Laws and generalizations serve as the first premise, and initial conditions as the second. The conclusion is the event or process explained. In some cases, the conclusion is itself a "lower level law" and the premises just consist of those laws from which it is derived.
On this model, laws of some sort or necessary to explanation. The classic model has deterministic laws and initial conditions serving as the explanans (that which explains), suggesting that sciences without such laws (nomological or law-like) cannot provide "genuine" explanations (often said of the social sciences, for example). But more contemporary models recognize probabilistic or statistical laws as generating explanations, though not of course by way of deductively valid arguments -- only probabilistic arguments.
Hempel: explanations (genuine ones) meet two criteria:
- explanatory relevance: a genuine explanation provides good grounds for assuming that the phenomenon to be explained did or does indeed occur
- testability: the statements constituting a scientific explanation musst be capable of empirical test (with testability understood as "in principle," if not as matter of fact)
L1, L2, ... Ln
C1, C2, ... Cn
----------------
E
Again, E (the explanandum) may be a particular event, or a regularity, or an empirical law.
Fulfilling the criterion of explanatory relevance:
The explanandum is derived (in a deductively valid argument) from the explanans (laws and initial conditions)
Fulfilling the criterion of testability: the explanans implies (again, deductively) that the explanandum will occur.
An explanation can provide new insight in:
- The discovery of a new fact, which by virtue of laws already accepted, explains that fact
- The discovery of a new fact which leads to the discovery of a new (covering) law
- Showing how a new fact can be explained by reference to data about which particular facts are already available
(Universal) Laws vs. accidental generalizations
Laws, whether they explain the link between particular circumstances C and some fact, E, or more comprehensive uniformities, are universal statements: If F, then G.
But they need not assume that a generalization is exception-less to view it as a law; many hold only approximately.
Nor do all statements of a universal form, even if true, qualify as laws of nature. A universal statement can be true (all the people in this room have 4 limbs) but accidentally so (nothing in the basic laws of nature precludes the possibility that there could be someone in this room with less than or more than 4 limbs).
Proposals for what distinguishes laws from accidental generalizations:
- Laws can support counterfactual conditionals
- Laws can support subjunctive conditionals Laws can serve as a basis for explanation
- A universal statement will qualify as a law if it is implied by an accepted theory.
for the philosopher interested in the philosophy of biology:
among biologists, there are two general positions:
Autonomism: the aims of biology and methods of achieving them are fundamentally different from those of the physical sciences; these differences trace their roots to the domain of biology -- living organisms -- and the non-deterministic processes that characterize the domain.
The provincialist (and the philosophers who share this view) takes molecular biology to be the locus at which biology and the physical sciences come together: shared subject matter and techniques, and molecular biology viewed as enabling intra-theoretical reduction with the biological sciences.
For Rosenberg, it is the debate among biologists concerning autonomism/provincialism which is the more significant; indeed, he claims, "if the provincialist is correct, much of biology must be recast, if not cast down".
break with earlier approaches
to autonomism is possible if one loosens the strictures banning metaphysics
(posed by positivists), recognizing that metaphysical commitments are part
and parcel of science, and develops an alternative to vitalism. (Rosenberg
makes the claim this way: "... an autonomist who repudiates vitalism must
find a replacement for it" and he claims "one must find an argument that
will show that the epistemological differences between physics and biology
are unbridgeable in principle, even though there are no important metaphysical
differences between them". Rosenberg seems to think this is an impossible
argument to make, but Mayr's constitutive reductionism is precisely
such a position. The question is: is it successful?