Return
to Human Knowledge and Human Nature contents
10. EVOLUTIONARY NATIVISM
AND A PRIORI KNOWLEDGE
In this chapter I shall consider how much
remains of the traditional conflict between empiricism and rationalism, given
my proposed account of the core of the former.
Nativism and the A Priori
In chapter 9 I argued that contemporary
empiricists should have no objection to evolutionary versions of nativism. It
follows that they should also have no objection to some forms of a priori
knowledge. For knowledge which is innate will at the same time be a priori, at
least in the sense of not having been learned from experience. This is, if
correct, a startling result. It means, not only that one can remain an
empiricist while accepting that there are innate information-bearing mental
structures, innate concepts, and innate knowledge, but also while accepting the
existence of substantive a priori knowledge. Indeed, you may be inclined to
object that this is too good to be true. For there would appear to be a real
danger, now, that I have construed empiricism in such a way that the
differences between empiricism and rationalism may disappear altogether.
For
example, if it were to turn out that an evolutionary nativism can provide an
adequate defence of platonism, then on my account a contemporary empiricist
would probably have no objection of principle to any of the traditional
rationalist doctrines. In fact it would turn out that the early empiricists'
opposition to rationalism derived entirely from the poorly-developed state of
science at the time, in particular from the absence of any sort of theory of evolutionary
selection of inherited characteristics. Some might find such a conclusion so
implausible as to warrant a rejection of my account of empiricism. Others,
accepting that account, might take the conclusion to mark the demise of
empiricism as a distinctive movement in philosophy. I shall make no attempt to
decide which of these responses might be the more reasonable, since I believe
that much does in fact remain of traditional disputes between empiricists and rationalists,
even given my account of the essential concerns of the former.
If
the course of evolution has provided us with substantial innate knowledge,
either of our own psychology or of other aspects of the natural world, then it
will at the same time have provided us with substantial a priori knowledge.
However, it is important to realise that the status of such knowledge as a
priori will be quite different from what rationalist philosophers have
generally had in mind when they employed the phrase 'a priori knowledge'. In
particular, the knowledge in question will not have been acquired merely
through the thinking of it, by a process of reasoning alone. For example,
innate a priori knowledge of folk-psychology would not consist of beliefs which
can be discovered to be true just by thinking of their subject-matter; nor
would their content be intuitively certain. Rather, we would find ourselves
with those beliefs already; and then provided that they are both true and
reliably acquired through evolution, this would be sufficient to constitute
them as knowledge. Hence at least one remaining difference from traditional
rationalism, is that the most plausible cases of innate knowledge for a
contemporary empiricist will not be knowable by thought alone.
A
related point is that the question whether a given belief is a priori, in the
nativist sense, will not itself be an a priori one. It will rather be
empirical. For recall our rejection of the KK thesis (that knowledge requires
knowledge that you know) in chapter 5. Knowing something (whether a priori or
not) is one thing, knowing that you know it may be something quite different.
In the case of folk-psychology, for example, if our beliefs are, as a matter of
fact, both innate and reliably acquired through evolution, then they will be
known a priori (in the sense of not having been learned from experience). But
to know their status as such will require empirical investigation, and argument
from contingent premisses, such as we provided in chapter 8.1 This
is in stark contrast with the traditional rationalist conception of a priori
knowledge where, since the belief in question can be known to be true by
thought alone, it is often felt that the knowledge that we know it a priori can
also be established by the very same method.
With
these remaining differences between rationalism and my contemporary empiricism
established, let us turn to consider those domains of belief which have
traditionally been regarded as a priori, such as mathematics and logic. And let
us consider the platonist construal of the subject-matter of these disciplines,
which entails that our knowledge of them is not only a priori but genuinely
substantive, being concerned with a reality independent of our minds. Our
question is going to be whether an evolutionary nativism can be pressed into
the service of platonism, to provide a naturalistic account of how the process
through which we acquire beliefs about the abstract realm may be a reliable
one. But first let us consider what such a nativism might look like.
Perhaps
the most plausible form of nativistic platonism is that defended by Jerrold
Katz.2 His approach combines evolutionary nativism with more
traditional appeals to intuition as a warrant for our a priori knowledge of
logic and mathematics. The advantage of this combination of views is that it is
obviously not the case that we simply find ourselves with mathematical beliefs,
as is perhaps so with folk-psychology. Any adequate account of mathematical
knowledge should find some place for the concept of intuitively-acceptable proof
of new mathematical propositions. But then on the other hand, intuition,
without the backing of a rich nativism, cannot be thought of as a form of
direct access to the abstract realm, as we saw in chapter 3.
Katz's
idea is that we have an innately given faculty of intuition, whose constituent
structure mirrors that of the abstract realm. This faculty enables us to
construct mental representations of items in the abstract world, and to intuit,
on this basis, what has to be true about those things. But intuition is not to
be thought of on the model of sense-perception. His claim is not that abstract
objects exert a causal influence on our minds via the faculty of intuition,
those objects being directly responsible for our beliefs. It is rather that the
innately-given structure of the faculty of intuition causes us to have true
beliefs about items in the abstract realm, by determining our intuitive
judgements when we manipulate our mental representations of those things. On
this account our knowledge of the abstract realm is genuinely a priori. For it
is arrived at by thought alone, this being the only process directly involved
in our acquisition of new mathematical or logical beliefs. But on the other
hand, the intuitions we arrive at by such thought will only be reliable guides
to the way things are in the abstract realm, if the structure of the faculty of
intuition has been reliably determined in evolution. So the knowledge that we
know is not equally a priori.
What
emerges from this discussion of Katz's platonism is this. Even if it were to
turn out that empiricists can now accept the existence of substantial knowledge
obtainable by reason alone, through accepting some form of evolutionary
nativism, they can still maintain their traditional opposition to the
characteristic rationalist thesis that we may know, by reason alone, that we
are capable of obtaining substantial information about the world (or about
ourselves) by reason alone. For they will continue to demand a naturalistic
explanation of how reason can have acquired the power reliably to generate
truths about things outside of itself. It seems certain that no adequate answer
to this can be provided, in turn, by reason alone. It will rather require the
backing of some evolutionary (and hence empirical) explanation, if reason's
claims to knowledge are to be allowed. So at least one further aspect of the
traditional debates between empiricists and rationalists will survive my
proposed account of the core of empiricism. Only if our a priori knowledge
concerns nothing which is independent of our minds (that is to say, only if it
is analytic) can we know a priori that we possess such knowledge.
Nativism and Mathematics
The only issue now remaining, is whether an
evolutionary nativism can be used to render acceptable to contemporary
empiricists some of the particular beliefs traditionally defended by
rationalists, such as platonism. For example, can a platonist respond to the
challenge to provide a naturalistic account of the process through which we
acquire knowledge of the abstract realm, by maintaining that such knowledge
arises out of the structure of a faculty of intuition which has been determined
through evolutionary selection? The difficulty for such accounts which I propose
to raise first, relates to the supposed innateness of our knowledge of
mathematics (the subject-matter of which is understood platonistically).
How
can it be maintained that such knowledge is innate, given that a developed form
of mathematics has only been in existence for a few thousand years? For it is
obviously not the case that all humans now living are descended from a common
ancestor, or single group of ancestors, of a few thousand years ago. On the
contrary, human beings were already widely dispersed around the globe at the
time when mathematical knowledge first began to make its appearance. Yet there
appear to be no human beings who lack mathematical ability. Even those
primitive tribes who (until recently) had no knowledge of mathematics at all
(perhaps counting 'one, two, three, many') are able to grasp it quite easily
when introduced to the ideas. So if mathematical knowledge is to be innate in
its own right, it would have to be supposed that it evolved separately in all
the different groups of humans at about the same time. This is extremely
unlikely, given that evolution operates by random gene mutation.
One
response to this problem would be to suggest that our innate knowledge of
mathematics should really be subsumed within our knowledge of logic. For it is
clear that all human communities, however remote in the past, will have had
knowledge of the truths of logic (though whether those truths would concern
platonic objects is of course another matter). Rational planning of a strategy
of action would be impossible if one could not rely upon such truths as 'If I
will not get A unless I do B, and I cannot do B unless I do C, then I will not
get A unless I do C'. It would be tempting, then, for a platonist to maintain
that it is the logic-faculty which is most directly innate, the truths of
mathematics being somehow derivable from those of logic. The cost of this
response to our difficulty, however, is that it commits us to the logicist
programme of reducing mathematics to logic, discussed in chapter 3. Yet few
philosophers today believe that such a programme can succeed.
Chomsky,
too, faces the difficulty of explaining how our knowledge of mathematics can be
innate, since he is another who believes that it is so (but without endorsing
platonism). His solution is to suggest that mathematics can be seen to be at
least partially innate, by virtue of its similarities with the structures
inherent in our innate language-faculty.3 In particular, the grammar
of any human language will contain rules which may be applied recursively an
indefinite number of times. (For example, the rule for 'and' enables you to
reiterate conjunctions indefinitely: 'A and B and C and D and so on'.) Chomsky
suggests that we may exploit this feature of natural language, which is
innately known, in constructing a system of numbering.
Can
a platonist, similarly, make use of Chomsky's suggestion? I believe not. For
the idea is not that the truths of mathematics may be reduced to those
of grammar. This would be absurd. Rather, it is that there is a formal similarity
between the two, which can be exploited to enable us to construct the one on
the basis of knowledge of the other. But a platonist, of course, cannot accept
such an account as it stands, since platonists do not believe that mathematical
truths are constructed. Rather, those truths are held to relate to an objective
abstract realm, which exists independently of our minds. So a platonist would
have to claim that the evolutionary selection of the language-faculty is also
reliable in giving us knowledge of mathematics, in virtue of the formal
similarities between the structures of grammar and the structure of the
mathematical realm. But in fact a mere similarity between a realm to which we
do have reliable access, and one to which we do not, cannot be sufficient to
confer knowledge of the latter.
For
example, my perceptions are generally reliable in giving me beliefs about the
features of the planet Earth. Now let us suppose that there is another planet
in the galaxy which is in fact structurally similar to Earth. Do I thereby have
knowledge of the features of that planet? Obviously not, at least until I learn
on other grounds that the two planets are indeed similar. Equally, then, in the
case of mathematics: the mere fact that there are structural similarities
between the mathematical realm and the rules of grammar cannot be sufficient
for my innate knowledge of the latter to give me knowledge of the former, until
I know on other grounds that the two domains are indeed similar (and in what
respects they are similar). But in order to know this, of course, I should
first have to know something of the mathematical realm; which is precisely what
we were trying to explain.
There
are therefore real difficulties, for a platonist, in appealing to evolutionary
versions of nativism to explain our knowledge of mathematics. For on the one
hand it is implausible that such knowledge should be innate in its own right.
And yet on the other, attempts to explain such knowledge in terms of our knowledge
of logic, or in terms of our knowledge of grammar (both of which are very
probably innate) are beset with difficulties. However, it is worth considering
whether there are any further - more general - arguments against nativistic
platonism. To this topic I now turn.
Evolutionary Platonism?
The main problem for a platonist, as I see
it, is to explain why natural selection of a faculty of intuition should be a reliable
process, leading us to have true beliefs about the abstract realm. For what
difference could the way things are in such a realm make to our chances of
survival in the physical world? If we are to appeal to evolution to explain our
innate knowledge of the abstract realm, then it must be supposed that
possessing truths about such a realm will make a difference in survival. Those
early humans who happened (as a result of random mutations) to have true
beliefs about such a realm would have to have had a better chance of surviving
than those who did not. But what difference can abstract objects make to our
survival? You cannot eat an abstract object, nor poison yourself on one. Nor
can ignorance of an abstract object lead you to be eaten by a tiger or drowned
in a flood. So how can true beliefs about the abstract realm (or a faculty
whose constituent structure enables us to obtain true beliefs about that realm)
have been reliably selected in evolution?
In
reply, it might be said that mathematical truths are extremely useful to those
who possess them. For example, they can enable you to work out what would be a
sufficient store of food to see your family through a winter. Ignorance of a
truth such as 'If we need 6 kilos of rice a day, we shall need 900 kilos of
rice to last through a winter of 150 days' may lead someone to die of
starvation. Something similar will be true of logical and grammatical truths as
well - the former underlying all planning and reasoning; knowledge of the
latter being what makes communication possible.
Empiricists
have traditionally provided a number of explanations of the usefulness of
mathematical truths, ranging from the suggestion that they concern internal
relations between our concepts, to the idea that they consist of high-level
empirical generalisations. But what is hard to see, is how facts about platonic
objects could enter into the explanation of the usefulness of mathematics and
logic. If abstract objects cannot causally affect the natural world, then
presumably there is at most some sort of structural correspondence between the
two realms, in virtue of which pure mathematics can become applied. Then
knowledge of these structural features of the natural world itself would
suffice for whatever practical advantage is to be gained from mathematics.
Thus
it surely cannot be ignorance of the properties of the abstract objects 6, 900
and 150 as such, which would lead me to die of starvation in the example above.
Rather, it would be ignorance of the general structural features of the
physical world in virtue of which it is true that 150 parcels of 6 kilos of
rice makes 900 kilos. It would therefore be sufficient for me to survive that I
should know some general truth from which it follows that, in order to have a
parcel of rice for each day of the winter, each of which when placed on the
scales will tip the needle to the '6 kilos' mark, I must collect a body of rice
which will tip the needle to the '900 kilos' mark. This is not knowledge of
abstract objects of such, but rather knowledge of the manner in which bodies of
matter in the physical world may be collected together or divided up. (The
platonist view, in contrast, must be that these properties of matter are
structurally isomorphic with the properties of objects in the abstract realm,
so that knowing the latter thereby provides us with knowledge of the former.)
It
therefore remains problematic why evolution should be reliable in selecting
beliefs about the abstract realm. For it is not truths about that realm as such
which are an aid to survival, but rather truths about the structural features
of the physical world in virtue of which the truths of mathematics can be
applied to that world. These truths about the physical world could possibly
have been reliably selected through evolution. But to go beyond this, to have
true beliefs about the abstract realm itself, would confer no additional
advantage in survival. In which case it seems that a platonist cannot explain
in evolutionary terms how knowledge of the abstract realm can come to be
innate. This will then leave an empiricist, who rejects claims to knowledge which
cannot begin to be explained, with sufficient reason for denying that we can
have knowledge of mathematics, if its subject-matter is construed
platonistically.
Now
the point here is not that evolution cannot secure more than is necessary to
meet the needs of survival. Plainly it can, if random mutations are what fuel
the evolutionary process. It is rather that evolutionary selection of beliefs
is only a reliable process to the extent that it is the truth of the
beliefs which confer advantage in survival. When it comes to beliefs whose
contents go beyond those structural features of the natural world which may
matter for survival, natural selection is just as likely to produce beliefs
which are wrong as right. It will not matter for survival if innate mathematical
beliefs are false with respect to the platonic realm, provided that they work
successfully in the natural world.
Remember
that what evolution would in the first instance deliver, would be a faculty for
manipulating representations of platonic objects, finding some but not others
of these representations intuitively acceptable. But since the survival-value
of truths about platonic objects derives entirely from the supposed structural
correspondence between the natural and platonic realms, evolution could only be
reliable in respect of those structural features of the natural world. That is,
it would lead us to find '2 + 2 = 4' intuitively acceptable when, and only
when, it is true that things may generally be divided up, and combined, and
counted in such a manner.
In
fact we can easily imagine worlds where '2 + 2 = 4' would not be selected for,
but rather, if anything, '2 + 2 = 5'. For example, a world where two apples and
two pears put into a lunch-box produce a fifth piece of fruit, or a world where
the very act of counting the union of the two sets produces an extra member,
and so on. So how could we know that our innate belief that 2 + 2 = 4 was not
of this sort? Maybe, in the platonic realm, 2 + 2 is really 5, but the
structural facts about the natural world are such as to make '2 + 2 = 4' apply
to it.
The
point here needs stating with some care. For if evolution selects a faculty of
mathematical intuition in virtue of the access which the latter gives us to
structural facts about the natural world, and if there is in fact a
structural correspondence between the natural and platonic worlds, then
evolution will be reliable in delivering us truths about the platonic realm.
Hence we would, in these circumstances, have innate first-order knowledge of platonic
objects. And we cannot here object, against the thesis that evolution is
reliable with respect to the platonic domain, that if the mathematical facts
had been different, evolution would still have selected for us the very same
mathematical beliefs. For we cannot intelligibly suppose that truths which are
necessary (as are those of mathematics) can be other than they are.
It
seems that an appeal to evolution can, after all, explain how innate knowledge
of platonic objects is possible, given a reliabilist conception of knowledge.
But what appeals to evolution cannot do, is provide us with any reason
for thinking that there is a structural correspondence between the
platonic and natural realms. For we cannot, in this context, take our knowledge
of 2 + 2 = 4 for granted. While this proposition may strike us as intuitively
obvious, our question is precisely whether we have any reason to think that our
intuitions are a good guide to the states of the platonic realm. Appeals to
evolution therefore cannot give us the knowledge (second-order) that we have
knowledge of platonic objects. On the contrary, it can only generate scepticism
on the issue.
If
we had no other option but to accept the platonist interpretation of the
subject-matter of arithmetic, then the only proper conclusion would be that we
have no reason to think that we have any knowledge of that subject-matter
(namely, of platonic objects). But of course there are other options available,
as we saw in chapter 3. Appeals to evolution therefore cannot render platonism
acceptable to a contemporary empiricist. While evolution may possibly explain
how innate knowledge of platonic objects is possible, it cannot provide us with
any reason for thinking that such knowledge actually exists. While this is the
case, it will remain necessary for an empiricist - who insists that claims to
knowledge should only be allowed where we can begin see in natural terms why we
are reliable in making those claims - to continue to reject platonism.
Other Rationalist Beliefs
It turns out that the empiricist objection to
platonism remains intact, even given my construal of the core of empiricism,
and the consequent empiricist acceptance of evolutionary nativism. But what of
other aspects of the traditional debate between empiricists and rationalists?
What of empiricist objections to the claims of rationalists, that we can have a
priori knowledge of the existence of God, of the existence and immortality of
the soul, or of the freedom of the will? On my account, the most fundamental of
these objections is that no naturalistic explanation can be provided of the
manner in which we might acquire such knowledge. Can this, too, survive the
empiricist acceptance of nativism?
Here,
as before, the main issue is whether the beliefs in question, or the faculty of
thought which enables us to intuit the truth of those beliefs, could have been
reliably selected through evolution. For the sake of simplicity, let us confine
ourselves to consideration of only one of these forms of nativism. Let us
consider the suggestion that we possess an innate faculty of mind which enables
us to discover the truth of the beliefs in question by thought alone. Our
conclusions will then readily generalise, to cover the possibility of innate
belief as well.
Consider
first, the suggestion that we possess an innate capacity for acquiring
knowledge of the existence of God. Can we explain how such a faculty of mind
might have been reliably selected in evolution? Note that we cannot appeal to
the social or psychological benefits of religious belief in this connection
(supposing that there are any). For such benefits would be independent of the truth
of the belief in question. Selection of a belief on grounds unrelated to truth
is not a reliable method of acquiring true beliefs. So even if the faculty
enabling us to arrive at belief in God were innate, and that belief were true,
still our belief would not constitute knowledge if the explanation of its
innateness were in terms of benefits unrelated to truth.
The
only truth-related way in which belief in God could conceivably have value in
survival, would be if such a belief enabled one to secure, through prayer,
God's positive intervention in the world. But this violates the empiricist
constraint that the explanation of a belief-acquisition process should be a
natural one, mentioning only processes which operate in accordance with causal
laws. For God's intervention in the world, for example to save a life, would
involve a violation of such laws. So I conclude that even an empiricist who
accepts nativism will still have an objection of principle to the rationalist
claim that we can have a priori knowledge of the existence of God.
Consider,
now, the suggestion that we have an innately given faculty of reason which
enables us to discover a priori the existence and immortality of the soul, or
the freedom of the will. Once again it is difficult to see how a faculty with
such powers could have been reliably selected through evolution. For it is hard
to see what possible difference the possession of true beliefs on these matters
would make to our survival. (Just as in the case of belief in God, it it not
sufficient to show that the beliefs in question have survival-value -
perhaps belief in immortality can keep us from despair, for example, and
perhaps belief in freedom of the will is a necessary condition of our having
moral beliefs. It needs to be shown that it is the possession of a true
belief which aids survival, since otherwise the belief-acquisition process will
not be a reliable one.) It therefore follows that an empiricist will object to
such knowledge-claims, on the grounds that no naturalistic account can be given
of the manner in which we might acquire that knowledge.
Many
philosophers have claimed to be able to demonstrate the existence of the soul,
for example, believing that they could establish the conceptual impossibility
of our lacking a soul. In fact such arguments generally turn out to be
unsound.4 But even supposing that some such argument were to
succeed, empiricists would have a further line of defence. For suppose that it
had been demonstrated that we cannot conceive otherwise than that we possess a
soul. An empiricist will still demand to know how we come to have a faculty of
reason which is capable of discovering a priori the fundamental nature of
ourselves. How can thought alone have acquired such powers? Unless some
evolutionary explanation can be discovered, an empiricist will wish to exploit
the gap between our inability to conceive of something being otherwise (a fact
about us and our concepts), and the fact of something being so (a fact about
the world). Even if we cannot conceive of ourselves otherwise than possessing a
soul, this does not mean that we know of the existence of the soul. On the
contrary, this latter knowledge should be denied of us.
It
may be worth stressing in conclusion of this section, that the empiricist
objection is not to the idea that reason, as such, has value in survival. This
would be absurd. It is obvious that the human ability to reason is at least one
of the distinctive features underlying the success of the species. The
objection is only to the survival-value of a faculty of reason which would
contain within itself the power to generate true beliefs about the matters
generally beloved by rationalists, such as the existence of God or the soul.
Transcendental Arguments
Finally, let us consider how empiricists
should respond to various forms of transcendental argument, which attempt to
show a priori what must be the case about ourselves or the world if experience
is to be possible at all. Such arguments were developed most famously by
Immanuel Kant in his Critique of Pure Reason, though he has had many
imitators. His position is often thought to provide a kind of synthesis of the
empiricist and rationalist traditions. For example, suppose it were
successfully argued that it is a condition for the possibility of our
experience that all events should have causes. That is, suppose it were shown
to be conceptually necessary that all events must have causes, if there is to
be any experience at all. Is this something that my sort of empiricist could in
principle accept? The answer will in fact depend very much on how the
conclusion of such an argument is to be understood.
Suppose
first, that the conclusion purports to be a truth about a mind-independent
reality (many contemporary philosophers have constructed transcendental
arguments of this sort). Then in my view an empiricist should object to the
possibility of our having a priori knowledge of it. For the fact (supposing
that it were a fact) that we cannot conceive of the world otherwise than
governed by universal causality, is not enough to establish that all real
events do have causes. Nor is it at all obvious how a faculty for obtaining
such knowledge could have been reliably selected through evolution. No doubt a
true belief in universal causality might be useful in survival. But given that
only a handful of people have ever employed their supposed innate faculty in
order to demonstrate the existence of universal causality, it is hard to see
what advantage the possession of such a faculty can have conferred in
evolution.
Kant
himself saw, at least partly, the empiricist challenge to rationalism. (He
regards it as a challenge to show a priori how reason can have the power
to generate knowledge of anything outside of itself, rather than, as I propose,
a demand for a natural explanation of reason's reliability on such
matters.) He put forward his doctrine of Transcendental Idealism in an attempt
to overcome it. Kant's view is that the structure of the human mind imposes an
order on the world of our experience, and what we may know a priori is limited
to that world, considered as an object of possible experience by us. On this
view, our proof that all events have causes would be guaranteed to be reliable,
because it is a structure which reason itself imposes on the course of our
experience.
However,
there are two different ways in which Kant's Transcendental Idealism can be
understood, each finding some basis in his writings. The first regards it as a
sort of dual aspect realism. On this view, there is a realm of objects existing
independently of our minds (hence 'realism'). But those objects can be thought
of under two different aspects - firstly, as possible objects of our experience
(as Phenomena), and secondly, as they are in themselves, independently
of our experience (as Noumena). All our knowledge, whether empirical or
a priori, would be confined to the realm of Phenomena. And the truth
that all events have causes would be imposed upon Phenomena by the
structure of our reason. But then there is one item of knowledge claimed on
this account which can not be explained as a structure imposed by our
reason on objects as Phenomena - namely, the claim that there are
objects independent of our minds (the claim that there are Noumena). So
if this is how Kant is to be interpreted, he is still wide open to the
empiricist challenge.
Alternatively,
Kant's Transcendental Idealism may be understood as a form of sophisticated
phenomenalism. On this view, it cannot be known that there are objects
independent of our minds, and all a priori knowledge concerns only patterns
which the mind imposes on the course of our experience itself. On this account,
at least one of the empiricist objections of principle to the existence of
transcendental arguments collapses.5 For the a priori knowledge that
all events have causes, while being informative (that is, new), would not be
genuinely substantive (that is, mind-independent). The truth that all events
have causes would in fact only concern the course of our experience itself.
Admittedly, this truth would be, to a degree, substantive, since it is a
contingent fact that there should be any experience at all. (Recall from
chapter 1 that knowledge may be substantive by virtue either of being
contingent or of concerning some reality independent of our thoughts.)
But the substantive content of the claim that all events have causes would not
reach any further than the bare, but substantive, claim that there is
experience. If this is how Kant is to be interpreted, then in attempting to
surmount the empiricist challenge he has, in reality, surrendered to it. For he
has failed to show how substantive a priori knowledge is possible. He would
only have been prevented from seeing this because he does not distinguish
clearly between knowledge which is informative and knowledge which is
substantive.6
In summary, I conclude that my proposed account of the core of empiricism leaves much of the traditional dispute between empiricists and rationalists intact. Both sides may now accept the existence of innate knowledge. But they remain divided over the general claim that it is possible for us to know that we have substantial knowledge through thought alone, as well as over many specific claims to a priori knowledge.