You are on page 1of 5

Michael Lacewing

A priori knowledge
A PRIORI AND A POSTERIORI KNOWLEDGE
A priori knowledge is knowledge of propositions that do not require (sense) experience
to be known to be true. Propositions that can only be established through experience are
a posteriori. The a priori/a posteriori distinction is about how to check or establish
knowledge. It is not a claim about how we acquire the concepts or words of the
proposition. Babies are not born knowing that all bachelors are unmarried! Yet this is a
truth that clearly doesnt need testing against experience; we know it is true just by
knowing what it means. Of course, we first have to learn what it means, but that is a
different issue from how we check if it is true.
This contrast between a priori and a posteriori knowledge should be distinguished from
another contrast, between analytic and synthetic propositions. A proposition is analytic if
it is true or false just in virtue of the meanings of the words. Many analytic truths, such as
all bachelors are unmarried, are obvious, but some are not, e.g. your mothers brothers
fathers nieces sole female cousin is your mother (think about it!). A proposition is
synthetic if it is not analytic, i.e. it is true or false not just in virtue of the meanings of the
words, but in virtue of the way the world is, e.g. snow is white. This is not, in fact, the
way that Kant, who invented the distinction, describes it; but this is in part because of
criticisms of Kants distinction made by people like Ayer, so well discuss this below.
RATIONALISM AND EMPIRICISM ON A PRIORI KNOWLEDGE
Philosophers do not agree on precisely how to define rationalism and empiricism.
Obviously, rationalism gives an important role to reason, and empiricism to experience,
but the terms range across theories of knowledge, theories of concept acquisition,
theories of justification, and historical schools of thought. Thinking of them just in terms
of reason versus experience is much too simple why should we think that we have to
choose? Most rationalists (the exception is Plato) do not deny that experience can
provide us with knowledge under certain circumstances; and empiricists clearly use
reasoning, based on experience, to construct arguments about what we know.
Furthermore, we need to qualify just what is meant by reason and experience.
Nevertheless, there is a way of contrasting rationalism and empiricism that makes them
exclusive (no one is both a rationalist and an empiricist), and which goes to the heart of
the historical debate. Stephen Law puts it like this: Rationalism claims that we can have
synthetic a priori knowledge of how things are outside the mind. Empiricism denies this.
In other words, rationalists argue that it is possible for us to know (some) synthetic
propositions about how the world outside our own minds, e.g. about mathematics,
morality, or even the material world, is without relying on sense experience. Empiricists
argue that it is not. Notice that they dont deny that all a priori knowledge no empiricist
claims that you have to check whether all bachelors are unmarried to see if it is true!
They simply claim that all a priori knowledge is of analytic propositions. If we dont

know if a proposition is true or false just by the meaning of the words, we have to use
sense experience to find out whether it is true or false.
(The clause how things are outside the mind is necessary. Many propositions about my
mental states are synthetic, e.g. I feel sad or I am thinking about unicorns. But they
dont require sense experience to be known; in fact, does knowing my own thoughts
involve experiencing them at all? We dont need to worry about this. Rationalists and
empiricists alike do not deny that we just do know that we have certain impressions and
ideas, thoughts and feelings. The argument is about knowledge of things other than our
own minds.)
EMPIRICISTS ON A PRIORI KNOWLEDGE
Empiricists deny that there is any substantive a priori knowledge of how things stand
outside the mind. So for any area of knowledge, they have three options:
to deny that we have any knowledge in that area;
to say that any knowledge we do have is based on experience; or
to say that any knowledge we have is analytic.
Hume argues that we can have knowledge of just two sorts of thing: the relations
between ideas and matters of fact. His distinction was developed by later philosophers,
and is now understood in terms of the two distinctions, mentioned above:
analytic/synthetic and a priori/a posteriori. Hume argued that all a priori knowledge
must be analytic, while all knowledge of synthetic propositions is a posteriori. In other
words, anything we know that is not true by definition, every matter of fact, we must
learn and test through our senses.
Ayer follows Humes contention, doing so on the basis of meaning. Before we get to the
question of what we can know, we must have a criterion for what is meaningful. There is
no question that we cannot know what is meaningless; knowledge is of true propositions,
and propositions are meaningful statements. Ayer argued for the verification principle
as a criterion of meaning. The principle claims that that all true meaningful statements
are either analytic (true in virtue of the meanings of the terms used) or empirically
verifiable (can be shown by experience to be true or to be probably true). These are the
only two ways of establishing the truth (and hence the falsity) of a statement. Any
statement that cannot be shown to be true or false in these ways is meaningless; so the
only two classes of meaningful statements are analytic and a posteriori. So there can be
no a priori knowledge that is not analytic. But an area of knowledge that has proved
problematic for empiricists is mathematics, which I discuss below.
Morality and religion
It is difficult to argue that knowledge of God and knowledge of morality is gained just
through experience, especially if this is limited to sensory experience. Hume and Ayer
deny that there is any moral knowledge at all, because moral judgements dont express
propositions that can be true or false. Ayer also argues that all talk of God is literally
meaningless.
John Locke, somewhat surprisingly, argues that truths of morality and the existence of
God can be established by reason because they are truths by definition. He uses a form
of the cosmological argument to derive the idea of God and to prove that we can know

God exists. We need only the fact, from experience, that we exist, and some truths that
Locke claims are analytic: We know that we exist and that something cannot come from
nothing. So something must always have existed, and everything else which exists must
have come from this. As we have knowledge and intelligence, we may deduce that this
original being is a knowing intelligence.
He argues that we can know moral truths in a similar way. From our knowledge of the
existence and nature of God, and of ourselves as creations of God, we can deduce what
our moral duties are. It is only because it is not obvious, and we dont reason well, that
people have ever disagreed on such matters.
John Stuart Mill, by contrast, argues for moral knowledge on the basis of observation
and experience. He argues that the only evidence we have for what is good is what we
desire. Everyone desires happiness, and so there is no better final aim for action than
happiness. This argument doesnt establish that we should desire and aim at each others
happiness, since each person desires their own; but Mill assumes that morality is
concerned with all persons equally, and this is an analytic truth. So we know that
everyones happiness is what we should aim at.
KANT AND MATHEMATICS
Kant invented the analytic/synthetic distinction, and first defined analytic propositions as
those in which the predicate doesnt add anything to the subject. Instead, the predicate
breaks the subject into its constitute concepts (hence analysis analytic). Kant talks of
the predicate being contained in the subject, and in developing this idea, argues that the
concept 12 is not contained in the idea of 5 + 7 hence mathematics must be
synthetic rather than analytic.
Most commentators now agree that Kant has misled himself here. He argues that you
can think of 5 + 7 without thinking of 12; but this is to do with how we (psychologically)
understand subject and predicate (what Ayer refers to as their subjective intension, p.
104). But because we are capable of understanding a concept without our minds
immediately being led on to all its implications and components, it is a bad test of the
relation between subject and predicate. His other test can we deny the proposition
without contradicting ourselves? is the better one. Hence Ayer argues that we should
understand analytic propositions as those whose validity depends only on the symbols
they contain; and synthetic ones as those whose validity is determined by facts of
experience.
Locke, Hume, and Ayer argues that mathematical propositions are analytic. To show this,
Ayer says, consider how we deal with potential counterexamples: if I count what I think
are 5 pairs, and only arrive at 9, we dont take this as evidence against the truth of 2 x 5 =
10. We say either I was mistaken at some point or that the number of objects changed.
The same applies to the rules of formal logic. We dont allow principles of logic or
mathematics to be false. We cant abandon them without contradicting ourselves, as they
are true in virtue of the meanings of the terms.
Geometry
Geometry provides a serious challenge to Ayers claim that all mathematics is analytic.
Kant argued that geometry was a set of truths about physical space, yet it is clearly a
priori (the truths of geometry, e.g. properties of triangles, is not arrived at by induction).

The fact that it takes at least three lines to enclose a space in two-dimensions seems to be
a truth about the nature of space, rather than the concept of space. Yet it has mathematical
certainty, and can be proved by mathematical geometry. How could such certainty come
from sensory experience alone? Hence there can be a priori truths about the world. Since
Kants time, as Ayer notes, different systems of geometry, known as non-Euclidean
geometries, with different initial axioms about lines and figures, have been developed by
mathematicians, and in the process, the question arose as to which geometry best
described physical space. This question, Ayer argues, is empirical. Geometries remain
analytic they start with definitions (axioms) and derive theorems by deduction.
Mathematical discoveries
If mathematics is all analytic, how are mathematical discoveries possible? How can we
discover something that is true by definition? Ayer argues that analytic truths (in general,
not just maths) can give us new knowledge in two ways. First, they draw attention to
linguistic usage that we may not have been aware of (e.g. the claim about colour); second,
they can reveal unsuspected implications. Analytic knowledge doesnt need to be
obvious; mathematical truths are very complex, so it takes work to establish that they are
true. But that doesnt mean they are not true by definition. The mathematician Henri
Poincar couldnt believe that all the books of maths are just roundabout ways of saying
A=A. But Ayer comments that this is a result of the limitations of our reason. So they
are useful to us in both these ways. They are also useful in enabling us to make sure
synthetic propositions form a self-consistent system, by rooting out contradictions in our
use of words.
RATIONAL INTUITION?
Not all mathematicians and philosophers agree with Ayers argument. If we return to
Kant, setting aside his particular arguments regarding mathematics: his claim was that
statements that are a priori yet synthetic define the structure of experience, as it is possible for
us. This structure is manifest in our accepting certain judgments as non-logically
necessary unlike analytic judgments, they rest on something outside the circle of
concepts, unlike a posteriori, not on experience itself.
Ayer argues that those statements in metaphysics that are not meaningless, such as
Nothing can be coloured in different ways at the same time in the same part, although it
seems to be a statement about material objects, is not a synthetic truth at all. It is analytic,
recording our determination to count the different colour as belonging to different parts
of the object. The certainty of a priori knowledge derives from the fact that it cant be
refuted (or confirmed) by any fact of experience, not from some guarantee provided by
rational insight into the nature of things.
Kants response would be that our experience is ordered, but not, as Ayer would argue,
by arbitrary linguistic rules, that we could change if we wish. If Kant is right, and the
debate is still live, then there is the possibility of synthetic a priori knowledge.
However, by what means, Ayer asks are we supposed to gain knowledge of metaphysical
truths? Rationalists claim that reason (and what reason is here is unclear) can discover
non-analytic truths independent of experience. From Plato onwards, there is a long
tradition of arguing for rational insight, whether into the Forms, the structure of
possible experience (Kant), or moral values (Moore). (The alternative is that we know
certain truths innately, as part of our rational nature.) Hume argued that many of the so-

called truths previous philosophers claimed were known by rational intuition were
actually just assumptions, propositions unjustifiably taken for granted; Kant sought to
rebut that charge.
So rationalists owe us an account of what intuition is and exactly how it can provide
knowledge. Knowledge needs to be secure, it needs justification, it needs to be reliable.
Our senses are a reliable basis for beliefs because sense experiences are caused by
material objects. But what justifies the claim that rational intuition is reliable?
However, rationalisms best form of defence may be attack. Empiricism has struggled to
show that all our knowledge is either analytic or comes from the senses. If there are good
arguments against empiricism, then rationalism can claim that, if we are not to fall into
scepticism, we must accept that we have rational intuition, even if we dont know what it
is or how it works. One way to display its continued relevance is to ask how we know
what a reason whether for a scientific or a value judgment is.

You might also like