Statistical laws
I have tried to show, in the previous two chapters, that determinism and
free will are compatible, that one need not postulate a breakdown of determinism
to accommodate free will. I have tried to show that the problem comes about
through a mistaken conception of what physical laws are; the solution to
the problem does not require abandoning Determinism, but neither does it
require embracing Determinism.
Freed of the enticement of warranting indeterminism
to solve the free-will problem, we can now go about investigating the matter
of probabilistic laws in a dispassionate way. At the very least, the solutions
to the weighty problems of free will and moral responsibility will not
hang in the balance.
The Copenhagen interpretation (1927) of quantum
mechanics … historically served as the catalyst for the contemporary
debate about the status of statistical laws. That the debate should derive
its principal impetus from physics is unfortunate from the point of view
of intellectual history. For it bolsters the belief that the natural sciences
are somehow more fundamental than the social sciences and that the “true” nature
of physical laws is to be learned from what physics and chemistry reveal,
rather than from such sciences as economics and/or sociology. Certainly,
the view still prevails that the “laws” of sociology are but
the logical consequences of the “fundamental” laws of physics
and chemistry. Such a view carries the corollary that, were the “laws” of
sociology, economics and so on to be statistical rather than universal,
this fact would not be decisive, or even for that matter particularly relevant,
in answering the question of whether any “real” physical laws
are statistical or whether all physical laws are – without exception – universal. “Real” laws,
in this view, are the preserve of physics, and it is to physics and physics
alone that one must turn to answer questions of whether “real” laws
are statistical.
This way of approaching the question of whether
physical laws can be statistical makes it look as if what were at issue
were an empirical question, as if it were a matter to be settled in the
physicists’ laboratories whether physical laws might have a certain
property. But surely this is a mistake. To proceed in
the manner imagined, one would have to have an independent way
of recognizing what a physical law is, and then one would check to see
whether any members of this class were statistical rather than universal.
We have only to put the matter this way to see immediately that the question
is not empirical but conceptual. It falls to us to decide whether,
and if so under what circumstances, we might want to allow that a statistical
proposition is to be regarded as a physical law. Certainly, it is an empirical
matter to discover which of a certain class of contrary propositions, of
potential laws, is in fact a law; but to decide what the criteria
are by which a proposition comes to be in this class of candidates for
lawfulness in a conceptual problem.
What “laws” certain sciences – for
example, sociology, economics, pharmacology, linguistics – adduce
are nearly always statistical. Is this because the “real” laws
are universal, but incapable of explicit formulation, perhaps because of
the enormous number of variables, the unethicalness of performing controlled
social-scientific experiments, prohibitions of cost, irreproducibility
of initial conditions, etc? Or might it be that the “real” underlying
laws of social events are genuinely statistical?
If some physical laws are statistical, then it
logically follows that some statistical propositions are physical laws.
I want to go considerably further. I want, now, to argue that not
just some, but all statistical propositions that satisfy all other requirements
for physical lawfulness – being true, contingent, conditional, and
purely descriptive in their nonlogical and nonmathematical terms – are
physical laws.
Swartz, Norman. The Concept of Physical Law, Cambridge University Press, 1985, pp. 171-173