isn't? Why isn't it just as reasonable for the naturalist

to take it as one of the tenets of naturalism that our cognitive systems are on the whole reliable (especially since

it seems to be in our nature to have it as a basic belief)?[22]

D. The Maximal Warrant Objection (William Alston, Timothy O'Connor, William Craig, others) According to this

objection, R has a great deal of intrinsic warrant for us. This proposition has warrant in the basic way: it doesn't get its

warrant by way of being accepted on the evidential basis of other propositions. It has so much intrinsic warrant, in fact,

that it can't be defeated--or at any rate can't be defeated by the fact that P(R/N&E) is low or inscrutable. A variant of

this objection (Van Fraassen) addresses and rejects the argument's implicit premise that if the right attitude towards

P(R/N&E) with respect to that hypothetical population is low or inscrutable, then the same goes for that probability with

respect to myself (ourselves).)

E. The Dreaded Loop Objection (Richard Otte, Glenn Ross, David Hunt, others).

Following Hume (and Sextus Empiricus) I said that if the devotee of N (or N&E) is rational, then he will fall into

the following sort of diachronic loop: first, he believes N&E and sees that this gives him a defeater for R, and hence for

N&E; so then he stops believing N&E; but then he loses his defeater for R and N&E; then presumably those beliefs

come flooding back; but then once again he has a defeater for them; and so on, round and round the loop. In this loop

N&E keeps getting alternately defeated and reprieved--i.e., at t1 it is defeated, at t2 undefeated, at t3 defeated, and so on.

And then I went on to say that his falling into this loop gives him an ultimately undefeated defeater for N&E.

According to the objector, the problem is two-fold. First, suppose the devotee of N&E were to fall into such

a loop and doggedly plod around it: he wouldn't thereby have an ultimately undefeated defeater for N&E. What he

would have instead is a defeater that is not ultimately defeated--a different matter altogether. An ultimately undefeated

defeater would be one such that at a certain point it is undefeated, and remains undefeated thereafter. But, says the

objector, that doesn't happen here: here the devotee of N&E alternately has and loses his defeater for N&E. For every

time at which he has a defeater for N&E, there is a subsequent time at which that defeater is defeated; this alternation is

terminated only by death or disability. Hence, obviously, he does not have an ultimately undefeated defeater for N&E.

But further, why think in the first place that rationality requires him to fall into this appalling loop? The fact is rationality

requires that he stay out of the loop, or at least get out of it after a couple of tours around it. Can't he see in advance what is

coming? He'd have to be (at best) extremely imperceptive to keep on slogging round and round that loop.

III Defeaters and Defeat