Elements of metaphysics

by A. E. Taylor

Now this assumption is even more evidently than the preceding a postulate. We have to make it, if we are to calculate the course of events, but we have no guarantee that it will succeed beyond the fact of its actual success. If it fails anywhere, as we shall hereafter contend that it does in the critical case of the sequence of psychical on physical events, and vice versâ, two results will follow, one practical, the other speculative. The practical consequence of the failure is that in such cases we cannot apply the concept of continuous process, and have to fall back upon the cruder notion of causal sequence. Thus, in attempting to create a science of Psychophysics we cannot hope to exhibit the whole of a psychophysical process as the continuous realisation of a single principle; we must be content to establish laws of causal connection between the physical and the psychical sides of the process. The speculative consequence is that the principle of Ground and Consequence is only imperfectly represented by the conception of a continuous process, inasmuch as that conception is only applicable where qualitative differences within the consequences of a single ground can be disregarded. This hint will prepare us for subsequent criticism of the concept of continuity when we come to deal with the metaphysics of the time-process.[^[110]]

#§ 8 (b). The Indefinite Regress. The defects of the causal postulate as a principle of explanation may also be exhibited by showing the double way in which it leads to the indefinite regress. The indefinite regress in the causal series is an inevitable consequence of the structure of time, and, as we please, may be detected both outside and inside any causal relation of two events, or two stages of a continuous process. For it follows from the structure of the time-series (a) that there are an indefinite number of terms of the series between any two members, between which there is a finite interval, and (b) that there is also an indefinite number of terms before or after any given member of the series. Like the series of real numbers, the time-series, because it satisfies the definition of a continuous infinite series, can have neither a first nor a last term, nor can any member of it have a next term.[^[111]] Applying this to the case of Causation, we may reason as follows:—

(1) The same reasons which lead us to demand a cause A for any event B, and to find that cause in an assemblage of antecedent events, require that A should be similarly determined by another assemblage of antecedent events, and that this cause of A should itself have its own antecedent cause, and so on indefinitely. Thus the causal principle, logically applied, never yields an intelligible explanation of any event. Instead of exhibiting the transition A-B as the logical expression of a coherent principle, it refers us for the explanation of this transition to a previous instance of the same kind of transition, and then to another, and so forth without end. But it is impossible that what is not intelligible in one instance should become intelligible by the mere multiplication of similar unintelligibilities.

(2) Similarly if we look within the transition A-B. This transition being continuous must have its intermediate stages. A becomes B because it has already become C, and the transition A-C-B is again “explained” by showing that A became D which became C which became E which became B. And each of the stages A-D, D-C, C-E, E-B can be once more submitted to the same sort of analysis. But in all this interpolation of intermediate stages there is nothing to show the nature of the common principle in virtue of which the stages form a single process. We are, in fact, trying to do what we try to do wherever we establish a relation between terms, to answer a question by repeating it. And we decided at the end of our last chapter that this kind of repetition is never an answer to any question. How entirely it fails to answer the question we ask whenever we look for a cause is obvious. We want to know why B exists, and we are told that B exists because it is determined by the previous existence of A. But why does A exist? Because of the previous existence of C. And so ultimately the existence of everything depends on the existence of something else, and this again on the existence of still something else. If this is so, since nothing can exist until its cause has existed, and the cause again not until its cause has existed, then, as this unending series has no first term, nothing can ever come into existence at all. This inevitable introduction of the indefinite regress whenever we try to think out the causal principle to its logical consequences, has sometimes been treated as proving the inherent defectiveness of the human mind. What it proves rather is that Causality is not a proper formulation of the real principle of the unity of all experience.

A word may be said about the attempts which philosophers have made to extricate themselves from the difficulty without giving up Causality as an ultimate principle of explanation. The least philosophical method of escape is that of arbitrarily postulating a first cause with no preceding cause, which amounts to the same thing as a beginning of existence or a first moment of time. This way out of the difficulty obviously amounts to an arbitrary desertion of the causal principle at the point where it becomes inconvenient to remain faithful to it. Whatever the nature of the event you pitch upon as your “first cause,” the causal principle, if logically valid at all, is just as applicable here as anywhere else Your “first cause” must have had a previous cause, or else the whole causal scheme must be, as we have contended that it is, the illogical and imperfect perversion of a genuine principle of systematic connection, useful and indeed indispensable in practice, but quite indefensible in theory.

It would not help you out of the difficulty to distinguish between a first event and a first moment of time, by postulating a first cause with an indefinite lapse of empty time before it. For the causal principle would then require you to look for the determining conditions of the first event in the preceding lapse of empty time. But this lapse, because merely empty, cannot contain the determining conditions for any special occurrence in preference to others. This is why the conception of a beginning of the causal series in time, with an empty lapse before it, has always led to the insoluble riddle, “Why did God create the world when He did, rather than at some other point of time?”

Nor can the difficulty be escaped by taking refuge in the continuity of the stream of events. For (1) as we have seen, the recognition of events as continuous processes necessarily leads to the surrender of the causal principle as inadequate to express the real connection of facts. Causality, as a special form of the category of Ground and Consequence, must stand or fall with the view of occurrences as sequences of discontinuous events. And (2) even apart from this consideration, the appeal to continuity can at best only be worked as a rejoinder to the internal analysis of a sequence into an infinite process. When it is urged that, on the causal principle, there must be an infinite number of intermediate stages between the cause and the effect in any given case, it is possible to retort that the stages are not “really” distinct, but only distinguished by an artificial abstraction, that the process is actually one and continuous, and therefore does not involve an infinite regress, except for the logician who erroneously construes it as discontinuous. But with the external regress in indefinitum you cannot deal in this way. The absence of a beginning follows as necessarily from the principle of explaining the later stages of a continuous process as conditioned by the earlier, as it does when the stages are taken to be distinct events. (This is easily seen from the simple consideration that the time-variable in your formula for the successive stages of the process may have an unlimited range of possible values from -[infinity] to +[infinity].)

In short, whether the succession of events be taken as continuous or not, the attempt to translate the axiom that whatever happens has its ground in the nature of the whole system to which it belongs, into the doctrine that the posterior in time is completely determined by and dependent on the prior, leads straight to the infinite regress. And, as we said in our last chapter, the occurrence of the infinite regress is always a sign that there is imperfection somewhere in the thought which sets it up. For it always implies the formal contradiction of the actual summation by successive increments of an infinite series. Further considerations on this point may be deferred till we come to treat of the continuity of time.[^[112]]

§ 9 (c). The Plurality of Causes. The indefinite regress may be shown to be inherent in Causation by a different line of argument, without appealing to the principle of the continuity of time. As the reader is doubtless aware, it was a favourite doctrine of John Stuart Mill, that whereas the same cause is always followed by the same effect—in the absence of counteracting circumstances—the same effect need not be preceded by the same cause. An effect may be “produced” on different occasions by entirely different sets of antecedents. Thus death may be due either to disease or to violence, and both the disease and the violence may have very different forms, yet the result is the same, namely, death. Heat may ensue from friction, percussion, chemical combination, and so forth. This doctrine of the Plurality of Causes is an obvious result of generalisation from the important practical consideration that different means will often lead us to the same end, so that where we cannot employ one we can often fall back on another.

Mill’s critics have not failed to point out that his doctrine is based on the rather illogical combination of a concrete cause with an abstract effect. He considers the “effect” in its utmost generality simply as a state or quality, e.g., “heat,” “death,” and rightly contends that this general state or quality may issue on different occasions from different combinations of conditions. But he fails to observe that in any concrete case this effect exists in a special form, and with special modifications corresponding to the special character of the antecedents. Death, for instance, may result from a thousand circumstances, but the total effect in each case is never mere death, but death in some one special shape. A man who is shot and a man who is drowned are both dead, but one is dead with the special symptoms of death by drowning, the other with those of death by shooting. The water will kill you and a bullet will kill you, but death with a bullet-hole does not come from drowning, nor death with one’s lungs filled with water from a gunshot. If you take cause and effect at the same level of concreteness, they are always strictly correlative. Any variation in the one must have a corresponding variation in the other, for circumstances which vary without affecting a result are by definition no part of its conditions.