Chemistry Reference
In-Depth Information
draw on what Strevens calls a “basing generalization”, which expresses the
existence of a contingent pattern of phenomena required to casually entail the
explanatory target.
2
This brief synopsis of Strevens
account of scientific explanation hardly begins
to do justice to the power and comprehensiveness of his ideas. And while the focus
of this chapter is relatively narrow because it is concerned with idealized explana-
tion, this is nonetheless an important and revealing aspect of his thinking on
scientific explanation. Idealization arises in the context of regularity explanation
and is “any misrepresentation of the causal process that improves the explanatory
power of a model by comparison with its veridical counterpart” (ibid, p. 300).
The explanatory power of idealized models is to be assessed relative to a “veridical
counterpart”: a model that has not been subject to the kairetic procedure of
optimization. Since it hasn
'
t been optimized, a veridical counterpart falsely implies
that
non
-difference-makers as well as difference-makers are explanatorily relevant.
To say that idealization
improves
explanatory power relative to its veridical coun-
terpart might, however, sound counter-intuitive because idealized models distort
the causal story. But Strevens
'
tactic is simply to assert that idealizations are not to
be taken literally; they should not be taken as implying that
non
-actual properties
are causally relevant to the production of an event or regularity. Rather, they draw
our attention to the fact that some actual property, while causally salient, is a
non-difference-maker. An idealized model will only distort some subset of
non
-
difference makers, and what remains are only those causal factors that make
a difference to the explanatory target. Therefore an idealized model has more
explanatory power than its veridical counterpart because it does not contain an
excess of explanatory irrelevances.
More provocative is Strevens
'
claim that “[a]n idealizing explanation is in one
important respect explanatorily optimal: it cannot be further improved” (ibid).
Idealized models explicitly communicate the idea that certain features of a system
do not make a difference to the causal entailment of the explanandum. Compare
idealized models to what Strevens calls a “canonical model”, which is a causal
model that has been subject to the kairetic procedure of optimization and contains
'
2
For example, in order to explain why all normal ravens are naturally black, it is not enough
to posit a mechanism described by a causal model that cites the relevant biochemical laws,
and physiological and environmental conditions. To causally entail, and thus to explain the
blackness of ravens, one must also cite the basing generalization “All normal ravens have
P
”,
where
P
stands for the appropriate physiological properties (ibid, pp. 228-229). Basing general-
izations can be physically contingent, physically necessary or perhaps metaphysically necessary,
but one thing to note is that they need not be causal generalizations. Hence non-causal factors can
play a role in the explanation, a theme I will reprise below. A basing pattern of phenomena
corresponding to a basing generalization play an explanatory function in a manner like initial
conditions in event explanation. A major difference between basing patterns and initial conditions
is that since the former concern regularity explanation, they include not just actual but also
counterfactual states of affairs (p. 235). For basing generalizations to play an explanatory function
requires that they are subject to counterfactual constraints, the details of which I omit here (but see
