. 4
( 8)


Human minds, then, are not the only supernatural entities that exist.
They do not come from nowhere. Each has come into Nature from
Supernature: each has its tap-root in an eternal, self-existent, rational
Being, whom we call God. . . . [H]uman thought is . . . God-kindled.117

Incorporating this idea, and putting aside the super¬‚uous early stages
of the Chapter 3 argument, it seems to me that Lewis™s argument
from reason, as it appears in the revised version of Miracles, amounts
to the following:

Beyond Nature

Lewis™s Argument from Reason

1. If Naturalism is true, then knowledge exists only if natural selec-
tion could produce a capacity for knowledge starting with crea-
tures with no such capacity.
2. But natural selection could not produce a capacity for knowledge
starting with creatures with no such capacity.
3. So: If Naturalism is true, then knowledge does not exist (from 1
and 2).
4. But knowledge does exist.
5. So: Naturalism is false (from 3 and 4).
6. If knowledge exists and Naturalism is false, then there is a super-
natural, eternal, self-existent, rational Being that is the ultimate
source of all knowledge.
7. Therefore, there is a supernatural, eternal, self-existent, rational
Being that is the ultimate source of all knowledge (from 4, 5,
and 6).

How might this argument be resisted? The second premise is one
that many contemporary nontheists would be inclined to reject. Let
us take a closer look at how Lewis supports this crucial premise.
Lewis claims that it is “not conceivable” that evolutionary pro-
cesses could produce creatures capable of knowledge from creatures
incapable of knowledge.118 To evaluate this claim, we must distin-
guish two ways of understanding it. One interpretation has it that
Lewis is claiming that he (and perhaps the reader as well “ perhaps
everyone) cannot conceive of any way in which evolutionary pro-
cesses could produce beings capable of knowledge. Let us say that
when something is inconceivable in this sense, it is weakly incon-
ceivable. Consider, for example, the process by which the letters
inscribed on the keys on my computer keyboard appear on the screen
when I press down the keys. This process is weakly inconceivable
(by me).
Another interpretation has it that Lewis is claiming that he can see
that it is impossible for evolutionary processes to yield beings capable
of knowledge. Let us say that when something is inconceivable in this

God and the Reach of Reason

sense, it is strongly inconceivable. Consider, for example, the concept
of a round square, discussed in the previous chapter. Such a shape
is inconceivable, not merely in that I cannot conceive of a process
that would produce such a shape (although that is true), but also
in that I can see in a rather direct way that no such shape could
exist. Indeed, I can give a proof that such a shape is impossible. Here
it is: A square must have exactly four corners, and a circle must
have exactly zero corners. So a round square must have exactly four
corners and simultaneously have exactly zero corners. But this is
plainly impossible; hence there cannot be a round square.
When Lewis claims that the production of beings capable of
knowledge by way of evolutionary processes is inconceivable, does
he mean to say that it is weakly inconceivable or strongly incon-
ceivable? The former claim seems too weak for his purposes. This
is because the fact that a given process is weakly inconceivable is
perfectly consistent with the occurrence of the process in question. I
cannot conceive of the process by which the letters I type appear on
the screen; nevertheless, it is happening right now! Even if everyone
who understood the process were to forget how it works so that no
one at all could conceive of the process, it would not follow that the
process is impossible. It seems, therefore, that Lewis must be claiming
that he can see that the production of beings capable of knowledge
by evolutionary processes is impossible. What support does Lewis
offer for such a claim?
Lewis remarks that if Naturalism is true, then “all our thoughts
once were . . . merely subjective events, not apprehensions of objec-
tive truth.”119 What exactly is the contrast Lewis is attempting to
highlight with this distinction between merely subjective thoughts
and thoughts that are apprehensions of truth? An earlier passage
sheds some light on this question: “Acts of thinking . . . are a very
special sort of events. They are ˜about™ something other than them-
selves and can be true or false.”120 The later passage suggests that
Lewis™s view is that not all mental states have intentionality (are
about something). Some mental states do have intentionality; for
example, mental states that are apprehensions of objective truth.

Beyond Nature

There are other mental states, purely subjective mental states (e.g.,
pain), that are not about anything.
It is shortly after the passage about subjective and objective
thoughts that Lewis claims that “[t]he relation between response
and stimulus is utterly different from that between knowledge and
the truth known.”121 It seems that when Lewis speaks of stimulus
and response in this context, he has in mind a very speci¬c type of
response “ speci¬cally, responses involving mental states that are not
about anything (lack intentionality). Immediately after the sentence
I just quoted, Lewis provides the following example to illustrate his

Our physical vision is a far more useful response to light than that of
the cruder organisms which have only a photo-sensitive spot. But nei-
ther this improvement nor any possible improvements we can suppose
could bring it an inch nearer to being knowledge of light.122

I think that what Lewis has in mind here is that the human visual
system is capable of producing a much broader range of visual expe-
riences than is a photosensitive spot. Despite this difference, how-
ever, neither system produces mental states that are about anything.
Visual experiences are not intentional states (according to Lewis, at
any rate). Furthermore, no increase in the range of experiences a
given visual system can produce can ever render the system capable
of producing intentional mental states. The mere addition of more
and more nonintentional mental states will never somehow “add
up” to intentionality. This means that no increase in the range of
experiences a given visual system can produce can ever turn it into a
system capable of producing knowledge. This is because knowledge
requires intentionality.
It seems to me, therefore, that the fundamental problem Lewis
sees with the notion that knowledge could arise via evolutionary
processes is that he thinks it is impossible that intentional mental
states could be created by evolutionary processes. Nature alone can-
not produce intentionality; for this, you need something outside of
nature, something that Lewis calls “reason.” Without supernatural

God and the Reach of Reason

reason, there could be no thinkers capable of thinking about the nat-
ural universe:

[A]cts of reasoning are not interlocked with the total interlocking sys-
tem of Nature as all its other items are interlocked with one another.
They are connected with it in a different way; as the understanding of
a machine is certainly connected with the machine but not in the way
the parts of the machine are connected with each other. The knowl-
edge of a thing is not one of the thing™s parts. In this sense something
beyond Nature operates whenever we reason.123

Lewis presents a complicated and challenging argument. Let us con-
sider how a naturalist might defend her position against the argu-
ment. One potential weakness in the argument lies in Lewis™s attempt
to establish the impossibility of the production of intentional states by
evolutionary processes. The attempted proof runs roughly as follows:
If evolutionary processes could produce intentional states at all, they
could do so only by increasing the variety of nonintentional mental
states produced by an already-existing response system. But such
increases can never generate intentional states; hence, evolutionary
processes cannot produce intentional states. The weakness in this
argument is in the ¬rst premise. Why should the naturalist accept
that if evolutionary processes can produce intentional states at all,
they would have to do so in the speci¬c way described by Lewis?
From the fact that evolutionary forces couldn™t accomplish the task
in that particular way, it does not follow that they couldn™t accomplish
it at all.
Despite this weakness, the argument highlights a real puzzle for
naturalism, and drawing attention to this puzzle is among Lewis™s
most important contributions to contemporary philosophy.124 By
Lewis™s own account, doubt about the compatibility of naturalism
and knowledge was one of the main intellectual components of his
abandonment of naturalism and eventual conversion to Christianity.
Lewis credits his friend Owen Bar¬eld with drawing his attention to
the dif¬culty. He writes that Bar¬eld “convinced me that the posi-
tions we had hitherto held left no room for any satisfactory theory
of knowledge.”125 Questions about the compatibility of naturalism
Beyond Nature

and human knowledge are also prominent in the contemporary
debate between theists and naturalists; the Christian philosopher
Alvin Plantinga has proposed a much-discussed argument that owes
much to Lewis™s argument in Miracles.126
In the previous chapter I suggested that a certain kind of suffering,
while compatible with the existence of the Christian God, constitutes
evidence against such a God because the presence of such suffering
is less surprising on the hypothesis that atheism is true than it is
on the hypothesis that the Christian God exists. While Lewis™s argu-
ment from reason fails to establish that intentionality is incompati-
ble with naturalism, it may show that the existence of intentionality
constitutes evidence against naturalism in much the same way that
certain kinds of suffering constitute evidence against the existence
of the Christian God: The presence of intentionality is, perhaps, less
surprising on the hypothesis that an eternal Reasoner exists than it
is on the hypothesis that naturalism is true.
How might a naturalist address this sort of challenge? One obvi-
ous strategy is to ¬nd a plausible explanation of the emergence of
intentionality in the context of naturalism. Tellingly, explaining how
natural forces alone might produce intentional states is one of the
central projects of contemporary philosophy of mind. The literature
on this topic is vast, indicating that contemporary naturalists are
keenly aware of the problem that bothered Lewis.127
Another strategy is available. This other strategy involves the claim
that naturalism actually predicts that it will be dif¬cult or impos-
sible for us to understand how intentionality could be produced
by evolution. The idea underlying this strategy is that if a given
theory predicts a certain fact, that fact cannot constitute evidence
against the theory in question. This kind of strategy has been pur-
sued by some theists in connection with the following line of reason-

1. The world contains evil for which we can discern no justi¬cation
(inscrutable evil).
2. Inscrutable evil is probably evil that has no justi¬cation (pointless
God and the Reach of Reason

3. Therefore, the world probably contains pointless evil (and hence
God probably does not exist).128

In response to this kind of argument, some theists point out that
given the large gap between the cognitive abilities of God and those of
human beings, it is not at all surprising that we cannot ¬nd an expla-
nation for every instance of suffering, and hence that the second
premise of the argument just given “ the inference from inscrutable
to pointless evil “ is no good.129 Indeed, Lewis himself pursues this
kind of strategy, as we will see in Chapter 4. Now consider the fol-
lowing line of reasoning:

1. The world contains mental phenomena (e.g., intentionality)
such that we can discern no evolutionary process that could pro-
duce such phenomena (evolutionarily challenged mental phe-
2. Evolutionarily challenged mental phenomena are probably
mental phenomena that were not produced by evolution
(nonevolved mental phenomena).
3. Therefore, the world probably contains nonevolved mental phe-
nomena (and hence naturalism is probably false).

Can the naturalist respond to this kind of argument in a way that
parallels the theistic response to the previous argument? I believe
so. Colin McGinn has taken such an approach with respect to con-
sciousness. Although consciousness as understood by McGinn is
importantly different from intentionality, consideration of McGinn™s
approach can give us a sense of the kind of strategy I have in mind.
According to McGinn, the human mind is structured in such a way
that it is incapable of grasping the connection between the brain and
conscious experience. Furthermore, the fact that the human mind
is so constituted is to be explained at least in part by evolutionary
theory. The following passage summarizes this aspect of McGinn™s

I maintain that the perennial puzzlement surrounding conscious-
ness and its relation to the body is an indication that we are on
the edge of what we can make comprehensible to ourselves. Human

Beyond Nature

intelligence . . . is an evolutionary contrivance, designed with purposes
far removed from the solution of profound philosophical problems,
and it is not terribly surprising if it lacks the tools to crack every prob-
lem. . . . There is no product warranty inscribed on our brains reading,
“This device is guaranteed to solve any problem it can formulate. If
not completely satis¬ed, please return to Philosophical Products Inc.
for a sincere apology and your money back.”130

A central theme of McGinn™s approach is that contemporary evo-
lutionary theory predicts that in general, human beings will tend
not to be very good at solving philosophical problems.131 McGinn
does not stop here; he proceeds to offer a fairly developed account
of the speci¬c capacities of the human brain and why these capac-
ities fail to provide us with tools suf¬cient to understand the
brain-consciousness relationship. His theory (“New Mysterianism”)
depends on certain properties peculiar to conscious experience
and so cannot be straightforwardly applied to intentionality, but
McGinn™s more general theme can: Understanding how evolution-
ary processes could produce intentional states is (at least in part) a
philosophical problem, just the sort of problem evolutionary theory
predicts our brains will be bad at solving. So, just as the presence
of evil for which we can discern no justi¬cation should not lead us
to infer that such evil has no justi¬cation, the presence of mental
phenomena for which we can discern no evolutionary explanation
should not lead us to infer that such phenomena have no evolution-
ary explanation.
Christianity and contemporary naturalism both predict that
human beings will have certain signi¬cant cognitive limitations. In
Christianity, this prediction is grounded in the gap between the cog-
nitive abilities of human beings and those of God. In contempo-
rary naturalism, this prediction is grounded in the nature of the
processes that produced the human mind. This feature of the two
worldviews provides them with similar strategies for dealing with
certain objections. If these strategies are successful, each side seems
to have a philosophical card that can be used to evade challenges
from the other side “ like this: The naturalist says, “Look, there™s
this evil for which we can discern no justi¬cation. This kind of evil is
God and the Reach of Reason

evidence against your view.” The Christian replies, “Ah, but my view
predicts that there will be evil for which we can discern no justi¬ca-
tion (even though such evil is justi¬ed). After all, ¬nite, limited, and
¬‚awed beings such as ourselves cannot be expected to have com-
plete comprehension of the ways of an in¬nite, perfect God. So such
evil is not evidence against my view after all.” Or the Christian says:
“Look, there™s this mental phenomenon for which we can discern
no evolutionary explanation. This kind of phenomenon is evidence
against your view.” The naturalist replies, “Ah, but my view pre-
dicts that there will be phenomena for which can discern no evolu-
tionary explanation (even though such an explanation exists). After
all, ¬nite, limited, glori¬ed monkeys cannot be expected to be able
to solve every philosophical problem; our brains were selected for
other purposes. So such phenomena are not evidence against my
view after all.”
Both sides seem to recognize that this second kind of strategy (our
worldview predicts this kind of puzzle) is far less satisfying than the
¬rst kind of strategy (here is a solution to the puzzle); theists continue
to propose theodicies, and naturalists continue to seek accounts of
how intentionality could arise by way of evolution. This means that
even if pointing to such puzzles fails to settle the issue of which
worldview is correct, it does stimulate thinkers on both sides to
investigate further. Lewis™s argument from reason and related argu-
ments, while not deserving of the rarely achieved title of “proof,”
should at least provoke re¬‚ective naturalists to engage in some seri-
ous thought. As Victor Reppert observes, “[g]reat thinkers are always
the ones that make us think harder for ourselves, not thinkers who
do our thinking for us. And the same is true for Lewis.”132


In his autobiography, Lewis credits a state of mind he labels “Joy”
with playing an important role in his conversion to Christianity. The
importance of Joy in his conversion is indicated by the title of his
autobiography, Surprised by Joy. Early in that work Lewis remarks

Beyond Nature

that “in a sense the central story of my life is about nothing else.”133
He describes Joy as “an unsatis¬ed desire which is itself more desir-
able than any other satisfaction.”134 Lewis encountered Joy peri-
odically throughout his life, but apparently found the experience
¬‚eeting and devoted much effort to recapturing it. He puzzled over
its nature for many years, eventually coming to believe that it “was
not a deception” but instead involved “the moments of clearest con-
sciousness we had.”135 He ultimately concluded that Joy is a desire
for God.136 At one point he says that God had been shooting “arrows
of Joy” at him “ever since childhood.”137 He attributes one of his
crucial insights into the nature of Joy to his reading of Samuel
Alexander™s 1920 book Space, Time, and Deity. In that work, Alexander
describes a “distinctive religious appetite, comparable to the appetite
for food or drink.”138 Alexander claims that this religious appetite is
caused by God and that experiencing it provides insight into God™s
Several commentators have argued that Lewis™s writing suggests
an argument based on Joy. This argument has come to be known
as “the argument from desire.”140 The argument has captured the
heart of at least one commentator: Peter Kreeft declares that “[n]ext
to Anselm™s famous ˜ontological argument,™ I think [the argument
from desire] is the single most intriguing argument in the history of
human thought.”141
The following lines from Mere Christianity indicate the nature of
the argument:

Creatures are not born with desires unless satisfaction for those desires
exists. A baby feels hunger; well, there is such a thing as food. A
duckling wants to swim: well, there is such a thing as water. Men feel
sexual desire: well, there is such a thing as sex. If I ¬nd in myself a
desire which no experience in this world can satisfy, the most probable
explanation is that I was made for another world.142

In light of Lewis™s various remarks on Joy and discussions of the
argument in the secondary literature, I believe that the essence of
the argument may be captured as follows:

God and the Reach of Reason

Lewis™s Argument from Desire

1. All normal human beings have an innate, natural desire (Joy)
that is for some thing, x, where x lies beyond the natural world.
2. Every desire that is innate and natural to all normal human
beings can be satis¬ed.
3. So: Joy can be satis¬ed (from 1 and 2).
4. If Joy can be satis¬ed, then there is something that lies beyond
the natural world.
5. Therefore, there is something that lies beyond the natural world
(from 3 and 4).

The most interesting premises here are the ¬rst two. In support of
the ¬rst premise, Lewis remarks that “[m]ost people, if they had
really learned to look into their own hearts, would know that they
do want . . . something that cannot be had in this world.”143 Steve
Lovell offers the following:

Given that even the best the material world has to offer leaves many
of us deeply unhappy, that even some of the most staunch of atheists
admits a desire for ˜something more,™ that over 90% of the world™s
population engage in some form of religious practise, and that the
longing for transcendence is a recurring theme in both religious and
secular literature it would be, at the very least, reasonable to posit a
natural desire for something beyond this world as an explanation of
these facts.144

What about the second premise? Why should we think that all our
natural, innate desires can be satis¬ed? Russell appears to reject such
a notion outright: “The fact that I feel a need for something more than
human is no evidence that the need can be satis¬ed, any more than
hunger is evidence that I shall get food.”145 In his 1941 essay “The
Weight of Glory,” Lewis writes:

A man™s physical hunger does not prove that man will get any
bread. . . . But surely [it] proves that he . . . inhabits a world where eat-
able substances exist. In the same way, though I do not believe . . . that
my desire for Paradise proves that I shall enjoy it, I think it a pretty
good indication that such a thing exists and that some men will. A

Beyond Nature

many may love a woman and not win her; but it would be very odd if
the phenomenon called “falling in love” occurred in a sexless world.146

Lewis™s remarks indicate that Russell™s hunger example does not do
much to support the rejection of the second premise of the argument
from desire, but Russell nevertheless makes a relevant point: In gen-
eral, we do not think that the mere fact that something is desired
indicates that the desired thing exists. What reason is there to think
that natural desires do point to the existence of the desired object?147
Three interesting kinds of arguments for (2) (the claim that all nat-
ural, innate desires are capable of being satis¬ed) exist. The ¬rst of
these is an inductive argument based on our experience of natural
desires other than Joy. As Lewis notes, we observe that we natu-
rally desire food, and food exists; we naturally desire sex, and sex
exists. It is not hard to add to this list: We naturally desire sleep, and
sleep exists. So there is lots of inductive evidence that our innate,
natural desires can be satis¬ed; thus, it is likely that Joy can be sat-
is¬ed as well.148 A second line of argument is based on the idea that
“if natural desires did not have correlating objects, there would be
something fundamentally wrong, awry, disjointed, illogical, unfair,
twisted, fraudulent, or out of kilter about reality.”149 If (2) is false,
then the universe is absurd in a certain way; but the universe is not
absurd in this way, so (2) is true. Finally, it might be suggested that
(2) or something that entails it is self-evident and can simply be seen
to be true.150
Let us consider these three approaches in reverse order. For my
part, I cannot simply see that (2) is true; after sincere re¬‚ection, it
does not strike me as self-evident. Russell apparently thought he
could see it to be false. To my way of thinking, (2) is precisely the
sort of claim that needs to be argued for or against. Readers who ¬nd
(2) to be obviously true may as well skip ahead to the concluding
section of this chapter.
What of the absurdity of the universe if (2) is false? Here we must
keep in mind the distinction between absurdity (of the relevant sort)
and inexplicability. To deny (2) is not to assert that the universe or
some feature of it cannot be explained. To deny (2) is rather to assert
God and the Reach of Reason

that there is something “out of kilter” about the universe in that it
is part of human nature to desire something that does not exist. As
I have suggested elsewhere, the question of whether the universe
fundamentally makes sense is one of the central questions about
which theists and atheists tend to disagree, and hence any argument
put forth by either party to the debate that simply assumes that the
universe does (or does not) make sense in this way is not particularly
helpful.151 That the universe fails to conform to our natural desires is
an implication the atheist is unlikely to ¬nd surprising or implausible.
It is natural to wonder about the explanation of a natural desire that
cannot be satis¬ed; I will make a proposal about this later.
Finally, let us consider the inductive case for premise (2). Recall
the natural desires to which Lewis initially draws our attention:
hunger, sexual desire, and a duckling™s desire for water (or to swim).
The inductive case goes like this: All examined natural desires can be
satis¬ed; therefore, all natural desires (examined or unexamined)
can be satis¬ed.152 Since Joy is a natural desire, it follows that Joy
is capable of satisfaction, and hence some sort of supernatural entity
that can satisfy it exists.
I see two weaknesses with this inductive approach. One feature
of simple inductive arguments of this sort is that they can be coun-
tered by other simple inductive arguments for an opposed conclu-
sion. Consider the following example. Imagine a sports team with a
long history that has never lost a home game and has never won on
a Tuesday. Despite playing many games at home and many games
on Tuesdays, the team has never played a home game on a Tues-
day. Let us suppose that for the ¬rst time the team is scheduled to
play at home on a Tuesday. If all we have to go on are the facts I
have just mentioned, can we draw any conclusions about whether
the team is likely to win the Tuesday home game? I do not think
so. The reason we cannot is that we have two con¬‚icting simple
inductive arguments that nullify each other. On the one hand, we
can reason that because the team has won all past home games, it is
likely to win this one. On the other hand, we can reason that because
the team has lost all past Tuesday games, it is likely to lose this one.
The two arguments cancel each other out, leaving us with no idea
Beyond Nature

whether the team is more likely to win or lose the Tuesday home
I believe that a similar situation obtains with respect to the induc-
tive case that Joy can be satis¬ed. In the example I just described, the
problematic game belongs to two relevant classes of games: games
played by the team at home, and games played by the team on
Tuesdays. Its membership in the ¬rst class suggests victory, whereas
its membership in the second class suggests defeat. Similarly, Joy
belongs to two relevant classes of desires. One of these classes is
the class of natural desires, and its membership in this class (let us
suppose for the moment) suggests that Joy can be satis¬ed. But Joy
also belongs to a second class of desires, namely, the class of human
desires for things that are not part of the known natural universe.
And Joy™s membership in this class of desires suggests that it is not
capable of satisfaction. The reason for this is simply that most desires
in this second category are for things that do not exist. Throughout
human history, people have had desires involving all kinds of entities
that do not belong to the known natural universe, and it is clear that
the vast majority of these desires involve objects that do not exist.
So, we have two opposed inductive arguments that cancel each other
out. On the one hand, we can reason that because Joy belongs to the
class of natural desires, Joy probably can be satis¬ed. On the other
hand, we can reason that because Joy belongs to the class of human
desires for things that are not part of the known natural universe,
Joy probably cannot be satis¬ed. The inductive case that Joy can be
satis¬ed therefore fails, just as the inductive case that the team will
win its Tuesday home game fails.
A second (independent) weakness of the inductive approach lies
in the fact that, putting aside the issue of whether Joy can be satis¬ed,
it is clear that Joy is quite different from the other natural desires
that Lewis mentions. Those other desires share a number of features
that Joy lacks. For instance, they are all desires for things that are
part of the natural world. Furthermore, they are all such that their
satisfaction never brings permanent contentment and ful¬llment.
In maintaining that Joy is a desire for union with God, Lewis is
committed to the view that Joy is not a desire for something that
God and the Reach of Reason

is part of the natural world and that it is a desire whose satisfaction
brings (or would bring) eternal bliss. It is clear that Lewis conceives
of Joy as being unlike other natural desires in some important ways.
What, then, is the basis for maintaining that it is like them in that it is
capable of being satis¬ed? The inductive argument for the conclusion
that Joy can be satis¬ed seems to be countered by what we might
call a “meta-inductive argument”: All examined inductive arguments
from the nature of natural desires to the nature of Joy fail; therefore,
all such inductive arguments fail. Joy™s unique status among natural
desires seems to undermine the inductive case for the second premise
of the argument from desire.
Robert Holyer comments on this type of criticism as follows:

[O]ne could question the extent to which Joy is like other natural
desires. . . . However, even if the challenger has some success with this
line of argument, the most he could do is to diminish the similarity
between Joy and other natural desires and hence weaken the argu-
ment. The fact that it is still a desire, albeit of a somewhat different
sort, would seem to limit the extent to which this sort of challenge
could vitiate the argument.153

Holyer seems to be suggesting that no matter how different from
all other natural desires Joy turns out to be, the fact that it is a
natural desire will still constitute some reason to believe that Joy can
be satis¬ed. But this seems wrong. If Joy is suf¬ciently different from
all other natural desires, the fact that it is a natural desire will tell
us nothing at all about whether it is a desire that can be satis¬ed.
Here is a simple example that illustrates this point: If we know of a
particular swan that it is unlike all other swans in many signi¬cant
ways, the fact that it is a swan may give us no reason to believe
that it is white, even if all other swans we have observed are white.
The many known differences between this swan and other swans we
have observed means that the fact that it is a swan tells us little about
its particular characteristics.154 I conclude that the inductive case that
all natural desires can be satis¬ed is very weak at best, and further
that the second premise of our initial formulation of the argument

Beyond Nature

lacks adequate support. I see no good reason to believe that Joy, even
if it is part of human nature, is capable of satisfaction.
There is one ¬nal way of understanding the argument from desire
that merits consideration. We might construe it as based on an infer-
ence to the best explanation. As I noted earlier, it is natural to wonder
what explains the fact (if it is a fact) that Joy is part of human nature.
That God crafted us so that we would experience Joy would certainly
account for Joy™s existence. So we might understand the argument
this way:

Lewis™s Argument from Desire (Revised)

1. All normal human beings have an innate, natural desire (Joy)
that is for some thing, x, where x lies beyond the natural world.
2. The best explanation of (1) is that a mindlike Higher Power
instilled Joy in human nature.
3. Therefore, there is a mindlike Higher Power that instilled Joy in
human nature (from 1 and 2).

Is there a plausible explanation for this (alleged) feature of human
nature that does not involve a supernatural entity? I believe that
we can develop the basic elements of just such an explanation by
drawing once again on evolutionary psychology. Interestingly, Lovell
considers a suggestion along these very lines:

We can . . . understand how a creature acquires its natural traits in
evolutionary terms. Evolution will encourage those traits that are an
aid to survival and reproduction. If the possession of a particular desire
would be an aid to these, then evolution will favour it. In this manner
the trait will become prevalent in, and perhaps therefore natural to,
the descendants of the creature in which it arose. But many desires
would not be at all useful to survival or reproduction unless they are
desires for things that can really be obtained. . . . [I]n an environment
that contains lakes or is near to the sea the desire to go swimming
might (for some reason) confer an evolutionary advantage. But in an
environment void of such expanses of water, the desire might only
encourage a fruitless search for places to swim, which would be a
waste of time and energy and would therefore be no aid to survival

God and the Reach of Reason

or reproduction. In this way, we come to see why it is that so many
natural desires have correlating objects, but we do so in a way that
need not require that they all do. Suppose a desire for communion
with God . . . does confer an evolutionary edge; why would its doing
so require the existence of God . . . ?155

This is a promising start, but a crucial element is missing: We need
some plausible account of how the desire for something beyond the
natural world provides those beings that have it with an evolutionary
advantage over those that lack it even if no such transcendent object
Two important facts about Joy as it is described by Lewis consti-
tute the foundation of such an account. The ¬rst important fact is
that one of the main effects of Joy is that it prevents a person from
deriving lasting contentment from earthly things. This fact is impor-
tant because deriving lasting contentment from earthly things can be
quite disadvantageous, evolutionarily speaking. Dissatisfaction can
bene¬t us in the long run. This idea is evident in Ronald Dworkin™s
criticism of the use of psychotropic drugs as a “treatment” for ordi-
nary unhappiness (as opposed to genuine psychological disorders).
Dworkin labels the happiness produced in this way “Arti¬cial Happi-
ness” and observes that “[p]eople with Arti¬cial Happiness don™t feel
the unhappiness they need to move forward with their lives.”156 To
see the evolutionary drawbacks of lasting contentment, consider a
male human who is perfectly content as long as his basic needs (food,
shelter, and sex) are satis¬ed. Once such needs are satis¬ed, he will
have no motivation whatsoever to acquire additional wealth, power,
status, or success; indeed, he will have no motivation to do anything
at all, other than perhaps ensure that his basic needs continue to be
satis¬ed. Contrast this male with a second male who has the same
basic drives but who never achieves lasting contentment, regardless of
his earthly accomplishments. Everything else being equal, the second
male will likely do better than the ¬rst in the competition for limited
resources and access to the most desirable females. Indeed, a male
who would derive lasting contentment from some level of earthly
success, no matter how high, always stands the risk of being bested
by a male who would never derive lasting contentment from any
Beyond Nature

amount of earthly goods. Evolutionarily speaking, a good strategy is
never to be entirely satis¬ed with one™s lot in life. Lasting content-
ment breeds inaction, which in turn breeds reproductive failure “ at
least when the competition is not entirely content. Robert Wright
describes the basic idea here as follows:

[W]e are designed to feel that the next great goal will bring bliss, and
the bliss is designed to evaporate shortly after we get there. Natural
selection has a malicious sense of humor; it leads us along with a series
of promises and then keeps saying “Just kidding.” . . . Remarkably, we
go our whole lives without ever really catching on.157

The second important fact about Joy is that what Joy is a desire
for is not at all obvious. Lewis spent years trying to understand the
nature of Joy. This fact is important because if one wants something
but doesn™t know what, one is likely to conclude that what one wants
is more of some earthly good. So as long as the true object of Joy
is unclear, Joy functions as a sort of catalyst for one™s more fun-
damental desires. And those fundamental desires, of course, have
been instilled by evolution because they tend to produce behav-
iors that lead to evolutionary success. In light of all this, it is easy
to see why the disposition to experience Joy might be selected for
by evolution: No matter how many earthly goods we acquire, Joy
relentlessly whispers, “It™s not enough.” Joy prevents us from falling
into the genetically disadvantageous trap of lasting contentment; it
ensures that our basic ¬tness-enhancing drives will never cease to
operate entirely. It is Lewis himself who suggests perhaps the sim-
plest and clearest illustration of this concept: “Joy is not a substitute
for sex; sex is very often a substitute for Joy.”158 And Lewis™s very
next sentence captures the more general relationship between Joy
and the basic drives that I am attempting to describe here: “I some-
times wonder whether all pleasures are not substitutes for Joy.”159
Joy can function in the way I have suggested even if its object does
not exist. By causing us to strive for the in¬nite, it prevents us
from being entirely satis¬ed by the ¬nite, and in this way causes
us to survive and reproduce more successfully than we otherwise
God and the Reach of Reason

One objection that might be raised against this account is that it
is merely a “just so story.” The term “just so story” is taken from
Rudyard Kipling™s book by that name. The book consists of fanci-
ful accounts of how various animals got to be the way they are “
for example, “How the Whale Got His Throat,” “How the Leopard
Got His Spots,” and “The Beginning of the Armadillos.” The point of
labeling an evolutionary explanation of some trait a “just so story”
is to suggest that the explanation is merely a plausible (or plausible-
sounding) possibility that is not supported by empirical evidence.
With respect to the evolutionary explanation of Joy that I have
sketched, the point is well taken, as I have provided no empirical
evidence to support my account. However, in the present context,
this point is not damaging. The reason is that the theistic explanation
of Joy lacks empirical evidence as well.160 What we have are essen-
tially two competing “just so stories.” If this is right, then we have
a stalemate. And a stalemate in this context means failure for the
argument from desire, which is supposed to provide some positive
reason to believe that a transcendent object exists.
It seems, then, that we are left with two possible explanations for
the fact that Joy is part of human nature. The ¬rst explanation is
that the disposition to experience Joy has been selected for by evo-
lution because it enhances the ¬tness of those beings that have it,
even though Joy™s object does not exist. The second explanation is
that “we cry out for eternity, because God has put such desire into
our hearts. Our hearts are restless until they rest in God, because
they were designed by God to rest in God alone.”161 The ¬rst of
these explanations may seem a bit odd or even disturbing in that it
involves the claim that it is in our nature to desire something that
does not exist. As Lovell observes, “[r]ather than helping the athe-
ist avoid the conclusion that the world is ultimately irrational, this
objection simply offers us a way of understanding how it is that the
world came to be so absurd.”162 This is exactly right; but I do not see
that the mere fact that a view is disturbing or implies that the world
is absurd (in the relevant sense) constitutes a reason for thinking
that the view in question is false. Indeed, evolutionary psychology
predicts many similar oddities. For instance, it predicts that human
Beyond Nature

beings will tend to hold a number of false beliefs, that families will
tend not to be perfectly harmonious units but instead will produce
various kinds of con¬‚ict among their members, and that contempo-
rary humans will be susceptible to various phobias involving things
that, for most of us, represent little or no danger, like spiders and
small animals.163 Some thinkers attribute more extreme absurdities
to evolution. For example, Arthur Koestler suggests that evolution-
ary processes are responsible for a lack of coordination between older
and newer parts of the human brain and that this lack of coordina-
tion is one important cause of, among other things, the remarkably
violent history of the human race. Koestler writes: “[W]hen one con-
templates the streak of insanity running through human history, it
appears highly probable that homo sapiens is a biological freak, the
result of some remarkable mistake in the evolutionary process.”164
From the point of view of evolutionary psychology, then, the pres-
ence in human nature of a desire that cannot be satis¬ed is not
particularly surprising.
The upshot of all of this is that the fact (if it is a fact) that Joy is part
of human nature does not constitute much of a reason for thinking
that a transcendent entity exists. The initial version of the argument
from desire that we considered is unconvincing because none of the
ways of supporting its second premise (that all natural desires can be
satis¬ed) succeeds. The revised version of the argument fails because
there is a naturalistic explanation for the (alleged) fact that Joy is part
of human nature that is at least as plausible as the explanation that
involves a transcendent being.


Toward the beginning of this chapter I suggested that none of the
three arguments we have considered was intended by Lewis to con-
stitute a decisive proof of the existence of a Higher Power but rather
that he intended the three together to amount to a solid cumulative
case for the existence of such a Power. Now that we have examined
each of Lewis™s three arguments in some detail, it is worth consider-
ing the overall weight of Lewis™s cumulative case. As should be clear
God and the Reach of Reason

by now, my own view is that the cumulative case constituted by the
three arguments is not terribly weighty. It seems to me that recent
work in evolutionary psychology has done much to weaken Lewis™s
overall argument.
Nevertheless, Lewis™s arguments were ahead of their time in at
least two ways. First, they all rest on a substantive conception of
human nature. The rise of behaviorism in psychology in the 1920s
marked the rejection by many psychologists of the very idea of
human nature, a rejection that lasted for around ¬fty years.165 It is
only relatively recently that the idea of human nature has returned to
the psychological mainstream, and, interestingly, this is due in large
part to the rise of evolutionary psychology.166 Evolutionary psychol-
ogy and Lewis™s Christian apologetics therefore rest on a common
axiom: That there is such a thing as human nature. They differ, of
course, when it comes to explaining how this human nature came
to be.
Second, two of Lewis™s three arguments “ the moral argument
and the argument from reason “ involve topics that are the sub-
ject of much work in contemporary philosophy. Whether and how
morality and intentionality can ¬t into a nontheistic universe are
hotly contested issues on the current philosophical scene. Whether
they realize it or not, many contemporary naturalists are grappling
with challenges posed by Lewis.
Having examined Lewis™s arguments for the existence of a Higher
Power, we are now ready to turn to his attempt to provide at least
a partial case for the conclusion that this Higher Power is in fact
the God of Christianity. This attempt is contained in Lewis™s book
Miracles, in which he tries to establish the plausibility of miracles in
general and of the Resurrection of Christ in particular. Miracles is, in
part, a direct response to Hume™s famous essay “Of Miracles.” Thus,
in the next chapter we ¬nd ourselves once again in the midst of a
clash between those twin titans, Hume and Lewis.




Graham Greene˜s short story “The Second Death” centers on the
deathbed fears of a sinful man. His fears stem from a mysterious
event that happened while he was a child. He had been declared dead
and was being carried out to be buried “when a doctor stopped them
just in time.”1 Re¬‚ecting on this childhood incident, the dying man

[W]hen I came around that other time, I thought I™d been dead. It
wasn™t like sleep at all. Or rest in peace. There was someone there all
round me, who knew everything. Every girl I™d ever had. Even that
young one who hadn™t understood. . . . it must have been a dream,
mustn™t it? The sort of dream people do get when they are ill. And I
saw what was coming to me too. I can™t bear being hurt. It wasn™t fair.
And I wanted to faint and I couldn™t, because I was dead. . . . [S]uppose
it was true. Suppose I had been dead. I believed it then, you know,
and so did my mother. But you can™t trust her. I went straight for a
couple of years. I thought it might be a sort of second chance. Then
things got fogged and somehow . . . [i]t didn™t seem really possible. It™s
not possible.2

The narrator of the story is the dying man™s companion, who has
rushed to his bedside to try to comfort him as he faces death. He
assures the dying man that “[m]iracles of that sort don™t happen
nowadays” and “anyway, they aren™t likely to happen to you, are

God and the Reach of Reason

they?”3 The dying man seizes on his friend™s suggestions but ulti-
mately is unable to drive away his fears before dying:

“There were some others,” he said. “But the stories only went round
among the poor, and they™ll believe anything, won™t they? There were
lots of diseased and crippled they said he™d cured. And there was a man,
who™d been born blind, and he came and just touched his eyelids and
sight came to him. Those were all old wives™ tales, weren™t they?” he
asked me, stammering with fear, and then lying suddenly still and
bunched up at the side of the bed.4

Among the questions raised by the story is this one: Is it ever reason-
able to believe that a miracle has occurred? The story suggests that
it can be dif¬cult even for those directly involved in miracles to be
sure that a miracle has actually taken place.5 Surely the dif¬culty is
even greater when one is trying to assess the accuracy of a second-
hand report of a miracle. Is it ever reasonable to believe that such
reports are true?6 This question is the central topic of the present


In eighteenth-century Europe, a debate about miracles raged. More
precisely, a debate about the miracles associated with Christianity,
particularly the Resurrection of Christ, raged. The debate arose as a
consequence of the fact that many Christian philosophers of the time
maintained that the occurrence of the Christian miracles, particularly
the Resurrection, provide the basis for a proof of the central claims
of Christianity. Samuel Clarke bluntly put it this way in 1705: “The
Christian Revelation is positively and directly proved, to be actually
and immediately sent to us from God, by the many infallible Signs
and Miracles, which the author of it worked publicly as the Evidence
of his Divine Commission.”7 One of the more popular presentations
of this sort of argument at the time was Thomas Sherlock˜s The Tryal
of the Witnesses of the Resurrection of Jesus, which employed the clever
format of a jury trial on the question “Whether the Witnesses of the


Resurrection of Christ are guilty of giving false evidence or no?”8
Following presentations by “Mr. A,” who argues in favor of the guilt
of the apostles, and “Mr. B,” who defends their innocence, the jury
returns a unanimous verdict: “Not guilty.”9
The earthly stakes in this debate were high, particularly for those
who criticized the argument for Christianity based on miracles.
Thomas Woolston, a well-known critic of the argument, was con-
victed of blasphemy for his work Six Discourses on the Miracles of Our
Savior. Unable to pay the ¬ne for blasphemy, he died in prison in
1733.10 Another critic, Peter Annett, who wrote a critical response
to Sherlock™s Tryal, was eventually sentenced to be pilloried for the
views he put forth in a later work.11
Hume™s own contribution to this debate is section X, “Of Mira-
cles,” of his Enquiry Concerning Human Understanding, the ¬rst version
of which appeared in 1748, thirty-one years prior to the publica-
tion of his Dialogues Concerning Natural Religion. In “Of Miracles,”
Hume never explicitly mentions the Resurrection at all, and he
refers to Christianity by name only in the last two paragraphs.
He casts the topic of his essay as miracles in general rather than
any particular alleged miracle. Despite this, the ongoing debate
about miracles at the time together with the fact that Hume repeat-
edly refers to the dead being raised makes it clear that the cen-
tral miracle with which Hume is concerned is the Resurrection of
These days, “Of Miracles” is widely considered to be one of the
most important texts on miracles ever written, even by those who
¬nd it unconvincing. It is standard reading for courses in the philos-
ophy of religion. And it apparently impressed Lewis enough that he
wrote a book-length response to it, for, as we shall see, the central
argument of Lewis™s Miracles culminates with a response to Hume™s
essay. But before examining “Of Miracles” and Lewis™s reply, it will
be useful to consider another anti-miracle argument suggested by
Hume as well as Lewis™s response to it. In the course of responding
to this argument Lewis introduces an analogy that is central to his
response to Hume™s argument in “Of Miracles.”

God and the Reach of Reason


Speakers using the expression “Hume™s argument against miracles”
are typically referring to the argument found in Hume™s essay “Of
Miracles.” But Hume™s writing contains another argument worthy
of this title. In The Natural History of Religion, Hume describes a view
often called “Deism”:

Many theists, even the most zealous and re¬ned, have denied a par-
ticular providence, and have asserted, that the Sovereign mind or ¬rst
principle of all things, having ¬xed general laws, by which nature
is governed, gives free and uninterrupted course to these laws, and
disturbs not, at every turn, the settled order of events by particular

According to Deism, God sets up the laws that govern the physical
universe and then simply lets nature take its course according to
these laws. Deists deny that God engages in “particular volitions”;
once nature is up and running, God does not intervene in any way.
In short, the God of Deism does not work miracles.
Why accept Deism? On the same page Hume suggests that
“[c]onvulsions in nature, disorders, prodigies, miracles” are “the
most opposite to the plan of a wise superintendent.”13 The idea seems
to be that the only reason God could have for performing miracles
would be to rectify some previous error or oversight. Miracles are
improvised patches or “¬x-its.” They are crude devices for which
an all-knowing God would have no use. This suggests the following
argument against miracles:

Hume™s Deistic Argument

1. An omnipotent, omniscient, and morally perfect God would
never make mistakes.
2. If (1), then such a God would never perform miracles.
3. Therefore, an omnipotent, omniscient, and morally perfect God
would never perform miracles.

In Miracles, Lewis provides what is, by my lights, a devastating
refutation of this argument. Lewis™s strategy is to reject the claim

that justi¬es the second premise of the argument, the claim that the
only possible function of a miracle is to rectify a divine blunder.
The strongest component of Lewis™s response to the argument
introduces an analogy that lies at the heart of his response to Hume™s
argument in “Of Miracles.” This is “the analogy between God™s rela-
tion to the world, on the one hand, and an author™s relation to his
book on the other.”14 Lewis ¬rst observes that “miracles or abnor-
mal events may be bad art, or they may not.”15 They are bad art (and
indicate an error or lack of skill on the part of the author) when they
are used as ad hoc plot devices: “The ghost story is a legitimate form
of art; but you must not bring a ghost into an ordinary novel to get
over a dif¬culty in the plot.”16 On the other hand, they are not bad
art (and hence do not indicate an error on the part of the author) if
they are “what you are really writing about.”17 The application of this
point to Hume™s Deistic argument is as follows. If the Resurrection,
for instance, is really one of the main things that God™s “story” is
about “ if it is one of the main purposes of the entire Creation “ then
its occurrence, even though it involves divine interference with the
course of nature, does not indicate a previous error on God™s part.
And this is exactly Lewis™s position:

Some people probably think of the Resurrection as a desperate last
moment expedient to save the Hero from a situation which had got out
of the Author™s control. . . . [On the contrary,] Death and Resurrection
are what the story is about; and had we but eyes to see it, this has been
hinted on every page, met us, in some disguise, at every turn. . . . If you
have hitherto disbelieved in miracles, it is worth pausing a moment
to consider whether this is not chie¬‚y because you thought you had
discovered what the story was really about? “ that atoms, and time
and space and economics and politics were the main plot? And is it
certain you were right?18

Miracles might be crude patches to keep the universe running along
as it must so that God can secure some further goal, or they might
be among the things for the sake of which everything else exists.
Hume™s Deistic argument assumes that the former option is the
only possibility; therefore, that argument fails. Lewis™s response here
seems to be me to be entirely convincing. However, as the title of
God and the Reach of Reason

this section suggests, this is but a preliminary skirmish. Hume™s main
assault on miracles is to be found in “Of Miracles.” It is to that work
that we now turn.


Hume™s “Of Miracles” has been the object of an astounding amount
of commentary. There has been tremendous debate over what the
argument or arguments of the essay are, and whether those argu-
ments are any good. Contemporary philosophical work on the essay
has failed to produce anything even remotely resembling a consen-
sus concerning these issues. In his 2000 book Hume™s Abject Failure,
John Earman declares that the argument of “Of Miracles” is “largely
derivative” and “almost wholly without merit where it is original.”19
Earman™s critique ends with the assertion that Hume™s essay is “a
confection of rhetoric and schein Geld.”20 On the other hand, in his
2003 book A Defense of Hume on Miracles, Robert Fogelin defends the
view that “Hume™s treatment of miracles, when properly understood,
exhibits a level of richness, subtlety, coherence, and force not gen-
erally appreciated.”21
It is with considerable trepidation that I enter into such con-
tentious (and crowded!) philosophical waters. In what follows I
attempt to formulate what I take to be Hume™s central argument
in “Of Miracles” as clearly as possible. I will provide textual support
for my interpretation where appropriate, and I will sometimes draw
on the words of other commentators to explain or emphasize par-
ticular points, but material pertaining to scholarly debates about the
proper interpretation of Hume™s work will be largely relegated to the
notes. The task before us is to get as clear as possible on what Hume
tries to prove in “Of Miracles” and how he tries to prove it.
Let us begin with the ultimate conclusion of Hume™s central argu-
ment. The conclusion of the argument is not that miracles are impos-
sible. It is not that miracles never in fact occur. It is not that a perfect
God would never perform miracles (as in the Deistic argument). It
is not that it is never reasonable for anyone to believe that a miracle
has occurred. It is not even that it is never reasonable for anyone to

believe that a miracle has occurred solely on the basis of testimony.
Hume™s conclusion, rather, is that “no human testimony can have
such force as to prove a miracle, and make it a just foundation for any
system of religion.”22 The second clause of this statement is impor-
tant. It indicates that Hume™s conclusion applies not to all testimony
but only to a particular kind of testimony. Speci¬cally, it applies to
what we may call religious testimony, which we may de¬ne as testi-
mony by human beings that is intended to support a particular “sys-
tem of religion.” Hume™s claim is that it is never reasonable to believe
that a miracle has occurred on the basis of religious testimony alone.
To see how Hume endeavors to establish this conclusion, we will
need to get clear on his understanding of some key concepts, partic-
ularly the concepts of proof and miracle. Let us begin with the former
concept. Early in “Of Miracles,” Hume describes the proper procedure
to follow when it comes to forming our beliefs about what he calls
“matters of fact.”23 Into this category fall all claims that cannot be
known a priori, or independently of experience. Historical claims “
including, for instance, Christian claims about the Resurrection of
Christ “ fall into this category. Hume writes:

A wise man . . . proportions his belief to the evidence. In such conclu-
sions as are founded on an infallible experience, he expects the event
with the last degree of assurance, and regards his past experience as a
full proof of the future existence of that event. In other cases, he pro-
ceeds with more caution: He weighs the opposite experiments. . . . A
hundred instances or experiments on one side, and ¬fty on another,
afford a doubtful expectation of any event; though a hundred uniform
experiments, with only one that is contradictory, reasonably beget a
pretty strong degree of assurance.24

In every observed instance, night has been followed by a sunrise. This
observed constant conjunction constitutes what Hume in the passage
just quoted calls “infallible experience.” Therefore, the observed con-
stant conjunction constitutes a proof that the next night will be fol-
lowed by a sunrise. The important thing to note here is that the claim
that experience provides us with a proof that the sun will rise tomor-
row does not imply that the relevant experiences render the prob-
ability that the sun will not rise tomorrow zero.25 Rather, the claim
God and the Reach of Reason

that experience affords us with a proof that the sun will rise tomor-
row implies that the probability that the sun will not rise tomorrow
is extremely small “ so small that, under most circumstances, it is
perfectly reasonable to put aside the remote possibility that it will
not.26 Thus, in the light of his past experience, a wise man will be
extremely con¬dent in his belief that the sun will rise tomorrow. At
the other extreme, suppose I am drawing colored marbles from an
urn. I have drawn one hundred marbles, ¬fty of which are white,
and ¬fty of which are black. Under these circumstances, I should
not believe that the next marble I draw will be white, nor should
I believe that it will be black; rather, I should suspend judgement
(although I can be con¬dent that it will be white or black rather than
some third color).
In the midst of the passage just quoted, Hume makes the fol-
lowing remark: “All probability . . . supposes an opposition of experi-
ments and observations, where the one side is found to overbalance
the other, and to produce a degree of evidence, proportioned to the
superiority.”27 On the basis of this claim and the rest of the passage,
I attribute to Hume the following principle:

Probability Principle: We should rate the occurrence of event A as
more probable than the occurrence of event B if and only if: The
evidence provided by our experience supports the occurrence of
A to a greater extent than it supports the occurrence of B.

Next, let us consider how Hume conceives of miracles. He offers the
following de¬nition: “A miracle is a violation of the laws of nature.”28
This is somewhat tricky; If laws of nature are to be understood as
true, exceptionless generalizations (e.g., the dead never rise), then
the very concept of a miracle would be self-contradictory (since true
generalizations, by de¬nition, are never violated). This understand-
ing of laws of nature would provide Hume with a quick but utterly
unconvincing disproof of the very possibility of miracles. Fortunately,
the text makes it pretty clear that Hume did not intend to defend this
disappointing argument.
A better way of understanding how Hume conceives of laws of
nature in this context is proposed by both Earman and Fogelin.29

The idea is that when Hume speaks of “laws of nature” here, he is
thinking of propositions or statements to the effect that certain reg-
ularities hold without exception. Such propositions merit the title
“laws of nature” only if there is a lot of experience-based evidence
that supports them and none (putting aside any religious testimony)
that tells against them; they are, as Hume puts it, supported by “a
¬rm and unalterable experience.”30 A miracle, then, is any event that
violates one of these well-supported generalizations. On this under-
standing of miracles, they are possible, but the available experience-
based evidence (again, excluding religious testimony) indicates that
they are very unlikely. As Hume puts it, “the proof against a miracle,
from the very nature of the fact, is as entire as any argument from
experience can possibly be imagined.”31 In light of all of this, we can
state the ¬rst premise of Hume™s argument as follows:

1. For any miracle M, our experience provides us with a (Humean)
proof that M did not occur (excluding any religious testimony
that supports the occurrence of M).

It should be noted that this premise is consistent with the possibility
of there being testimony that would make it reasonable to believe
that a miracle has occurred. However, such testimony would have to
conform to very high evidential standards in order to outweigh the
other evidence against the occurrence of the miracle. Hume remarks
that “no testimony is suf¬cient to establish a miracle, unless the testi-
mony be of such a kind, that its falsehood would be more miraculous,
than the fact, which it endeavors to establish.”32
To arrive at the next main premise of Hume™s argument, let us
consider some of what Hume has to say about testimony. He begins
by noting that it is experience that tells us how reliable testimony
is: “[O]ur assurance in any argument of this kind is derived from
no other principle than our observation of the veracity of human
testimony, and of the usual conformity of facts to the reports of
witnesses.”33 Thus, testimony constitutes evidence that the reported
events actually occurred only to the extent that testimony has been
accurate in the past. If all observed human assertions had also been
observed to be true, then a single piece of testimony would constitute
God and the Reach of Reason

a proof (in the Humean sense) that the reported events had occurred.
At the other extreme, if all observed humans were known to be
pathological liars, then testimony would not count as evidence for
the occurrence of the reported events at all. As it stands, the reliabil-
ity of testimony lies somewhere between these two extremes, and
some kinds of testimony have been observed to be more reliable than
In Part II of “Of Miracles,” Hume argues that past experience indi-
cates that religious testimony in support of miracles is a particularly
unreliable brand of testimony. This is an important part of Hume™s
argument, and it indicates a limitation on the scope of his argument.
The argument is not intended to show that it is never reasonable to
believe that a miracle has occurred on the basis of testimony; rather,
it is intended merely to show that it is never reasonable to believe
that a miracle has occurred on the basis of a certain kind of testimony “
namely, religious testimony, as de¬ned earlier.34
The heart of Hume™s position here is straightforward: History is
¬lled with reports of miracles that support one system of religion or
another, and most of these reports have subsequently been revealed
to be false. “How many stories of this nature, have, in all ages, been
detected and exploded in their infancy? How many more have been
celebrated for a time, and have afterwards sunk into neglect and
oblivion?”35 In addition, Hume notes that religious testimony about
miracles is “observed chie¬‚y to abound among ignorant and bar-
barous nations” but that “as we advance nearer the enlightened
ages” such accounts are viewed with increasing skepticism.36 Finally,
Hume observes that the various systems of religion each have their
own reports of miracles that support the system in question. How-
ever, since (at least in the case of monotheistic religions) the sys-
tems are incompatible with each other, the majority of these alleged
miracles must not have happened. All of these considerations are
intended to show that we can be con¬dent that most religious testi-
mony in support of miracles is false.
Hume also offers a suggestion about human nature that helps
to explain why such accounts are often widely believed, at least
for a time. He proposes that this is a consequence of “the strong

propensity of mankind to the extraordinary and the marvelous.”37
In short, people simply love wild stories. Hume provides two illustra-
tions of this aspect of human nature. The ¬rst is the “greediness” with
which “the miraculous accounts of travelers are received,” including
“descriptions of sea and land monsters, their relations of wonderful
adventures, strange men, and uncouth manners.”38 This passion for
stories about land and sea monsters continues unabated in our own
time, as evidenced by the persistent myths of Bigfoot and Nessie,
the Loch Ness monster. We may add to this catalog of wild stories
accounts of space monsters in the form of tales of alien abduction.
Hume™s second illustration of our propensity to believe juicy stories
also persists to the present day:

There is no kind of report, which rises so easily, and spreads so
quickly . . . as those concerning marriages; insomuch that two young
persons of equal condition never see each other twice, but the whole
neighborhood immediately join them together.39

Moreover, Hume argues, this propensity is least restrained when it
comes to religious testimony: “[I]f the spirit of religion join itself to
the love of wonder, there is an end of common sense; and human tes-
timony, in these circumstances, loses all pretensions to authority.”40
As Fogelin puts it, “for Hume, it is an empirical fact, amply illustrated
by history, that testimony concerning religious miracles is notori-
ously unreliable.”41 This gives us Hume™s second premise:

2. For any religious testimony, T, to the effect that miracle M
occurred, our experience contains much evidence that T is false.

From premises (1) and (2) we may infer:

3. So: For any religious testimony, T, to the effect that miracle M
occurred, the evidence provided by our experience supports the
falsity of T to a greater extent than it supports the occurrence
of M.

Recall Hume™s Probability Principle:

Probability Principle: We should rate the occurrence of event A as
more probable than the occurrence of event B if and only if: The
God and the Reach of Reason

evidence provided by our experience supports the occurrence of
A to a greater extent than it supports the occurrence of B.

Hume illustrates the Probability Principle in these lines:

When anyone tells me, that he saw a dead man restored to life, I
immediately consider with myself, whether it be more probable, that
this person should either deceive or be deceived, or that the fact, which
he relates; should really have happened.42

From the Probability Principle and premise (3), we may infer:
4. So: For any religious testimony, T, to the effect that miracle M
occurred, we should rate the falsity of T as more likely than the
occurrence of M.
And this together with Hume™s maxim that “a wise man . . . pro-
portions his belief to the evidence” implies that a wise man will,
when confronted with religious testimony for a miracle, believe that
the testimony is false rather than that the miracle occurred.43 This
gives us Hume™s conclusion that “no human testimony can have
such force as to prove a miracle, and make it a just foundation for
any system of religion.”44
Here, then, is the complete formulation of Hume™s central argu-
ment in “Of Miracles”:

Hume™s Argument against Miracles

1. For any miracle, M, our experience provides us with a (Humean)
proof that M did not occur (excluding any religious testimony
that supports the occurrence of M).
2. For any religious testimony, T, to the effect that miracle M
occurred, our experience contains much evidence that T is false.
3. So: For any religious testimony, T, to the effect that miracle M
occurred, the evidence provided by our experience supports the
falsity of T to a greater extent than it supports the occurrence of
M (from 1 and 2).
4. So: For any religious testimony, T, to the effect that miracle M
occurred, we should rate the falsity of T as more likely than the
occurrence of M (From 3 and the Probability Principle).

5. Therefore, it is never reasonable to believe that miracle M
occurred solely on the basis of religious testimony T to the effect
that M occurred (from 4 and the maxim that a wise man pro-
portions his belief to the evidence).

It is worth observing that both premises (1) and (2) are required to
establish premise (3). Without premise (2), the possibility of religious
testimony that renders the occurrence of a miracle more probable
than the falsity of the testimony in question is left open. Without
premise (1), the possibility that the occurrence of a given miracle is
quite plausible independently of any religious testimony in support
of it is left open. And if this is a possibility, then it might be reasonable
for us to believe some such miracles on the basis of testimony of a
sort that is, in general, unreliable. Here is a simple example that
illustrates this point. Suppose I have been working on this chapter
in a windowless of¬ce all night. Larry the Liar, whose testimony is
generally unreliable, enters and informs me that the sun has just
risen. Under these circumstances, it may be reasonable for me to
believe what he says because the reported event is independently
plausible.45 Fogelin aptly describes how premises (1) and (2) work
together as follows:

[P]art 1 ¬xes the appropriate level of scrutiny for evaluating testimony
with respect to miracles; part 2 considers the quality of the testimony
that has hitherto been brought forth in support of religious miracles
and concludes that it comes nowhere near to meeting the appropriate
standards. . . . [T]he wise reasoner is fully justi¬ed in rejecting all testi-
mony given in support of a miracle intended to serve as the foundation
of a system of religion.46

As I suggested earlier, the nature of the eighteenth-century debate
about miracles together with Hume™s repeated reference to the rais-
ing of the dead indicates that, though he never mentions it specif-
ically, the religious miracle that “Of Miracles” is really about is the
Resurrection of Christ. Hume™s central message is: It is not reasonable
to believe that Christ was raised from the dead solely on the basis of
religious testimony (e.g., the Christian Bible) alleging that He was.
God and the Reach of Reason

As we have seen, Hume™s conclusion is that “no human testimony
can have such force as to prove a miracle, and make it a just foun-
dation for any system of religion.”47 The presence of the word “can”
here makes it tempting to attribute to Hume the view that there are
no conceivable circumstances under which religious testimony could
make it reasonable to accept that a miracle has occurred. It should
be clear by now that this is not Hume™s view. In ordinary speech, we
often say that certain things can or cannot occur, where such claims
are conditional on certain tacit assumptions. The context typically
makes it clear what these assumptions are. For instance, suppose I
launch into a diatribe outlining the mishmash of errors, distortions,
lies, and bad reasoning that typically issues forth from cable news
political “analysts.” I might conclude my diatribe with the assertion
that “no argument offered by a cable news political analyst can have
such force as to prove its conclusion.” What I mean, of course, is that
given the horri¬c track record of such arguments, it could never be rea-
sonable to accept the conclusion of such an argument on the basis
of the argument itself. Hume™s position, similarly, is that given the
horri¬c track record of religious testimony supporting the occurrence of mir-
acles (together with the evidence against the occurrence of any given miracle
provided by experience), it could never be reasonable to believe that a
miracle has occurred on the basis of such testimony alone. For all
Hume argues in “Of Miracles,” Christ may have risen from the dead.
Hume™s position is that, be that as it may, it is not reasonable for us
to believe that the Resurrection occurred on the basis of testimony
alleging that it has. Supporters of religion have cried “Miracle!” too
many times.


The opening chapter of Miracles contains the following lines:

Many people think one can decide whether a miracle occurred in the
past by examining the evidence ˜according to the ordinary rules of
historical inquiry™. But the ordinary rules cannot be worked until we
have decided whether miracles are possible, and if so, how proba-
ble they are. For if they are impossible, then no amount of historical


evidence will convince us. If they are possible but immensely improb-
able, then only mathematically demonstrative evidence will convince
us: and since history never provides that degree of evidence for any
event, history can never convince us that a miracle occurred. If, on the


. 4
( 8)