Michael Lacewing

Direct realism
Direct realism is the natural starting point for theories of perception. It is common sense to say that we
perceive physical objects, and these exist independently of our minds. Physical objects include tables,
books, our own bodies, plants, mountains. (We could also call them material objects. But physics shows
that matter and energy are interchangeable. So physical objects is better, because physics is the science
that studies what such things are, ultimately, made of.) Cosmology and the theory of evolution suggest that
physical objects, such as stars and planets, existed for billions of years before minds existed to experience
them. It is part of our idea of physical objects that they exist objectively in space and time. They continue
to exist when we dont perceive them. When I leave my study, all the physical objects the desk, the
chairs, the books, and so on remain just as they are.
According to direct realism, what we perceive through our senses are just these very things, physical
objects, together with their various properties. When I perceive my desk, for example, I perceive its
size, shape, colour, smell and texture (Ive never experienced its taste, but I could, I suppose!). So,
direct realism claims that what we perceive are mind-independent physical objects and their properties.

The argument from perceptual variation

Russell, The Problems of Philosophy, Ch. 1 A little reflection suggests that what we perceive isnt quite
the same as what is out there. Russell uses the example of looking at a shiny, brown desk. We say it is
brown, but it doesnt actually look an even colour all over: depending how the light falls, some parts are
lighter than others, and some are even white from the shininess. So Russell objects that saying the table is
brown means no more than that it looks brown to a normal spectator from an ordinary point of view under
usual conditions of light but why think that this colour is more real, more a property of the table, than
any of the other colours that you experience? Just what colour any part of the desk looks to you depends
on where you stand. If you and someone else look at the table together, you will see different patterns of
colour. Suppose a shiny spot on the table looks light brown to you but white to the other person. The table
cant be both brown and white in the same spot at one time.
Russell then runs the same argument, appealing to variations in our perceptual experience, for the
properties of texture and shape. The table might be smooth to touch, but at a microscopic level, there are
all kinds of bumps and dips so should we say that when we touch the table, the smoothness we feel is a
property of the table? And the shape that something appears to have, like its colour, varies with the angle
from which you view it. A rectangular table, from every angle except 90 degrees, does not look perfectly
rectangular.

These examples draw our attention to a distinction between appearance and reality. Obviously, much
of the time, we talk as though things are just as they seem. But, clearly, we also distinguish between
appearance and reality and Russell remarks that having any skill as a painter requires that one
does.
All this perceptual variation causes a real problem for the direct realist. The direct realist says I perceive
physical objects and their properties, in this case the desk, directly, as they are. Another way of putting
this is to say that the immediate object of perception is the physical object itself. The argument from
perceptual variation runs like this:

1
2
3
4
5

There are variations in perception.

Our perception varies without corresponding changes in the physical object we perceive. (For instance,
the desk remains rectangular, even as the way it looks to me changes as I look at it from different
angles.)
Therefore, the properties physical objects have and the properties they appear to have are not identical.
Therefore, what we are immediately aware of in perception is not exactly the same as what exists
independently of our minds.
Therefore, we do not perceive physical objects directly.

We now need a name for talking about what we are immediately aware of in perception, e.g. the colour
and shape of the desk as I see it now. Russell calls these sense-data (singular: sense-datum). When I
look at the desk, I have a (visual) sensation I am immediately aware of something. The content of my
sensation what I am immediately aware of is sense-data (on Russells view). We can also think of
sense-data as appearances (how things appear to us to be). Sense-data are distinct from the table. The table
exists independently of my perception of it, while sense-data are defined as what it is that I perceive so
they depend on my perception. If I close my eyes, the colour and shape of the table as seen by me, cease to
exist. And the colour and shape of the table as seen by me varies from where I look at it, while we dont
want to say that the table itself varies in this way. We can summarise the argument so far by saying that
perceptual variation shows that what we directly perceive are not physical objects, but sense-data.
Objections We can challenge Russells claim that there is no good reason to say that one of the colours we
experience the table as having is more real than the others. As he notes, what we mean by the colour of an
object is the colour that it appears to have when seen by normal observers under normal conditions. That
we dont always see this colour that our perception of its colour varies doesnt show that direct realism
is false: we can still say that we see the table, and its colour, under normal conditions. After all, we do all
see it as some shade of brown (shading to white), rather than some of us seeing it as brown, others as red,
others as blue. So, in seeing its colour (as some variant of brown), we see the desk and its properties.

With shape, we have an even better reason to privilege the claim that the desk is rectangular, rather than
obtuse we can use its shape to perform various actions, like getting it through a narrow doorway, which
will only succeed if it is rectangular and not obtuse.
But direct realism does need a more sophisticated account of what it is to see the desk and its properties.
In perception, we can be aware of a range of properties, some of which the object has independent of our
minds, and some of which it has in relation to being perceived. For instance, a rectangular desk has the
property of looking obtuse. The property of looking obtuse is a distinct property from being obtuse
so a desk can be rectangular and look obtuse. The property of looking obtuse is a relational property, in
this case, a property the desk has in relation to being seen. (Another relational property is being to the
north of the desk has this property in relation to me when it is to the north of me.) Looking obtuse is a
property the desk has, claims direct realism, not the property of a sense-datum. And we can even explain
why the desk has the property of looking obtuse (to us) in terms of its being rectangular plus facts about
light and vision.
Direct realism claims that what we perceive are physical objects (not sense-data), but it doesnt have to
claim that all their properties, as we perceive them, are mind-independent. This response challenges the
inference from (4) to (5) above.

The argument from illusion

The appearance/reality distinction challenges direct realism in cases of illusions and hallucinations.
Illusions first: if you half-submerge a straight stick in a glass of water, it looks crooked; but it isnt. We see
a crooked stick, but the stick isnt crooked. However, just from what you experience, you cant tell
whether you are seeing an illusion or not. Someone who doesnt know about the crooked stick illusion
thinks they are seeing a crooked stick. It looks just like a crooked stick in water. Illusions can be
subjectively indistinguishable from veridical perception.

1
2
3
4
5
6
7
8
9

We perceive something having some property, F (e.g. a stick that is crooked).

When we perceive something having some property F, then there is something that has this property.
In an illusion, the physical object does not have the property F (the stick is not crooked).
Therefore, what has the property F is something mental, a sense-datum.
Therefore, in illusions, we see sense-data, and not physical objects, immediately.
Illusions can be subjectively indistinguishable from veridical perception.
Therefore, we see the same thing, namely sense-data, in both illusions and veridical perception.
Therefore, in all cases, we see sense-data, and not physical objects, immediately.
Therefore, direct realism is false.

Direct realism can give the same reply as the above. When the stick in water looks crooked, there is
nothing that is crooked; (2) is wrong. Instead, the stick has the property of looking crooked when halfsubmerged in water. There is a difference between the property being straight and the property looking
straight. Usually, of course, something looks straight when it is straight. But the two properties can come
apart, and something can look crooked when it is straight. So, sometimes we perceive the looks
properties of physical objects, sometimes we experience the properties they have that dont relate to how
they are perceived. In both cases, we directly perceive physical objects and their properties.

The argument from hallucination

We can experience perceptual hallucinations not just visual ones, but auditory and olfactory
hallucinations as well.

In a hallucination, we perceive something having some property F.

2
3
4
5
6
7
8

When we perceive something having some property F, then there is something that has this property.
We dont perceive a physical object at all (unlike the case of illusion).
Therefore, what we perceive must be mental sense-data.
Hallucinations can be experiences that are subjectively indistinguishable from veridical perceptions.
Therefore, we see the same thing, namely sense-data, in both hallucinations and veridical perception.
Therefore, in all cases, we see sense-data, and not physical objects, immediately.
Therefore, direct realism is false.

The disjunctive theory of perception Direct realisms reply to the argument from illusion wont work here.
We cant say that what is seen is how some physical object looks, because no physical object is seen at all!
But there is a different way of challenging premise (2). According to the disjunctive theory of perception,
if something looks a certain way, then one of two quite different things is going on: either I directly
perceive a mind-independent physical object that is F or as in the case of hallucination, it appears to me
just as if there is something that is F, but there is nothing that is F. (An either/or claim is called a
disjunction.) Hallucinations and veridical perception are two completely different kinds of mental state,
because in hallucination, the person isnt connected up to the world. They can seem exactly the same, but
that doesnt prove that they are the same. We can use this to challenge (6). The fact that hallucinations are
subjectively indistinguishable from veridical perception tells us nothing significant about what perception
is. In hallucination, we dont perceive anything, we imagine it. To imagine something is not to perceive
something mental, such as sense-data, but not to perceive anything at all. So the argument from
hallucination doesnt show that in veridical perception, we perceive sense-data instead of physical objects.

The time-lag argument

As Russell notes in The Problems of Philosophy, Ch. 3, it takes time for light waves, or sound waves, or
smells, to get from physical objects to our sense organs.
For example, it takes 8 minutes for light from the sun to reach the earth. If you look at the sun (not a good
idea!), you are actually seeing it as it was 8 minutes ago. If it blew up, you would see it normally for 8
minutes after it had blown up it wouldnt even exist anymore, and youd still see it! Therefore, we could
argue, you arent seeing it directly.
But this doesnt show that what you perceive is actually a sense-datum of the sun. The image you see is
physical, carried in light waves. The light waves exist during those 8 minutes. So if you see the sun
indirectly, then it is because you see light waves directly. But then what we perceive immediately is not the
sun, but the light from the sun. We can generalise: what we perceive is the physical medium by which we
detect physical objects (light waves, sound waves, chemicals for smell and taste). So, we dont perceive
(ordinary) physical objects directly.
Direct realism can reply that this is a confusion between how we perceive and what we perceive.
Compare these two pairs of questions:

1
2

Can you see the lake? and Can you see the light reflecting off the lake?
Can you see the paper? and Can you see the light reflecting from the paper?

In (1), we can turn our attention from the lake to the light reflecting off it. So we can talk, literally, about
seeing the light. But in (2) there is no difference in what one is supposed to see. To see the light that the
paper reflects is just to see the paper. In fact, you cannot see the light itself only the paper. So, direct
realism can argue, except in special conditions, we dont perceive light waves directly and physical objects
indirectly. Light waves are part of the story of how we see physical objects.
The time lag means we see the physical object as it was a moment before, not as it is now. This means that
we see into the past. We always experience the world as it was a moment ago, or in astronomy, when we

look at distant stars and galaxies, we look into the distant past.

Direct realism and common sense

Describe what you see. You would usually do so by referring to physical objects: I see a desk, covered
with pens and paper, and a plant. If you perceive the world via sense-data, the immediate content of
what you perceive is mental. So try to describe your experience in terms of sense-data, without referring
to any physical objects. You could talk about coloured patches standing in spatial relations (above,
below, left, right, etc.) to each other. But this is very awkward, and it is virtually impossible for any
normal scene. What shape is that coloured patch on the left? well, plant-shaped! But plant refers to a
physical object. So our way of describing sense-data is dependent on concepts of physical objects. We
cant give an account of what we experience without referring to physical objects, even if we try.
What this shows is that our perceptual experience presents what we perceive as mind-independent objects.
That doesnt prove that we perceive mind-independent objects, but it does make such a claim highly
intuitive. Only direct realism holds onto this basic intuition. It is very counter-intuitive to think, then, that
what we perceive are sense-data. Any theory that claims we perceive sense-data has to say that perception
is not what it seems to be. It has to say that it seems that we perceive mind-independent objects, but we
dont. We need very strong reasons to accept that perception is misleading in this way.