Chemistry Reference
In-Depth Information
idealized models are equivalent in this respect. So, in the ontological sense of
explanation, an idealized explanation is just as good as the canonical explanation.
But Strevens is swift to reject the explanatory equivalence of idealized and canon-
ical models.
4
Idealized models are explanatorily deficient because they plumb
treacherous waters by getting into the issue of the irrelevance of some causally
salient factor. They may be deficient because they make a causal irrelevance
explicit without explaining
why
it is irrelevant. To take one of Strevens
central
examples of idealization, by ignoring molecular collisions in explaining the ideal
gas law one might construct an idealized model that explains the law but does not
tell us why molecular collisions are irrelevant to the causal entailment of the
explanatory target. Or by assigning an extreme value to a parameter, idealized
models falsely imply that a causally salient factor makes no difference at all.
5
So an
idealized model can increase explanatory power relative to its veridical counterpart
but it is not as good an explanation as a canonical model.
'
13.4
Intrinsic and Extrinsic Functions of Explanation
An important issue concerns the extent to which Strevens
account succeeds in
delivering an account of idealized explanatory models. Where idealized models
really come into their own is not in the ontological mode of explanation at all.
The crucial difference between canonical and idealized models, and the explanatory
virtue of idealization itself, is really appreciated once one shifts attention from the
ontological sense of explanation to what Strevens calls the “communicative” or
pragmatic sense of explanation. Explanation in the communicative sense is not
concerned with the objective explanatory facts but rather the “explanation
'
s means
of representation” (op. cit., p. 320). It is in the communicative sense that idealized
models are individuated and prove to be “valuable scientific all-rounders” in three
ways (ibid, p. 321). First, they highlight causally salient details that are nonetheless
irrelevant to the causal entailment of the explanandum, thereby helping us to
appreciate those irrelevances by explicitly distorting them. By the conspicuous
omission of the bonds forming the underlying carbon skeleton, one is in a sense
drawn to explanatorily irrelevant but nonetheless causally salient factors. Second,
idealized explanations are simpler than veridical models, and they can also be
'
4
According to Strevens, individuating models that explain in the ontological sense must be carried
out by attending to not only the setup but also to the “follow-through” - a deduction of the
explanatory target from the setup, which “represents the way in which [the difference-makers]
make a difference” (ibid, p. 319). By taking the follow-through into account one can individuate
canonical and idealized models because the former “contains more objectively explanatory
information than the idealized model” (ibid, p. 320).
5
In other words, it does not mean that the value for the parameter makes no difference; it is just
that whatever value it does take it will not make a difference to the occurrence of the explanatory
target, or that the value must fall within a certain range (ibid, p. 320).
