thoughts. But the second – semantic – part of the claim is deWnitely
incorrect, as we will now brieXy argue.
Images, of themselves, unsupplemented by much prior knowledge (and
hence thought) on the part of the thinker, are conWned to representations
of appearance. To have a visual mental image is to represent to oneself
how something would look; to have an auditory mental image is to
represent to oneself how something would sound; and so on. This gives rise
to an immediate problem for the imagist theory of thought, since many of
our words and concepts do not stand for the kinds of thing whichhavean
appearance. For example, and more or less at random, consider logical
concepts likeand,or,not; temporal concepts liketomorrow,yesterday,
year; concepts for abstract properties likeinXation(of money),prime(of
numbers); and number-terms likesixteen,orsixty-four. In none of these
cases is there any mental image which seems even remotely appropriate to
express what we mean.
Moreover, we have many concepts which represent things which dohave
an appearance, but which do not represent themin virtue oftheir ap-
pearance. Consider, for example, the conceptbus-stop. Bus-stops do, of
course, have a characteristic appearance (though diVerent in diVerent
parts of the country, let alone in diVerent parts of the globe). But if I steal a
bus-stop sign and erect it in my garden as an ornament, that does not turn
my garden into a bus-stop. Rather (very roughly), a bus stop isa place
where buses are supposed to stop. How is this to be expressed in an image?
Even if my image is of a bus stopping at a bus-stop sign, with people
getting on and oV, this does not attain to the generality of the idea of a
place where buses (in general) stop; nor does it touch the normativity
implicit in the ideasupposed to stop.
In the light of points such as these, it is plain that no image, or sequence
of images, can, of itself, carry the content of even a simple thought such as,
‘All grass is green’, let alone of a complex proposition such as, ‘Life may be
discovered on Mars in the next ten or twelve years.’ Yet it may be replied
that there can be nothing to stop people using images assignsto express
their thoughts, somewhat as words are used. However, it will not, then, be
the representational content of the image, as such, which determines the
content of the thought. Of course it is true that someone may employ
images of objects with conventionally determined conditions of applica-
tion, somewhat like a hieroglyphic or ideographic script. But then this is
not really distinct from the claim that thought involves some sort of
language, since such an image-system would presumably mirror the struc-
tural properties and combinatorial powers of a language.
Despite the points just made, we certainly do not want to claim that
mental images can never play a part in anything which might properly be
192 Forms of representation