Philosophy of Perception

What it asks

Look at a tomato. You seem to be directly aware of a red thing out there, not of an image or a representation.

Now consider a case where nothing red is out there - an illusion, a reflection, an afterimage, a dream. Something is going on in each of those cases and it can be indistinguishable from the good case. If it is the same kind of state, then even the good case is not direct contact with the tomato; if it is a different kind of state, the indistinguishability needs explaining.

That is the whole problem, it is three hundred years old, and every position on it costs something.

The recurring problems

The argument from illusion. A white wall in red light looks red. Something looks red and nothing red is present, so you are aware of something other than the wall. Generalise and direct realism is gone.

What experience represents. If perception has content that can be accurate or not, the question is how rich that content is. Do you literally see that something is a tomato, or only see shapes and colours and judge the rest? Everything from expert perception to eyewitness testimony turns on the answer.

Molyneux's problem. Someone blind from birth learns to tell a cube from a sphere by touch, then gains sight. Could they tell them apart by looking, without touching? Posed in 1688, it is the oldest question in the field with an empirical answer, and the answer turned out to be complicated.

Cognitive penetration. If what you believe and expect changes what you SEE rather than what you conclude, then perception cannot be the neutral tribunal that empiricism needs it to be.

How it developed

The early modern picture was indirect. Locke's primary and secondary qualities put colour, sound and taste in the perceiver rather than the object, and Berkeley pushed the argument to its conclusion: if all we ever access is our own ideas, talk of a material world behind them does no work.

The early twentieth century built the sense-datum theory, on which we are directly aware of mind-dependent particulars and only indirectly of objects. It handled illusion elegantly and created worse problems than it solved, chiefly what sense-data are made of and where they are.

Austin's Sense and Sensibilia, delivered as lectures and published in 1962, attacked the argument from illusion at its root by denying that looking red is a state that requires something red.

Since then the two live options are representationalism, on which experience is a way of representing the world and illusions are misrepresentations, and disjunctivism, which denies that the good case and the bad case share a common element at all - a position that sounds evasive and turns out to be very hard to refute.

The positions within it

Naive or direct realism, indirect realism and sense-datum theory, adverbialism, representationalism and disjunctivism. All are in the Schools section.

Idealism is the position that there is nothing behind the perceptions at all, and it is a live option here more than anywhere else in philosophy.

Where it stands now

Disjunctivism and representationalism are the two serious contenders, and the dispute has become closely engaged with vision science.

Predictive processing accounts, on which perception is largely the brain's model of the world corrected by error signals, have made the question urgent again in a new form: if what you experience is substantially a prediction, the line between perceiving and inferring is not where anyone assumed.