3.4 Varieties of realismWe have been defending realism of intention with respect to folk psychol-
ogy. But what about the initial thought which we canvassed at the start of
this section, concerning theshallownessof our folk psychology? Can what
we have just been saying be reconciled with a lack of commitment to brains
and their properties? Here it is important to distinguish between three
diVerent varieties of realism of intention about the mental, namely (1)
token physicalism, (2)compositional realism, and (3)nomic(or causal)
realism. Only (2) and (3) are committed to the reality of mental properties
and mental state kinds, hence licensing the idea that mental states really do
exist in the natural world asnatural kinds. But (3) does this in a way which
need not carry any commitments for the neural composition of mental
state kinds. So it is (3) which we should endorse.
(1) Token physicalism has already been discussed in connection with
Davidson’s views in section 2 above, where we concluded that it could be
classiWed as a weak form of realism. If each token mental state is identical
with some token brain state, then mental state tokens really do exist in the
natural world – for those brain states will certainly be real. But this is as far
as it goes: there need be no reality to the various mental statetypes, and one
could, consistently with token physicalism, deny that the world contains
any psychological natural kinds.
(2) Compositional realism is perhaps the orthodox picture of what it
takes to be a natural kind. Under the inXuence of Kripke (1972) and
Putnam (1975a) we are tempted to think that what it is for a kind at the
level of a ‘special science’ to be real is a matter of its having a common
underlying structure – for example, water is real because it is H 2 O. (The
‘special sciences’ are those which operate at a diVerent level from, and have
a more restricted range of application than, basic physics. Special sciences
include chemistry, biology, psychology and economics.) But besides this
being an over-simpliWcation (for how tight does the basic similarity of
constitution have to be? are gases not a real kind? and solids?), it is not the
only way in which the reality of a kind can be vindicated. This is fortunate
for folk-psychological realism, because folk-psychological states are not
likely to be real in virtue of sharing a common composition (remember the
multiple-realisation objection to type–type identity theory).
(3) Nomic realism is the sort we endorse. Kinds can be real in virtue of
the fact that the terms for them mark things which are similar in their
causal interactions in a law-like way. Fodor (1983) makes this point about
aerodynamics and aerofoils: provided it is rigid enough, all that matters is
that an aerofoil should have a certain shape, not what it is made of. The
point should be obvious enough from physics, anyway (see Blackburn,
The case for realism about folk psychology 39