Metaphysics and Induction’: Reply and Rejoinder

by James W. Felt

James W. Felt, S.J., is Associate Professor of Philosophy at the University of Santa Clara, having received the Ph.D. from Saint Louis University in 1965. He has published “Whitehead and the Bifurcation of Nature,” Modern School-man 45 (1968), “Invitation to a Philosophic Revolution,” New Scholasticism 45 (1971).

The following article appeared in Process Studies, pp. 179-180, Vol. 1, Number 3, Fall, 1971. Process Studies is published quarterly by the Center for Process Studies, 1325 N. College Ave., Claremont, CA 91711. Used by permission. This material was prepared for Religion Online by Ted and Winnie Brock.


The author replies to Gary Gutting’s article “Metaphysics and Induction,” stating that Gutting’s criticisms do not quite succeed in making the case. In a rejoinder, Gutting claims (and Felt seems to agree) that metaphysical theories of causal efficacy and internal relations are neither necessary nor sufficient for a solution of the problem.

Whitehead’s fall into metaphysics, it has been observed, began with his biting into the apple of induction (2). Has even this turned mealy and sour in his mouth, as Professor Gutting implies? Illuminating as Gutting’s criticisms are, I do not think they quite succeed in making the case he supposes.

The key to Whitehead’s understanding of the problem of induction and of its solution lies in his recognition of our direct experience of causal efficacy, that vector aspect of experience by which the immediate past is felt as imposing limitations on the present, and the present felt as making a difference to the future. In overlooking this dimension of immediate experience Hume rendered inductive inferences radically unsupportable. At the same time he reduced causality to a purely inferential, and in fact groundless, relation posited between observed events or their appearances. Whitehead, who is not given to exaggeration, asserted: it is impossible to overemphasize the point that the key to the process of induction, as used either in science or in our ordinary life, is to be found in the right understanding of the immediate occasion of knowledge in its full concreteness" (SMW 64). It should be noted that this right understanding of the vector characteristic of immediate experience is asserted to be "the key to the process of induction," not a solvent for the practical problems of inductive inference.

Professor Gutting does a real service in underlining the distinction between the metaphysical and the epistemological aspects of the inductive problem. Why then does he proceed to write off the importance of the metaphysical problem, as at the beginning of his second section where he explicitly equates the problem of induction with the epistemological problem? This move, however, enables him to state that acceptance of a theory of genuinely productive causality is neither a sufficient nor a necessary condition for a solution to the problem of induction. But clearly that is true only of the epistemological problem. One may grant Gutting’s point with regard to the epistemological problem without in the least admitting that causality is neither a sufficient nor a necessary condition for the metaphysical problem of induction.

Now the whole efficacy of our practical solutions to the epistemological problem rests ultimately on there being a valid metaphysical ground underlying them, namely, a describable causal efficacy given within experience. It is true that metaphysics cannot inform me whether you have laced my bread with arsenic, but it can provide rational grounds for thinking that if you haven’t, then I can expect from the bread the usual nourishment and the usual fattening carbohydrates. It may be a perplexing legal problem to discover who is the rightful heir of an estate, but the whole importance of the investigation rests on the existence of an underlying inheritance law which will make a real difference indeed to the legally declared heir.

By shifting the entire discussion to the likelihood of determining when "other things are equal," Gutting has neglected Bertrand Russell’s admonition to keep our eye on the interesting doubt about induction. In seeking our justification for thinking the sun will rise tomorrow, Russell suggests that a reply might first be given in terms of the laws of motion of rotating bodies, such as the Earth. Then he goes on to say: "Of course it might be doubted whether we are quite certain that there is nothing outside to interfere [Gutting’s "other things equal"], but this is not the interesting doubt. The interesting doubt is as to whether the laws of motion will remain in operation until tomorrow" (1:61).

Granted that it may be difficult, perhaps impossible, to ascertain that "other things are equal," the more fundamental question still remains: supposing that other things are equal, what justifies our expectation even then that the future event will turn out like the past? It is only by tacitly supposing that there is a positive answer to this latter question that it makes any sense to attempt to answer the former. Hume himself grants this in his Enquiry when, speaking of the present fact and that which is inferred from it, he says: "Were there nothing to bind them together, the inference would be entirely precarious" (Sect. IV, Part I). Why inquire whether there is arsenic on our bread if even without it we can have no legitimate expectations about what the bread will do for us? In fact, unless we face Russell’s interesting doubt then we cannot reasonably entertain any expectations about the effect of arsenic itself.

Whitehead’s technical solution of the metaphysical problem of induction amounts to a careful description in terms of his categorical scheme of that vector characteristic of lived experience alluded to earlier. We find that in our experience there are no radical discontinuities This experiential fact constitutes a necessary, though not a sufficient, condition for the availability of a solution to the epistemological problem. Every solution to the epistemological problem implies the continuity of the future with the present and the past. Whitehead’s theory of internal relatedness through causal prehensions makes speculative sense out of this continuity.

Finally, is it true, as Gutting contends, that there can exist metaphysical justifications of induction which are independent of any doctrine of causality or of internal relations? I suggest that any such supposed justifications (for instance, those Gutting cites of Keynes or of D. C. Williams) are effective only within a quantified, static view of temporal process, in which future events are taken as already in some sense defined, like marbles in a bag. Given certain suppositions about the randomness of our sampling, we may formulate statistical expectations about the color of the marbles still remaining in the bag. But future events are not defined like so many already-given marbles. The future grows creatively and continuously out of the present, and if it does not grow under causal constraints already operative and in principle discernible within the present, then our expectations about the character of the future are as groundless as Hume said they were.


1. Bertrand Russell, The Problems of Philosophy, Oxford University Press, 1959.

2. Gregory Vlastos, "Whitehead, Critic of Abstractions," The Monist, 39 (1929), 170-203.


Rejoinder, by Gary Gutting

Let me begin by summarizing what seems to be a fairly wide area of agreement between Professor Felt and me. We seem to agree that the problem of induction is a problem of justifying inferences from what has happened in the past to what will happen in the future. As such, induction is a problem about what we can justifiably claim to know; this is why I term it an essentially epistemological problem. Of course what we can justifiably claim to know about the world will depend on what the world is like; therefore, I agree that the problem of induction has metaphysical aspects. Accordingly, I am willing to accept Felt’s suggestion that the metaphysical and the epistemological aspects of the problem of induction can be conveniently separated to give us two sub-problems: (1) How do we know that there are regularities in nature that persist through time (metaphysical aspect)? (2) Given a particular present event, how do we know which regularities are relevant to the prediction of its effects (epistemological aspect)? Felt admits that a doctrine of causal efficacy or internal relations is neither a necessary nor a sufficient condition for a solution to (2). In turn, I admit that such a doctrine is sufficient (but not necessary) for solving (1). Actually I make this admission only on the condition that the relevant metaphysical theory itself be established in some satisfactory way that does not itself presuppose the legitimacy of inductive inference. In particular, I think Whitehead’s hypothetico-deductive version of metaphysical method makes him very liable to the charge of circularity in his attempt to justify induction.

However, I disagree with two of Felt’s basic claims: (a) the claim that (2) is an "uninteresting" problem; (b) the claim that a theory of causal efficacy via internal relations is a necessary condition for a solution of (1).

With regard to (a): The only ways a problem can be uninteresting is for the issue it raises to be of no significance or for its solution to be trivially simple. Since (2) is crucial for the justification of inductive inference, it cannot be called uninteresting as long as we regard the problem of induction as interesting. Indeed, it seems that Whiteheadians should find (2) particularly significant, since they so greatly emphasize the concrete particulars of a situation as opposed to abstractions from the situation. To minimize the significance of (2) is precisely to ignore the fact that the concrete locus of the problem of induction is the particular inductive inferences made by individuals. (I owe this comment to my colleague, Neil Delaney.) And if Professor Felt could show that the solution to (2) is trivially simple, I am sure he would have done so in his reply.

With regard to (b): Felt argues that any proposed justification of induction which did not employ a theory of causal efficacy or internal relations would fail because it could not take account of the fact that the future involves genuine novelty and hence cannot be presently defined as something already given. I agree that a satisfactory justification of induction must take account of the presently indeterminate, genuinely novel character of the future. But I do not see that he has shown that the sampling methods of probability theory (supplemented by minimal metaphysical assumptions about, say, random samples) are not capable of taking account of this aspect of the future. Surely, it is not true as Felt suggests that the use of sampling techniques must presuppose that future samples are already given in a completely determined way. For example, sampling techniques can be used to predict the probable behavior of children as yet unborn. In fact, even in the case of drawings of already-existing marbles from a bag, the real subject of the probabilistic inference is not the marbles in se but the as yet nonexistent and indeterminate drawings of marbles. The possibility of radically different behavior due to completely novel characters or circumstances is taken into account by the very fact that any predictions about the future are regarded as at best probable.

To summarize my position: Its most central claim is that (2) as well as (1) is an essential element of the problem of induction. Once this is granted, I claim (and Felt seems to agree) that metaphysical theories of causal efficacy and internal relations are neither necessary nor sufficient for a solution of the problem. Finally, the only relevance to the problem of induction that I see for metaphysical theories of causal efficacy and internal relations is that they may provide one way of finding a sufficient condition for the solution of sub-problem (1).