Return
to Human Knowledge and Human Nature contents
5. IS INNATE KNOWLEDGE EVEN
POSSIBLE?
In this chapter I shall consider various
proposed accounts of the concept of knowledge, and how they bear on the
question whether innate knowledge is possible.
A Justificationalist
Argument
In chapter 4 we looked at Locke's reasons for
denying that there is in fact any innate knowledge, and found them wanting.
Some people have thought that he also had available to him a knock-down proof
of the impossibility of innate knowledge, deriving from the concept of
knowledge itself. But in fact this argument depends upon an account of the
concept which is probably incorrect. Moreover, that Locke did not employ such
an argument is evidence that he conceived of knowledge somewhat differently, in
a way that would allow innate knowledge to be possible in principle (even if it
is not actual). I shall first explain the conception of knowledge from which
the argument derives, and then develop the argument itself. I shall argue later
that the account of knowledge in question is incorrect.
Many
have held that knowledge can be defined as justified true belief. That is to
say: in order to count as knowing something, you must at least believe
it, what you believe must in fact be true, and you must have
sufficiently good reasons for your belief. Such an account is not
without its attractions. For example, it is capable of explaining why someone
who 'plays a hunch' that a particular horse will win a race does not know that
it will win, even though the horse does in fact win just as they believed that
it would. For such a person has insufficient reason for their belief - rather,
it was a mere guess. The account can also explain why someone who has excellent
reasons for their belief still does not have knowledge, if it should turn out
from another, or later, perspective that their belief is not in fact true. For
example, even someone who has just doped all of the other horses in the race
does not have knowledge that the remaining candidate will win, if it should
turn out, through some bizarre sequence of accidents, that one of the others
does.
(We
can leave open, here, the question of just how powerful the reasons for
believing some truth would have to be, in order to count as 'sufficiently good'
to qualify that belief as knowledge. For essentially the same issue - of the
degree of grounding required for knowledge - arises in connection with each of
the various candidate accounts of the concept which we shall be considering.
But in fact I am inclined to think that the standards for a belief to count as
knowledge are purpose-relative, varying depending upon what is at issue in the
context in question. What counts as knowledge in response to a casual enquiry
may not count as knowledge in a court of law, where someone's life may hang in
the balance. So I suggest that 'sufficiently good' might mean 'good enough for
the purposes in hand'. I shall return to this idea in chapter 11.)
If
knowledge is justified true belief, then an innate belief can count as innate
knowledge only if it has a justification which is also innate. This is because,
in order for someone to know something on this account, it is not enough that
there be a justification. Rather, the person in question must possess
that justification. For consider the following example. Playing a hunch, I
decide that the horse in the red and green colours will win the race. But
unknown to me, there is in fact a sound justification for my choice - that
horse has won all of her last ten races, whereas the rest of the field have
been markedly unsuccessful. Even supposing that my horse does in fact win, I
surely did not know that she would. For I was just guessing, and that is no
basis for knowledge on any account of the matter. So even if there are innate
beliefs, they will not count as innate knowledge if the evidence supporting
them only emerges from subsequent experience.
Your
reasons for a belief will presumably consist, in general, of other beliefs. If
you have a reason for believing that world deforestation will be disastrous,
then this must consist in some other belief or beliefs which imply or otherwise
support that belief. And these beliefs, in their turn, must be justified if the
original belief is to be justified. For you cannot provide justification for a
belief through beliefs which are themselves unjustified. So it looks as
if we have a regress: in order for one innate belief to be innately known,
there must be some other innate belief which justifies it; and this in turn
must have a justification which is also innate, and so on to infinity. In which
case innate knowledge is impossible. For we surely cannot have infinitely many
innate beliefs.
It
might be said that this regress could terminate with beliefs which are self-evident,
which provide their own justification. This may be so. But in fact there are
only two sorts of belief which are plausible candidates for such self-evidence,
and neither of these kinds will be innate. First, a belief may be self-evident
in virtue of concerning the subject's own immediate experiences (such as the
belief that I now feel a pain). But in this sort of case it would be absurd to
suppose that the belief in question is innate. Secondly, simple truths of
reason may be self-evident, such as the belief that 2 + 2 = 4. But in fact such
beliefs will only count as self-justified if they concern nothing beyond
themselves (that is, if they are analytic). For in this context,
self-evidence cannot be merely a matter of a belief striking us as intuitively
obvious, since there would remain the question of what reason there is for
thinking that such obviousness is a good guide to the state of the world in question.
To
see this, suppose that the subject-matter of the proposition '2 + 2 = 4' is
construed platonistically, as concerning necessarily-existing abstract
individuals. Although this proposition may seem to us undeniable, this is in
fact insufficient for it to count as self-justifying. For what reason have we
for supposing that obviousness-to-us (a feature of our psychology) is a good
guide to the way things really are in the abstract realm? If we lack any reason
for this belief, then the proposition will not really be justified, despite our
inability to deny it. And if we possess such a reason, then the justification
for the proposition will derive from some other belief of ours, and the regress
will continue. So innate self-evident propositions can only halt the regress if
their subject-matter concerns nothing beyond themselves (that is, if they are
analytic).
What
this argument appears to show, is that it is impossible that there should be
innate knowledge which is at the same time substantive (which concerns a
reality which is independent of our minds). This is not quite the same as
saying that innate knowledge is impossible altogether. But in fact, given any
developmental construal of innateness (two versions of which were outlined in
chapter 4), that a belief is analytic will undercut all warrant for supposing
it to be innate. Only if we were prepared to believe that analytic propositions
are stored in the mind as such from birth could we accept that there is innate
knowledge of any sort, if the argument above is sound.
The
reason why acquired analytic beliefs would not be counted as innate, is that
there will be no problem in explaining how people might come to possess them
through general learning mechanisms. Since knowledge of simple (self-evident)
analytic truths requires only that we be capable of discerning simple relations
between our own concepts (ideas), there will be insufficient reason for saying
that the knowledge in question is innate. Even if the constituent concepts are
innate, our knowledge of the relations between them need not be. For in order
to explain how we come to possess such knowledge, we only need to suppose that
we have the general ability to compare and contrast our own ideas.
A Coherentist Suggestion
It should be noted that the argument
developed above against the possibility of innate knowledge presupposes the
truth of foundationalism in the theory of knowledge. For it assumes that the
justification-relation between beliefs is a linear one. The argument
takes for granted that one belief will be justified by another, which is
justified by another, and so on until we reach beliefs which somehow justify
themselves (either reports of immediate experience or simple analytic truths);
these latter beliefs constituting the foundation for the rest. But what if we
were to embrace a coherentist account of justification instead? What if we were
to take the view that a belief may be justified by virtue of forming part of a
mutually-supporting network of beliefs, each member of which receives its
justification from its relationship with the others? This would be to picture knowledge, not as a pyramid built on
secure foundations, but rather as a web held together by the relationships
between its component parts. If this account were adequate, then there would be
no reason why we could not possess innate knowledge. For we might be born with
(or born determined to develop) a set of consistent and mutually supporting
beliefs on some subject-matter. In which case, provided that those beliefs were
true, we could be said to have innate knowledge. For given a coherentist
conception of justification, those innate beliefs would not only be true but
justified.
However,
coherentism faces considerable difficulties if taken at face value. For there
may be any number of coherent bodies of belief which are mutually inconsistent
with one another. For example, I suppose that both Christianity and Hinduism
are internally coherent. Yet they cannot both be true, since there cannot both
be one God and many. But in that case it is very hard indeed to see how the
mere fact of the coherence of Christian belief could render it justified, given
that there are other incompatible beliefs which are equally coherent. For there
would then be no justification for being a Christian rather than a
Hindu.
Even
if coherence is construed in such a way that a body of belief must be more
coherent than any incompatible set of beliefs in order to count as justified,
there will still be problems. For suppose that I had read Jane Austen's Emma
as a child, under the mistaken impression that it was the biography of an
historical individual. Then as an adult I still retain a great many beliefs
about Emma Woodhouse, but have forgotten how I came to have them. Since these
beliefs constitute a consistent mutually-supporting set, a coherentist seems
committed to saying that they are justified. But this is counter-intuitive. I
think we should be strongly inclined to deny that a coherent set of fictional
beliefs could count as justified in this example.
Coherentists
are no doubt correct to stress, as against some (but not all) forms of
foundationalism, that a justification for a belief can appeal to
non-self-evident principles, such as inference to the best over-all explanation
of a given range of phenomena. But the beliefs within an explanatory network
must surely be anchored somewhere. They cannot just 'float free' of all
constraint, as do my beliefs about Emma Woodhouse in the example above. On the
contrary, I suggest that there must be some beliefs given to us in experience,
providing the data which the coherent network explains. Just such a weakened
form of coherentism (which may alternatively be seen as a weakened form of
foundationalism) will be considered in more detail in chapter 12.
If
some suitably weakened form of coherentism is true, then it will certainly be
possible that there should be innate knowledge. For suppose that some set of
true beliefs were innate, serving to provide a coherent explanation of the data
provided by subsequent experience. Then provided that those beliefs could not
have been (or at least were not) learned from the experiences in
question, they would count as innately known. They would be true beliefs whose
justification is triggered by experience (either generally or locally), rather
than derived from it.
However,
this suggestion cannot serve to explain why Locke (and other empiricists)
failed to deploy the argument against the possibility of innate knowledge
sketched earlier. For it seems certain that none of the classical protagonists
in this debate - whether empiricist or rationalist - endorsed a coherentist
conception of justification. A better explanation is that Locke employed some
other account of knowledge, which would leave open the possibility of innate
knowledge even when conjoined with foundationalism. This idea is also supported
by considerations of charity, since there are powerful reasons for thinking
that the justificationalist conception of knowledge is false. To this issue I
now turn.
Against Justificationalism
While a conception of knowledge as justified
true belief may raise a difficulty for the possibility of innate knowledge,
that conception has itself come under increasing pressure in recent decades.
One problem with it, is that we are then constrained to deny knowledge to
those, such as children, who may be unaware of the justification for their true
beliefs. Yet it is counter-intuitive to insist that a child does not know that
Father is washing-up dishes (even though the child can see that he is doing
so), on the grounds that the child is as yet incapable of providing any sort of
justification for their belief. (After all, the child may be just as reliable in
reporting such matters as the rest of us.) Equally, suppose that you are unable
to respond adequately to a sceptic who demands to know what reason you have for
believing that you are not now dreaming. Then if justificationalism is correct,
you do not in fact know that you are awake. It will only be the privileged few,
who can answer the sceptic, who may be said to know that they are not dreaming;
the rest of humankind, in lacking a justification, will lack knowledge also.1
This
problem is really very widespread. For few of us are capable of articulating
anything like a convincing justification for more than a handful of our
beliefs. (Try asking an ordinary person for their reasons for one of their
everyday beliefs - for example, their belief in their mother's birthday - and
see how quickly their feeble attempts at justification will run out.) Yet we
cannot avoid the difficulty by saying that knowledge only requires that there be
a justification, not that the subject be aware of it. For as we saw earlier,
this would then permit many beliefs which are really just guesses to count as
knowledge. Neither will it help to say that knowledge only requires that a
justification be constructable from amongst the subject's other beliefs,
not that the subject need actually, themselves, have effected such a
construction. For suppose that I had lain a bet on the winning horse because I
liked the colours in which her jockey was dressed. This was surely a mere guess
rather than knowledge, even if I did in fact know (but failed to recall) that
she had won all of her last ten races.
Not
only does justificationalism conflict with our common-sense views about the
circumstances in which people may be said to possess knowledge, but it is far
from clear why we should want to insist that knowledge requires
justification. For our main interest in the question whether or not someone
knows something, is that a positive answer will warrant us in adopting that
belief for ourselves. And while the fact that they are justified in holding
their belief may be sufficient to give us such a warrant, it is unclear why it
should be necessary.
The
main use which we have for the concept of knowledge, is that if someone knows
something, then we may safely add what they know to our own stock of beliefs.
We therefore need to be able to establish that they have knowledge
independently of assessing, for ourselves, the truth of their belief. (If we
could do that, we should not need to know that they know.) And the basic
fact we need to establish is simply that their belief is, in the circumstances,
likely to be true. Whether or not they themselves can provide
reasons for their belief is of no particular importance. All that really
matters is that the believer should, in the circumstances, be reliable
(or reliable enough) on the topic in hand. Indeed, it seems to me that from the
point of view of our practical interest in knowledge, the appropriate concept
is a reliabilist one, of the sort to be defended in the next section.2
Even
if justified true belief is not necessary for knowledge, it may still be
sufficient. But this thesis, too, is vulnerable to counter-example. Thus
suppose that I had turned on my radio in order to hear a live account of the
1988 Olympic 100-metre final, having just read in the paper that such a
programme was due to be broadcast at that time. I duly hear the commentator
describe a race in which Ben Johnson beats Carl Lewis, setting a world-record
in the process - which is exactly what was in fact happening. But unknown to me
there had been a fault in the satellite transmission from Seoul, and they were
broadcasting repeat-commentary from the previous year's World Championships
instead, in which a similar race took place. Here I have a justified true
belief that Johnson has just set a world record in beating Lewis, but I surely
do not know it. There are many examples of this general sort.3
What
this example shows, is at least that the account of knowledge as justified true
belief needs to be supplemented in some way. Perhaps we might require in
addition that the belief should be non-accidentally connected with the fact
which it concerns. For the distinctive feature of the example is that the truth
of the belief in question is entirely fortuitous, despite being justified. But
our earlier arguments (concerning the knowledge of children, and your knowledge
that you are now awake) suggest something much more radical. For if a belief
can count as knowledge although the believer lacks any justification for it,
then we should reject the justification-condition altogether, replacing it with
some other clause.
Causal Theories and
Reliabilism
Some have developed an account of knowledge
which is causal in character. They have said that knowledge is true belief
which is caused by the fact which it concerns.4 This can explain
why, in the example above, my belief that Johnson has broken the world record
does not count as a case of knowledge. For the fact of his having broken the
record fails to play any role in the causation of my belief. The account can
also explain why the child knows that its father is washing-up, despite lacking
a justification. For the child's belief is in fact caused by perceiving Father
at work at the sink. Similarly, you may be said to know that you are not now
dreaming, despite your inability to answer the sceptic, provided that your
belief that you are not dreaming is in fact caused by the fact that you are
awake.
Although
the causal theory of knowledge is in many ways attractive, it runs into trouble
if it is to allow for the possibility of knowledge of the future, and of
knowledge of unrestrictedly general statements. For future states of affairs
surely cannot be causes of present belief, and it is implausible to suggest
that our beliefs about general laws of nature are in fact caused by those laws
themselves. Thus suppose that I have just set light to the fuse on a fire-work.
I know that it is a rocket of reliable manufacture, which has never failed in
the past. I know that it has been stored in dry conditions; that the weather itself
is now dry; and that there is no wind. Surely in these conditions I may know
that the rocket will shortly take off. But it is not the fact that the rocket
will take off which causes my belief that it will (as the causal theory of
knowledge would require). This would involve backwards causation, which is
impossible. My belief is rather caused by the facts which I have mentioned,
such as my belief that the fire-work is dry. Equally, consider my belief that
all massive bodies attract one another (the law of gravity). This may surely
count as knowledge. But my belief is not caused by the fact that all
bodies attract (past, future and distant), but rather by the bodies which I
have observed, and by the reports of other observers and scientists. So again,
we appear to have a case of knowledge without causation by the fact which it
concerns.
We
can keep all of the advantages of the causal theory, while avoiding its
difficulties, if we say that knowledge is true belief which is caused by a
reliable process. A reliable process is one which generally issues in true
beliefs, and which also serves, in the particular case in hand, to reliably
discriminate truth from relevant falsehood.5
This
can explain how we may have knowledge of the future, if the process of
inference from past facts and present tendencies is (in the circumstances) a
reliable one. It can also account for our knowledge of laws of nature, if the
processes of induction and inference to the best explanation employed by
scientists are also generally reliable. Yet we can still explain why I would
lack knowledge of Johnson's victory, in the example discussed above. For while
listening to radio-broadcasts may be a generally reliable method of acquiring
true beliefs, in this case my belief would have been the same even if Johnson
had not won. In the circumstances the process which causes my belief is
not reliable in discriminating truth from relevant falsehood. We can also still
explain the fact that the child knows its father is washing up, as well as your
knowledge that you are not now dreaming. For the child's perceptions in
circumstances such as this (with the observed events taking place not too far
away, in good lighting, and not too swiftly) will generally result in true
beliefs. And your belief that you are now awake results from a process which
involves your current conscious awareness which is also reliable.
While
the thesis that knowledge implies causation by a reliable process is beginning
to look plausible, it might be claimed that we have been too hasty in dropping
the justification-condition. For consider the following example.6 I
am taking part in a psychological experiment, where the experimenter has told
me that I have been given a drug which will totally distort my visual
perception. Nevertheless, I perversely continue to believe, on the basis of my
perceptions, that there is a pink rabbit sitting on the desk in front of me.
But in fact the experimenter had given me a placebo, and there really is a pink
rabbit on the desk, which has escaped from the genetics laboratory next door.
This is a case in which my belief appears to be caused by a reliable process
(undistorted perception), and is in fact true. But, it may be urged, I surely
do not know that there is a rabbit on the desk. This is because it is,
in the circumstances, unreasonable of me to trust my visual perceptions.
In
fact, however, the correct response to this example is that I do have
knowledge that there is a pink rabbit on the desk before me. This is because I
am, in the circumstances, a reliable informant on such matters. The intuition
that I do not have knowledge only arises through a confusion of levels. It is
true that, representing the example to myself in the first person singular, I
cannot correctly claim to have knowledge of the presence of the rabbit. But
others may correctly claim this of me, and I may correctly claim it of myself
at a later time when the details of the situation emerge. This will be made
clear in the next section.
Orders of Knowledge
Reliabilism can only be acceptable if we are
prepared to deny what is sometimes called 'The KK thesis', which holds that in
order to know something, you must at the same time know that you know it. For
the process which gives rise to a belief can in fact be reliable even though
you yourself do not know that it is; indeed, you may not even know what that
process is. Then according to reliabilism your belief will count as
knowledge, but you will not know that it is knowledge.
There
is some reason to think that it is the KK thesis which gives rise to
justificationalism. For if knowing something requires that you know that you
know it, then you will only have knowledge if you know that your beliefs have
been formed in a way which renders them likely to be true, thus in fact possessing
a justification for the belief. But why should the KK thesis be accepted?
Surely knowing (first-order knowledge) is one thing, knowing that you know
(second-order knowledge) is another. Indeed, the thesis threatens to degenerate
into a regress. For if taken generally, then you will only know that you know
something, in turn, if you know that you know that you know it; and so on.
It
seems to me likely that both justificationalism and the KK thesis result from
conflating our basic interest as epistemologists (theorists of knowledge) with
what it is that we take an interest in, namely knowledge. I propose to
argue for this in a number of steps. First, I claim that the central question
of epistemology is what we may know ourselves to know (a second-order question).
This is easily overlooked. For if I ask myself 'What do I know?' the term
'know' only figures in the question once. But in fact, in returning a positive
answer to the question whether I know that such-and-such, I should be tacitly
claiming to know that I know it. For suppose that I assert 'The ozone layer is
shrinking'. This is in fact a claim to know that the ozone layer is
shrinking. Any unconditional assertion is, in effect, a tacit knowledge-claim.
For to respond to it by saying 'But you don't really know that' is to
challenge the speaker's right to say what they have said. Similarly, then, if I
assert 'I know that the ozone layer is shrinking'. This is in fact a tacit
claim to know that I know it.
My
second thesis is that the only generally reliable method of acquiring
second-order beliefs about what I know (that is, of acquiring second-order
knowledge) is by a process of reasoning. The only way to know that you know, is
to construct a justification for claiming to have knowledge. There could not,
for example, be any process of introspection which would be reliable in
obtaining for us beliefs about our own states of knowledge. For while
introspection might have access to my first-order states of belief, it
cannot have access to the fact that those beliefs were produced by a reliable
process, when they were. It follows that when we come to do epistemology
(seeking to know what we know) justificationalism and reliabilism converge -
each will arrive, by different routes, at the view that we need to seek
justification for our claims to knowledge.7 It would then be quite
natural that philosophers should mistakenly slip into thinking that knowledge
itself (whether first-order or second-order) requires justification.
This,
then, is how we should respond to the example of the pink rabbit presented
earlier: since my belief in the presence of a rabbit is in fact produced by a
reliable process, I may truly be said (by others or by myself at a later time)
to have knowledge. For someone who knows that I have been given a placebo (and
so who knows that my perception will most likely be accurate) may reasonably
treat me as a reliable informant on the matter. It is only when I raise the
question about myself at the time - 'Do I know that there is a pink rabbit here?'
- that the answer has to be negative. For the question is, tacitly, a
second-order one. An affirmative answer will claim to know that I know. But as
we have just seen, such second-order claims have to be based on a process of
reasoning. And it is plain that, in the circumstances, I lack any justification
for saying that I know.
It
is important to note that the relevant concept of justification, in this
context, has to do with the reasons which are available to the person whose
knowledge is in question. Whether or not a belief is justified has to be
determinable from the perspective of the subject themself.8 This is
because the main use which we have for the concept of justification is in
determining what we ourselves should believe. The question whether someone else
is justified in holding a belief is generally of very little interest, except
in so far as believing on the basis of a justification is one form of reliable
belief-acquisition, or except by way of providing a moral applicable to our own
case. We therefore want a concept whose conditions of application are defined
in such a way as to be available to us. (This is in contrast with the case of
knowledge itself, where our main use for the concept is a third-person one - on
the question whether someone knows something will turn whether or not I myself
should believe that thing. Here all that matters is that the person in question
should be reliable on the issue at hand.)
In
the light of the above considerations it seems to me quite possible that
classical philosophers such as Locke and Descartes may have been implicitly
working with a reliabilist conception of knowledge, despite their overt concern
with justification. For as epistemologists, they were interested in the
question of what we could know ourselves to know. To seek an answer to this
question is to look for reasons in support of our various (first-order) claims
to knowledge. But the fact that the process of acquiring second-order knowledge
of this sort involves justification, is quite consistent with the thesis that
knowledge itself is reliably acquired true belief.
Indeed,
Descartes' strategy in the Meditations, for example, is quite naturally
interpreted in this light. For what he wishes to know, is whether the processes
through which he acquired his beliefs were reliable. He believes that he can
answer this question affirmatively, by virtue of proving the existence of a
veracious God. But there is no suggestion that ordinary people do not have
perceptual knowledge until they, too, have grasped such a proof (that is, until
they, too, have provided a justification for their knowledge-claims). Rather,
they have knowledge because perception is in fact reliable. What they
lack is only the knowledge that they know.
Innateness and the A Priori
If reliabilism is acceptable as an account of
knowledge, then the possibility of innate knowledge is left open. Innate
beliefs will count as known provided that the process through which they come
to be innate is a reliable one (provided, that is, that the process tends to
generate beliefs which are true). There are two possible candidates for such a
process: Divine intervention on the one hand, and evolution on the other. We
could maintain, as most classical rationalists did, that innate beliefs are
directly implanted in the mind by a veracious God. Or we could hold that innate
beliefs have been acquired through evolution, via natural selection. Each of
these candidate processes would most probably be reliable. We shall return to
consider them in more detail in later chapters.
I
have argued for a conception of knowledge which would allow innate knowledge to
be possible, no matter whether we have a foundationalist or a coherentist
conception of justification. And I have suggested that the reason why Locke did
not employ the argument against innate knowledge sketched at the outset of this
chapter, is that he too may have conceived of knowledge in reliabilist terms.
But would innate knowledge, if it existed, at the same time be a priori? In one
sense at least it would be. For as we noted in chapter 1, one thing which can
be meant by saying that something is known a priori is that our knowledge of it
is independent of empirical support. In this sense, a priori knowledge is
knowledge which has not been learned from experience. And innate knowledge
(reliably caused innate belief) would apparently possess just such a status,
neither having been learned from experience, nor requiring support from
experience to qualify as knowledge.
However, to say that innate knowledge would be a priori is one thing, to say that we may know that we know it a priori is quite another. Knowledge which is a priori in this sense will not, unlike what is generally held concerning the other main traditional category of the a priori, be knowable to be such by a process of thought alone. On the contrary, to know that a belief is both innate and reliably produced, and hence that a given item of knowledge is a priori, may require argument from empirical premisses. An appropriate defence of nativism will not itself be a priori but broadly empirical. This topic - the defence of nativism - will now occupy us over the next three chapters.