Philosophy of perception

Perception is one of our most important mental processes. If we could not perceive anything, then we would know nothing except the contents of our own minds. Because it is our window onto the world, it is important for us to know some basic facts about perception. For example, does our perception let us experience the world as it really is? What are the immediate objects of perception? The philosophy of perception tackles these difficult questions.

Table of contents
1 Categories of Perception
2 Perception as a Cognitive Act
3 What are the immediate objects of perception?
4 See also

Categories of Perception

We can categorise perception as internal or external.
  • Internal perception (proprioception) tells us what's going on in our bodies. We can sense where our limbs are, whether we're sitting or standing; we can also sense whether we are hungry, or tired, and so forth.
  • External or Sensory perception (exteroception), tells us about the world outside our bodies. Using our senses of sight, hearing, touch, smell, and taste, we discover colors, sounds, textures, etc. of the world at large.

The philosophy of perception is mainly concerned with exteroception. When philosophers use the word perception they usually mean exteroception, and the word is used in that sense everywhere below.

Perception as a Cognitive Act

As is often necessary in
philosophy, some simple observations put the whole discussion in context.

Some obvious features of perception spring from the nature of the mind in general. Perception is a cognitive process. Instances of the use of perception are mental events because perception is something that occurs, and it occurs in the mind. I may look outside and see that it is raining. Seeing is a process; and when I looked outside and saw the rain, a perceptual mental event occurred. Or we might say a visual mental event occurred, since the event was one of my sense of sight.

Philosophers often refer to perceptual mental events as acts of perception. This implies that perception is not merely something that occurs, but that it is something that we do, sometimes proactively. We open our eyes and actively examine things, listen for particular sounds, and so forth. So we can proactively choose to perceive things. But a good deal of the time we do not choose to direct our senses in one direction or another. For example, I am sure I just happened to notice that it was raining a few days ago; I didn't, as it were, actively choose to look out the window. Our senses may be distracted here and there, so then perception is passive and not active. Nonetheless philosophers will often refer to perceptual events as acts, although that need not imply that we are always proactive when we perceive; obviously sometimes we are passive.

According to philosophers who use the word act to refer to perceptual events, acts always have objects. When we act, or do something, there is always something else on which we act, or something to which we do it. Talk of acts requires talk of objects. So if a perceptual event is an act, it has an object, something that the act is directed upon. And of course that's certainly the case with at least some kinds of perception. For example when I see the rain, the object of my visual act is the rain. And of course more generally, whenever we see, or hear, or touch, it at the very least appears that there is something that we see, or hear, or touch. The things that we perceive, that perception is of, are the objects of perception.

What are the immediate objects of perception?

The rest of this article addresses this question. To understand the question, we must agree what immediate means.

Suppose you see the President on television. You aren't seeing the President immediately, or directly. You're looking at a television screen. On that screen is a moving, colored shape that greatly resembles the President, but that shape is not the President himself. It is only a picture that represents the President. Does that mean that you aren't seeing the President at all? Not in any sense? Of course not. You are seeing the President indirectly, by means of seeing a representation of him.

When you see the President on television, your visual act involves two real-world objects; the more direct or immediate object is a moving picture on the television screen; the indirect or mediate object is the President himself.

The immediate object of perception is whatever we see first. By seeing it, we may be able to see other things with the same act of perception.


The concept of sense data (singular: sense datum) is very influential and widely used in the theory of perception.

Many philosophers have said that the most immediate objects of perception are mental objects -- objects in the mind. So within the mind there are two different items; there is a perceptual act, and there is a mental object, which may represent things outside the mind.

For example, according to this widely held view, when I see the President on TV the very first thing I perceive is an image of the President in my mind. This image represents the moving picture on the television screen, and that moving picture on the television screen in turn represents the President himself.

Immediate mental objects of perception such as the image of the President in my mind are known as sense-data, ideas or percepts.

Perceptual Irregularities and Illusions

Why think that Sense-data, internal objects of perception, even exist? Why not think that we just immediately, or directly, perceive external objects? Why think that we perceive internal objects at all? Several arguments, for this view go under one general title, the argument from illusion. This argument concludes that the immediate objects of perception are sense-data. What reasons can be given to support this conclusion?

Consider several examples of what we might call perceptual illusions.

  • Take a perfectly straight stick, and partially submerge it in a pond: then the stick may appear to bend where it passes into the water. The bend in the stick is illusory.

Suppose you want to say that the immediate or direct object of your perception is the stick itself. Then it appears that you're saying that the stick itself was first straight, and then bent. You say it's the stick that is the direct object of your perception. Something did look straight, and then bent. So if the direct object of your perception was the stick, why then it must have been the stick that was first straight, and then bent. But that's ridiculous if we know nothing happened to change the stick.

We could say, instead, that when you were looking at the stick, you had a series of stick sense-data; and the sense-data were at first straight, and then bent. The stick itself, all the while, was perfectly straight.

  • Driving across the countryside, you see some hills in the distance, which appear bluish. Then as you approach nearer to the hills you see that they are actually green. The bluish color the hills originally appeared to have wasn't real; it was illusory.

You want to say that, when you were looking at the hills, something looked bluish, and then it looked green; but it wasn't the hills that actually were bluish and then green; so we could say that you immediately perceived hill sense data, and the sense data were bluish, and then green. The hills, all the while, were the same color.

These examples relate only to visual perception, to the sense of sight, but there are similar examples of such "illusions" for the other senses.

In all of these cases we can see that it is very convenient, on first glance anyway, to suppose that we immediately perceive sense-data. Therefore, the immediate objects of perception are sense-data. This includes all sense-perception -- sense-data occur not just when illusions occur, but whenever we perceive the external world.


Various theories about perception use the concept of sense-data. One such theory is called the representational theory of perception, or simply Representationalism. This is the doctrine that in any act of perception, the immediate (direct) object of perception is a sense-datum, a mental object that represents an external object.

This immediately leads to a question. If sense-data represent external objects, properties, and events, how well do they represent them? At least sometimes, they do not represent them at all well. When, for example, the stick sense-datum is bent, it does not perfectly represent the stick, which is straight. And similarly with the other illusions - to the point of somebody hallucinating pink elephants, whose sense-data do not represent anything at all.

Primary and Secondary Qualities

Any claim that sense-data are always right, that they always give us accurate information about what the world is like, would clearly be over-simplistic. Several good examples above show it is wrong. If there are sense-data at all, they definitely do not always accurately represent the world as it is. That's at least the case with perceptual illusions.

Both Descartes and Locke recognized this. Beside various perceptual illusions, they said that another category of sense-data also did not adequately represent the external world. They distinguished two kinds of properties they called primary qualities and secondary qualities. They said that sense-data of primary qualities can accurately represent external qualities in things; but sense-data of secondary qualities cannot accurately represent external qualities in things.

The distinction Descartes and Locke drew is a fairly common sense sort of distinction.

  1. Some qualities things appear to have, the primary qualities, really are in the things.
  2. Other qualities things appear to have, the secondary qualities, are only in our minds; they are just aspects of how we perceive the world.

Some examples:

  • There is a real difference between the actual temperature of a kettle of water, and how warm the water feels when I put my hand in it. The water might be only lukewarm; but if my hand is extremely cold, when I put it in the water, the water is going to feel hot. So for the sake of accuracy we should distinguish between the actual temperature of the water, which is a matter of how fast the water molecules are vibrating, and the perceived warmth of the water, which is not a quality of the water, but instead of how we perceive the water. So then we could say: actual temperature is a primary quality of the water, but perceived warmth is a secondary quality of the water.

  • We can say that a physical body's mass, or its shape, is a primary quality; but the perceived weight, or how heavy a body feels, is a secondary quality. The perceived weight of a thing depends on who is lifting it, how tired they are, and so on. But the mass of a body does not depend on who is lifting it.

  • Tastes and smells are often given as good examples of secondary qualities. When you bite into a lemon, and say it is sour, do you mean that the lemon itself is sour, or only that in your perception of the lemon, there is this element of sourness? You can say that the lemon is acidic, and that is a primary quality. The lemon's acidity does not depend on who or what is tasting it. But whether the lemon is sour or how sour it tastes, can indeed depend on who is tasting it. So acidity (pH) is a primary quality, but sourness is a secondary quality.

We can have ideas or sense-data of both primary and secondary qualities. But, say Descartes and Locke, and many who came after them, sense-data of primary qualities can actually represent primary qualities; but sense-data of secondary qualities do not represent secondary qualities, because really there are no secondary qualities in the object at all. We might just say that "secondary quality" refers to exactly the same thing as "sense-datum of a secondary quality."

Examples again will help clarify this.

  • We can have an idea of the actual pH of a lemon, its degree of acidity; and that idea can perfectly well represent the lemon's acidity. But when we get a sense-datum of sourness, it is not that there is something properly called "sourness" in the lemon. Acidity is all that is in the lemon; the sourness is just part of how we perceive the lemon, on account of its acidity.

  • In the same way, we can have an accurate idea of the mass of a box; but when we lift the box and call it "heavy," that heaviness is just in the sense-datum we have of the box. There is nothing properly called "heaviness" in the box itself. Heaviness is just how the box feels to us. Mass is what the box itself has; heaviness is part of how we experience the box.

Critical Realism

According to Locke and Descartes, some sense-data, namely the sense-data of secondary qualities, do not represent anything in the external world, even if they are caused by external qualities (primary qualities). This makes it natural to adopt a theory of
critical realism, which may be defined as follows:

Critical realism is the theory that some of our sense-data (for example, those of primary qualities) can and do accurately represent external objects, properties, and events, while other of our sense-data (for example, those of secondary qualities and perceptual illusions) do not accurately represent any external objects, properties, and events.

This theory has been held by some very reasonable philosophers. But it does, by its talk of sense-data and representation, depend on or presuppose the truth of representationalism. If critical realism is correct, then representationalism would have to be a correct theory of perception.

So, how well do sense-data represent external objects, properties, and events? Critical realism says: well enough. Some of our sense-data don't represent any external qualities, but through our sense-data of primary qualities, we can still know what the external world is like.


Suppose a rather extreme skeptic asks: "But how do you know that your sense-data represent primary qualities? Prove it." If we start to answer, the skeptic may interrupt. So let's listen now to the skeptic's argument, which is due to that influential Scotsman, David Hume:
"There is no proof that your sense-data represent primary qualities in the external world. Such a proof would be impossible. You have already concluded that whenever we perceive, we immediately perceive sense-data. That means that the only way we can perceive any external world is through sense-data. Perception is our external-world-information-gathering-faculty, and we have no other. So perception is the only way we can learn about the external world.

"Now you think that you can prove that your sense-data represent the external world. But how could you prove that? You would need to have some evidence that your sense-data represent the world. But have you ever perceived the connection between your sense-data and the world? Obviously not! You say that the only way we can perceive any external world is through sense-data. So we obviously could not use sense-data to prove that sense-data represent the world well, poorly, or at all.

"And it's even worse than that. Now, you think that you must always immediately perceive sense-data whenever you perceive anything. But that means that you cannot perceive the external world directly, nor can you perceive the connection between sense-data and the world. So how on earth could you even prove that the external world exists? How do you know that there is a world there on the other side of the sense-data? If perception is like a window onto the world, then how do you know that there is a world on the other side of the window? You can't prove it; all you have are sense-data. So we might as well give up all claims to having knowledge of any external world at all!"

This sort of skeptical challenge is very hard to meet. This article does not meet it or discuss it, because we are talking about theories of perception; and skepticism is not a theory of perception.


A common reaction among philosophers to the foregoing skepticism has been simply to deny that an external world exists.

Philosophers who hear the skeptic's challenge -- "There's no reason to think an external world exists" -- reply, "Well, no, I guess there isn't any reason to think that an external world exists. All there is, is sense-data. Physical objects are bundles of sense-data. When I hold up my hand, and I see it, I'm not seeing something external to my mind; I'm seeing a series, a whole bundle, of hand sense-data, and there is no hand apart from those hand sense-data. That's what my hand is -- a bundle of sense-data." Such philosophers get around skepticism, not by replying to the skeptic and proving the existence of an external world, but instead by saying that there is no external world.

This approach is called phenomenalism, the view that physical events are nothing more than a special kind of mental event. But it doesn't have to be stated in terms of just events. More generally, it can be defined as follows:

Phenomenalism is the view that physical objects, properties, events, etc. (whatever is physical) are reducible to mental objects, properties, events, etc.; hence, ultimately, only mental objects, properties, events, etc., exist.

In particular, we may reduce talk of physical bodies to talk of bundles of sense-data. This logically goes with the bundle theory of objects. And the philosopher who is most famous for advocating both the bundle theory of objects, and phenomenalism, is the 18th century Irish philosopher, George Berkeley. Berkeley's version is more commonly called "subjective idealism".

This article does not discuss Berkeley or phenomenalism at any length. It may however be interesting to note just how far from common sense the doctrine of representationalism has led.

For an argument against phenomenalism, please see the phenomenalism article.

Direct Realism

Scottish philosopher Thomas Reid lived at the same time as Hume. Reid argued strenuously against the notion that ideas, or sense-data, are the immediate objects of perception at all -- he rejected representationalism.

One of Reid's arguments was very simple, and went like this: If representationalism is correct, then we are forced to either skepticism or phenomenalism. But skepticism and phenomenalism are both absurd; there surely is an external world, and we surely do have knowledge of it. So, by reductio ad absurdum, we must reject any theory that would force us to accept either skepticism or phenomenalism. So, we must reject representationalism.

What would it mean to reject representationalism? It would mean accepting that we do not perceive sense-data at all. When I look at my hand, I do not immediately perceive a bundle or series of hand sense-data, which represent my actual hand. No, I immediately perceive my hand. I do not perceive any hand sense-data at all. So the view up for consideration now is that we immediately, directly perceive the external world.

This view is called direct realism, which Reid championed brilliantly:

Direct realism is the view that the immediate (direct) objects of perception are external objects, qualities, and events.

Do not confuse direct realism with the more naive view discussed earlier, that the world is exactly as we perceive it to be. Obviously, sometimes we misperceive the world. The direct realist does not deny that there are perceptual illusions. The claim is, rather, simply that when we do perceive something, what we directly perceive, the immediate object of perception, is in the external world, not in the mind.

Nonetheless, the argument from illusion can be taken as an argument against direct realism because the argument from illusion shows the need to posit sense-data as the immediate objects of perception. How might direct realism answer the argument from illusion?

One strategy is to show how all those different cases of misperception, failed perception, and perceptual relativity -- all those hard cases -- do not really make it necessary to suppose that there are sense-data. Those cases might be explained without having to talk about sense-data.

Take first the case of the stick that looks bent in the water. Direct realism doesn't say that the stick actually is bent; it says, rather, that the stick, which is straight, can, in some unusual circumstances, look bent. And to say that it looks bent is just to say that the light, which is reflected from the stick, arrives at our eyes in a crooked pattern. So the stick can have more than one appearance. But the appearance of a stick isn't a sense-datum in my mind. It's a pattern of light, the sort of things that physicists can study, that arrives at my eye. What's mysterious about that? A similar sort of thing can be said about the bluish color of the hills in the distance. Hills, and everything else, can appear with all sorts of different colors; but the color is simply the wavelength of light as it reaches my eye. If the light from the green hills has to traverse many miles, then it may be bluish when it arrives at my eyes. There's no need to suppose I am seeing bluish sense-data: nope, what I'm seeing is bluish light, which comes from the hills. The hills would reflect green light to my eyes if I were closer to them.

Now the case of pressing on my eyeball, and getting a double image. Well, it's undeniable that, when I cross my eyes and seem to see two fingers, there are two of something. But of what? Why say there are two sense-data? Why not, instead, say that I have two eyes, and each eye gives me a different view upon the world. Usually the eyes are focused in the same direction; but sometimes they're not. And as a result, each eye sees things in a different way. That doesn't mean that I see two visual sense-data in my mind; but it does mean that there are two slightly different acts of vision going on. One for each eye! What's mysterious about that? Nothing, as far as I can tell. And similar things can be said about the coin that appears both circular and oval-shaped: so the same coin can reflect different patterns of light to my eye. Does that mean that I perceive two different sense-data? No, all it means is that I perceive the same coin in two different ways.

Now as for Mary's vivid hallucination of the pink elephants. It was so vivid that the elephants were just as real as real elephants. We said this was evidence for thinking that she is perceiving sense-data; she sure as heck isn't perceiving elephants, and yet she seems to be hallucinating something. So maybe it's elephant sense-data that she is hallucinating. Well, that seems like a pretty tough case to deal with. It definitely does seem that there is an object, in some sense, of Mary's hallucination; but this object is only in her mind. Isn't that what we'd call sense-data?

The direct realist might reply to that case as follows: Mary was not perceiving anything at all; she was hallucinating. That's a different, though related, mental process. So maybe Mary has visual images of some sort when she is hallucinating; that wouldn't mean that she has such images when she engages in actual sense-perception.

This may not be a particularly strong reply. If there are visual images when we hallucinate, it seems reasonable to think that there are visual images when we see. It's the same way with dreams: if there are visual and auditory images of some sort in our minds, when we dream, it seems reasonable to think that there are visual and auditory images, or sense-data, when we are awake and perceiving things. We might ask for a better reply.

Some people end up denying that there are any such things as mental images at all, but this is rather hard to maintain, since we seem to be able to imagine all sorts of things: for example, here's something that will give you an image: imagine a square, then imagine the top of the square popping off and disappearing, and the two sides of the square collapsing together at a point, to make a triangle. Even if it should happen that perception does not involve images, other mental processes, like imagination, certainly seem to.

Closing note on mental images

The topic of mental images is very complicated and controversial.

One considered view is similar to Reid's. It is that, in some sense, we do indeed have images of various sorts in our minds when we perceive, and dream, and hallucinate, and use our imaginations, but when we actually perceive things, our sensory images, or sensations, if you will (that's Reid's word), cannot be considered the objects of perception, or attention, in any sense whatsoever. The only objects of perception are external objects. Even if perception is accompanied by images, or sensations, it's wrong to say we perceive sensations.

See also

A cognitive process is a mental route of absorbing knowledge and subsequent recollection.

Two cognitive processes are organization and integration of thought. Organization of thought sorts and makes comprehensible. Integration of thought builds to application or purpose.

copyright © 2004