From the Routledge Encyclopedia of Philosophy

Innate Knowledge

Elliott Sober

Philosophical Concept

If innate knowledge exists, there must be innate beliefs and those beliefs must count as
knowledge. In consequence, the problem of clarifying the concept of innate knowledge divides
in two: to explain what it is for a belief to be innate and then to connect that account with a
characterization of what knowledge is. Modern biology requires changes in traditional
philosophical conceptions of innateness; and two quite different theories of knowledge entail that
innate beliefs will often fail to count as knowledge.

1. Innateness
Within some species of bird, individuals produce their characteristic song even if they are reared
in silence. Among others, a bird sings its song only if it first hears that song performed. And
among still others, a bird produces its characteristic song only if it hears some sort of song or
other (Gould and Marler 1991). We are inclined to apply the concept of innateness to the first
type of bird and to withhold it from the second. But what of the third? Should we say that these
birds learn their songs by hearing any of a set of quite dissimilar songs? Or should we say that
the song is innate – that it is ‘in’ the bird, only awaiting an environmental trigger for its release?
Both options seem unsatisfactory. How can a given song be learned from hearing a quite
different song, if the target song would have emerged regardless of which song the bird had
heard? The option of judging the song innate also has its pitfalls. What does it mean to say that
the song is ‘in’ the bird from the start? This spatial metaphor sounds suspiciously like the
preformationist doctrines of eighteenth century embryology, according to which foetal
development simply involves an increase in size of the fully formed individuals found in newly
fertilized eggs. Arguably, the song is ‘in’ the bird from birth no more than adolescent pubic hair
is ‘in’ a new-born baby.

It is important not to confuse the claim that the song is ‘in’ the bird from birth with the much less
controversial claim that the new-born bird is disposed to acquire the song once the bird passes
various developmental landmarks and experiences various environmental cues. The latter claim
does not entail the former. Indeed, the latter formulation offers a completely de-natured reading
of the idea of innateness; it becomes trivially true that all beliefs are innate. Surely this
formulation short-circuits the controversy rather than solving it.
(In Notes on a Certain Program, Descartes suggests that ideas are innate in precisely the same
sense that diseases are sometimes innate; this does not mean that foetuses suffer from diseases in
utero, but that individuals are born with a certain ‘disposition’ to contract them. The challenge to
innatism, laid down by Locke in the Essay Concerning Human Understanding, is to explain what
‘disposition’ means so that some traits, but not others, turn out to be innate. Leibniz takes up this
challenge in the New Essays.)

Contemporary biology does not exploit the philosopher’s traditional distinction between innate
and learned. First of all, the biological opposite of innate is not ‘learned’, but ‘acquired’.
Learning is one way to acquire a characteristic, but there are others; sunburns are acquired, but
they are not learned. It is worth asking what distinguishes learning (and mislearning) from other
processes of belief acquisition; if beliefs could be acquired by swallowing pills, arguably that
would not count as learning. In any event, the present point is that biologists do not think of
‘innate’ and ‘learned’ as exhaustive alternatives.

Perhaps the most important contrast between the biological and the traditional philosophical
concept is that biology does not treat the difference between innate and acquired as a dichotomy;
rather, it is a matter of degree. Traits differ with respect to how dependent their emergence is on
environmental details. The three groups of birds described before can be arrayed on a continuum;
there is no need to draw a line that separates two of them from the third.

It is also important to distinguish the initial appearance of a trait from its subsequent
maintenance. A trait may emerge across a wide range of environments and yet be modifiable
once in place. For example, an Egyptian vulture, when first confronted with an ostrich egg and a
stone, will break the egg with the stone. However, if the vulture repeatedly discovers that the
eggs it breaks open are empty, eventually it will stop breaking eggs (Pulliam and Dunford 1980).
In this case, the behavioural disposition might be termed innate. All the same, ‘innate’ does not
mean ‘unmodifiable’.

Yet another gap between the traditional philosophical concept and modern biology concerns the
issue of universality. Philosophers often expect innate beliefs to be universal in our species; they
are part of ‘human nature’ and so should be present in all individuals who have enjoyed a normal
development. (Thus we find Locke, in the Essay (bk I, ch. III, §§8–18), arguing that the idea of
God is not innate because it is not found in children and ‘savages’.) A related expectation is that
if two people have a trait, then the trait must be innate for both if it is innate for either.
Neither of these assumptions is unproblematic. The biologist’s concept of a ‘norm of reaction’
describes how the phenotype which a genotype will develop depends on the environment it
experiences. For example, consider a particular fruitfly genotype; let us ask how that genotype’s
attainment of a particular phenotype (for example, bristle number) depends on some
environmental variable (for example, nutrition). Different genotypes often exhibit different
degrees of environmental contingency with respect to the same phenotype. One fruitfly genotype
may produce the same bristle number regardless of variation in nutrition, whereas another may
produce different numbers of bristles in response to different amounts of food. Because of this, it
is quite possible for two flies to have the same number of bristles, but to have achieved that
result by quite different developmental pathways. Bristle number may be ‘innate’ for one and
‘acquired’ for the other.

In view of all this, it is a reasonable hypothesis that the most that can be salvaged from the
ancient concept of innateness is this: a phenotypic trait is innate for a given genotype if and only
if that phenotype will emerge in all of a range of developmental environments. What counts as
the appropriate range of environments is left open in this proposal. Perhaps there is a uniquely
correct answer to this question; then again, maybe the range is determined pragmatically. It is
difficult to see how the latter conclusion can be evaded.

2. Innateness and the a priori

The above definition opens a large gap between the concepts of innateness and a prioricity. A
proposition is a priori if it can be known or justified independently of experience (see A priori).
A prioricity does not concern what causes an organism to formulate a belief; rather, it has to do
with the status the belief enjoys once it is formulated. Arguably, the belief that all bachelors are
unmarried is not innate; it emerges in some social and linguistic environments, but not in others.
Still, once an individual has this belief, it may be true that the belief can be justified without
appeal to experience.

Conversely, it is also possible for a belief to be both a posteriori and innate (see A posteriori).
Perhaps human cognitive development inevitably leads people to believe that a physical world
exists external to their own minds. This is as natural and invariant a developmental outcome as
the acquisition of pubic hair during adolescence. Yet, it remains for the philosopher to judge
whether solipsism is a priori false. Perhaps our natural anti-solipsism is either unjustifiable or is
justifiable as an inference to the best explanation; maybe the postulate of an external world
makes sense because it explains regularities that obtain in the flow of experience. If so, we would
have here the case of an innate belief that is a posteriori (see Inference to the best explanation).

3. Knowledge
If we have beliefs whose emergence is relatively invariant across a range of developmental
backgrounds, will those beliefs count as knowledge? To begin with, if S knows that p,
then p must be true. Beyond that, what does knowledge require? Let us explore the suggestion
that S’s belief that p counts as knowledge only if S is justified in believing p.
The justification that S has for believing p presumably must come from other beliefs (or thoughts
or sensations) that S has. In particular, it is not enough that those other beliefs (or other items)
bear some abstract relation of justification to the proposition p; after all, it might be true
that S believes q and that q is evidence for p even though Sbelieves p for no reason at all or for
bad reasons. What is required is that S’s belief in p be related ‘in the right way’ to S’s justifying
belief in q.

What might this right way be? One possibility is that S’s belief in qcauses S to start believing p.
Another is that S continues to believe in pbecause S believes q. A third is that S’s belief
in q makes it harder for Sto stop believing p. The list could continue (see Justification,
epistemic;Knowledge, concept of §§1–6).

Regardless of how this condition is interpreted, it seems clear that innate true beliefs can fail to
count as knowledge. Consider a belief that people inevitably have and cannot abandon; perhaps
our spontaneous anti-solipsism is an example. Such propositions may admit of inventive
justifications, but this does not mean that philosophical pronouncements can influence whether
ordinary folks believe in an external world.

Let us turn, instead, to a concept of knowledge that abandons the requirement of justification.
Reliability theories of knowledge demand that the act of believing should be related to the world
in the right way. If S knows that p, then it must be true that, in the circumstances
that Soccupied, S would not have believed p unless p had been true. If Sknows that p, then the
requirement is that the probability that p is true given that S believes that p, is equal to
1.0. S must be a perfectly reliable detector of the truth of p, if S’s belief that p is to count as
knowledge (Dretske 1981) (see Reliabilism).

It is easy to see why an innate belief can evolve so as to violate the requirement that the
probability that p is true given that S believes thatp, is equal to 1.0. For example, suppose that
snakes are usually, although not always, dangerous. An organism that confronts a snake thus
faces a problem of decision-making under uncertainty. It can believe without further ado that the
snake is dangerous; or, it can try to find some further observational evidence on which to base its
decision. In the former case, the organism might have the innate belief that snakes are dangerous.
In the latter, its opinion will depend on the details of present and past experience.

Imagine that the organism in question is not very good at determining from observable cues
whether a snake is dangerous. Imagine further that the organism does better by erring on the side
of caution (Stich1990). If the snake is dangerous, it is very important to believe that this is so;
but if the snake is not dangerous, it is of much smaller moment whether one realizes this fact.
These assumptions all favour the evolution of the innate belief that snakes are dangerous.
Innateness will be a better strategy than learning in this instance (Sober 1994).

Notice that the factors that predict the evolution of innateness provide no guarantee of the
reliability of the resulting beliefs. The organism believes of each snake encountered that it is
dangerous. The organism therefore has true beliefs only as often as snakes in fact are dangerous.
This need not be a terribly high frequency; it clearly need not equal 100 per cent. For this reason,
it is easy to see how innate beliefs can fail to count as knowledge from the point of view of the
reliability theory of knowledge. Indeed, there is no guarantee that the belief must be true more
often than not; pervasive error may be a consequence of the evolution of innate beliefs.

References and further reading

Descartes, R. (1648) Notes on a Certain Programme, in The Philosophical Works of Descartes,

trans. E.S. Haldane and G.R.T. Ross, Cambridge, Cambridge University Press, 1931, 431–450.
(Descartes responds to various misunderstandings of his views, including his position on innate
ideas.)
Dretske, F. (1981) Knowledge and the Flow of Information, Cambridge, MA: MIT Press.
(Elaboration of the reliability theory of knowledge.)
Fodor, J. (1981) ‘The Current Status of the Innateness Controversy’, in Representations,
Cambridge, MA: MIT Press, 257–316.
(Argues that the dispute between empiricists and rationalists should be understood in terms of
whether various simple beliefs are learned or triggered.)
Gould, J. and Marler, P. (1991) ‘Learning by Instinct’, in D. Mock (ed.) Behavior and Evolution
of Birds, San Francisco, CA: Freeman, 4–19.
(Describes a variety of learning strategies that birds use to acquire their songs.)
Leibniz, G.W. (1703–4) New Essays Concerning Human Understanding, trans. with notes by
A.G. Langley, La Salle, IL: Open Court, 3rd edn.
(Point-by-point commentary on Locke’s essay by the greatest exponent of continental
rationalism.)
Locke, J. (1689) An Essay Concerning Human Understanding, ed. P.H. Nidditch, Oxford:
Clarendon Press; New York: Oxford University Press.
(The founding document of British empiricism, accessible to the general reader.)
Pulliam, H. and Dunford, C. (1980) Programmed to Learn, New York: Columbia University
Press.
(Argues that different learning strategies will evolve depending on how the environment is
structured.)
Sober, E. (1993) Philosophy of Biology, Boulder, CO: Westview Press.
(Clarifies concepts such as norm of reaction and heritability.)
Sober, E. (1994) ‘The Adaptive Advantage of Learning and A Priori Prejudice’, in From a
Biological Point of View: Essays in Evolutionary Philosophy, Cambridge: Cambridge University
Press.
(Presents an evolutionary model for predicting when a proposition will be an innate belief and
when it will depend on experience for its emergence.)
Stich, S. (1975) Innate Ideas, Berkeley, CA: University of California Press.
(Anthology of historical and contemporary essays, some of which address issues concerning
Chomskian linguistics. Of special interest are the essays by A. Goldman relating the causal
theory of knowledge to the question of whether innate beliefs can count as knowledge; by W.
Hart, arguing that innateness is irrelevant to the issue of a priori knowledge; and by S. Stich,
which proposes a clarification of what innateness means.)
Stich, S. (1990) The Fragmentation of Reason, Cambridge, MA: MIT Press.
(Argues that evolutionary considerations make it plausible to suspect large amounts of
irrationality and unreliability in belief formation.)