BROMBERGER ON WHY-QUESTIONS
In his famous paper on ‘Why-Questions’, in Mind and Cosmos (1966), Bromberger criticized Hempel’s deductive model of explanation by showing, with the aid of many instances, that not all deductions are explanations.
His aim was to supplement Hempel’s model with the aid of the logic of questions and answer. Before following him through I want to discuss his criticism because it has puzzled me for some time. Surely Hempel knows that not all deduction is explanation. Hempel knows that any statement is logically equivalent to its double-negation but the two do not explain each other. Bromberger has laboured to construct many non-explanatory instances with a universal and an existential premiss and an existential conclusion, even a conclusion which does not follow from the existential premiss alone. I dare say Hempel knows this too. Moreover, there is really no difference, as far as explanation is concerned, or as far as deduction is concerned, between the case of a universal and an existential premiss with an existential conclusion, and the case of a universal premiss with a conditional conclusion composed of the two existential ones in the proper order. The question really is, do Bromberger’s examples constitute a criticism of Hempel’s model? What is Bromberger saying Hempel has overlooked? Did Hempel ever claim that all deduction is explanation? Is Bromberger’s criticism also relevant to any other model of explanation, from Aristotle’s to Descartes’, to Newton’s, to Whewell’s, or to Popper’s?For Aristotle, explanation is deduction from a definition or from a statement of an essential quality. When we explain Socrates’ mortality by means of ‘Man is a featherless biped’ our explanation is defective; but when we use ‘Man is a rational animal’ together with ‘All animals are mortal’, etc., our explanation is satisfactory provided our deduction is valid.
In spite of all savage attacks on Aristotle, his metaphysics, his vagueness, and what-have-you (attacks which are often just, in my opinion), Aristotle wins every time. Take the following: All metal responding properly to a touch-stone is gold; Tom’s coin responds properly to a touchstone; therefore Tom’s coin is gold. This is a valid inference, the premisses are eminently reasonable; the application of the inference for ages and ages has not been criticized by any historian of science, of technology, or of economics. Yet anyone who calls this inference explanatory will be met with an amused smile. Take, per contrast, all metal which is yellow and whose average specific gravity is dAu is gold, Tom’s coin etc., therefore Tom’s coin is gold.
This comes closer home. Go deeper: All metal with specific gravity dAu is gold, etc., and you are still closer home. Still closer is, All and only atoms with the nuclear charge ZAu are gold; atoms with nuclear charge ZAu have the nuclear mass NAu; the number of atoms in a cubic centimeter of solid metal under normal circumstances is, etc. etc.... therefore Tom’s coin is gold. In other words, we do recognize certain deductions as explanations, classical or modern, an inaccurate modern (without isotopes) and an accurate (with isotopes) modern, and we grade them as to their degree of satisfactoriness. Incorrect as certain classical explanations surely are, they nonetheless are explanations proper, though less satisfactory than other explanations we have; whereas deductions from accidental statements referring to touchstones are not viewed as explanations, even when endorsed. Why then did Hempel, as well as others before and after him, refrain from dividing those deductions which are explanations from those which are not?
That some deductions are not explanations is a fact with a history too well-known to be forgotten. Soon after Aristotle offered his theory of explanation as deduction from essences, a new idea of deduction was used not as explanation but to save the phenomena.
It is commonly well-known that Ptolemy did not consider his deductions explanatory but purely mathematical. All these distinctions, between explanatory theory, or philosophical theory, and mathematical theory or purely descriptive and/or predictive theory, never died. The Copernican tradition was opposed to the mathematical tradition and in favor of a philosophical tradition. Galileo, and later on Descartes, make it clear that in physics the philosophical is the geometrical. What is the geometrical is less clear - it is, I suppose, Cartesian metaphysics or some such system. Equally clear - to Galileo, Descartes, and others - was the fact that theory had to have deductive force to be explanatory. Thus, though Descartes’ astronomical theory did comply with his metaphysical principles, it was not satisfactory in so far as it was not deductive. Newton’s theory of gravity seemed to be the reverse: deductive it surely was, and to a high degree; yet it did not comply with Descartes’ metaphysical theory. Perhaps it might have been considered a mathematical theory, purely descriptive, designed merely to save the phenomena. Orthodox Cartesians did, indeed, tend to consider Newton’s theory purely mathematical and not at all philosophical. Even when he was an orthodox Cartesian, Newton resented this. His principles were mathematical principles of natural philosophy, even though as causal explanations they were not yet satisfactory. He confessed he could not deduce his theory of gravity from a satisfactory causal theory.Causal explanation now becomes a mystery. In Aristotle, causal explanation is deduction from definitions of essences. In Descartes, causal explanation is deduction from a hypothesis conforming to Cartesian metaphysics. (I have discussed the theory of a hypothesis conforming to metaphysics in Chapter 9 above.) When Newton was a Cartesian he agreed; even then the status of his own theory of gravity was obscure; when he ceased to be a Cartesian the status of his own theory became even less clear.
Meanwhile two things happened which led to the elimination of causality from causal explanation altogether. The first was Newton’s answer to the charge of assuming occult qualities, the second was Hume’s attack on causality.The attack on occult qualities was a confused critique of the moderns, especially the Cartesians. They charged the Aristotelians with two charges as if they were one. The first charge is that Aristotelian explanations in terms of occult qualities are circular or ad hoc. Moliere’s example is the paradigm: opium puts you to sleep since it has vis dormativa. The second attack, to use Hosper’s terminology, was that Aristotelians explained the known phenomena in terms of unknown or hidden (etymologically, occult=hidden) essences. Somehow, it was felt, if the explanatory principles were better known, better comprehended, than the explained phenomena, then the circularity too will vanish. There is much force to this idea. Yet the demand to explain by the known only, should not be confused with the refusal to allow for circular explanation: at best the one covers the other; they are certainly not identical. The requirement to use only known explanations may be sufficient for the exclusion of circularity: it certainly is not necessary.
Newton did not clarify matters; rather, and not for the only time, he exploited the confusion of his critics in order to repel the attack: his theory of gravity, he said, was not of an occult quality, since it was not circular. His disciple Roger Cotes even declared gravity to be an essential quality, thus making Newton conform to the Aristotelian-Cartesian theory of explanation, though not to Cartesian metaphysics. Kant later declared Newtonian gravity self-evident.
I shall not dwell on Hume’s attack on all cause, all essence, all substance. Let me merely say that those who took some heed of his criticism, yet refused to go all the way with Kant, quite naturally found in Newton’s idea of non-circularity the only traditional element still available for epistemologists.
IX.