Justification is what makes a belief reasonable to hold. But what gives a belief justification? Usually another belief: I believe it will rain because I believe the forecast said so, and I believe the forecast is reliable. But what justifies those beliefs? This question leads to a famous puzzle.
The regress problem
If every belief must be justified by another belief, there seem to be only three possibilities. The argument goes back to the ancient skeptics, is preserved in the writings of Sextus Empiricus, and is often called Agrippa's trilemma:
- The chain of reasons goes on forever (an infinite regress).
- The chain circles back on itself (circular reasoning).
- The chain stops at some beliefs not justified by others (which looks arbitrary).
The skeptic says all three options fail, so no belief is justified. The main theories of justification are attempts to answer this challenge.
Foundationalism
Foundationalism accepts the third option but denies that it is arbitrary. Some beliefs are basic: justified without needing support from other beliefs. All other beliefs are justified by resting on these foundations, like a building on its base.
What could the foundations be? René Descartes wanted foundations that were absolutely certain, such as his knowledge that he exists while he is thinking. Empiricists looked to immediate experience: “I seem to see something red” is hard to doubt even if I am wrong that something red is there. Modern moderate foundationalists accept that basic beliefs, such as ordinary perceptual beliefs, can be justified without being certain.
Criticisms: few beliefs are certain enough for Descartes's purposes, and if foundations are not certain, it is unclear what justifies them without appeal to other beliefs. Critics also doubt that so much of what we know can be built up from such a narrow base.
Coherentism
Coherentism rejects the building image. Beliefs are justified by fitting together in a system where they support each other, more like a web than a pyramid. The philosopher Otto Neurath compared us to sailors who must rebuild their ship on the open sea, replacing planks while staying afloat, never able to start again from the bottom. A belief is justified if it coheres well with the rest of what we believe.
Criticisms: a coherent system could still be cut off from reality entirely, the same objection we met for the coherence theory of truth. Coherentism also seems to give experience no special role, though perception is our main contact with the world.
Reliabilism
Alvin Goldman proposed in 1979 a different approach, reliabilism: a belief is justified if it was produced by a reliable process, one that tends to produce true beliefs. Good eyesight in good light, careful memory and sound reasoning are reliable; wishful thinking and guessing are not.
Reliabilism is an externalist view: what justifies a belief may lie partly outside the believer's awareness. Foundationalism and coherentism are usually internalist: justification depends on reasons the believer can access by reflection.
Criticisms: imagine a person with a perfectly reliable ability to sense distant events but no reason to trust it. Is their belief justified? Internalists say no. Imagine also a person deceived by a demon, whose beliefs are carefully reasoned but produced by an unreliable process. Many say they are just as justified as we are.
Mixing views
Many epistemologists now combine elements, for example treating perception as a source of basic justification while also valuing coherence among beliefs.
Worked example: the eyewitness
A woman sees a man run from a store that has just been robbed. Later she picks him out of a lineup, and she is confident. Is her belief that he was the robber justified?
A foundationalist traces her belief to a basic perceptual experience: she saw a man's face. Ordinary perception provides justification unless something defeats it, and her belief rests on that foundation plus her memory.
A coherentist asks how well the belief fits with everything else known. Did his clothing match? Was he near the store? Does he have an alibi? A belief that fits many independent pieces of evidence is better justified than one that stands alone.
A reliabilist asks whether the process that produced her belief is reliable, and here psychological research matters a great deal. Studies have found that eyewitness memory is less reliable under stress, at a distance, in poor light, when a weapon draws attention, and when witness and suspect are of different races. Confidence often grows after the event, especially when a witness receives encouraging feedback. Many wrongful convictions later overturned by DNA evidence involved mistaken eyewitnesses.
The reliabilist can say something the others find harder to say: the witness may be sincere, careful and reasonable from the inside, yet her belief may still be poorly justified, because the process that produced it is less reliable than it feels. Courts in some places now instruct juries about these factors.