Identity, Structure, and Causal Representation in Scientific Models - Mechanism and Causality in Biology and Economics - page 54

identity by closely matching relevant characteristics of different applicants - other
than the target characteristic of sex or race. Significant knowledge is obtainable in
such ways and there is no need to stigmatize its causal bona fides .
Woodward's argument that it is the beliefs of the employer not the being female
that is the relevant causal variable does not persuade. The beliefs of employers may
be causes of the detailed outcomes of the discrimination - for example, not hiring or
paying a lower salary. The relevant question here, however, is what causes those
beliefs. The most common reason that we believe someone to be female is that she
is female. Woodward could object that it is not being female that causes the belief
but the appearance of being female. But this is just the analogue of saying that it is
not the “moneyness” of the nickel, but its physical characteristics that cause it to fall
into slot 3. The physical characteristics are the causally relevant ones. What is it to
be female in a causally relevant sense? It is to have a sufficient number of
stereotypical female characteristics. For purposes of the employment process, a
sufficiently plausible transvestite counts as female. Characteristics of actual
females determine the female stereotype - that is, why the values of certain
variables bundled in a certain way convey the appearance of femaleness. And
without the stereotype, variables with those values would not be causally salient.
(The same issue arises with the coin sorter: a slug is to a nickel as a transvestite is to
a female.)
Some entities may have an essence that cannot be changed while maintaining
existential identity. But if, as the structural account would have it, token
manipulations are not essential to defining cause, then it is better in these cases to
say that causal identity is determined by possession of the explicitly causally
relevant characteristics and not by some sine qua non. It is also pragmatically
superior if we accept the view that knowledge is acquired in a piecemeal fashion.
As we have seen, variables such as sex or race are avoided through substantial extra
articulation and refinement of causal mechanisms, which may lack clear conceptual
or evidential foundations. What, for example, are the detailed mechanisms by
which the appearance of being female translates into beliefs and how are they
causally relevant? We may not have grounds for knowing such details and we may
not need to know them to know what is pragmatically relevant about discrimina-
tion. And if I am correct that the coin sorter is really no different in principle than
the sex discrimination case, then we would be forced to look for such extra
articulation in a vast array of cases. But if the structural account is plausible, we
can do happily without

it and without the troubling requirement of token

intervention.

4 Causation and Representation

The focus of this chapter has been on representing causes. Of course, the ultimate
importance of causal analysis is not located in its representation but in discovering
what causal relationships actually obtain in the world and using those causes to