This point about the hypothetical character of a priori sciences has not always been appreciated. In many classical discussions of the subject, the assumption was made that a system of this kind will start from as well as terminate in truths and that necessity will attach to premises and conclusions alike. Aristotle and Descartes both spoke as if this must be the case. It is clear, however, that in this they were mistaken. The form of a typical argument in this field is as follows: (1) p is taken as true or given as true; (2) it is seen that if p, then q; (3) q is deduced as true, given the truth of p. There is no need here for p to be a necessary or self-guaranteeing truth; p can be any proposition whatsoever, provided its truth is granted. The only necessity that needs to be present is that which characterizes the argument form, “If p is true, and p implies q, then q is true,” that is [p · (p ⊃ q)] ⊃ q, in which · symbolizes “and,” and ⊃ means “implies”; and this is a formula that belongs to logic. It is this fact that makes philosophers say, misleadingly, that a priori sciences are one and all analytic. They are not because their premises need not answer this description. They, nevertheless, draw their lifeblood from analytic principles. Metaphysics as an a priori science
It is clear that metaphysical philosophers have sometimes aspired to present their results in the form of a deductive system, to make metaphysics an a priori science. For this purpose they have taken a deductive system to require not just that the premises entail the conclusions but further that they themselves be necessarily true. Spinoza thus began the first book of his Ethics by laying down eight definitions and seven axioms whose truth he took to be self-evident and then proceeding in the body of the text to deduce, as he thought with strict logic, 36 propositions that follow in order from them. He repeated the procedure in the rest of his work. That philosophical conclusions should thus be capable of being set out “in the geometrical manner” was something that Spinoza took as axiomatic; to be worthy of attention at all, philosophy must issue in knowledge as opposed to mere opinion, and knowledge proper had to be exempt from the possibility of doubt, which meant that it must either be intuitively evident or deducible from what was intuitively evident. Spinoza took this conception of knowledge from Descartes, who had himself toyed with the idea of presenting metaphysical arguments in the geometrical manner. Descartes, however, pointed out that, although there was no difficulty in getting agreement to the first principles of geometry, “nothing in metaphysics causes more trouble than the making the perception of its primary notions clear and distinct”; the whole trouble with this discipline is that its students fail to see that they must start from what are in fact the basic truths. Descartes himself spoke as if the problem were no more than pedagogical; it was a question of making people see as self-evident what is in itself self-evident. His own “analytic” approach in the Meditationes was chosen to overcome these difficulties; it was, he said, “the best and truest method of teaching.” But it may well be that this account is too optimistic. The difficulty with a system such as those of Descartes and Spinoza is that there are persons who cannot be brought to see that the primary propositions of the system are self-evidently true, and this not because they are lacking in attention or insight but because they see the world in a different way. This suggests that in any such system there will necessarily be an element that is arbitrary, or at least noncompulsive. However cogent the links that bind premises to conclusions, the premises themselves will lack a firm foundation. If they do, the interest of the system as a whole must be greatly diminished; it can be admired as an exercise in logic but not valued for more than that.
Spinoza, Benedict deBenedict de Spinoza, painting by an anonymous artist; in the Herzogliche Bibliothek, Wolfenbüttel, Germany.Fine Art Images—Heritage Images/age fotostock
To avoid this unpalatable conclusion, two expedients are possible. The first is to say that the first premises of a metaphysical system must be not merely self-evident but also self-guaranteeing; they must be such that any attempt to deny them can only result in their reaffirmation. Descartes believed that he could satisfy this requirement by grounding his system in the cogito, though strictly this was the primary truth only from the point of view of subjective exposition and not according to the objective order of things. Aristotle somewhat similarly had argued that the logical principle of noncontradiction, which he took to express a highly general truth about the world, must be accepted as axiomatic on the ground that its correctness is presupposed in any argument directed against it.
Even the Idealists Bradley and Bernard Bosanquet at times spoke as if the first principles of their system were in some way logically compulsive; as Bosanquet put it, one had either to accept them or recognize that one could know nothing. Whatever the position may be about particular metaphysical propositions, however, it seems clear that not all truths that are taken as basic in metaphysics have the characteristic of being self-guaranteeing. A Materialist takes it as fundamental that whatever occurs happens as a result of the operation of natural causes; a theist sees things in the world as finite and thus as pointing beyond themselves to the infinite being who is their ground. No contradiction is involved in denying these positions, though of course for those who accept them the denial necessarily involves commitment to falsehood. It is, however, one thing for a proposition or set of propositions to be false, another altogether for it to be necessarily false. If the first principles of metaphysics were really self-guaranteeing, only one system of metaphysics could be coherent, and it would be true just because it was coherent. The very fact that there is an apparent choice between competing metaphysical systems, which may differ in plausibility but agree in being each internally self-consistent, rules this possibility out.
The alternative is to argue that fundamental metaphysical propositions, though not self-guaranteeing, are nevertheless not arbitrary; they have or, to be more cautious, can have a firm foundation in fact. Metaphysical speculation is not, as some opponents of metaphysics have suggested, essentially idle—that is, the mere working out of the logical consequences of premises that the metaphysician chooses to take as true. Or, rather, it does not necessarily answer this description because a metaphysician can have insight into the true nature of things and can ground his system on that. This second position in fact involves arguing that metaphysics is not an a priori but an empirical science. Metaphysics as an empirical science
If metaphysics is an empirical science, the question of whether or not to accept a metaphysical theory must be answerable, in part at any rate, by reference to experience. It will not depend on experience alone, any more than does the acceptability of a scientific theory, because here, as in the scientific case, thinking comes into the reckoning too. A metaphysician can be mistaken in his deductions, just as a scientist can. But even if these are impeccable, he will not necessarily succeed on this view of his undertaking. It may be that he argues correctly from premises that are unacceptable—unacceptable because they lack the necessary foundation in fact. He will then be like a scientist who puts forward a hypothesis and deduces its consequences without mistake only to find that experience fails to confirm the supposition on which he is working.