© 2007 The Philosophical Forum, Inc.

KANT, HUME, DARWIN, AND DESIGN: WHY

INTELLIGENT DESIGN WASN’T SCIENCE BEFORE
DARWIN AND STILL ISN’T

JONATHAN LOESBERG

There are two historical claims about the end of the argument from design and
they do not fit together well. According to the first, Hume’s Dialogues Concerning
Natural Religion made the argument untenable and marked its end. Although this
claim might be logically true, we know it not to be historically true since the
standard version of the argument usually cited, William Paley’s Natural Theology,
did not appear until 1802, well after Hume’s work first appeared in 1779 and in
full cognizance of it. And the Bridgewater Treatises, although they certainly had
a mixed reception, were written two decades later still. Thus, we get the second
claim, that Darwin’s theory of natural selection provided a natural mechanism that
explained the appearance of design in living things without recourse to a super-
natural being and thus undid the argument from design. Of course both these
claims might be taken to be true in a certain sense. One could argue that Hume
undercut the logic of the argument, making it philosophically untenable. Unfor-
tunately, however, philosophical arguments, even good ones, do not always sway
people. Darwin, by providing a mechanism to explain how organic function came
to be, thus provided a more sociologically powerful disconfirmation of the argu-
ment. One almost never finds this position, however. Philosophers who credit
Hume with ending the argument do not refer to Darwin while those who think
Darwin ended the argument from design usually find Hume’s argument less than
complete.
This debate has a relevance beyond getting an intellectual history articulated
correctly because the claim that Darwin undid the argument from design has led
to the current Intelligent Design movement. This movement maintains, first, that
the argument from design still holds as a scientific proof that one must construe at
least elements of nature as designed by an intellect and not resulting from natural

95
 JONATHAN LOESBERG

mechanisms and, second, that this renewed design argument makes Darwin’s
theory of natural selection scientifically questionable. And one must give the
position of Intelligent Design this much: If a scientific theory can in principle
disconfirm evidence for Intelligent Design, then it follows that design theory does
make a scientific claim in the sense of a claim that is empirically disconfirmable.
And further, if Intelligent Design could make its case, it would then call Darwin’s
disconfirmation and thus the sufficiency of natural selection as an explanation for
the order of biological organisms into question. Politically, this would mean that
as long as one discounts Hume’s arguments and holds natural selection to be
relevant to disconfirming design theory, and as long as Intelligent Design can hold
tenably that natural selection’s explanation of order in biology is less than fully
persuasive, then Intelligent Design will have earned a place in the scientific
curriculum. The response of most scientific polemicists to this problem has been,
as we will see, at least logically contradictory. The same authors will argue both
that Intelligent Design is not science and that science disproves its validity.1
I want to argue, in contrast, that the arguments of Kant and Hume are sufficient
to show that natural theology and thus Intelligent Design as well do not success-
fully make scientific arguments. In its original embodiment, natural theology
claimed to ally religion with the contemporaneous scientific theories rather than
refute them, to ground religion in empirical evidence. Both Hume, in Dialogues
Concerning Natural Religion, and Kant, not only in The Critique of Pure Reason
but more fully in The Critique of Judgment, undid this claim. They did not explain
the material cause of the appearance of design in nature, but showed that that
appearance could not function as evidence for an extra-natural intelligent
designer.2 Indeed, what Kant and Hume show, I think, is that limiting oneself to
seeking natural causes for natural effects is not—as Intelligent Design theorists
from Philip Johnson through Michael Behe and William Dembski have
insisted—a metaphysical principle with no inherent grounding in science but
rather a disciplinary condition of doing science, the only way to get the particular

1
As I will argue further down, it is consistent to hold that scientific claims made, for instance, by
Michael Behe that cellular mechanism cannot be explained by natural selection are incorrect and also
that whether or not they are incorrect, Behe’s theory of those mechanisms as proving the existence
of an intelligence that created them will not hold logically as a scientific claim. But Behe’s critics
have almost all held that the first error is evidence of the second, which is a different matter.
2
The Critique of Judgment actually goes further than either The Critique of Pure Reason or Hume
toward accepting nature as manifesting end-directed design since it assents to the intellectual
necessity of seeing nature as purposive and in providing that necessity a regulative role in making
moral judgments. Since Kant’s argument for the extent to which we may allow teleology to play a
role in our apprehension of nature and even the moral necessity of our giving it this role, though,
specifically eschews scientific claims, that part of his argument—the main part really—concerns us
here only to the extent that it indicates how someone sympathetic to the perspective of natural
theology and moved by the kind of “evidence” it can offer, nevertheless denied that it made its case.

96
 KANT, HUME, DARWIN, AND DESIGN

kinds of answers that science seeks within the terms of the evidentiary warrants it
demands.3 Consequently, the problem with Intelligent Design is not that it is
scientifically inaccurate but that it fails to have any scientific relevance at all. It
will follow, however, that Darwinists who use natural selection to question Intel-
ligent Design also fail in their attempt. It is not only logically possible both for
natural selection to be true and for the world to have been designed, but more to
the point, the status of Darwin’s theory cannot change the nature of the “evidence”
Intelligent Design cites. As a matter of both rhetoric and logic, it is as important
to recognize that Darwin does not provide counter-evidence to the argument from
design as it is to recognize that a design argument is not in fact a scientific one. In
fact, Intelligent Design is about as relevant to a course in biology as quantum
mechanics is to a course on Victorian literature.
Before discussing this intellectual history, I will give two classic statements of
the argument from design, noting the features from them that give the argument its
strength and its problems. The first, Thomas Aquinas’s fifth proof for the existence
of God in the Summa Theologica is admirably direct:

The fifth way is taken from the governance of the world. We see that things which lack intelligence,
such as natural bodies, act for an end, and this is evident from their acting always, or nearly always,
in the same way, so as to obtain the best result. Hence it is plain that not fortuitously, but designedly,
do they achieve their end. Now whatever lacks intelligence cannot move towards an end, unless it
be directed by some being endowed with knowledge and intelligence; as the arrow is shot to its
mark by the archer. Therefore some intelligent being exists by whom all natural things are directed
to their end; and this being we call God.4

3
Various scientific opponents of ID (e.g., see Niall Shanks, God, the Devil, and Darwin: A Critique
of Intelligent Design Theory [New York: Oxford UP, 2004] 142–45; and Mark Perakh and Matt
Young, “Is Intelligent Design Science?” Why Intelligent Design Fails: A Scientific Critique of the
New Creationism, ed. Matt Young and Taner Edis [New Brunswick, NJ: Rutgers UP, 2005] 185–96,
at 189) have claimed that science does not disallow supernatural explanations in principle but just
always has found the evidence for them insufficient. This claim is importantly wrong. To the extent
that science evaluates evidence importantly in terms of the repeatability of experimental results, its
system of justification disallows evaluating the likelihood of supernatural or non-natural one-time
interruptions to natural regularities. This may be described as a disciplinary limit rather than a
necessary limit to knowledge: philosophers and literary critics are constrained by their goals to be
open to different non-experimentally verifiable forms of evidence with less rigorous warrant and yet
that constraint, by itself, would not disallow them making accurate statements and true arguments,
though it probably demands holding the conclusions as less certain (assuming that both scientific and
other disciplines can make true statements, a problem on which this article has no bearing). And no
doubt the argument from design will always be taught in courses in philosophy and intellectual
history—probably as an exploded argument but not as one without relevance to those disciplines. But
even if scientific warrants are disciplinary limits and set boundaries to the kinds of questions science
may answer, history and the success of scientific inquiry have shown that they are not artificial limits.
4
Thomas Aquinas, The Summa Theologica of St. Thomas Aquinas, 2nd ed. Vol. 1. trans. The Fathers
of the English Domincan Province (London: Burns Oates & Washbourne, Ltd., 1914) 17.

97
 JONATHAN LOESBERG

One should note that Aquinas does not in fact speak of any kind of design in the
sense of complex patterns or interrelationship of parts. The word “designedly”
designates end-oriented structure or process. It means things that manifestly act
“for an end,” things that can be seen to be structured to serve a purpose (the Latin
is “ex intentione”). The having of ends is the manifestation of an intelligence
behind those things because unintelligent things cannot have ends unless directed
by some intelligence outside themselves. Since this issue will matter, one should
note that there is at this point no element of an argument from analogy (the simile
regarding the archer is illustrative but not really logically necessary). The argu-
ment presumes that the structure of natural things as end-directed is self-evident
and that things with such structures are self-evidently the products of intelligence.
The argument from analogy, in which one first looks at humanly created products
and specifies what about them, makes us conclude that they were designed and
then notices that natural objects have the same features, which both Hume and
Kant contest, gets articulated through the 17th and 18th centuries, first as a logical
shoring up of Aquinas’s case, and second as part of the enhanced ambition of the
argument.
We see the first purpose of the analogy if we jump forward to Paley, who,
writing in the wake of Hume’s critique of the argument from analogy, begins his
Natural Theology by insisting on that analogy. The book begins with a famous
walk on a heath in which one first encounters a stone and then a watch. The stone,
Paley says, for all we know, might have been there forever and offers no evidence
of how it was created. But the watch is a different case: “When we come to inspect
the watch, we perceive (what we could not discover in the stone) that its several
parts are framed and put together for a purpose, e.g. that they are formed and
adjusted as to produce motion, and that motion so regulated as to point to the hour
of the day.”5 Although Paley also points to the end-directed structure in the watch,
he adds an element to that structure: It has several parts and those parts work
together to produce the end. The watch thus shows evidence of contrivance, a
word that becomes central to Paley’s argument. Contrivance rather than merely
end-directed structure evidences that the watch is a product of an intelligent agent
because, while purpose evidences intelligence, as Aquinas argues, contrivance
inescapably evidences purposive form, for Paley, as mere end-directedness does
not. The appearance of end-directedness could occur by accident, but contrivance
shows a complexity that implies an intelligent planning needed to produce that
end. Paley, then, famously goes on to instance the human eye as exhibiting exactly
this contrived structure, as indeed working like the humanly produced telescope
but in a much more complex and articulated manner. Paley, as we will see, uses

5
William Paley, Natural Theology; or, Evidences of the Existence and Attributes of the Deity Col-
lected from the Appearances of Nature (Boston, MA: Gould, Kendall and Lincoln, 1850) 5.

98
 KANT, HUME, DARWIN, AND DESIGN

complexity to counter Hume’s suggestions that the analogy between the design of
human products and the design of natural objects is an uncertain one.6
Paley uses analogy, in the tradition of the 18th-century natural theologians such
as Bishop Butler, to make a more far-reaching argument than Aquinas, one that
claimed to use Enlightenment reason to prove that the world showed itself to be
ruled by a moral governor and that the features of that governor were Christian.
But he also needs analogy to make Aquinas’s conclusion seem more certain. The
appearance of end-directedness in nature could be accidental, but contrivance in
the service of end-directedness makes its occurrence by accident much less likely.
Thus, whenever we see both end-directedness and complexity in human activity,
we see intention and the same will be true in nature. Both Hume and Kant hone in
directly on both the analogy and its basis as inadequate as evidence and their
critiques are much more devastating than both sides of the current debate between
biologists and Intelligent Design theorists realize.

II

In going back to Hume to refute the argument from design, I am well aware that
his argument has not seemed that persuasive to numbers of philosophers. Hume’s
core argument is taken to be that: (1) the argument from design depends on an
analogy between human products and natural products that is less than certain,
and (2) to the extent that the analogy is an induction, it is even less certain,
inasmuch as it is based on a single example (ours is the only world we know).
Elliott Sober argues against Hume by reconstructing the argument from design as
not an analogy but the choice of the likeliest of two competing inferences (either
complex organisms came together by chance or they were designed).7 But this
argument is not very persuasive because it evades the question Hume asks, which
is why intelligence seems a likelier cause of design than mechanical causation or
accident. The answer to that question depends on how much the structures of
natural organisms appear analogous to the structures of human, purposive objects
and whether that appearance is reliable evidence. The question at issue thus
remains whether this is a good analogy or not and, then, further, what other
consequences, desirable or not, it might entail. Because Dialogues Concerning

6
In instancing contrivance, Paley does not introduce a new element to the argument. Hume was
already aware of its role since it was a common element in the 18th-century natural theology. But
Paley, whose knowledge of biology was far superior to Hume’s, sought to overcome Hume’s logical
hesitations with an overwhelming account of the intricacy of contrivance in the features of natural
organisms. The closeness in the language that both Paley and Darwin use to describe features of
nature and the wonder, albeit of different kinds, both convey at the spectacle of nature’s intricacy,
perhaps accounts for our sense that Darwin’s way of thinking about nature has replaced Paley’s.
7
Elliott Sober, Philosophy of Biology, 2nd ed. (Boulder, CO: Westview Press, 2000) 34–35.

99
 JONATHAN LOESBERG

Natural Religion is in the form of a dialogue, in which positions are tested by

dialogic argument and also, at times, by dramatic irony, Hume’s position must at
times be teased out. Virtually no writer on the issue of Darwin and design give
satisfactory accounts of the course of the argument and how he addresses the
problem of the analogy.8 To get to that argument more clearly, it will help to start
with Kant, who poses a more direct, logical argument, without the divagations of
dramatic dialogue. Here again, we need to go beyond the usually quoted para-
graph in The Critique of Pure Reason to the far more extensive discussion in The
Critique of Judgment.9 The analysis of Kant will show three parts to his argument:
First, in the familiar argument, the first Critique seems to allow the argument from
design to establish a lesser intelligent force but not a Divine Creator of the world.
But, second, even though Kant accepts that some objects have what he calls
objective purposiveness, only explanations of their structure via natural mecha-
nisms can be taken as material explanations of that structure and one should
always carry such explanations out as far as they will go because, third and most
importantly, the analogy between human and non-natural intelligent cause fails
precisely where it needs to succeed if it is to be meaningful. There is one final part
to Kant’s argument, however, that goes to the more extravagant claims on behalf
of the theory of natural selection made by Dennett and Dawkins: Natural expla-
nations of purposive structure can never really give a satisfactory explanation of
the final source of purposiveness any more than the argument from design can
adequately conclude the existence of a designer. Consequently, non-scientific
arguments may not be refuted scientifically. Following the reading of Kant and
Hume, I will then turn to how they first show the arguments not only of natural
theology but of Intelligent Design to fail to have scientific purchase and second
also show how Darwin therefore must be irrelevant to disproving the claims of
Intelligent Design theorists.
Even Kant’s basic argument against natural theology, in The Critique of Pure
Reason, read carefully is much more comprehensive than it appears:

[. . .] the purposiveness and harmonious adaptation of so much in nature can suffice to prove the
contingency of the form merely, not of the matter, that is not of the substance in the world. To prove
the latter we should have to demonstrate that the things in the world would not of themselves be
capable of such order and harmony, in accordance with universal laws, if they were not in their
substance the product of supreme wisdom. But to prove this we should require quite other grounds

8
Neither Dennett, Dawkins, and Ruse on the one side nor Dembski and Behe on the other (who show
little sign of having read Hume carefully at all) have anything like even accurate summaries of what
Hume says.
9
Only Ruse discusses Kant’s Critique of Judgment and he astonishingly seems to think that Kant
meant to argue that final purposes were material features of objects. See Michael Ruse, Darwin and
Design: Does Evolution Have a Purpose? (Cambridge, MA: Harvard UP, 2003) 46–50.

100
 KANT, HUME, DARWIN, AND DESIGN

of proof than those which are derived from the analogy with human art. The utmost, therefore, that
the argument can prove is an architect of the world who is always very much hampered by the
adaptability of the material in which he works, not a creator of the world to whose idea everything
is subject. This, however, is altogether inadequate to the lofty purpose which we have before our
eyes, namely, the proof of an all-sufficient primordial being. To prove the contingency of matter
itself, we should have to resort to a transcendental argument, and this is precisely what we have set
out to avoid.10

Kant’s argument, of course, is directed at Christian natural theologians, who took

the argument from design to prove the existence of, if not a specifically Christian
god, at least a monotheistic creator with recognizably human-like intention. He
thus seems to allow that the argument might admit the modest conclusion of a
more limited creator (a seeming possibility that, as we will see, Intelligent Design
will want to take advantage of). The problem with this reading is that it does not
really respond to the implications of Kant’s argument. The architect, he posits, is
not merely a smaller, more limited divinity. It is no divinity at all but a human-like
force struggling with obdurate matter. And that figure is unsuitable not only to the
“lofty purpose” of natural theology but even to any argument for a non-natural
creator.11
Kant’s objections to the argument from design become much more pointed in
his Critique of the Teleological Judgment precisely because the aim of that second
half of The Critique of Judgment is to justify judging natural objects in terms of
their having an objective purposiveness. In other words, Kant wants to grant what
natural theologians argue, that nature may be productively analyzed in ways that
see it as designed to produce ends, as teleologically structured. And yet he
rigorously denies that one may infer a designer from nature’s purposiveness. And
his reason for doing so explains better than any scientist does the basis for what,
as we will see, Dembski and Johnson claim is science’s mere prejudice in favor of
naturalism. Kant writes well before Darwin and has no knowledge of natural
selection as a proposed cause of the appearance of design. And he explicitly
rejects evolution not as theoretically impossible but as empirically unlikely.12 And
yet he insists that the fact that one may not know what mechanically caused the
appearance of purposiveness does not allow us to stop considering mechanical

10
Immanuel Kant, The Critique of Pure Reason, trans. Norman Kemp Smith (New York: St. Martin’s
Press, 1965) 522.
11
A version of the argument from design that asserted that beings from outer space created the
purposiveness of terrestrial organisms would be a scientific argument, albeit not, I think, a very good
one since we would expect natural beings to leave natural traces of their work. And even if such an
argument for a cause of design in terrestrial organisms gained acceptance, the question with regard
to the design of the extra-terrestrial MacGuffins would remain unchanged.
12
Immanuel Kant, The Critique of Judgment, trans. Werner S. Pluhar (Indianapolis, IN: Hackett
Publishing Company, 1987) 305–06.

101
 JONATHAN LOESBERG

causation as a possible explanation and his reason goes to why naturalism is not
a prejudice but a regulating principle for empirical thought. In discussing why no
ultimate explanation for the appearance of design in nature can be satisfying, Kant
ends with the theistic explanation, which he considers in many ways as the best of
the ones he proposes. It is quite clear that Kant has a certain respect for the
argument from design and his reservation is all the more powerful for his instinc-
tive sympathy with it:

For [theism to succeed in its explanation,] we would first of all have to prove, adequately for
determinative judgment, that the unity of a purpose, which we find in matter, could not possibly
result from the mere mechanism of nature. Only such a proof would entitle us to postulate
determinately that the basis of this unity lies beyond nature. But in fact all we can make out is that
the character and limits of our cognitive powers (which give us no insight in the first, inner basis of
even this mechanism) force us to give up any attempt to find in matter a principle [that implies]
determinate references [of this matter] to a purpose, so that we are left with no other way of judging
nature’s production of things as natural purposes than in terms of a supreme understanding as the
cause of the world. That basis, however, [holds] only for the reflective and not for the determinative
judgment, and is absolutely incapable of justifying any objective assertion.13

Kant’s logic here effectively constructs a logical basis for limiting scientific
explanation to natural causes, even without regard to issues of evidentiary warrant.
He argues that what may seem compelling evidence of a non-natural cause of a
material effect will only be compelling on the basis of certain knowledge about the
basic principle of matter and that knowledge is empirically unavailable. The most
that natural theologians can say they have shown is that the understanding is
incapable of thinking otherwise about purposive form than that it results from an
intelligent designer and that conclusion Kant says cannot justify making “deter-
minative” or scientific claims about the sources of seeming purposiveness.
But Kant has a further point to make about naturalism, one that will finally go
to the whole basis of making an analogy from human to non-natural design and
that will provide a conceptual background for understanding some of the turns in
Hume’s more empirical argument. Explaining purposiveness by reference to intel-
ligence, Kant notes, does not really explain anything:

Reason is tremendously concerned not to abandon the mechanism nature [employs] in its products,
and not to pass over it in explaining them, since without mechanism we cannot gain insight into the
nature of things. Even if it were granted that a supreme architect directly created the forms of nature
as they have always been, or that he predetermined the ones that in the course of nature keep
developing according to the same model, still none of this advances our cognition of nature in the
least; for we do not know at all how that being acts, and what its ideas are that are supposed to
contain the principles by which natural beings are possible, and [so] we cannot explain nature by

13
Ibid: 276–77.

102
 KANT, HUME, DARWIN, AND DESIGN

starting from that being [. . .] Or suppose we try to explain by ascending (in other words a
posteriori), i.e., we start from the forms of objects of experience because we think they display
purposiveness, and then, to explain this purposiveness, we appeal to a cause that acts according to
purposes: in that case our explanation would be quite tautologous and we would deceive reason with
[mere] words [. . .]14

Kant’s point, which will bring us to Hume, also brings us to the core of why
neither natural theology nor Intelligent Design theory can have any coherent
empirical claim to make. He notes that the only explanations of natural events that
actually give information about those events are natural causes and so we should
never abandon mechanistic explanation. Given that we cannot know anything
about a non-natural designer (not merely a Supreme Creator but any non-natural
designer), we cannot know how the designer did what he or she did or why or how
the designed object came into being materially.15 In effect, concluding the exist-
ence of a designer does not tell us anything new about the object, but merely
moves tautologously from the pointing out that one perceives purposiveness to
claiming one sees purpose.
Kant expands his refusal to disallow mechanistic explanations in principle to a
direct attack on the principle of analogous reasoning from human design to
non-natural design:

[. . .] from the fact that in the case of beings of the world we must attribute understanding to the
cause of an effect that we judge to be artificial, we can in no way infer by analogy that the same
causality that we perceive in man can also be ascribed to the being that is wholly distinct from
nature, with nature itself as [its effect] [. . .] The very fact that I am to think of divine causality only
by analogy with an understanding (a power we do not know in any being other than man, who has
a condition in the sensible) forbids me to attribute to this being an understanding in the proper sense
of the term.16

The argument here is that when we judge a thing to be made by humans, we do

more than merely move from a form of purposiveness to the inference of an
intelligent cause. We know how that cause works in the material world (what Kant
calls “the sensible”), what kind of effects it can and cannot achieve, what kinds of
evidential marks it leaves in addition to the abstract concept of purposiveness.17
Nor should we allow Sober to object that he does not argue by analogy but only
14
Ibid: 295.
15
When it suits him, as for instance, when it allows him to argue against concluding from flawed
designs an absent or flawed creator, Dembski agrees that we have no knowledge about a non-natural
designer other than its existence that would allow us to move from the fact of design to a judgment
of motive or capability. He does not recognize what this argument does to his larger aim though.
16
Kant (1987): 357–58.
17
Niall Shanks and Karl H. Joplin, in “Of Mousetraps and Men: Behe on Biochemistry,” 2–3
〈http://www.etsu.edu/philos/faculty/NIALL/Mousetraps.and.Men.htm〉 make a similar point in
more empirical terms.

103
 JONATHAN LOESBERG

by inference on the basis of natural hypothesis. Every case of known intelligent

causes of purposiveness is known because they are human and Kant’s point is that
we do not know in advance, nor can we know whether that is a relevant set from
which to draw an inference because we cannot know that a non-human creator has
an even remotely human-like understanding and the analogy is based on the
working of the human understanding.
In the light of Kant, we may start to make sense of the various divagations of
Hume’s argument (Kant detests Hume’s argument as disrespectful but he never
disputes its conclusions about natural theology). Hume’s claims that the argument
from design is based on first a questionable analogy and second an insufficient
induction all occur in Part II. If he thought that was his entire case, he might have
left the rest of the book unwritten. In fact, Cleanthes, the proponent of arguing
from design to intelligence, at the opening of Part III, strengthens the case from
analogy to look very much like Sober’s claim that intelligent causation is the only
likely conclusion to the fact of design. And he concludes by addressing the critic
of design theory, Philo: “Choose, then, your party, Philo, without ambiguity or
evasion; assert either that a rational volume is no proof of a rational cause or admit
of a similar cause to all the works of nature.”18 We could hardly imagine a stronger
claim for the validity of the design inference as, in Sober’s terms, the likeliest
hypothesis, and, indeed, Philo manifests at least a momentary assent to that
conclusion in terms of a chapter long silence.
But Hume buttresses the uncertain critique of the strengths of the analogy by
noting the difficulties that occur if one accepts the hypothesis on the basis on which
it is made. Hume does not, like Kant, divide sharply between a sensible or material
and a transcendent world, so he does not precisely criticize natural theology for
trying to get from a material effect to a transcendent cause, but his opening
argument against analogy and the insufficient basis of the induction is essentially
the same. He argues that we can only reliably conclude from effects enough to
account for their existence and that the argument from design attempts to conclude
too much. Cleanthes argues that the appearance of design in natural objects is so
exact that we are more nearly arguing from identity than analogy. When Philo
re-enters the argument, he attempts, as he says, to show Cleanthes “the inconve-
niences [. . .] in your anthropomorphism.”19 Philo proceeds to argue that if we grant
an extreme connection between human contrivance and the appearance of natural
design, we have a right to make the same kinds of inferences from that design that
we would make from human contrivance. We can no longer appeal to the fact that
the divine intelligence is unknowable because we have claimed to know it to be as
close to human in its effects as would be a book growing on a plant. Philo thus

18
David Hume, Dialogues Concerning Natural Religion (New York: Hafner, 1966) 28.
19
Ibid: 37.

104
 KANT, HUME, DARWIN, AND DESIGN

proceeds to formulate all the types of human-like designers we could construe from
the appearance of design. His point is not to show that the world is an imperfect
place but that the design analogy only works if the designing intelligence is thought
of as human, and when we think of human contrivances, we conclude much more
than merely the fact of contrivance and the fact of human causation. This, in effect,
reverses the order of Kant’s argument: Instead of contesting the analogy and leaving
it with an unsatisfactory alternative, Hume momentarily stipulates the analogy in
order to show the unfortunate implications that follow from it. If Kant’s argument
with regard to the basic analogy is better, Hume’s point is more telling with regard
to Kant’s allowing an architect: The basis for allowing that architect, rather than
being a modest first step toward monotheism, equally allows various other, far less
acceptable alternatives while failing to resolve in any final way where design comes
from. Accepting an architect, while still not really resolving the question about the
ultimate cause of design, thus becomes far less attractive for a theist than admitting
absence of proof. Much of the rest of the dialogue can be described as Philo’s
struggle to get Cleanthes to see that either the analogy is less than persuasive or that
it is less than happy in its results.
Neither Kant nor Hume wants to replace the Enlightenment vision of a designed
world with the concept of an accidental or undesigned world. Kant, indeed, thought
that thinking about the world purposively has a philosophic role to play in thinking
about ultimate meanings. He just refuses it a scientific role to play in investigating
natural objects insofar as they are natural. Hume’s battle is with what he takes to be
the assumptions of the Enlightenment that one can reason one’s way into meta-
physical certainty. Both of them show, therefore, not that the conclusion of the
design argument is either right or wrong about ultimate questions, but that it simply
has no ability to say anything meaningful about the material world. If that critique
is accurate, then far from confronting natural selection as an opposed scientific
theory, Intelligent Design should not be able to succeed in grappling with it at all.
And its failure should result from the kinds of weaknesses Kant and Hume point out,
in particular that the assertion of an intelligent designer turns out not to change our
understanding of nature in any empirically meaningful way and that the connection
between human design and non-natural design fails for the same reasons that make
the theory unable to confront natural selection. And a look at the Intelligent Design
arguments will show, I think, that these precise flaws disable their case.

III

This final section has two ends, each of which is part of the larger claim that
neither is Intelligent Design part of scientific discourse nor can scientific discourse
provide a disproof of Intelligent Design. Before proceeding to the discussion of
Intelligent Design and some of its most significant materialist opponents, Dennett

105
 JONATHAN LOESBERG

and Dawkins, though, I would like to begin with Kant’s classifications for all
ultimate explanations of the appearance of purposiveness, or design, and the
reasons he gives for their unsatisfactoriness since they nicely structure the kinds of
objections I will be making. Kant divides the explanations for the appearance of
purposiveness in nature into two main groups, each of which has two analogous
sub-categories.20 He calls the main responses idealist (the purposiveness is not
intentional and thus only appears to be purposiveness) and realist (the purposive-
ness is intentional and is thus really purposive) and the sub-groups into physicalist
(the explanation of the appearance or the reality is an entirely physical or material
one) or hyperphysicalist (there is a force behind material reality causing either the
appearance or the reality of purpose). And he explains why each of these expla-
nations is unsatisfactory.21 The fourth of these categories, hyperphysical realist,
refers to natural theologian and the flaws he finds in this argument have been
sufficiently detailed for us to apply them, with the aid of Hume, to Intelligent
Design. The middle two categories—the Spinozist hyperphysical idealists and the
hylozoist physical realists—do not describe any group among modern controver-
sialists and so need not detain us here. The first of the categories, the physical
idealists, who think the appearance of purposiveness to be mere appearance and to
be explicable in terms of natural or mechanical causes, should be the group to
which scientists belong. Kant faults this group with failing really to explain how
even the appearance of purposiveness or design occurs in the first place. And, by
attempting to offer such an explanation—instead of following Kant’s warning that
one cannot offer any ultimate explanation for the appearance of purpose—
Dawkins and Dennett will, interestingly, fall back into the hyperphysical realist
camp of natural theology—though in their fashion.
Intelligent Design means to escape the kinds of critiques that occur in Kant and
Hume by making three claims, each of which intends to revise and strengthen
natural theology. Each of these claims fails, either by failing to achieve persua-
sively the revision of natural theology it intends or by failing to escape Kant’s and
Hume’s critique regardless of the revision. These claims are first that Intelligent
Design’s conclusion of an intelligent designer has no theological content, makes
no claims about the identity of that designer, but posits the mere bare necessity of
a non-natural intelligent designer of some kind and, second, that nothing empiri-
cally or scientifically authorizes a limitation of empirical investigation to natural
causes of natural effects.22 These two claims in effect, as we will see, Kant has

20
Kant (1987): 270–73.
21
Ibid: 273–77.
22
This argument is not a direct revision of natural theology since those theologians did not recognize
a distinction between natural and non-natural causes and effects. Establishing that distinction is one
of Kant’s achievements and so disputing it must become part of the modern theory.

106
 KANT, HUME, DARWIN, AND DESIGN

already responded to. The core of the argument is the third contention that
Intelligent Design is a rigorously empirical claim. Responding to this argument
will entail going into both William Dembski’s and Michael Behe’s claims in some
detail. Before proceeding to dispute these claims, however, it is worth noting first
their political and then their intellectual roots since these non-logical elements
condition the shape their argument has taken.
The challenge to teaching Darwin in the classroom begins of course with the
Scopes trial in 1925. Once laws disallowing the teaching of Darwin were struck
down, however, the modern attempt has been to have some version of Christian or
theistic creation taught as an alternative “scientific” explanation to natural selec-
tion and/or evolution as an account first of human origin and more recently of
biological function. The most immediate political sources of Intelligent Design
begin with the legal rulings in the 1980s stating that creation science was not
science and could not be taught in science classrooms. The core decision in these
cases was the 1987 Supreme Court ruling in Aguilard, which saw creation science
as essentially religious in content and therefore forbidden from the classroom by
the First Amendment’s Establishment Clause.23 This decision helps explain Intel-
ligent Design’s unlikely claim that its conclusion of an intelligent designer has no
theological and thus no religious content. But the earlier ruling by Judge Overton
in McLean, in 1982, has also been important because of Overton’s listing of
features that make a discipline a science, among which were the criteria that
science concern itself with explanations in terms of natural law and that it draw its
conclusions from empirical evidence.24 From this element in the ruling, comes the
attack on what Intelligent Design theorists refer to as metaphysical naturalism.
Intelligent Design, as a political movement, is usually dated to the publication of
Philip Johnson’s Darwin on Trial, in 1991, which gave considerable attention to
the Overton ruling25 and took as one of its themes the idea that Darwinian
evolution maintains “naturalism” as an extra-scientific philosophical commit-
ment.26 Since, Intelligent Design theorists argue, that philosophical doctrine is not
itself scientifically grounded, empirical evidence that led to the conclusion of a
non-natural intelligence behind nature would have to deserve scientific consider-
ation, at least to the extent that science claims to be a system of empirical

23
See Edward J. Larson, Trial and Error: The American Controversy over Creation and Evolution,
3rd ed. (New York: Oxford UP, 2003) 179–80.
24
William R. Overton, “McLean v. Arkansas. United States District Court, Eastern District of Arkan-
sas, Western Division. Opinion of William R. Overton, U.S. District Judge,” reprinted in Science,
Technology, & Human Values 7 (Summer 1982): 28–42, at 36.
25
Phillip E. Johnson, Darwin on Trial (Washington, DC: Regnery Gateway, 1991) 112–13.
26
Intelligent Design theorists define “naturalism” as a metaphysical commitment to the proposition
that all things can ultimately be explained by natural causes and see this as the philosophical
assumption behind a theory of evolution by means of natural selection.

107
 JONATHAN LOESBERG

explanation and not a metaphysical position. Further, as long as this argument did
not presume any explicit religious doctrine but only claimed evidence for a
designing intelligence, it could not be found to contravene the Establishment
Clause’s prohibition of teaching religion in the classroom. This article intends to
take Intelligent Design theory explicitly on its own terms and not as a politically
reconstituted creationism designed to get religion back in the classroom, even
though there seems little doubt that it has had the explicit aim of changing the high
school curriculum and of operating as a political force.27 Logically, of course, the
political ends of an organization are not relevant to the intellectual quality of the
arguments its adherents make. Still, one can easily see the sources for the three
claims Intelligent Design makes, outlined above, in the history of rulings against
creationism.
But the sources of Intelligent Design’s challenge to Darwinian natural selection
were by no means all political. In 1986, between the McLean and the Aquilard
rulings, Richard Dawkins published The Blind Watchmaker: Why the Evidence of
Evolution Reveals a Universe Without Design. Ironically, as Dawkins declared the
argument from design dead, he aided in its revivification. Phillip Johnson’s move
toward articulating Intelligent Design as a theory opposing Darwin’s began in part
with his reading of Dawkins.28 And one can see why, since Dawkins certainly laid
down a challenge:

Paley knew that [the appearance of design in nature] needed a special explanation [. . .] it is
sometimes said that [David Hume] disposed of the Argument from Design a century before Darwin.
But what Hume did was criticize the logic of using apparent design in nature as positive evidence
for the existence of a God. He did not offer any alternative explanation for apparent design, but left
the question open. An atheist before Darwin could have said, following Hume, “I have no expla-
nation for complex biological design. All I know is that God isn’t a good explanation, so we must
wait and hope that somebody comes up with a better one.” I can’t help feeling that such a position,
though logically sound, would have left one feeling pretty unsatisfied, and that although atheism
might have been logically tenable before Darwin, Darwin made it possible to be an intellectually
fulfilled atheist.29

27
See Barbara Forrest and Paul R. Gross, Creationism’s Trojan Horse: The Wedge of Intelligent
Design (New York: Oxford UP, 2004). This book more than adequately outlines the political
workings and aims of the organizations that sponsor Intelligent Design theorists.
28
Ibid: 17.
29
Richard Dawkins, The Blind Watchmaker: Why the Evidence of Evolution Reveals a Universe
Without Design (New York: W.W. Norton, 1996) 6. One should note that although Darwin did think
his theory undercut Paley’s, he did not think that that was because natural selection offered a better
explanation of complex design. He argued, in The Origin of Species and the Descent of Man (New
York: Modern Library, n.d.), that natural selection could tenably explain what he called “organs of
extreme perfection and complication” (133–35), but he classed these organs as difficulties to his
theory that the theory could deal with. He did not think those explanations were the strength of the
theory. Rather, in his Autobiography, Darwin pointed to the conclusion of Variation of Domestic

108
 KANT, HUME, DARWIN, AND DESIGN

Dawkins sets up an intellectual history in which Paley’s argument survives

Hume’s critique (written, of course, before Paley), slightly damaged but more or
less intact. Darwin’s theory then offers a positive explanation for the design.
Dawkins’s claim about natural selection’s importance as an explanation of
complex organization among living things sets up a clear pathway for design
theory to re-enter scientific discourse. If Darwin’s theory can be shown to be, at
the very least, an improbable explanation of complex organization, then the
philosophical objections to the argument having been found more or less wanting
by Darwin’s allies, Intelligent Design re-emerges as a positive explanation for that
complexity. Even a cursory reading of Michael Behe and William Dembski will
show that most of their arguments are concerned specifically with why natural
selection will not explain biological complexity rather than more generally with
natural complexity as implying Intelligent Design. Dembski does offer a positive
argument in favor of design, but, as we will see, it is both thin and problematic. He
is quite explicit, though, that he thinks that if Darwin fails, then Intelligent Design
re-emerges, and that these are the only two competing explanations.30 Darwin
having been shown unlikely, and naturalism disposed of as a philosophical pre-
supposition, inferring an intelligence as causing biological complexity becomes
an empirical argument claiming its place in the scientific curriculum.
The first claim of Intelligent Design theorists, and most significantly William
Dembski, that the inference from design to an intelligent designer, need have no
theological content if it remains sufficiently modest, may be dealt with fairly
quickly. We have seen Kant’s distinction between what natural theology could
prove, a limited architect, and what it cannot, a force from outside nature that
brings it into existence. Dembski seems to think that Kant, by allowing an
architect of the world, is criticizing only what Dembski thinks of as immodest
claims for a theologically specified intelligent designer that he himself would not
make.31 Since Intelligent Design, he says, makes no claims to know the nature or

Animals and Plants as the reason Paley’s argument “fails” (Charles Darwin, Autobiography [New
York: W.W. Norton, 1958] 87–88). And that conclusion argues that natural selection manifests itself
in the way in which natural mechanisms operate with a dependence on what falls to hand, rather
than by the strictures of contrivance Paley discussed (Charles Darwin, Variation of Animals and
Plants Under Domestication, 2 vols. [New York: Appleton, 1890–92] II, 431). In other words, unlike
Dawkins, Darwin thought natural selection was evidenced by natural contingency, though it could
explain complex organization when it occurred. Although John Stuart Mill shares Dawkins’s logic
in his Essays on Ethics, Religion, and Society, Vol. 10 of Collected Works, ed. John M. Robson
(Toronto: U of Toronto P, 1969) 449–50, it is really far more characteristic of the late 20th-century
argument than of the mid-19th-century abandonment of natural theology.
30
William A. Dembski, The Design Revolution: Answering the Toughest Questions About Intelligent
Design (Downers Grove, IL: InterVarsity Press, 2004) 265–66.
31
Ibid: 68.

109
 JONATHAN LOESBERG

extent of the designer, it can be quite satisfied with Kant’s architect and use
further, theological reasoning to determine the unitary, Christian nature of that
designer. Aside from the fact that this argument is politically disingenuous, it also,
as we have seen, does not really grasp the distinction Kant makes. Because the
evidence for intelligence is the intention behind contrivance, contrivance can only
evidence the workings of an intelligence that needs to work through contrivance,
effectively a natural designer with perhaps greater than human powers. Dembski,
though, will not accept a natural designer (aliens from outer space, for instance)
because they do not really answer the kinds of ultimate questions he wants to
answer. Further, as we have seen, Hume’s extension of this argument for a
human-like intelligence entails accepting—as Dembski explicitly does not—
drawing inferences from the imperfections of natural design about the intentions
or abilities of the designer. In effect, to accept a design inference sufficiently
modest to meet the objections of Kant and Hume, Dembski would have to accept
that he is arguing not for a non-natural first cause but merely for a natural cause
in the form of a natural, though non-human, designer. In addition to being a weak
scientific conclusion, as we have seen (though not a non-scientific one), such an
argument does not exit from the naturalism that is Intelligent Design’s complaint
against Darwinism. But a conclusion of a non-natural designer will fall afoul of
the objections to natural theology that Dembski claims to escape.
And this brings us to the argument against metaphysical naturalism. One can
see what is at stake by turning to Michael Behe’s version of the argument from
design. Behe effectively allows all of Darwinian theory above the level of molecu-
lar development,32 but then steps back to argue that natural selection cannot
explain cellular function. He adduces notoriously, although not solely, bacterial
flagella, which have multiple moving parts that work together to produce a
function, propulsion. Paley argued that relations of parts to whole in the eye
evidenced contrivance. Although Behe’s argument is slightly different in the light
of Darwin’s theories, his point is the same. He argues that in complex organiza-
tions, all features need to be in place for a given function to result. Behe charac-
terizes such organizations as having irreducible complexity and he explicitly
claims that Darwin cannot explain it:

Irreducibly complex systems like mousetraps, Rube Goldberg machines and the intracellular
transport system cannot evolve in a Darwinian fashion. You can’t start with a platform, catch a few
mice, add a spring, catch a few more mice, add a hammer, catch a few more mice, and so on: The

32
Michael Behe, Darwin’s Black Box: The Biochemical Challenge to Evolution (New York: Free
Press, 1996). Behe explicitly accepts common descent (5) and, although he does not say anything
about the role of natural selection in speciation, he at least seems to stipulate the Darwin–Dawkins
explanation of the eye at the level of the features they describe—retinas, lenses, etc.—only demur-
ring about the cellular make-up of those features (38–39).

110
 KANT, HUME, DARWIN, AND DESIGN

whole system has to be put together at once or the mice get away. Similarly, you can’t start with a
signal sequence and have a protein go a little way toward the lysosome, add a signal receptor
protein, go a little further, and so forth. It’s all or nothing.33

Behe, by insisting on the relation of parts to a function which will only occur in
the presence of all of parts working together, adduces precisely what Paley
adduces about watches and eyes.34 And, like Paley, he also argues that irreducible
complexity is a positive indication of Intelligent Design. However, unlike Paley,
who merely argues from design to a designer, Behe’s argument has two parts: (1)
Darwin’s theory of natural selection will not explain certain features of cells, and
(2) in the absence of natural selection, the only explanation left is that of an
intelligent designer.
Since I will be arguing that Intelligent Design is not science because it has
insufficient purchase on material reality to make an effective scientific claim, at
this point, I need to face an objection that both Behe and Dembski make with some
justice. If Intelligent Design is not scientific because it makes no measurable
scientific claims, how is it that scientists have mounted scientific refutations of its
claims? As Behe notes: “Now one can’t have it both ways. One can’t say both that
ID is unfalsifiable (or untestable) and that there is evidence against it. Either it is
unfalsifiable and floats serenely beyond experimental reproach, or it can be criti-
cized on the basis of our observations and is therefore testable.”35 And, indeed,
many of the objections of scientists to Intelligent Design, as we will see, are more
metaphysical than they would admit. But Behe’s objection points nicely to the
logical separation of the two parts of his argument. The first of these claims, that
natural selection will not explain certain features of cells, is quite obviously a
claim within biology and therefore theoretically capable of being refuted within
that discourse’s terms. Arguments in favor of how natural selection might have
produced bacterial flagella are thus refutations of Behe’s scientific claim. Whether
they are effective refutations this article will not judge. Indeed, it does not have to
make that judgment. We have seen Kant argue that only what he calls mechanical
principles advance our scientific knowledge and that the argument for a designer
amounts to abandoning the search for that knowledge rather than a scientific
inference. The distinctness of Behe’s two arguments indicates the way the second
claim departs from the kind of argument made by the first. His second claim rests
precisely on the metaphysical leap Kant showed Paley to have made. Behe’s

33
Ibid: 110–11.
34
Behe’s critique of Paley for missing his point about the irreducible complexity of the watch
(215–16) by the way is quite unjust. Paley’s point about contrivance captures Behe’s point and also
accommodates evidence Behe wrongly thinks is beside the point.
35
Michael Behe, “Philosophical Objections to Intelligent Design: Response to Critics,” 2 〈http://
www.trueorigin.org/behe06.asp〉.

111
 JONATHAN LOESBERG

argument thus shows us the split between the metaphysical and the biological
argument. Dembski’s argument will show us why the metaphysical argument fails
to refute Darwin and also, in the attempt, fails to make the empirical case it claims
to make.
In order to show the ways in which Dembski’s argument both fails to disallow
natural selection and fails to avoid the critiques of Kant and Hume against natural
theology, we need to start by recognizing the revisions he makes to natural
theology in an attempt to make it a more completely empirical argument. Aquinas,
as we have seen, argued that the existence of means to ends relationships in natural
entities entailed a designer since such relationships are only caused by intelligent
planning. Paley went a step further and argued that the relation of different parts
working together to produce a function indicated contrivance, which is a specific
feature of intelligence. But arguing from contrivance to a contriving intelligence
led to difficulties in natural theology that Dembski wants to avoid. He thus
proposes what he offers as a more direct, empirical, and modest proposition, one
that has no claims about the kind of intelligence at work, but merely posits the
existence of intelligence as a necessary conclusion. In the place of purposive form
or contrivance, Dembski articulates the concept of complex, specified information
(CSI), which he defines as follows: “An object, event, or structure exhibits speci-
fied complexity if it is both complex (i.e., one of many live possibilities) and
specified (i.e., displays an independently given pattern). A long sequence of
randomly strewn scrabble pieces is complex without being specified. A short
sequence spelling the word “the” is specified without being complex. A sequence
corresponding to a Shakespearean sonnet is both complex and specified.”36
Because the example of the sonnet here is also an example of intended meaning
and thus purposive, this definition must be refined a little further. Dembski’s
concept of specification entails a pattern that is either explicitly stated beforehand
(an archer specifies the target she shoots at in Aquinas’s example) or may be taken
as sufficiently special in some sense as to count as specification: A sequence of 50
coin tosses resulting in heads is no more specified than any other sequence of 50
results, but if we saw it happen, we would think it an improbable enough event to
demand further explanation, possibly in terms of human intervention in the form
of a loaded coin.
But in the absence of a contingent end, such as winning a bet, tossing fifty heads
in a row does not by itself imply an external end in the way that the relationship
among parts in the mammalian eye suggests an explanation of that relationship as
having the function of producing sight. Coin toss results by themselves do not
evidence intentions or contrivance, even if, when they fall in certain patterns, we

36
William A. Dembski, No Free Lunch: Why Specified Complexity Cannot be Purchased without
Intelligence (New York: Rowman & Littlefield, 2002) xiii.

112
 KANT, HUME, DARWIN, AND DESIGN

do start to make guesses about loaded coins. Because he does not refer to qualities
of intelligence such as contrivance or intention, Dembski asserts the connection
between CSI and intelligence as one of purely empirical, causal sequence: “The
justification for [the] claim [that CSI always indicates Intelligent Design] is a
straightforward inductive generalization: in every instance where specified com-
plexity obtains and where the underlying causal story is known (i.e., where we are
not just dealing with circumstantial evidence but where, as it were, the video
camera is running and any putative designer will be caught red-handed), it turns
out that design is actually present.”37 One should note that this claim can be taken
as making a buried analogy. Since all cases in which we already know the causes
of specified complexity are not only cases where the causes are intelligence but
cases where the causes are human intelligence, the induction that Dembski claims
can be taken as, like Paley’s, to point to examples of humanly created design and
argue that natural objects share certain significant features with those cases.
Dembski vigorously denies that Paley’s argument was essentially one of analogy
and that his is in any sense an analogy (following Sober’s critique).38 In order for
this denial to make sense, however, Dembski must be claiming simply that CSI
just is an inescapable empirical sign of Intelligent Design. The fact that all the
cases we know happen to be humanly produced is for him irrelevant. What
matters, he says, is that in all cases we know, except possibly for the case of
biological organisms, we would as a result of the presence of CSI infer an
intelligence as the cause. Since Dembski sees no basis for excepting biological
organisms (and we must grant that we cannot except them a priori without
begging the question), CSI would then be a positive, empirical indication of
Intelligent Design.
In a certain sense, as I will argue further down, the argument could be left here,
without it being relevant to any scientific explanation for functions in organs. If
CSI just did indicate intelligence, then even if natural selection could produce CSI
as a mediate cause, this fact could not disprove the prior argument. It would only
mean that natural selection was the natural means that intelligence used to produce
CSI. But, as we have seen, both Darwin’s allies, such as Dawkins and Dennett,
and his enemies, Johnson and Dembski, accept natural selection as an alternative
explanation that must be disproved. And the argument of Intelligent Design
against it has been that it cannot sufficiently explain complexity. Ever since
Darwin, that argument has regularly entailed accounting for the structure of the

37
William A. Dembski, “The Logical Underpinnings of Intelligent Design,” Debating Design: From
Darwin to DNA, ed. William Dembski and Michael Ruse (Cambridge: Cambridge UP, 2004)
311–30, at 320.
38
William A. Dembski, Intelligent Design: The Bridge between Science and Theology (Downers
Grove, IL: InterVarsity Press, 1999) 271–75.

113
 JONATHAN LOESBERG

mammalian eye, but both Darwin and Dawkins do offer a picture of how natural
selection could produce the eye.39 And although Dembski does not really grant
Dawkins’s argument, he does not expend much energy trying to refute it, offering
rather an argument that simply makes an end run around natural selection.
We have seen that, according to Kant’s claim that one can never rule out of
court a mechanical cause of a natural effect just because it is unknown, even if
Behe were right that natural selection could not explain irreducible complexity or
purposive structure at the cellular level, that would not support the leap to a
non-natural designer. Dembski shows us the inability of concluding the existence
of a designing intelligence even to rule out natural selection. In order to show that
CSI cannot be the consequence of natural causes, Dembski argues, in No Free
Lunch, that simply as a matter of fact, no natural cause, by itself, can ever produce
specified complexity. To deal with cases of CSI that seem to be produced by
necessary causation, he refines this claim to say that natural causation may, in a
sense, shuffle CSI from one place to another, thus giving the appearance of having
created it. This allows Dembski to argue that “even if Darwinian evolution is the
means by which the panoply of life on earth came to be, the underlying fitness
function [the algorithmic formulation that describes how selection produces its
results] that constrains biological evolution would not be a free lunch and not a
brute given, but a finely crafted assemblage of smooth gradients that presupposes
much prior specified complexity.”40 To take this statement out of Dembski’s
probabilistic formulation, his claim is that natural selection will not work unless
there are already prior conditions that allow it to work and those prior conditions
enable its success. The prior conditions he posits will hardly disturb any evolu-
tionary theorist (that the earth be in physical existence, that DNA work to repli-
cate, that there be some general survival advantage to complexity to ward off
selection’s tendency to simplify). Nor would they really look like the kind of
specifications that demand inferring Intelligent Design, unless again one presumes
that specified complexity must evidence intelligence even if we have knowledge
of secondary causes that might have produced it. Thus, in effect, rather than
disproving natural selection, as Dembski usually claims to do, in order to deal with
the more general problem of bracketing all examples of necessarily produced CSI
as the results of secondary causes, the design inference must stipulate natural
selection’s possibility and declare it irrelevant.
But the fact that Intelligent Design cannot overturn natural selection or any
other theory offering natural explanations of biological complexity that might

39
Darwin’s explanation occurs in The Origin of Species, 133–34. Dawkins’s more fleshed out
discussions are first in The Blind Watchmaker, 84–86 and then in Richard Dawkins, Climbing Mount
Improbable (New York: W.W. Norton, 1996).
40
Dembski (2002): 212.

114
 KANT, HUME, DARWIN, AND DESIGN

replace or augment natural selection does not, by itself, show the inference to a
designer to be an insufficient one on its own terms. It is likely that if my contention
were accepted that design theory has no grip on scientific conclusions and thus
had no particular claim to the time of at least high school science curricula, the
political impetus for the movement would end and with it the movement itself. But
the intellectual claims of the movement remain separable from its political aims.
And, in particular, Dembski’s theory, whether he meant it to or not, may be taken
to respond to one aspect of Kant’s claim that the basic analogy at the core of the
argument from design does not hold and thus to the inconveniences that Hume
claimed to find in its anthropomorphism. Kant responded to the claim of all design
theorists up to his time that what one saw in nature was evidence of purpose or
intent and so he noted that the analogy from humanly artificed object to natural
object presumed we could get from human intelligence to the working of divine
intelligence or understanding.41 Dembski’s theory of CSI, however, explicitly
intends to eschew reference to the problem of other minds. He distinguishes CSI
from semantic information, claiming that CSI is a formal description of the
external features of objects, without reference to the purpose, meaning, or func-
tion of the object42 and asserts that unembodied designers are no more a difficulty
to his theory than embodied ones because the theory does not assert anything
about a designer’s “mental processes.”43 If he can make good on the argument that
CSI entails Intelligent Design, without reference to how the CSI was effected or
why, then, even if Intelligent Design would not have much to say in a science
classroom, it would at least remain an empirical conclusion of sorts.
The problem with this formalization of the evidence for design is that it
produces problems on one side of the analogy and wildly improbable descriptions
of how evidence works on the other. In other words, on the one hand, there are
natural patterns that look enough like CSI to fit Dembski’s criteria but do not look
like things to which we want to attribute design to make it unclear that in fact CSI
is as uniformly attached to design as purposiveness or contrivance seemed to be to
Paley. And, on the other hand, Dembski’s descriptions of how we conclude human
design from the evidence of CSI are instructively unlikely. To start with the first
problem, scientists have offered two examples of patterns that seem to meet the
criteria of being CSI but that, they say, do not entail a designer: Bénard cell
formations (when heated under certain conditions, water forms hexagonally and

41
Elliott Sober, in “The Design Argument,” Debating Design: From Darwin to DNA, ed. William
Dembski and Michael Ruse (Cambridge: Cambridge UP, 2004) 98–129, at 105–14, makes a similar
argument but does not see it working in either Hume or Kant.
42
Dembski (2002): 147.
43
Ibid: 147–48.

115
 JONATHAN LOESBERG

pentagonally shaped cells)44 and the fact that leaves on some plants form patterns
around the stem that follow a Fibonacci sequence (a sequence of numbers in
which each number in the series is the sum of the prior two numbers).45 Dembski’s
replies to each of these examples have been different. He has simply denied that
Bénard cell formation makes it through his explanatory filter because it is the
result of “causal specificity,” thus an example of something occurring necessar-
ily.46 In the case of the Fibonacci sequence of leaf arrangement, Dembski is less
clear. On the one hand, he suggests that the sequence may be necessary and thus
not necessitating a design inference. On the other hand, he also suggests that the
necessity may result from some prior biological cause that would, as inputting
CSI, result in the design inference.47 Dembski’s responses are not, on their own
terms, very satisfying. With regard to the Fibonacci sequence of leaves, no one has
suggested a biological cause to the sequence, so in order to get to the design
inference, he has to propose more than he knows. And, since his explanatory filter
never really explains why causally necessary patterns are not specified complex
information regardless of their origins, the relegation of Bénard cells to causal
specificity seems no more satisfying. Moreover, we are left with the question of
why Dembski accepts Fibonacci sequences as designed but rejects Bénard cells.
Why not just have said that both exhibit CSI and both are therefore designed? And
why did his opponents think that Bénard cells would be exceptions rather than
designed patterns for Dembski and why were they at least partially right?
The answer, I think, has to do precisely with the lack of contrivance or purpo-
siveness these effects seem to display. Unless we imagine that a designer provided
Bénard cell patterns for our aesthetic pleasure (in which case we imagine a
purpose for them), we see no reason for the patterns, no end that they serve. In the
case of the leaves, that which allows us to recognize specified complexity, the
mathematical pattern, is not that which benefits the leaves, assuming there is a
benefit to the pattern, and this differentiates the pattern from the ostensibly
irreducible complexity present in the eye for Paley or bacterial flagella for Behe.
One must say that the scientists proposing these patterns seem no more aware of
why they present the possibility of counter-examples than Dembski does and so
though they tax his answers with scientific inaccuracy or logical contradiction,
they do not tax them, as I am here, with missing the point of the problem. And yet

44
See Shanks (2004): 125–28; and Niall Shanks and Istvan Karsai, “Self-Organization and the Origin
of Complexity,” Why Intelligent Design Fails: A Scientific Critique of the New Creationism, ed.
Matt Young and Taner Edis (New Brunswick, NJ: Rutgers UP, 2005) 85–106, at 91–94.
45
See Korthof 〈http://home.planet.nl/~gkorthof/kortho44.htm〉.
46
Dembski (2002): 243.
47
Ibid: 12, 14. In The Design Revolution, 90, Dembski more straightforwardly attributes the sequence
to a prior designed biological complexity.

116
 KANT, HUME, DARWIN, AND DESIGN

that the abstract, non-intentional element of the CSI is the problem one may see,
again, by asking why Dembski does not simply assert both patterns as evidence of
Intelligent Design. To do so would make his argument more logically coherent but
also more obviously detached from the biological features he ostensibly seeks to
explain.
Still the design inference, although with a weakened and more arbitrary
seeming basic claim, might hold formally if Dembski simply accepted Bénard
cells and Fibonacci sequenced leaves as examples of CSI demanding the existence
of an intelligent designer. The Fibonacci sequenced leaves, however, in the dis-
tinction between what gives their sequence complex specified information (that
the leaves are in a numerical sequence), and whatever about that sequence might
create a function, indicate another problem with Dembski’s theory, which is that
it really does not apply, as he thinks it does, to the cases where we do know or
should know that design occurs. To show the problem, I want to look at one of
Dembski’s examples of how we make a judgment about CSI. Dembski imagines
a probability professor assigning some of his students to produce a report of the
results of flipping a coin 100 times and assigning others to produce simulated
random sequences that they mentally construct without actually flipping coins.
The professor can almost always differentiate between actual reports of tosses and
constructed reports because there are arcane features of coin flipping sequences
that unknowing students imagining a sequence would not create. Dembski then
imagines a student knowledgeable in probability who sets out to fool the professor
into guessing wrongly by knowingly recreating all the arcane features of
sequences of 100 coin tosses in his or her report. At first, Dembski imagines the
professor being fooled. But the professor then looks at the reported sequence more
closely, reading it, instead of in terms of heads and tails, in terms of 0s and 1s. She
then notices that the sequence created by 0s and 1s is the sequence that reproduces
counting to 100 in a binary system and concludes that the report was constructed
instead of real.48 Should she have so concluded? For Dembski, the point of the
anecdote is that she has noticed a specified complexity and thus knows that she is
not in the face of chance. But if the features of counting to 100 in binary numbers
only accidentally reproduces the features of sequences of 100 tosses of a coin and
is not one of the tests for such features, then that conclusion is not a clear one. Why
would the knowledgeable student trying to fool the professor have used that
sequence, except as an act of ego to see if he could get away with it? Why would
the professor have tested for that sequence since it is normally irrelevant to the
features she is looking for (except that it happens to be one such sequence out of
many containing those features)? Even knowing the student to have been knowl-
edgeable in probability, we are not warranted, merely from the unlikelihood of the

48
Dembski (2002): 16–18.

117
 JONATHAN LOESBERG

sequence, to draw the conclusion that the student chose that sequence since there
is no reason for him to have done so. We need to imagine the student intentionally
encoding the binary sequence as a result of some motivation that goes beyond
merely fooling the professor. If we knew the student to enjoy, like master crimi-
nals in detective stories, leaving traces of his deceptions, we might then take the
presence of the binary code as evidence of his having worked up the sequence
rather than having flipped the coin. But this is precisely the kind of knowledge
Dembski denies we need to draw a design inference. Thus, not merely pattern but
actual intention and purpose do enter into the design conclusion here. One could
easily multiply such examples in Dembski of accounts in which he claims we
would infer design where we might not or of accounts where we would infer
design because of elements of motivation he explicitly denies are relevant.49 Kant
and Hume show us why the appearance of design does not constitute empirical
evidence of a non-natural human-like intention. Dembski’s attempt to avoid the
concept of intention shows us the failure of the argument without that concept to
make empirical sense in its own terms. So, not only does Intelligent Design fail to
disprove natural selection as at least a mediate cause of complex structure in
nature, it fails, even in its revised form, to make a case for itself as an empirical
conclusion as well.
There is one more step to this argument and it turns not on the argument from
design but on the claim by Darwinian controversialists—notably Dawkins and
Dennett, but also others—that Darwin in fact puts the nail in the coffin of natural
theology by providing an alternative explanation of design. My point will not quite
be Stephen Jay Gould’s of defining science and religion as non-overlapping
magisteria,50 in which their claims have nothing to do with each other ever. At the
very least, science and religion overlap in the disciplines of philosophy and
intellectual history, or this article would have no entry point into the debate over
Intelligent Design. And just as there are other reasons for endorsing theism rather
than natural theology, there are powerful reasons for being an atheist that do not
depend on Darwin’s having provided one with intellectual fulfillment. Rather, my
point is limited to undoing an overreaching among scientific controversialists that,
oddly, provides ground for the claim by Intelligent Design theorists that their
inference must be scientific at least to the extent that scientists think they can

49
One should note in this connection Gary Hurd’s contention that archaeology and forensic science,
fields that Dembski cites as unproblematically making design inferences based on CSI, actually
look for the kind of information about motivation and actual human, physical intervention, the
relevance of which he denies. See Gary Hurd, “The Explanatory Filter, Archaeology and Forensics,”
Why Intelligent Design Fails: A Scientific Critique of the New Creationism, ed. Matt Young and
Taner Edis (New Brunswick, NJ: Rutgers UP, 2005) 107–20.
50
Stephen Jay Gould, Rocks of Ages: Science and Religion in the Fullness of Life (New York:
Ballantine Books, 1999) 6.

118
 KANT, HUME, DARWIN, AND DESIGN

refute it. If scientists could refute Intelligent Design in the way they do the
empirical claims of older style creationists (that the earth is only 6,000 years old,
that homo sapiens lived in the time of dinosaurs), Intelligent Design might be false
science or bad science. But in fact scientists can never quite close with this
opponent because it is not theirs (in intellectual terms; it is theirs, of course,
culturally in wanting to presume on their discipline) and, again, Kant argues out
why in the four categories of explanation for design discussed in the opening of
this section.
Since natural theology is the fourth of these types (hyperphysical realist), it
would be convenient to include scientific explanations (as Kant thought he could)
within the first group (physical idealists) and accept the reason we will see that he
gives for the insufficiency of that explanation. But surely, Dawkins and Dennett,
who I will use as the most obvious examples here, would not entirely accept Kant’s
description of the physical idealist argument. As a consequence, their theories
appear at times to exhibit elements of each of the categories of explanation and
ultimately become not refutations of the argument from design but versions of it.
It should be the case that contemporary scientific belief in natural selection
would fit in Kant’s category of physicalist, idealistic explanation of purposiveness.
Science surely does not think that there is an intending force (although that may
depend, as we will see on how we interpret the concept of an intending force) and
it thinks that necessary physical laws explain the appearance of natural purpo-
siveness. Kant’s knowledge of physical laws was, on the evidence of The Critique
of Judgment, not very advanced, even for his own time, and limited to ontological,
philosophical explanations. The only physical idealist explanation he considered
attributed the appearance of purposiveness to laws of motion, in the manner of
Epicurus or Democritus. Kant considered this explanation “manifestly absurd.”
Thus, when he characterizes why a mechanist explanation will not really explain
even the appearance of purposiveness, Dennett and Dawkins might rightly ques-
tion whether his argument applies to them. Kant says that the physicalist idealists,
who explain purposiveness as an unintended consequence of the laws of motion,
“adopt blind chance to explain not only [nature’s] technic, i.e., why [nature’s]
products harmonize with our concepts of a purpose, but even nature’s mechanism,
i.e., how the causes of this production are determined to this production according
to the laws of motion. Hence nothing has been explained, not even the illusion in
our teleological judgments, so that the alleged idealism in them has by no means
been established.”51 But Dawkins, for instance, pointedly and repeatedly denies
that the natural process he describes is in fact “blind chance.” He describes natural
objects that appear to be purposive to be “designoid”52 because they have been

51
Kant (1987): 274.
52
Dawkins (1996): 6.

119
 JONATHAN LOESBERG

shaped and molded like a “designed” object, not by chance, but by a natural force
that reproduces purposive design53 selecting for functions that offer survival or
reproductive advantage and thus explaining the appearance of purposiveness.
But from Kant’s perspective, this answer is insufficient. He would surely and
happily admit that the Darwinian explanation for the appearance of purposiveness
is far from being manifestly absurd and might well admit to its being likely. But
from an ontological perspective, its explanation of the appearance of design only
works because it fiddles with the concept of “function.” In pre-Darwinian terms,
“function” described the function an object was intended to perform. Chairs may
be used to tame lions or to stand on and reach high objects, but they have the
function of being a thing to be sat on because, in addition to all the intentions any
given maker of chairs might have, being sat on is the intention that defines them
as chairs. But in biology, “function” is merely an etiological description. It
describes historically how adaptations developed. And although both Dawkins and
Dennett accept the concept of “exaption” as grudgingly as Dembski does, they
recognize that features that historically served one function may come to serve
another under different environments. A vigorously exaptationist position, in
which the workings of natural selection are saturated with the effects of historical
and environmentalist contingency, such as Stephen Jay Gould’s, would surely still
thus fall afoul of Kant’s objection of not really explaining, ontologically, how the
appearance of purposiveness occurs since it ends up abandoning function, nor-
mally so-called for accidents of history and environment.
But Dawkins, by implication, and Dennett, quite directly, do mean to offer an
ontological explanation.54 And thus Dennett vigorously defends an essentially
adaptationist form of explanation, in which the etiological and the normative
definition of function coincide. He also vigorously defends “function” as a
straightforwardly accurate description of what biological features have.55 In order
to make natural selection an ontological explanation, however, he departs not only
from a purely physicalist explanation of purposiveness, but even from an idealist
one, ending up in a curious place.
Dennett argues not for the appearance of purposiveness, but for its actuality.
This may sound surprising, and if one puts together various parts of his argument,
one gets a very surprising picture indeed. First, he really does not think that Paley
was wrong in attributing design to natural objects: “So Paley was right in saying
not just that Design was a wonderful thing to explain, but also that Design took

53
Ibid: 28.
54
Dawkins cites Darwin’s Dangerous Idea as a book that develops his position (see Dawkins [1996]:
x).
55
Daniel C. Dennett, Darwin’s Dangerous Idea: Evolution and the Meanings of Life (New York:
Simon & Shuster, 1995) 399–400.

120
 KANT, HUME, DARWIN, AND DESIGN

Intelligence. All he missed—and Darwin provided—was the idea that this Intel-
ligence could be broken up into bits so tiny and stupid that they didn’t count as
intelligence at all, and then distributed through space and time in a gigantic
connected network of algorithmic process.”56 Dennett does not deny that design in
nature is real or even that intelligence produces design; he just disputes with Paley
how that intelligence came about. For him, intelligence can be produced as the
sum of physicalist parts which participate in a formal process that algorithmically
produces certain ends. At this point, Dennett has moved out of Kant’s version of
a physicalist idealism to a Spinozist, hyperphysicalist idealism: He constructs a
unity of purpose—the algorithmic process—which has as its consequence, though
not its intention or purpose the purposiveness of natural objects. We need not here
outline Kant’s objection to the idea of unity of cause as falling back either into the
non-explicativeness of the mechanistic theory or into wordplay because Dennett
will not stay within this idealist explanation of purposiveness.
When Dennett follows Paley not only in attributing design to nature but in
explaining design in terms of intelligence, we ultimately find that he is not
speaking metaphorically. He of course does not think there is either an organic
mind or a disembodied mind behind natural purposiveness, but he denies that only
minds can intend things. In a discussion of artificial intelligence, he criticizes
Searle, not only for denying function to unintended things, but for denying that
things that exhibit the practices and have the effects of intelligent intention are in
fact effects of intelligent intention. Rather Dennett insists that complex processes,
either neural or mechanical, that produce the ends that intention produces are in
fact just what intention is. But if human intelligence is merely a mechanical
process at the neural level, then natural selection can also be described as an
intentional process. And so Dennett concludes by chafing Searle “for denying
Paley’s premise: according to Searle, nature does not consist of an unimaginable
variety of functioning devices, exhibiting design. Only human artifacts have that
honor and only because (as Locke ‘showed’ us) it takes a Mind to make something
with a function!”57 One might have thought that in contending that nature only
appeared to be designed and that biological function differed significantly from
human function in describing an etiology and not an essence or an intention,
Searle was being relatively uncontroversial. If Dennett finds him self-evidently
wrong, then one must ask what Dennett thinks does produce function and design
normally so understood.
Quite evidently, Dennett cannot be classed, in Kantian terms, as an idealist with
regard to intention. He thinks that if the parts add up to a function of the kind that
minds produce, then they were produced in the same way that minds produce

56
Ibid: 133.
57
Ibid: 400.

121
 JONATHAN LOESBERG

functions. And he thinks that what he calls “as-if intention” really just is intention.
So he regards purposiveness as produced by real intention—or, what for him
amounts to the same thing—intentional structure. It is just not neural intention.
Now Kant’s intentional realists, according to his logical alternatives, must be
either physicalists or hyperphysicalists. Physicalists must be endowing natural
matter with life in the manner of hylozoists since attributing natural purposiveness
to the physical working of matter, without recourse to intention or mind, entails
thinking of nature as having a life of its own. It is tempting but unfair to class
Dennett as a hylozoist. While he does not distinguish between algorithmic process
and intention, he does not seem to want to designate life to algorithmic forces as
well (though from Kant’s perspective, this distinction would be a nearly disap-
pearing one). We are left, then, with the hyperphysical, theist, argument from
design. And in what important way does Dennett differ from the design inference?
He certainly does infer design from biological structure. And he attributes that
design to a force to which he would endow the features of intention. Obviously
Dennett does not believe in an old man with a white beard or a separate force
external to nature. And he does not believe that intentional design in nature is
intelligent, except as one can speak of intelligence as the sum of unintelligent
parts. But these are distinctions that matter less than he thinks they do. Since even
human intelligence is, for Dennett, the sum of unintelligent parts, it is not clear
that his explicit denials that design results from intelligence really hold all the way
down. And Kant’s objections to the design argument, particularly in asserting an
analogy, not to say an identity, between artificed objects and natural objects in just
the place where the asserted analogy breaks down would apply to Dennett as well
as to Dembski.
My point here is as far as it can be from the claim of both old style creationists
and new style design theorists that Darwinian evolution is just another metaphys-
ics in disguise. I do, however, think that the claims that the truth of natural
selection disproves the argument from design and conduces toward or proves an
atheistic ultimate reality is a metaphysics mistaking itself, as the old argument
from design did, for having an empirical basis. Kant argued that purposiveness in
nature was, not supernatural, but extra-natural, something whose existence was
subject to neither empirical proof nor empirical disproof. Natural selection, as
Darwin proposed it, explained how organisms have evolved into different species
and the particular way that they have evolved. To do so, he certainly had to provide
a natural process that would produce complex biological features. But he did not
have to assert, nor, Kant argued, could he have proven, that that natural process
either was or was not an exhaustive explanation for complexity. The ambition to
provide such exhaustive explanations pushes Intelligent Design but also Darwin-
ian controversialists who think they can show Intelligent Design to be scientifi-
cally incorrect as opposed to simply not relevant to scientific discourse, beyond

122
 KANT, HUME, DARWIN, AND DESIGN

science into metaphysics. If Intelligent Design should be returned to the pile of

philosophic theories that the Enlightenment had rightly found wanting, it will be
a small cost to find that the same basis also forces us to consign more ambitious
Darwinian metaphysics to the same pile of theories. It has always been an odd
creature, materially atheist but metaphysically Leibnitzian. One can imagine it
establishing a basis for itself as a metaphysics to compete with other views of the
universe as essentially with or essentially without design. Both it and the equally
unlikely arguments for theism should be left to compete there so that biology can
get on with its business.

Department of Literature, American University, Washington DC

123