DISCUSSION PAPER: INDUCTION, EVOLUTION, AND ACCOUNTABILITY

The problem seems to have to d o with the process called “induction.” Let us adopt a simple definition and let intuition supply the rest. Induction involves the extraction of information from certain inputs. It involves real time, a n initial state in which certain information is demonstrably absent, and a final state in which the information is demonstrably present. In between is something very like trial and error. The extent to which this trial and error is constrained by factors other than pure chance is determined by initial conditions. Furthermore, the extent to which induction is constrained by factors other than pure chance is precisely the extent to which it is not induction. So anyone speaking about “pure” induction must have something else in mind, because according to the present definition, some sort of structurally constrained initial conditions must obtain if an initial state is to be specified at all. There is no question but that the above definition strongly equates inductive processes with statistical processes. Alternative contenders will have to come forward and be equally specific. Now, the central question that motivates such close scrutiny of induction: When is it incumbent upon a scientist to give an inductive explanation of an empirical phenomenon? Otherwise stated, what must be explained in terms of “how it got that way” (as opposed to what can be accounted for by a state description simpliciter)? Of course, “a statedescription simpliciter“ is usually neither a simple nor a simplistic matter. For example, if such a description is not t o be just an empty phenomenology, it must make reference to certain alternative states, not actual but “possible” ones, and the state description must specify the constraints that generate the actual states rather than the possible states. It is the rules that exclude or forbid the alternative-states which give power and substance t o a generative state description. Now to return to the question of what demands an inductive explanation rather than a “mere” (possibly generative) state description. Perhaps it would be simpler to state what does not demand an inductive explanation, for I claim that what does demand such an explanation constitutes by far the larger class. As an example of what does not require inductive explanation, 1 offer a reductio in the form of an absurd question: Surely it makes no sense to ask, in general, “how the universe ‘learned’ its laws”! The primary laws of physics (whatever they will ultimately turn out to be), as well as cosmological initial conditions, are the structural constraints on the “initial state” of the universe, and they are not t o be accounted for inductively; and this for totally nonmysterious, logical, and methodological (if one can speak of a methodology of theoretical explanation) reasons. Second, the “laws” of logic are themselves not to be accounted for inductively. They too are “givens.”