Charles Hartshorne "Creativity and the Deductive Logic of Causality" (1973)

From The Review of Metaphysics, 27:1, September 1973, 62-74.

Anthony Flood

July 29, 2009

Creativity and the Deductive Logic of Causality

Charles Hartshorne

A number of writers, sometimes called “process philosophers,” have held that be-coming is both creative and cumulative—“creative” meaning that the new in any moment of process was not wholly contained in or strictly implied or entailed by the previous moments, the initial conditions, and “cumulative” meaning that these conditions are in some sense contained in or strictly implied by the new. Each instance of becom-ing is a “creative synthesis” of the previous instances. Causality, on this view, is one-way inclusion or entailment. Bergson’s snowball image of duration neatly expresses the idea. The rolling ball acquires new layers while retaining the old ones. Or, to use a more literally applicable formula, an adult remembering (mostly with very indistinct consciousness) his childhood presupposes and somehow implies his childish experiences, though the child was in no comparable sense aware, or implicative, of the adult life of that individual.

It is less obvious but, according to Whitehead or to my “neoclassical” version of process philosophy, no less true that even in perception it is past, not absolutely simultane-ous, events which are given.

It seems not to have been noticed that deductive logic can show how, from the creative-cumulative view, it is intelligible that, although exact and unqualified prediction of future events is in principle impossible, still much about the future is predictable. Partial though not complete predictability, I shall show, is an entailment of the creative-cumulative view. This is the logical structure (not made explicit by him) of Whitehead’s “answer to Hume,” which I regard as the only answer that need have impressed Hume very much.

It may be well to say here that I assume, against Hume and also Einstein, but with Peirce, Bergson, Boutroux, Dewey, Montague, Whitehead, Popper, and many physicists that the genuine causal laws are all approximate or statistical, not deterministic in the classical sense. I believe that the arguments for this view are very cogent. Moreover, I am confident that quantum uncertainty is not the only aspect of nature in application to which classical determinism is at best useless. I believe Wigner is justified in expecting that the laws of organic systems, especially on the higher animal levels, will turn out to be more complicated than those of inorganic systems. We shall, it seems likely, have to go farther from classical concepts than Heisenberg has done, rather than be able to relapse into New-tonian rationalism, as even Einstein wanted to do. Moreover, the “dice-throwing God” that Einstein could not believe in is the only one Peirce and some of the rest of us could or can believe in. I should also say that while we must take the theories of physics seriously, I do not think they should be considered neces-sarily definitive in so fundamental or philoso-phical a question as the structure of time or causality.

Assume, as the meaning of cumulativeness cum creativity, that absolute knowledge of an event would necessarily include or entail ab-solute knowledge of its causal antecedents—but not vice versa. Let K stand for absolute knowledge (partially explicable as the unattainable limit of better and better human knowledge) of events preceding event E, and let K’ stand for absolute knowledge of E itself. And suppose one had K but not K’. Then (because of creativity) K’ could not be inferred; but still (because of cumulativeness) something about K’ could be inferred, namely, it would have to entail K. This is significant, for not any and every proposition, or state of knowledge, can entail a given proposition or state of knowledge. Moreover, since K is here logically very “strong,” being the absolute description or knowledge of certain concrete actualities, the limitations imposed upon K’ are substantial.

It must be understood that K’ is not a mere conjunction of K with additional propositions or bits of knowledge. Otherwise the point just made above would be merely trivial. For, given any proposition q, and any other proposition p, the compound proposition p and q will entail q. This is ruled out as an interpretation of process philosophy by its doctrine, found in all representatives, that concrete reality consists of unit-events each of which is genuinely singular, and not a mere conjunction of many events, though it intrinsically refers to many. Thus an experi-ence in which various events of the past are remembered or perceived is not those past events plus some new features which themselves do not entail the events. Rather there is a felt qualitative unity such that the new quality intrinsically refers to the old qualities. “The many become one and are in-creased by one” (Whitehead). This is the point of “creative synthesis.” Peirce’s theory of Firstness implies a similar doctrine. So does Bergson’s theory of becoming or durée. To this point we shall return.

My suggestion now is: the causal limita-tions which a present situation imposes upon its future are nothing but those that are logically implied by the principle that every future situation must strictly entail the pre-sent situation as belonging to its past. Predictability is cumulativeness read back-wards. Thus my childhood, with its world, determined that either I, as adult, or else the world years later without me, would be the kind of person or the kind of world that could refer back to just that childhood and that previous world.

The key to this reverse inference is found in deductive logic. Suppose X says to Y: “I am thinking of a class of propositions, call them the p’s, each of which entails, but is not entailed by, the proposition q*, which is, ‘There is an animal in the room’. What can you infer from this about the p’s?” Y: “I infer that the p’s are all at least as specific or definite (logically strong) as q*. Indeed, since they are stipulated to be non-equivalent to q*, they must be more definite, as in ‘There is a small animal in the room’, or ‘There is a dog in the room’. Also any p must have a subject matter closely related to that of q*; it cannot compare to q* as would ‘There is a rose or a pitcher of water in the room’.”

Suppose, instead of q* we take a logically stronger proposition, e.g., “There is a small terrier in the room;” or even, “Our little terrier Fido is in the room.” In this case the p’s must be much more narrowly limited than in the other case. Thus: “Our little terrier Fido asleep (or scratching himself) is in the room.” Entailment without equivalence is always a matter of dropping, in conclusion, some part of the logical strength, the information, con-tained in premises. The one-way view of cau-sal necessity is the limiting and most concrete case of this, with the description of the later situation, the outcome or effect, being the logically stronger premise and the antecedent situation or cause the logically weaker conclu-sion. Becoming is enrichment of reality, add-ing definiteness but not subtracting any. The cause-effect relation is to be fully understood only from the standpoint of the effect, in which alone is the complete determinateness.

The traditional view was very different. The cause was the superior entity, or if not, cause and effect were “equal.” Effects were to be known in their causes, as well as causes in their effects. And so, before an event existed to be known it was nevertheless completely defined and ready to be known. (Or else it existed before it happened.) This upside-down idea (I here begin a brief excursus from my main theme) was neatly embodied in the standard medieval theological view that God knows the world simply by knowing himself as its cause. Thus an effect was held to be implied by, logically contained in, though inferior to its cause, the cause minus something, and then what is the point to causal production? The only value which production can have is to enrich reality. The cause-with-the-effect must be superior to the cause alone. Also, if causes do not annihilate themselves, effects with causes are the only effects there are. And surely God does not annihilate himself in producing the world, and so the world with its divine cause is more than the cause alone. Either this effect is superior to God, or it is God as enriched by the world he produces. Neither implication was acceptable to medieval thinkers.

Spinoza took the medieval scheme seri-ously and deduced the catastrophic conse-quence that the world must be in God, and in him taken merely as a necessary and eternal cause, i.e., the world in all details must be as necessary as God. Therewith, as Wolfson rightly says, Spinoza destroyed (one chief aspect of) “the medieval synthesis.” For two centuries and more philosophers and theolo-gians hesitated to try the remaining theistic possibility, which is to admit that effects as such are more than their causes, and hence supreme Reality must be the universal or inclusive contingent effect, as well as the universal or primordial necessary cause. It must be influenced by, as well as influential upon, all things. To avoid contradiction one must admit (as Whitehead and others have done) a real distinction in God between his nature as primordial and his nature as consequent, or as influenced by the creation. (End of theological excursus.)

In our logical training, attention to the “fallacy of affirming the consequent” (and deducing the antecedent) has been allowed to obscure the important truth that “A certain proposition q is entailed by every member of the class C of propositions” does tell us some-thing about this class, and moreover, that the more definite or logically strong q is the more we thus learn about the class of propositions entailing it. One important group of such classes of propositions is what is meant by “real possibilities” in the normal sense, i.e., possibilities for the future of some given present. Here q is very strong indeed. And this is one reason why the transition from cu-mulativeness to the possibility of implications for the future has been overlooked. Ordinary textbook cases of implication involve much weaker instances of q. Suppose q is, “I see something moving.” Then “I see our dog Fido moving” is one of a fantastically wide variety of possible p’s entailing q. Causal implication can start from an incomparably richer base than a mere visible moving something. So the reverse nonstrict inference to the class of p’s can be incomparably more definite.

A full account would have to consider the distinction between law-like statements and those specifying particular “initial conditions.” It would also have to ask what, in “inanimate” nature corresponds to memory and percep-tion. I also happen to believe further (here comes a second theological digression) that the adequate understanding of law-like state-ments requires the idea of a cosmic law-giver. But still, all this would come under K, the knowledge or description of the antecedent situation from which the result, the effect, is to come. Even God’s law-giving action must be antecedent to the effects it influences, not simultaneous with them (nor yet eternal). Thus the ultimate principle of “cause” or “condition” is univocal, even though there is also a “difference in principle” between divine and other causation. Thus God as causal influ-ence in every event will (in one aspect of his being) be antecedent to every event, and as influenced by every event will be subsequent to every event. He alone is primordial and he alone is everlasting. That causes are logically independent of effects as, in normal cases of entailment, conclusions are independent of premises, will be universally true. (End of second theological excursus.)

Obviously one source of confusion is the metaphor that conclusions “follow” from pre-mises. In our knowledge, drawing conclusions may temporally succeed positing of premises. But this is the reverse of the ontological relations. Thus “animal” follows logically from “human,” but animals were there long before human beings. Events in their concreteness are never known in antecedent events, whereas they are habitually known in subse-quent ones (more or less directly) by either memory or perception. The past is cognitively derived from the present by abstracting from that in the present which is novel. To realize today what yesterday was like we must drop out part of what we know about today’s experiences. The full logical strength is in the present; it was in no sense in the past. Thus have multitudes of philosophers been misled, partly at least, by a metaphor. Causal predic-tion is reading the deductive relations back-ward, from a conclusion to the class of pre-mises capable of entailing it. We are asking, in Whitehead’s metaphor, what sort of future could “house,” or accommodate what is now going on as its past?

The chief mistake concerning causation has been acceptance of the false dilemma: either deductive logic cannot throw light upon the causal relations of events, or each new event is to be viewed as a new premise for new entailed conclusions. On the contrary each event is primarily a new premise for old conclusions. Only in a weakened sense, and purely derivatively from this, is it a premise for new conclusions. This is not particularly paradoxical. Each new experience of a grown man is a new premise from which that man’s birth and childhood logically follows. Or, after the death, say, of Washington, each new stage of the writing of American history is a new premise from which the one-time exis-tence of Washington logically follows (assum-ing, as we all do, that Washington can be and is denoted). The reverse inference, from the life of Washington to the future historians and their writings about Washington, is a weaker one, and merely gives us a class of possible or probable historians writing about him. And while the adult Washington presupposes the child Washington, the latter might have acci-dentally died before growing up. To assert the contrary is to leap into a “dark” whose degree of darkness cannot be exaggerated.

It is true that the laws of physics, as now known, do not suffice to validate all aspects of the above scheme. But we should not forget, I think, that physics is the most abstract form of knowledge of nature. We know what it is like to be an animal, in a sense in which we can never know what it is like to be an atom (ra-ther than a man experimenting with atoms). I cannot believe that physics goes as deeply into the nature of things as biology and psy-chology. It may be more accurate and more comprehensive than they, but at the price of not knowing what sorts of things it is dealing with, apart from human experiences of the things. Do we know what radiant energy or electricity are? We know some structural (mathematically expressible) truths about them, and we are familiar with certain human experiences partly traceable to them as causes, but this is all.

To the contention of some philosophers that the question, What is such and such, beyond its structural-causal status, has no meaning, I reply, It has a clear but in some cases not very definitely answerable meaning, which is: How does such and such differ from or resemble various aspects of our human experiences, and this in qualitative as well as merely structural respects? We know what we are asking, but since qualities are not definitely knowable except by intuition, and our sharply definite intuitions are limited by the capacities of our sense organs, there is much that we cannot know about the qualities of reality.

Two mistakes in traditional reflections upon causality have been the arbitrary assumptions (a) that causal conditioning is symmetrical or bi-conditioning (events equally requiring their antecedent conditions and their subsequent results) and (b) that the way to understand effects is to consider what it means to be a cause. By (a) either creativity is wholly excluded (determinism) or else no strict cumulativeness is allowed (Mead). By (b) one is trying to understand causal deduction by asking how its conclusions imply its premises, thus taking the affirmation of the logical con-sequent as the primary deductive procedure! Not causes but effects are the premises, the logically stronger terms. The past is found in the present, not vice versa; and only because the past is in the present is the future also, though only partially, in the present, as that which, whatever else later becomes true of it, must contain the present as its past. When we know what it is to be an effect, then we can also, by logical principles, derive what it is to be a cause, for that is the simpler case. A cause defines a set of possible effects, a set which, though it may not yet have an actual member, yet is bound to acquire one.

As soon as we see that the key to causa-tion is in the status of being an effect of ante-cedent conditions, we are ready to see also that both memory and perception are, by their very meaning, just such effects. And then we see what Hume overlooked. He sought causal connections between things perceived and/or remembered, rather than between perceivings or rememberings themselves and their data. More precisely, he sought connections bet-ween “impressions,” taken either as not in-trinsically referring to anything impressing itself upon experience, or else as things doing the impressing, while the results of this action, the real impressions, are dismissed from con-sideration. It was a strange error, repeated on a grandiose scale by Russell—and how many others!

That it was an error seems clear. If we can know that we know anything, we must be able to know what it is to experience, that is, to perceive or remember. And the time to look for relations between entities perceived or remembered but which perhaps, at least apparently, do not perceive or remember, is after we have done real justice to the internal relational structures of perceiving and remem-bering themselves. It is in this sense, and this only, that epistemology is prior to ontology. Our experiences are our privileged samples of reality. Humean positivism “misses the boat” at this landing as neatly as boats can be missed. To employ an old joke, the boat is missed by two, or rather three, jumps. 1) Humeans look for, and announce their failure to find, a symmetrical causal necessity, a bi-conditioning, whereas we know from propositional logic that the biconditional (equivalence) is a derivative and weakened relation, not the basic meaning of implication. 2) They look for necessity in the wrong place, in mere objects rather than in experiences as such. 3) Insofar as they take causality as a one-way (rather than symmetrical or direc-tionless) requirement, they take the inference to be primarily from cause to effect. Thus, in nearly all possible ways they beg the question against the most characteristic constructive metaphysical systems of recent times—Berg-son, Peirce, Whitehead, also Montague, Par-ker, Hartshorne, and others. And some idealis-tic critics of Humeanism, e.g., Ewing, make some of the same mistakes. It is indeed a comedy of errors, or of failure in communi-cation. For the creative-cumulative view of becoming is clearly implicit in doctrines that were held long before Hume (e.g., by the Socinians). Lequier held such a doctrine a hundred years ago. It is much more widely held now. And partisans of Hume’s view that causal relations are mere constant conjunc-tions have yet to tell us how they know that no such doctrine as the cumulative can be true.

Three admissions are in order. The process philosopher needs some solution of Kant’s First Antinomy. He must either suppose a first moment of all becoming, or admit an infinite or indefinite number of past events. My own view is that an actual infinity of already elapsed events is a tenable idea, and that mathematical finitism taken to exclude this is not obligatory. The second admission concerns a perhaps even greater difficulty. Granted, some will ask, that the past, insofar as remembered or perceived, is implied by the present, how far can this implication go, considering that much of the past is, so far as we can empirically know, neither remembered nor perceived? Here we confront the question of unconscious or faintly conscious memories and perceptions, also the question of various levels of nonhuman memory and perception, including for some of us divine memory and perception. I hold that a theist is here, as in other fundamental questions, in an advanta-geous position, provided he avoids certain mistakes in the idea of God, for example the idea that God is universal cause but not uni-versal effect, or that he is in every sense im-mutable. I also hold that a psychicalist, for whom all singular events (the momentary state of a stone is not one stony event but a vast number of molecular, atomic, or particulate events) involve memory or per-ception in at least some minimal sense, has the advantage over a materialist or dualist.

I believe that the difficulties mentioned in the previous paragraph are to be taken seri-ously, but are not insuperable. In contrast, the difficulties of Hume’s view, as well as those of materialism and also “absolute idealism” (holding either that every event implicates every other, past, contemporary, or future, or that events and temporal relations are not real), seem to me incurable and sufficient to justify the rejection of these views. Some version of process philosophy seems therefore the only hope of understand-ing causality. It alone can show how events require their antecedents, from which, as I have tried to show, it is deducible that they entail an important part of the natures of their successors. More than this is not needed for the purposes of life; indeed, the more than this which is affirmed by determinism is pragma-tically meaningless. It is intrinsic to the meaning of prediction that it be less than absolute. It is the very point of foreknowledge that it be qualified to allow for elements of choice between possible outcomes of present situations. Only the myth of the knower as mere spectator, outside the world of action and peering into it, as it were, through a metaphysical window, could ever have made determinism seem a sensible idea.

The third admission is that there is certain-ly considerable obscurity in the requirement that the character of later events must not consist of a mere conjunction of that of past events plus some independent new character, so that the backward deduction would be of the form p and q entails q. Rather p, no less unitary (though more complex than q) entails q. I must admit that this unity is not to be explicated with the devices of deductive logic. It is, as Peirce and Whitehead agree, a unity of feeling. Its principles are aesthetic. The various data from the past must be embraced in a unity of feeling. If the data are not sufficiently homogeneous, no unity of feeling, no aesthetic harmony, can occur; if the data are excessively homogeneous, if there is insufficient contrast, aesthetic achievement will also be impossible. Aesthetic value is the mean between mere diversity and mere unity. Its twin evils are boredom and suffering. At the limit in either direction no experience can occur. A blow on the head may make human consciousness impossible because the requisite set of sufficiently unified but sufficiently contrasting data are no longer presented. What then happens, however, ac-cording to the process view, is not that there will be no aesthetic achievement at all in the bodily system but that it will no longer be on the human level but will be confined to that of the cells and their individual feelings. That the process of achieving unity on some level can and will always continue I regard (this is my third theological digression) as due to the immanence of divine power ordering the world with infallible wisdom and power. “Ordering” does not mean determining events in their details, for in every case the creatures must achieve their own syntheses of antecedent data, the directives from God being merely the supreme example of such data. Thus there is in this view no “problem of evil” in anything like the classical form. The world order is not a matter of detail, but of outlines. And it cannot be shown that the outlines are bad.

Some light may be thrown upon the unity requirement by considering concrete experi-ences. Suppose I remember having just offended a taxi driver by telling him rather sharply that he was not taking the most direct route, and remember his having responded somewhat angrily. The memory is too imme-diate and its content too vivid to be simply dismissed. There are various aesthetically possible ways of synthesizing this memory with memories coming from my more remote past and with (at the moment only normally vivid) sensory contents coming (as I hold, with Whitehead) from the immediate past of my bodily organs). I can try to decide whe-ther to apologize to the man, I can wonder why I spoke so sharply, and so on. We have only vague understanding of how past experi-ences and present sensory conditions limit the possible ways in which a tolerable new state of feeling can be achieved. But do we not have reason to avoid two extremes: on the one hand supposing that there are no limitations at all, and on the other, that there is but one univocal possibility exactly defining the state of feeling which in fact takes place? How can many, indeed a huge number, of conditions in millions or billions of cells, and in a similar number of past experiences of mine, all, in combination with any psychological laws you please, strictly define and dictate a single new unitary experience? And also, how can the definite set of data available from memory and perception for that moment of my experiencing fail to influence the experience which occurs. Between these extremes lies the limited creative freedom (it may be small) we are seeking.

My attempt to show by an analogy with de-ductive logic how there can be partial predic-tability of the future, assuming that earlier events are precisely implicated by their suc-cessors, will be misunderstood if it is taken to be one more effort to show that we can have certain knowledge of the future. I am willing to grant just about all the qualifications as to the possibility of induction that Karl Popper proposes. I am a fallibilist, with him and with Peirce. What I am trying to do is not to show how we can know, simply know, the future, but only how there can be truth about it. For if there is not truth about the future we cannot even err in our predictions. I hold that truth about the future while it is future can only be truth about those determining tendencies, or in Popper’s word, propensities, which obtain when an assertion about the future is made. And Hume was, or should have been, asking if there are such tendencies, such limitations upon the real or causal possibility which is the referent of statements about the future. Our knowledge is as limited as Popper says it is, or something very like that, but the question of truth (here also Popper and I agree) is not at all the same as the question of human knowledge. What makes it true that the grass is green is the grass being green, no matter what we know of its greenness. Here I agree with Tarski and with Popper, but this does not tell us how to construe the truth of “the grass will be green.” I take this to be made true by there being already a propensity, either in grass now existing or in something capable of producing grass, such that the grass being green later on is implied as the only still open possibility. A “will be” is a present causal situation implying the sort of future event specified. This is a theory about truth and reality, not about our knowledge.

It does follow from my view that ideally adequate knowledge of the past and present world would precisely delineate the propensi-ties, the implications, of the causal situation. But only a modern Protagorean would suppose that this tells us much about our human knowledge. We shall never have anything close to the ideal knowledge I am talking about. We have the kind of knowledge we need to make reasonable, responsible deci-sions. If the solar system blows up tomorrow that will not be our fault, and our inability to know for certain that it will not need not trouble us. We shall have had our day and done our bit. And if I am right, we have our chance to know what it is that prevents us from knowing more definitely than we do the real possibilities and probabilities that consti-tute the future as such. We lack precise quan-titative knowledge of the past, also of remote contemporaries who will influence our remote future, and we have only vague qualitative knowledge of either past or contemporary realities, including that decisive part of the past which is the already formed divine ideals and purposes for the world. But we have some glimmer of understanding of how this so partially known past and present can consti-tute truth about the future. We can in a general way know what we are missing. And for those who like such general understanding, that is, for philosophically inclined persons, this is a source of some satisfaction. With the vast scope of our modern knowledge, we should be able to bear the thought that the vastness of what we do not know is of a higher order of magnitude, if not in some genuine sense infinite. Where we do not know we can hope or trust. Presumably the sun will not blow up in our time.

Hartshorne page