Internalism versus Externalism

A woman has an unusual talent. Shown a photograph of any street in a city she has never visited, she can say correctly whether it is in Lisbon or Porto. She is right every time. She has no idea how she does it, cannot point to anything in the photographs, and if pressed will say that Porto ones just look like Porto. She does not believe she has the talent, because nobody has told her, and she treats each judgement as a guess.

Does she know? The question splits epistemology along its widest fault line, and the two answers are not two theories but two conceptions of what a theory of justification is for.

The division

Stated that way the dispute sounds like a technicality about where to draw a boundary. It is not. Behind the two positions sit two different jobs that the word justification has been asked to do, and much of the literature's stubbornness comes from both jobs being real.

One job is evaluative in a broadly ethical sense. To say a belief is justified is to say the believer is doing well by their own lights, has not been careless, and is not open to criticism for holding it. That job requires the standard to be something the believer could in principle comply with, which pushes towards internalism.

The other job is to explain why justification matters. It matters because justified beliefs tend to be true. A standard that could be met perfectly by a brain in a vat, whose every belief is false, is not obviously tracking anything worth having. That job pushes towards externalism.

Two ways of being internalist

The internalist family divides over what internal means, and the division matters more than it first appears.

Access internalism is the historically dominant version and the one that gives the position its motivation. Justification is thought of as something a believer can be held to, and one cannot be held to a standard whose satisfaction is invisible.

Mentalism, defended by Richard Feldman and Earl Conee, is weaker and correspondingly harder to refute. It keeps justification inside the head without requiring that the believer can find it there. A stored memory trace that shapes what seems plausible to you counts as internal on this view even if no amount of reflection will turn it up.

The weakening is a genuine retreat, and it costs something. Access internalism explains why internalism should be true. Mentalism draws the boundary in a place that is easier to defend but leaves the motivating thought behind, since a mental state you cannot access is no more answerable to you than a fact about your visual cortex.

The case against internalism

The Gettier literature supplies the opening move. If justification is entirely a matter of what is going on inside, then the internal facts about Smith are the same whether or not Jones gets the job, and the same whether Henry is looking at a barn or a facade. Something outside the head is making the difference between knowledge and its absence, and the externalist takes this as evidence that the epistemically relevant properties were never all inside.

The stronger argument is the one about creatures. Small children have knowledge and cannot reflect on their grounds. Animals track their environment reliably and have no grounds to reflect on. Ordinary adults know a great deal that they could not defend if challenged: which way is north in a familiar city, that a friend is upset, that the engine sounds wrong. On access internalism all of this is either not knowledge or rests on reasons the believers do not have.

The woman in the opening case is designed to sharpen this. Her faculty is reliable. Her beliefs are true, and true because of the faculty rather than by accident. What she lacks is any inkling of why she should be believed, including by herself.

The case against externalism

BonJour's clairvoyance cases, published in 1980, remain the sharpest reply, and they work by making the third premise above cut the other way.

Norman is a reliable clairvoyant. He has no evidence that he has the power, no evidence that anyone does, and no evidence for or against the particular beliefs the power produces. One day he finds himself convinced that the President is in New York, and the President is in New York. By reliabilist lights Norman is justified, since the process is reliable. The intuition BonJour is pumping is that Norman is not justified but irresponsible, and that his conviction ought to strike him as something to be suspicious of rather than something to act on.

The generalisation is that externalism severs justification from responsibility. A believer who has no reason to trust a faculty and trusts it anyway is doing something wrong, and it is no defence that the faculty happens to work. The internalist adds that this is not a minor cost, because the whole point of epistemic evaluation is to guide believers, and a standard nobody can apply guides nobody.

The externalist has two replies. The concessive one is that Norman really is justified and our contrary intuition comes from importing background knowledge that clairvoyance does not exist, which gives Norman a defeater he ought to be responding to. Take that away, and Norman looks like a person with a sense he has not yet named. The unconcessive reply is that guidance was never the job. A theory of justification is a theory of what makes beliefs good, not a manual, and theories of biological function are not criticised for failing to help organisms function.

Where this leaves the woman

On mentalism she may be justified, if her sense that Porto looks like Porto counts as a mental state that supports the belief. On access internalism she is not, since reflection turns up nothing. On reliabilism she is justified, and knows. On Sosa's division she has animal knowledge and lacks the reflective kind, which is why she herself treats her judgements as guesses.

The one answer no position gives is that the question is unclear. All four say something definite, and they disagree, which is what a real fault line looks like.

Next

The leading externalist theory in detail, together with the virtue-theoretic variant that tries to keep the externalist's connection to truth while restoring the internalist's interest in the believer.