Atheism and Theism 97the grip of a false theory of human beings. All that really exists is bodily
movement arising from neurophysiological events; instead of being thinking
agents we are organic machines. Against this I contend that ordinary psy-
chological descriptions, including self-descriptions, are not parts of a proto-
scientific theory of the unseen and unknown inner causes of movement, but
(often observational) accounts of thought and action – in the latter case
interpretations of intentional behaviour.
According to the eliminativist the apparent action described as ‘Kirsty’s
writing a sentence’ is in truth no more than a sequence of causally related
physical events. Since ‘writing’ is an action-term implying intention, and
‘sentence’ is a grammatical description presupposing common linguistic con-
ventions, the eliminativist is committed to the possibility of replacing these
with non-psychopersonal terms in his account of events. It is now generally
agreed among philosophers that ‘type–type’ psychophysical identity theories
will not work. That is to say any hypothesis to the effect that psychological
items of type Ψ (psi) are identical with physical items of type Φ (phi) falls foul
of the fact that instances of the former type can be associated with many dif-
ferent sorts of physical set-ups. There are, for example, indefinitely many
ways of writing a paragraph and it would be crazy to think that this action
type could be correlated with any specifiable type of bodily movement, or
even with a disjunction of these, preparatory to identifying the former with
the latter (and eliminating the one in favour of the other). That is to say, any
suggestion of the form ‘action type Ψ is identical with physical type Φ, or
with one or other of the physical types Φ^1 ,Φ^2 ,Φ^3 ,Φ^4 ,.. .’ is refuted by
actual or easily imagined cases of the former that are not cases of the latter.
What I now want to add to this is the suggestion that it is equally impos-
sible to sustain a ‘token–token’ identity theory. Such a theory would insist
that while action types cannot be identified with physical event types, never-
theless individual instances (or tokens) of the former can be identified with
instances of the latter. In other words I am claiming that actions cannot be
identified with, reduced to, or eliminated in favour of movements of quantit-
ies of matter. Imagining an appropriate scene, ask yourself the question what
is the physical event that is the reality otherwise described as ‘Kirsty’s writing
a sentence’? No doubt some bodily movement, but which? As one begins to
think about individuating a series of events within a region of space–time the
nature of the problem starts to become clear. Consider all the movements
involving Kirsty’s body that might have occurred in the specified region:
heartbeats, hair quiverings, eye blinkings, nerve impulses, muscle contrac-
tions, desk impactings and so on. There is no prospect, not even ‘in principle’,
of identifying relevant movements save by non-dispensable use of action
concepts involving reference to Kirsty’s intentional behaviour. The relation
between the movements thereby identified and the action itself is not one of