84 J.J. Haldane
mutation plus ‘selection’ of features having adaptive utility. This is some-
thing to which I shall return in the next section, but as above my concern at
this stage is to query whether Smart’s conception of science is not ideologically
driven. Consider, then, the insistence upon reductionism. Like so many other
expressions used by philosophers this is a term of art in need of definition.
To begin with, let me distinguish between ontological and conceptual-cum-
explanatory reductions. These can go together but they need not.
An ontological reduction maintains that one purported category or class
of entities is a construct and that the things belonging to it are derived
from some more basic category. So, for example, the average weight of
members of a population is an artefact derived from a series of actual weights
upon which a mathematical operation has been performed: average weight
W= the sum of real weights (w^1 , w^2 , w^3 ,...wn) divided by the total number
n in the population. Therefore, we might say there are no such things
as average weights over and above real weights. Certainly some individual’s
weight may in fact be equal to the average; nevertheless his weight is real
in a way that the average is not. This comes out in the fact that there need
not be anyone whose weight equals the average; the latter is not an actual
scale-impacting weight, but rather an intellectual construct abstracted from
such. At this point, however, the ontological reduction might be pressed
further, since it may be claimed that actual weight is not a fundamental
category either, but is itself an artefact reducible to ‘real’ features such as mass
and gravitational acceleration. At some stage, however, the reductions will
have to come to an end and this amounts to an identification of the class
of basic entities.
In order to appreciate the difference between ontological and explan-
atory reductionism it is useful to distinguish between, on the one hand,
things or natures and, on the other, concepts or terms. Ontological
reductionism holds that what are identified as Xs are really Ys; explanatory
reductionism maintains that talk of ‘Xs’ can be replaced without loss of
content by talk of ‘Ys’. In the philosophy of mind, for example, there are
at least two kinds of behaviourism both of which involve reductionism.
Some behaviourists argue that mentalistic concepts such as ‘belief ’ and ‘desire’
classify patterns of actual and potential behaviour, and moreover that these
concepts can be replaced by overtly behavioural ones without loss of meaning.
In short, to say that A ‘believes’ something is not to describe or attribute
a state additional to his or her behaviour. It is precisely to refer to that
behaviour, and the same reference could be made using undisguisedly behavi-
oural terms. This claim combines ontological and explanatory reductions by
insisting both that there are no mental attributes over and above patterns
of behaviour, and that mental concepts can be translated into or replaced by
behavioural notions. However, while having reason to suppose that there are