Maverick Philosopher

1. The Issue

Which of the following is true? (A) Although events within the universe have explanations, the universe itself has no explanation; its existence is a brute fact. (B) The universe has a transcendent explanation in terms of a being external to it such as God. (C) The universe is self-explanatory: it has an immanent explanation, one in terms of itself. My view is that (C) is false: the universe cannot be self-explanatory. Here I go part of the way towards establishing the falsity of (C). If I am right, the universe either has no explanation, or has an explanation in terms of a transcendent being Although not every explanation is causal, by ‘explanation’ I here mean causal explanation.

To make things difficult for the theist, let us assume that (i) the universe, contrary to Big Bang cosmology, had no beginning in time, i.e., has an infinite past, and that (ii) the universe is not an individual in its own right, but a collection of individuals. Thus the universe is not something above and beyond its members, any more than a university is something above and beyond the sum-total of its buildings, roads, and playing fields, etc. By ‘universe’ I mean the totality of all concrete objects, whether physical or mental, except God (if He exists). It follows that abstracta (numbers, Fregean propositions, etc.) are not part of the universe. Since the universe can gain and lose members without prejudice to its identity, it is not plausibly thought of as a set or mereological sum; it is better thought of as a concrete aggregate. What exactly a concrete aggregate is, is part of the problem.

p. 2

I assume that the universe is modally contingent: it might not have existed. It exists in some but not all possible worlds, to employ Leibnizian imagery, and one of these is the actual world. If a thing is modally contingent, it does not follow that it is contingent upon something else, that it depends on something else for its existence. The qualifier ‘modally’ is meant to highlight this fact, but I will drop this qualifier from here on out to save breath. The contingency of the universe is not an inference from the contingency of its members — that would be the fallacy of composition. I merely posit as intuitively obvious the contingency of the universe. For if each member of the universe is contingent, it is difficult to see how the universe — which we are thinking of as a kind of collection — could be necessary. In any case, if the universe were necessary, it would not require an explanation. By the way, let us not confuse being necessary (existing in all metaphysically possible worlds) with being omnitemporal (existing at every time). For there is no reason to suppose that an omnitemporal universe might not have existed. These are clear and distinct, and clearly distinct, ideas: existing in every possible world vs. existing at every time in a given possible world.

I said that the universe exists in the actual world. This implies that the universe is distinct from the actual world. The universe is a concrete aggregate, but possible worlds, pace David Lewis, are abstracta, Fregean propositions on my view. A possible world is a maximal proposition, one that entails every proposition with which it is logically consistent. The actual world is the maximal proposition that is true. (The actual world cannot be the set of all true propositions, since one can prove using Cantor’s Theorem that there can be no set of all truths.)

Now David Hume, and more recently Paul Edwards, are proponents of (C). In his Dialogues on Natural Religion, IX, Hume argued against the cosmological argument for the existence of God roughly as follows: To explain the members of a collection is to explain the collection. So if the universe is a collection, an explanation of each member of the universe will suffice to explain the universe itself. Think of each member as a total state of the universe at a time. The Humean-Edwardian can say that each member is caused by (and is thus explainable in terms of) an earlier member. Since we are assuming, with Hume, that there is no first member, no initial state of the universe, EVERY member is caused and is thus explainable in principle. And since the universe is not something above and beyond its members, the fact that every member has a cause entails that the universe has a cause. The consequence is that the universe has an internal or immanent explanation. It follows, of course, that there is no need for an explanation in terms of a transcendent being. Since the universe is self-explanatory, causally self-contained, the cosmological argument collapses. Of course, in this context, ‘self-explanatory’ does not mean necessary: if Hume is right, the universe is a contingent being that explains itself. A fascinating idea — but can it be made to work?

2. A Putative Counterexample

Central to Hume’s argument is the premise that to explain the members of a collection is to explain the collection. But consider the components in a computer. Each component has its own manufacturer, reference to which explains its existence. But a computer is not just a number of components, but a number of components connected together in the right way. Therefore, to explain the computer, it does not suffice to explain each of its components; one must also explain their connectedness by invoking an assembler. So it cannot be true that every collection is such that it is wholly explained by an explanation of its members.

p. 3

Nevertheless, since many collections do satisfy this condition, the Humean may retreat to a restricted principle: to explain the members of a collection the unity of which is uncaused is to explain the collection. Consider the collection of Americans now in Ankara. To explain why each is here suffices to explain why the whole lot of them is here: there is no need for an additional explanation of why the collection is here.

This raises the question: Is the universe a whole the unity of whose parts is uncaused, or is it a whole the unity of whose parts requires a cause? To invoke an old Buddhist example, if you are given a bunch of chariot-parts, are you thereby given a chariot? No, for the reason that the parts can exist whether or not the chariot exists. A chariot is its parts PLUS their connectedness. (See Buddhist Studies to read more about chariots and where they can take us.) But whereas the parts of a chariot can exist whether or not the chariot exists, it is difficult to see how the parts of the universe could exist without the universe existing. In this maximal case, the existence of the parts is eo ipso the existence of the whole. The parts of the universe don’t need to be assembled: they are ‘automatically’ assembled. Or at least this is what I will be assuming so as to give Hume and Co. the benefit of the doubt. In other words, I dismiss the putative counterexample to Hume’s principle.

3. Hume’s Objection to the CA is Inconsistent with Hume’s Own Theory of Causation

Our question is why the universe exists. Does it exist without cause as a matter of brute fact? Does it exist because a transcendent being caused and causes it to exist? The Humean-Edwardian answer is that the universe exists because each of its members is caused to exist by a temporally preceding member, and the universe is just the collection of these members. This answer is a good one only if causation is production, i.e., only if causation is causation-of-existence. Only then will an immanent causal explanation of the existence of the universe be possible. On Hume’s regularity theory of causation, however, causation is not causation-of-existence. Thus what I will now argue is that Hume’s objection to the cosmological argument in the Dialogues on Natural Religion is inconsistent with the regularity theory of causation standardly imputed to him on the basis of passages in the Treatise and in the Enquiry. A regularity theorist cannot make use of Hume’s objection to the cosmological argument.

On the regularity theory, for any individual events x and y, x causes y if and only if (i) x and y are spatiotemporally contiguous; (ii) x temporally precedes y; and (iii) x and y are tokens of event-types X and Y that are “constantly conjoined.” Note that it is only event-types, and not event-tokens, that can be said to be constantly conjoined, and that this is just to say that it is an exceptionless de facto generalization that every instance (token) of X is followed by an instance (token) of Y. It would make no sense to say that a particular blowing is “constantly conjoined” with a particular extinguishing of a candle-flame. It is only the corresponding event-types that can be said to be “constantly conjoined.” The gist of the regularity theory is that what makes an event sequence a causal sequence is its instantiation of a regularity. Thus no event sequence is intrinsically causal: it is causal only in virtue of its instantiation of a regularity. This implies that a particular event-sequence's being causal depends on what happens elsewhere in the universe — something that ought to strike one as counterintuitive. Thus the regularity theory implies that a possible world in which there are only two events is a world in which neither can cause the other, and this for the reason that in such a world there are no regularities: the very sense of ‘regularity’ implies the multiple instantiation of event-types. (More on this later.) The point is not merely that causation is regular — same (type of) cause, same (type of) effect — but that regularity, or “constant conjunction,” is what constitutes causation.

p. 4

It seems clear that on this theory of causation, no event produces, brings about, or causes-to-exist any other event. For at the level of individual events, all there is is contiguous succession of Humean “distinct existences.” All there is is one event following another, with no real-world connection between them. The fact that there is “constant conjunction” or instantiation of a regularity does nothing to introduce production at the level of individual events. No individual event produces, brings about, or causes-to-exist any other individual event. Events simply fit a pattern: X-type events are, as a matter of contingent fact, always followed by Y-type events. But for individual events to fit a pattern, or instantiate a regularity, they must already (logically speaking) exist. It follows that one cannot account causally for the existence of an event by saying that it instantiates a regularity. To attempt such an account would be to move in an explanatory circle of embarrassingly short diameter. Events must exist if they are to instantiate regularities; hence, a causal explanation of why they exist cannot invoke their instantiation of regularities. At most, the regularity theory provides a criterion for distinguishing causal from non-causal event sequences given that there are events; it is useless when it comes to explaining why there are any events in the first place.

Since the regularity theory does not allow for genuine causal production, it cannot be used to explain the very existence (occurrence) of any event or state. Consequently, it cannot be used to explain the existence of the universe as the collection of all events. The upshot is that the regularity theory of causation is inconsistent with the thesis that animates Hume’s objection to the Cosmological Argument, namely, the thesis that the universe has an internal explanation of its existence. Hume’s theory of causation does not analyze our ordinary notion of causation as production, or causation-of-existence; it replaces our ordinary notion with a watered-down notion that conforms to empiricist strictures. For this very reason, Hume’s regularity theory of causation is useless when it comes to explaining the existence of the universe.

It is a further question whether there is any extant theory of event-causation that could be of use to those who seek an immanent causal explanation of the universe. I do not believe that there is — but that is a topic for another occasion.

4. The Regularity Theory is False In Any Case

The regularity theory is not only useless for purposes of rebutting the cosmological argument, it is demonstrably untenable in its own right. Here are four arguments.

p. 5

1. Consider a particular causal sequence S in the actual world. S might be the sequence consisting of the striking of a match followed by its lighting. Obviously, there is a very large finite number n of sequences of S’s type in the actual world. Now consider a series of merely possible worlds. The first such world, W(n-1), is the possible world that results from subtracting one of the sequences of S’s type (other than S) from the actual world, and making the required adjustments. The second such world, W(n-2), is the world that results from subtracting two of the sequences of S’s type (other than S) from the actual world, and making the required adjustments. And so on until we arrive at W(1), which is the merely possible world in which S is the only event sequence of S’s type. Since S is a causal sequence in the actual world, then surely it is also a causal sequence in W(n-1). One fewer sequence of S’s type can have no bearing on whether or not S is a causal sequence. And if S is a causal sequence in W(n-1), then surely it is also one in W(n-2), and so on. Iterating this procedure, we conclude that S is a causal sequence in W(1) even though S is the sole instance of its type in W(1). But if S is a causal sequence in W(1), then it is possible that there be a causal sequence that does not instantiate a regularity. This implies that the regularity theory is false: it cannot be instantiation of a regularity that makes an event sequence a causal sequence.

I have heard it said that in W(1), S instantiates a single-case regularity. But a single-case regularity is surely no regularity at all. It is no more a regularity than negative growth is growth. (‘Single-case’ is an alienans adjective.) If there were single-case regularities, then every event sequence would be a causal sequence: every event sequence could be claimed to instantiate a regularity, namely the single-case regularity consisting of x being contiguously followed by y. The whole problem with regularity theories is that they imply that e1's causing of e2 is not an ‘internal matter’ pertaining to that event sequence, but depends on what goes on elsewhere: only if there are other events of e1's and e2's types such that every event of the first type is contiguously followed by any event of the second type is e1 a cause of e2. But this is totally counterintuitive. One event’s causing another has nothing to do with anything going on elsewhere. Why should a hydrogen leak on a space shuttle depend on hydrogen leaks on earth? Of course, if e1 causes e2, then presumably every event of e1's type must be contiguously followed by an event of e2's type — but this is not to say that there are any events of e1's type (other than e1) or any events of e2's type (other than e2). I am not denying that causation is regular, that the same (types of) causes bring about the same (types of) events; I am saying that an event sequence’s being causal cannot depend on there being other event sequences of the same type.

2. At the time of the Big Bang, some 15 billion years ago, when according to current cosmology space-time-matter first exploded into existence, there were no regularities: regularities take time to show themselves. There were, however, causal processes, for example, those that brought the universe from the first instant of the Big Bang to the state it was in one nanosecond later. So if the regularity theory were true, we would have to conclude that events near the Big Bang were uncaused. I think it is better to conclude that the regularity theory is false.

3. It is plausible to think of life as an ‘emergent phenomenon.’ That is, there was a time at which biological events and processes did not yet exist, and a later time at which they began to exist. Presumably, the beginning to exist – the ‘emergence’ – of the first biological events was caused by events that were not biological but merely physical or chemical. If so, this is another case of causation without regularity, and hence a counterexample to the regularity theory of causation.

p. 6

4. So far, every instance of match-striking (with sufficient force, in an oxygen-rich environment, etc.) has been followed by an instance of match-ignition. But if this is merely an Humean regularity, i.e., an exceptionless pattern of experience not underpinned by anything (e.g., not underpinned by, and explainable in terms of, a relation between universals as on the Dretske-Tooley-Armstrong theory), then the course of future experience may well overturn the regularity. Suppose regularity R is refuted by experience at some future time. Suppose, for example, that tomorrow I strike a match (of the usual type, in the usual way, in the usual circumstances) and it does not light, but turns to stone instead. Perhaps every match struck after tomorrow turns to stone. Either way, R is refuted. But then no event sequence that instantiates R, whether the sequence be past, present, or future, is a causal sequence. That is, yesterday’s striking-ignition sequences were not causal, since the regularity whose instantiation was supposed to confer causality upon them is not in fact an exceptionless regularity. To put it another way, whether or not a given present causal sequence is indeed causal depends on the future course of experience –- which is surely absurd. My present lighting of my pipe either is or is not a causal process, and whichever it is cannot possibly depends on what happens in the future. The causality of a given present causal sequence is as independent of future events as it is independent of other present events, past events, and events in other possible worlds.

I am assuming that for a regularity to be exceptionless, it must admit of no exception at any time, past, present, or future. If you deny this, you will have to admit the possibility of a regularity which is merely a regularity-at-a-time (or over a short interval of time). Suppose at t1 all strikings are followed by ignitions, at t2 all strikings are followed by petrifications, and at t3 all strikings are again followed by ignitions. Even though there is an exceptionless regularity at t2, presumably no one will say that striking causes petrification.

The regularity theory thus issues in a very bizarre consequence, namely, that the very existence of causal sequences depends on the future course of experience. For on Humean principles it is possible that the future course of experience overturn every single regularity that has hitherto manifested itself.

5. Diagnosis

A complete refutation must not only show that the theory under consideration is false; it must also explain how the theory arose. A complete refutation must account for the etiology of error. In the case of the regularity theory, I conjecture that it arises through a confusion of epistemology with metaphysics, specifically, a confusion of the question as to how we come to recognize or detect causal sequences and distinguish them from noncausal ones, with the question as to what causation is. Perhaps the only way to know whether a sequence is causal is by seeing whether or not it instantiates a regularity. If so, no regularity, no knowledge of causation. But it would be a non sequitur to conclude: no regularity, no causation. That would be to confuse the knowledge of causation with causation. Analogy: One can determine whether a substance is an acid or a base by seeing whether it turns red litmus paper blue or blue litmus paper red. But it would be absurd to say that a substance’s being an acid or a base consists in its changing the color of litmus paper.

I tend to think that Hume is right that we cannot empirically detect causal production and causal necessitation. (This is not obvious: Harre and Madden in Causal Powers (1975) claim that one can literally see, e.g., an earthquake destroy a building.) But it does not follow that there is no causal production or causal necessitation. They might well be empirically undetectable.

Associated with E1 are perhaps three concrete physical particulars, BV, the board, and the stump. Associated with E2 is the table, a concrete physical particular. And yet neither BV nor the board nor the stump nor any combination of the aforementioned pushes or otherwise acts upon the table -- and this for the simple reason that there is no table to be acted upon independently of the operation of placing the board in the stump. If X pushes or otherwise acts upon Y, then both X and Y exist, and exist independently of the pushing or acting. The table, however, does not exist independently of the operation of placing the board on the stump.

This is a counterexample to the claim embodied in the above quotation. For I have just given an example of event-causation in which all of the associated concrete particulars are physical but in which the particulars associated with the cause-event do not push or otherwise act upon the particular associated with the effect-event.

How does this bear upon our debate about the probative force of the traditional objection to interactionist substance dualism? I agree with Spur about one thing. If one argues validly from P1 and P2 to C, then, in general, one cannot say that the argument begs the question against a denier of C at either P1 or P2. No doubt, as we say in the trade, "One man's Modus Ponens in another's Modus Tollens." But that is a merely formal-logical point. Dialectically, it may or may not be effective depending on how epistemically solid P1 and P2 are.

So if the claim quoted above were epistemically rock-solid (either self-evident, or universally accepted by philosophers or something like that), then my claim that the traditional objection begs the question against the interactionist dualist would be dialectically inept. But the claim quoted above is not epistemically rock-solid. The counterexample just given refutes it. And there are others.

Note that it does not matter what response Spur makes to my counterexample. It may be quite clever, but we have now got a controversy going, and that suffices to show that his major premise above is dubious. I have a good independent reason to reject it, a reason independent of the philosophy of mind.

So I persist in my claim that the traditional objection to interactionist substance dualism, as I explained this objection here, is probatively worthless. It is little more than a canard, endlessly repeated, copied from one book to the next, that begs the question against the interactionist. This is not to say that one couldn't beef it up with a detailed and comprehensive theory of causation. But then one will have moved far beyond the traditional objection. Let Spur develop his theory of causation and defend it against all competitors. If he convinces me that it is the best theory, then I will have a good reason to reject ISD.

For what is the essence of occasionalism if not the view that no natural or 'secondary' cause is a true cause, and that there is only one true cause, namely, God the 'primary' cause? Although occasionalism is often discussed as an ad hoc solution to the problem of mind-body interaction 'occasioned' by Cartesian dualism, occasionalism is a general theory of causation, and there was occasionalism long before Descartes. The doctrine can be traced back at least to al-Ghazali. Here is a good little article by William Hasker that fills in some of the details.

It is easy to see why occasionalism would be found attractive by Muslim thinkers. For the occasionalist, God is omnipotent not only in the sense that he can do anything (subject to certain logical and non-logical restrictions that I have discussed elsewhere); he is also omnipotent in the far stronger sense that anything that gets done gets done by God. He has not only the power to do all; he has and exercises all the power.

Of course, Hume does not bring God into his theory of causation. But for Hume, as for the occasionalist, no natural cause is a true cause since no such cause produces, or brings into existence, its effect. The relata of the causal relation are events, and the latter are Humean "distinct existences." To say that e1 causes e2 is just to say that (i) e1 temporally precedes e2; (ii) e1 and e2 are spatiotemporally contiguous, and that (iii) whenever an event of e1's type occurs, an event of e2's type occurs. That's all she wrote. There is no producing, no bringing about. Good empiricist that he is, Hume denies what he does not see. More precisely: he doubts the existence of that for which there is no corroboration in sensory impressions. Hume operates with a Principle of Significance according to which the validation of any idea requires tracing it back to sensory impressions; but such ideas as causal power, causal production, and causal necessity cannot be cashed out in terms of sense impressions.

These three conditions in Hume's empiricist analysis can be satisfied by events between which there is no real-world connection such as a transfer of energy or momentum. On the Humean scheme, all there is in the world is regular succession. This is tanatamount to saying that e1's occurrence is merely the occasion, not the true (productive) cause, of e2's occurrence.

Now common sense and Miss Anscombe both insist that the root idea of causation is that the cause is the source of its effect, that "Effects derive from, arise out of, come of, their causes." (Anscombe in Sosa and Tooley, eds. p. 92) Now if you accept Anscombe's point, and I do, then it seems accurate to say that Hume's regularity theory of causation is occasionalism without God. (David Lewis and others have noticed that there is also a counterfactual analysis of causation in Hume's texts; but let's not worry about that now.)

Surely there must be more to causation 'in the objects' than regular succession. If so, then perhaps theists would be within their epistemic rights in putting God to work in the theory of causation. As I once heard J. N. Findlay say, "God has his uses."

BV: To put it a bit more precisely: if the pairing problem (PP) can arise in a purely physical situation, then EITHER the PP constitutes an objection to mental-physical causation if and only if it constitutes an objection to physical-physical causation, OR the PP shows that the theory of causation that gives rise to the PP is false. Either way, the PP does not amount to a valid objection to ISD.

In summary (with small change in notation): We have experiment X1 with cause C1: bulb B1 is on for t seconds, and effect E1: temperature rises by d degrees; similarly experiment X2 with cause C2: bulb B2 is on for t seconds, and effect E2: temperature rises by d degrees. We can run X1 and X2 repeatedly and get reproducible results. We then run experiment X* with cause C*: both B1 and B2 on for t seconds simultaneously, and effect E*: temperature rises by 2d degrees. We view C* as a 'sum' of causes C1+C2 and E* as a 'sum' of effects E1+E2, so that C1+C2 ---> E1+E2. PP then asks do we still have C1 ---> E1 or could we have C1 ---> E2.

I think there are two problems with this. The first relates to the sum of the effects and the second to the sum of the causes.

Firstly, to what extent can we discern E1 and E2 within E* in order to ask the PP question? I don't see how we can. We can't examine an air molecule and ask how much of its extra energy came from B1 and how much from B2, as if B1 emitted red energy and B2 green. (Kinetic) energy is just motion and can't be 'labelled' in the required way. We can picture molecule M1 gaining energy by interacting with a photon from bulb B1, and likewise M2 from B2, but when M1 subsequently collides with M2, it makes no sense to ask how much of its 'B1 energy' is transfered to M2 and vice versa. So E1 and E2 are just not discernible within E* and we can't formulate PP.

BV: I agree that the effects E1 and E2 are not discernible within the total effect E*. But how does it follow that we cannot ask the PP question? Your words 'discern' and 'discernible' and 'label' have an epistemological flavor: they pertain to how we know or come to have reason to believe that such-and-such is the case. But in reality -- 'ontologically' if you will -- C1 has a definite effect, namely, E1, and C2 has a definite effect, namely E2. So I can ask the Pairing Question: what accounts for the fact that C1 pairs with E1 and C2 with E2 and not rather C1 with E2 and C2 with E1?

Now the causes and the effects all occur within the same spatial volume. So spatial contiguity cannot be used to distinguish the C1-E1 sequence from the C2-E2 sequence. Temporal contiguity and temporal precedence cannot be used to distinguish them either. Furthermore, both sequences instantiate the same law of nature. Yet C1 pairs with E1 and C2 pairs with E2. Since this pairing cannot be accounted for on the relevant theory of causation -- according to which an event-sequence is causal in virtue of its instantiation of a causal law -- this theory of causation is to be rejected. If we reject it, however, the PP fails to amount to an objection either to mental-physical or to physical-physical causation.

Let me try a simpler example. A bucket is supplied with ball bearings from two tubes T1 and T2. The ball bearings are all of the same size, composition, etc. T1 supplies only enough bearings to fill the bucket halfway. The same goes for T2. Suppose both tubes are emitting ball bearings starting at the same time, and at the same rate, and over the same interval of time. C1 = emission of n bearings from T1. C2 = emission of n bearings from T2. E1 = filling of bucket with n bearings from set S1 of bearings. E2 = filling of bucket with n bearings from set S2 of bearings.

Can one tell which bearings came from which emitter? No, but that is a merely epistemological point. Surely half of the bearings came from one emitter and the other half from the other. (Compare: if you have an indiscernible twin, and I can't tell you two apart, it doesn't follow that that there is no fact of the matter about who is who. If one of you must be executed, you will want it to be your twin, and he will want it to be you.)

So we have distinct causes and distinct effects, with C1 paired with E1 and C2 with E2. But can we deduce this pairing from the theory of causation in question? No, so the theory is defective, and the Pairing 'Problem' is not a problem.

I think issues concerning summation of causes and summation of effects are red herrings -- though it is my fault for not making this clear in my original post. What is relevant here is the possibility of purely physical set-ups in which the right causal pairings cannot be established by invoking spatial differences -- because there are none. Since summation is not the issue, I don't think your second point about summation of causes is relevant. So I won't respond to it now. In any case, brevity is the soul of blog and I have already gone on too long.

1. Assuming that causality is a relation (not entirely obvious!), the question arises as to what sorts of entity can serve as its relata. Following Schopenhauer, whom he cites, Curt Ducasse holds that in strict propriety only events can be causes and effects. An event is either a change or an absence of a change. Thus a tree's losing its leaves is an event, but a tree is not. In strict propriety, it makes no sense to say that Bill was killed by by a mountain lion. One has to say something like: Bill was killed by the attack of a mountain lion. In the attack the lion is the agent as Bill is the patient, but the latter is no more the effect than the former is the cause. The cause is the lion's attack, the effect is Bill's death. Some theorists distinguish between agent-causation and event-causation, but for Ducasse, there is no such thing as agent-causation: causation just is event-causation.

2. For Ducasse, events are not punctual but processual. No change or 'unchange' occurs at a time, but only over an interval of time, or at a point (or other cut of space), but only through a space between cuts. Thus causes and effects are processes.

3. Let C and K be particular changes, change-tokens rather than -types. If the occurrence of C suffices for the occurrence of K, then C causes K. Thus for Ducasse the task of defining causation is the same as that of defining this non-logical sufficiency. C causes K if:

A. The change C occurred during a time and through a space terminating at the instant I at the surface S. B. The change K occurred during a time and through a space beginning at the instant I at the surface S. C. No change other than C occurred during the time and through the space of C, and no change other than K during the time and and through the space of K. (127)

In brief, C causes K iff C alone occurs in the immediate environment of K immediately before.

4. Example. A rock collides with a window, breaking it. This is a case of causation if anything is. The cause is not the rock, but the rock's colliding with the window; the effect is not the window, but the window's breaking. The change in the rock's position is immediately followed by a change in the window's integrity, and the change in the rock's position is the only change in the immediate environment of the window immediately before the change in the window's integrity.

Think of it this way. (Cf. Sosa and Tooley's introduction, p. 18.) The immediate environment of the window is a continuous spatiotemporal volume. Slice through this volume on a plane perpendicular to the time-axis. Make the cut at the moment of collision. You now have two spatiotemporally contiguous sub-volumes. The processes, states, and events in the earlier sub-volume constitute the cause C, while the items in the later sub-volume constitute the effect K. C's causing K is nothing other than K's contiguously following C.

5. This is close to David Hume's theory, but without constant conjunction. There is spatiotemporal contiguity and temporal precedence. Now is the causal relation observable? Hume looked hard at his billiard balls but discerned no causing, no producing, no necessitating when the cue ball slammed into the object ball. He saw "distinct existences" but no causal connection. On Ducasse's view, however, Hume made the mistake of assuming that the connection, to be anything, must be a third entity, a third "distinct existence" separate from the cause and the effect. Ducasse: "Hume's view that no connection between a cause and its effect is objectively observable would be correct only under the assumption that a 'connection' is an entity of the same sort as the terms themselves between which it holds, that is, for Hume and his followers, a sense impression." (131)

Ducasse goes on to point out that even if there were a distinct empirically discernible link between cause and effect, this would not solve the problem but raise another: what connects this connector to what it connects? For Ducasse, the causal connection is not a sensory impression but a relation, the relation analyzed above. Moreover, it is an observable relation. "We observe it whenever we perceive that a certain change is the only one to have taken place immediately before, in the immediate environment of another." (132) Thus when I see a lightning bolt hit a tree and the tree burst into flame, what I am seeing is the causal relation.

Ducasse's view seems very close to this: If you can perceive change, you can perceive causation. To put it more precisely: If, in a circumscribed environment, you perceive one change contiguously following another, then you are perceiving causation. One change contiguously following another in an environment is itself a change. So it seems fair to say that, for Ducasse, the perception of causation is just the perception of change. Ducasse himself, in footnote 11, points out

. . . an essential connection between the two notions of Change and of Causation. For, taking any given change process, by specifying a space-time cut of it, one splits it into a cause and an effect; and, on the other hand, taking any given cause and its effect, by abstracting from the particular space-time cut in terms of which as common limit the cause process is distinguished from the effect process, one obtains a process describable as one change. This calls to mind Kant's very inadequately argued contention in the Second Analogy, that (objective) change involves the category of causation. (129)

6. Ducasse's theory raises many questions. Here is one. Ducasse's theory seems to imply that every case of change is a case of causation. But consider a moving shadow cast by a moving car. Since the shadow is moving, it is changing in respect of position as it travels from A to B. Now we just heard Ducasse say that "any given change process" can be split into a cause and an effect. So make a cut in the spatiotemporal volume mid-way between A and B, and call the earlier process the cause and the later the effect. But it is clear that the earlier process is not the cause of the later process. A moving shadow is not a causal process but a pseudo-process. So here it seems there is a case where there is more to causation than change. The shadow's being in a given position at a given time is not the cause of its being in a different position at a later time.

1. I want to set aside the Kantian question as to whether or not the concept of cause can be applied to the world of contingent beings as a whole. Kant famously answered that question in the negative. My concern is with a different question: Does it make sense to suppose that individual existing things, whether taken distributively or collectively, have a cause of their existence? Or is it rather the case that only states of, and changes in, individual existents can be terms of the causal relation?

2. Schopenhauer is one philosopher who maintains that only changes can serve as causal relata. His idea is that the causal relation holds between changes (states, conditions, properties...) of individuals, but cannot be extended to individuals themselves. Thus if a rock changes in respect of temperature, say going from cold to hot, one can legitimately ask for the cause of this change. But one cannot legitimately ask for a cause of the rock itself, or a cause of the existence of the rock. (See below for a refinement of this thesis.) Nor can one speak of a rock as a cause. It is not a rock that breaks a window, but a collision of a rock with a window that causes the window to break.

Causation is always and everywhere causation of change and never causation of existence. One can legitimately ask for a cause of a change in something that exists, but not for a cause of the existence of something that exists. If this is right, then cosmological arguments cannot succeed. For such arguments aim to demonstrate a causa prima, a First Cause, of the very existence of what (contingently) exists. Accordingly, the “law of causality” is to be formulated as “Every change has its cause in another change immediately preceding it.” (World as Will and Representation II, 42) It is not to be formulated as “All that is, has its cause.” (41) For “the mere existence of a thing does not entitle us to conclude that it has a cause.” (42) Since we cannot legitimately demand a causal explanation of the sheer being of anything, we cannot make this demand of the world; hence the cosmological argument, as presupposing the legitimacy of this demand, rests on a false presupposition. One illegitimately extends the concept of causality from states of things to the things of which they are the states.

3. But why does Schopenhauer restrict causality to changes? By ‘change,’ Schopenhauer understands alteration (Veraenderung). Indeed, for him alteration is the only kind of change. It is not obvious that alteration should be the only kind of change since there is a prima facie distinction between alteration and what we might call ‘existential change,’ i.e., coming-into-existence and ceasing-to-exist. Is there not a tolerably clear sense in which coming-to-be and passing-away are changes? Think of the coming-into-being of a batch of my famous lasagne, and its passing out of being when I devour it.

It seems obvious that if there is existential change, then it cannot be understood in terms of alteration: before a thing exists it is simply not available to suffer any alteration, and likewise when it ceases to exist. Coming-to-be is not gain of a property, but gain of a thing together with all its properties; ceasing-to-be is not loss of a property, but loss of a thing together with all its properties. But, as will emerge in a moment, Schopenhauer denies the very possibility of existential change with resepct to the ultimate substrata of alterational change.

4. Alteration may be defined by saying that x alters if and only if x has incompatible properties at different times. Change as alteration is always change in something which must remain the same through the change and which therefore, as the substrate of change, does not itself change in respect of its existence and identity. In a slogan, no change without unchange! The dynamism of change is erected upon a substructure of (relative) stasis. Without the latter there cannot be the former.

Thus there is alterational change only if the rock that was cold at t is numerically-existentially the same as the rock that is hot at t* later than t. Two rocks, one cold at t, the other hot at t*, do not a change make. For a thing to become different it must remain the same. If you deny this, then you are embracing a doctrine of Heraclitean flux. No doubt the world is 'fluxed up' but it is not that 'fluxed up.'

With respect to alteration, then, the slogan is ‘No change without an unchanging substrate of change.’ Schopenhauer’s point could then be put by saying that the substrates of change lie outside the causal nexus. They are presupposed by the causal nexus. As such, they are neither causes nor effects, and so they are not appropriate objects of causal explanation. One cannot meaningfully ask why there are substrates of change rather than none at all. And if this cannot be meaningfully asked, then “God caused them to exist” cannot constitute a meaningful answer.

5. For Schopenhauer, then, the substrates of change lie outside the nexus of cause and effect as its presupposition. But this thesis clearly needs some refining. A rock may be a substrate of change, as when it goes from cold to hot in the morning sun, but it is not an ultimate substrate. For surely there is a sense in which one can reasonably ask: Why does this rock exist? One can reasonably ask this for the simple reason that it has a reasonable answer: “It was caused to exist by the congealing of lava.”

So we need to distinguish between ultimate and non-ultimate substrates of change. A meatloaf is a non-ultimate substrate of alterational change: straight from the oven it is hot, later it is cool. Clearly, its coming-into-existence is not a change in it. But this existential change can be construed as an alterational change in some other thing or things, namely, its ingredients. They are alterationally changed by being combined and cooked with the result that something distinct from them, but composed of them, comes into being. Now if all change is alteration, then ultimate substrates of change cannot come into existence or pass away. And since these ultimate substrates are the basis upon which rests all causation, there can be no legitimate question concerning the cause of their existence.

6. Note that ‘This rock exists’ can be read in two ways. We can take it as an instantiation-claim, to wit, ‘The properties characteristic of rocks are instantiated here in this thing.’ But we can also interpret it as an existence-claim proper, namely, ‘This thing, which happens to instantiate the properties characteristic of rocks, exists.’ This duality of readings stems from the well-known ambiguity of ‘exists’ as between a second-level and a first-level use. If I say of a property that it exists, I am saying that it is instantiated; if I say of an individual that it exists, I am saying that it -- exists.

Accordingly, one question is: Why are rock-properties instantiated in this thing? And then the answer might be: because the volcanic lava of which this rock was formed cooled off, hardened, etc. The other question is: Why does this thing that has rock-properties exist? Clearly the first question, the instantiation-question, is legitimate. For this question asks why rock-properties as opposed to say lava-properties are instantiated in the thing, or perhaps in the space-time region before us. But the second question, for Schopenhauer, is not. For what the second question asks is why this property-bearer before us exists as opposed to not existing at all. But this question makes no sense on Schopenhauer’s assumption that causes and effects are changes (alterations). The sheer existence of an ultimate property-bearer cannot be a change from a previous state of nonexistence because every change is an alteration, and every alteration requires a substrate of alteration. But it is clear that a thing’s coming into existence cannot be an alteration of that very thing for the simple reason that before it exists it is not available to suffer any alterations. And the same holds at the other end: when a thing ceases to exist, it does not merely lose a property; it loses its very existence together with all its properties. It follows that if all change is alteration, then the ultimate substrata of change cannot come into existence or pass out of existence.

7. Thus we may impute to Schopenhauer something like the following argument:

1. The relata of the causal relation are changes.
2. Every change is an alteration.
3. Every alteration presupposes a substrate of alteration that does not change in respect of its existence and identity.
4. Some, but not all, substrates of alteration are non-ultimate: there must be ultimate substrates of alteration.
5. If there are ultimate substrates of alteration, then they do not come into existence or pass away.
6. Because changes occur, there are ultimate substrates of alteration.
7. There are ultimate substrates of change that do not come into existence or pass away, and are thus sempiternal.
8. The existence of these sempiternal substrata are at the basis of all causation.
Therefore
9. The existence of the ultimate substrata of change cannot have a cause, divine or otherwise.
Therefore
10. Cosmological arguments, presupposing as they do that causation of existence makes sense, are one and all unsound.

8. What can we say in critique of this argument? The crucial premise may well be (2). If every change is an alteration, then nothing can come into existence except by the tranformation of some thing or things already in existence. That is to say: there cannot be creation ex nihilo. But then Schopenhauer's argument may be said to beg the question against the theist at line (2) -- unless there is some independent way of supporting (2).

This is a good objection. I was a bit sloppy in my exposition. I stated the Regularity Theory in the way I did because I wanted to stay close to Hume, or at least Hume on one standard interpretation. That explains the spatiotemporal contiguity and temporal precedence clauses. These are relevant for the billiards balls cases that exercised Hume. His main concern was physical-physical macrocausation. But they drop out in the mental-physical case, or at least the spatiotemporal contiguity clause drops out. This is something I failed to make clear. Setting aside the historical Hume (or one interpretation of the historical Hume), the essence of the Regularity Theory is that one event causes another just in case the pair instantiates a regularity. Causation just is regular succession. This is hard to swallow, of course, but if this were all there is to causation, then there would not need to be any physical contact between cause and effect, or any transmission of a physical magnitude from cause to effect. There would be no need for a locus of interaction. If so, Sober-type objections (col)lapse.

Sober asked: "But if it [the mind] lacks spatial location, how can it be causally connected to the body?" This question presupposes that there is a locus of interaction, a place where the mental event makes physical contact with a physical event or the beginning of a physical process. But if causation just is instantiation of a regularity, then there is no need for a locus of interaction.

The main point of the post was that the sort of objection Sober makes, an objection endlessly and thoughtlessly repeated, is of value only if accompanied by a theory of causation according to which physical contact/transfer are essential to causation. But it is not clear that they are essential to causation. One cannot just assume that causation requires physical contact/transfer.

Smith assumes that if the universe has an internal cause, in virtue of each of its state’s having an internal cause, then it cannot also have an external cause on pain of causal overdetermination. But this isn’t obvious, even if we acquiesce in the Humean principle that to explain a collection it suffices to explain its members. A theist could grant that the universe is such that each state has an internal cause while also maintaining that it has an external cause. There are theories that allow this without overdetermination.

Suppose our theist adopts a regularity theory of causation. On such a theory, what makes an event-sequence causal is its instantiation of a regularity. As I have argued in mind-numbing detail elsewhere, ("The Hume-Edwards Objection to the Cosmological Argument," Journal of Philosophical Research, vol. XXII (1997), 423-443) on no such theory does the cause produce or bring into existence, the effect. For on a regularity theory, what you have out in the world is simply contiguous succession, just one event following another, and thus no producing of any event by any other. On this theory, what makes e₁ the cause of e₂ is simply the fact that e₁ and e₂ are tokens of event-types F and G which are related by a certain defeasible generalization, namely, ‘Whenever an F-type event occurs, it is followed by a G-type event.’

In other words, what the regularity theory does is provide a way of distinguishing causal from noncausal event-sequences -- given that there are events and sequences thereof. Thus the theory does not account for, but presupposes, the existence of events. Other theories do likewise. Counterfactual theories of causation, for example, analyze 'c causes e' in terms of 'Had c not occurred, e would not have occured.' There is no talk here of any producing, or bring-about, or causing-to-exist of e by c. Thus these two theories presuppose the existence of events and do no more than give us a way of sorting event-sequences into two classes, the causal and the noncausal.

Thus no theory having the above existence-presuppositional feature can yield an account of the existence of events or a fortiori of the existence of the universe. So a theist who is a regularity theorist could hold that while every state of the universe has a causal explanation in terms of earlier events, the existence of these states and thus of the universe is explained by God’s creative activity. There is no causal overdetermination here since although God is a cause of existence, Humean causes are not causes of existence.

Thus if Smith is to make good his claim that a self-explanatory universe excludes an external cause, he must specify a theory of natural causation according to which causes bring into existence their effects. It is invalid to infer straightaway that the universe cannot have an external cause just because each of its states has an internal cause. For it may be that these internal causes are not productive causes. But if he adds to his premises a theory of natural causation that precludes the possibility of an external cause, then he begs the question against the theist, who of course will adopt a theory of natural causation that does not preclude the possibility of an external cause.

Suppose my cat Zeno starts a ball rolling across the floor by hitting it with his paw. The rolling is a causal process with continuum-many instantaneous states, and this rolling can be viewed as having no first state. For one may reckon the moment of the ball’s transition from rest to motion as the last moment of the ball’s being at rest, which then entails, given the continuity of time, that there is no first state of the ball’s being in motion. Arguably, we must view the matter in this way: otherwise the ball would suddenly be moving at some positive velocity without having accelerated to this velocity from velocity zero through continuum-many instants.

And yet there is an external cause of motion, Zeno’s pushing the ball. Somehow the external cause has to ‘hook onto’ the rolling‑ball process. How? The last instantaneous state of the pushing cannot hook onto the first state of the rolling‑ball process, for there is no first state. And of course the last pushing state cannot hook onto some state of motion after the rolling has begun.

This suggests that the last state of the pushing connects to an entire causal sequence, namely the entire sequence of rolling-ball states. Thus a whole sequence would somehow be the effect of a cause. If so, the sequence or succession would not be logically supervenient upon its members, an "ontological free lunch," to borrow a charming phrase from D. M. Armstrong, but something in addition to its members. We would thus have a counterexample to Hume’s principle that a collection is explained when its members are. If something like this could be worked out in detail – clearly not a task for the present occasion! – then the way would be clear for the theist to argue that the entire universe is a sequence of states that is the effect of a cause.

The argument might go like this:

6. Ordinary causal processes within the universe (e.g., rolling balls) are not causally self‑explanatory: they have external causes.

7. The external cause of a continuous process cannot 'operate upon' any member (any instantaneous state) thereof, since each state has a sufficient cause in an earlier state or states within the process, and there cannot be causal overdetermination of any given state.

Therefore,

8. The external cause must 'operate upon' the entire causal process or sequence, which implies that causal processes have causal properties above and beyond those of their members. This undercuts the Humean principle that to explain a collection it suffices to explain its members.

Therefore,

9. The Humean principle gives us no reason to think that the universe cannot have an external cause. Note that this so whether the universe is finitely or infinitely old. For what counterexamples like the rolling ball process show is that there are infinite collections which cannot be explained by explaining their members.

Now,

10. The universe U is a finitely old continuous causal process with no first member.

Therefore,

11. If the rolling ball process has an external cause, but U does not, then there must be something that distinguishes U -- the universal causal process --from intramundane causal processes.

12. That U is a whole of which intramundane processes are proper parts does not constitute a difference that could explain why U, but no proper part of U, is caused but has no external cause.

13. There is no other difference betweeen U and intramundane processes that could explain why U, but no proper part of U, is caused but has no external cause.

Therefore,

14. If U has a cause, then U has an external cause.

1. Causation is a relation between concrete particulars.
2. Relations are existence-entailing: if a relation obtains, its relata exist.
3. Presentism is true: the present alone exists.
4. Most causes temporally precede their effects.

To say that a set of propositions is inconsistent is to say that the members of the set cannot all be true, where 'cannot' expresses logical impossibility. Clearly, the above quartet is inconsistent. To see this, note that if (3) and (4) are both true, then past causes do not exist. But if so, then either causation is not a relation between concrete particulars, or relations are not existence-entailing.

Let's not forget the context of the discussion. I claimed that a picture of X taken when X exists continues to represent X even after X is annihilated. Pictures of the philosopher Edith Stein such as the one below represent her despite the fact that she was murdered by the Nazis at Auschwitz in 1942. Those pictures did not cease to represent something, or come to represent something distinct from her, at the moment of her physical annihilation.

But at the moment of her physical annihilation there was nothing left for the photos to resemble. So representation cannot consist in resemblance. Resemblance is a relation, and relations are existence-entailing. Two things can't resemble unless they both exist. For if they resemble, then they resemble in respect of at least one property that both instantiate; but nothing can instantiate a property unless it exists, as I argued here. Representation, however, is not a relation, strictly speaking, since a representation can be of something that does not exist. The picture to the right is of Edith Stein's face despite the fact that her face no longer exists.

My argument presupposes the truth of (2) and (3). But if there are past causes of present events, then they will be counterexamples to (2) assuming the truth of (3).

To solve the problem posed by the above inconsistent tetrad, I propose we reject (1) by denying that singular causation is a relation between concrete particulars. What it is is a relation between propositions or proposition-like entities.

This will sound crazy to some. "Propositions are abstract, and the abstract is the causally inert; surely it is incoherent to say that singular causes and effects are causally inert!"

One cannot be pushed around by a proposition. But is it so obvious that the relata of the causal relation are pushers and shovers, twisters and benders? Jonathan Bennett (Events and Their Names, Hackett, 1988, p. 22) speaks of

. . . the false assumption that causal statements must report relations between shovers and forcers. I grant that facts [true propositions] cannot behave like elbows in the ribs, but we know what items do play that role — namely elbows. In our world the pushing and shoving and forcing are done by things — elementary particles and aggregates of them — and not by any relata of the causal relation.

Bennett bids us compare:

The vase broke because a heavy stone was dropped onto it.
The vase broke when a heavy stone sent shock waves through it.

The first sentence reports causation, while the second speaks of a pusher, an exerter of force, which is not a fact but a stone.

Glance back at the inconsistent tetrad. Consistency can be restored if we reject (1) and construe causation as a relation between propositions or proposition-like entities. Since propositions exist omnitemporally, their existence is consistent with the truth of presentism. For the same reason, they pose no threat to the principle that relations are existence-entailing. But if causes and effects are propositions, what would it mean to say that some causes precede their effects? Consider The vase breaks and A heavy stone is dropped onto it. To say that the latter precedes the former is to say that the latter is true before the former is true. A proposition cannot temporally precede or succeed another proposition; but one proposition's being true can precede another's being true.

Presentism is very attractive, but how can it accommodate the fact of causation? Consider the following argument:

1. If x is the cause of y, then both x and y exist. 2. Some temporally present items are caused by items that are not temporally present. Therefore 3. Some items exist that are not present.

Suppose the barking of the dog startles the cat. Both events must exist if they are to stand in the causal relation. (This is a special case of the proposition that the obtaining of any genuine relation entails the existence of its relata. Intentionality is not a relation.)But the barking is temporally prior to the being startled. So the barking exists even though it is not present — which contradicts presentism.

John Bigelow ("Presentism and Properties," Phil. Perspectives, 1996)deals with this objection by taking a page from the Stoics. He maintains that the relata of the causal relation are not events one temporally prior to the other, but presently true propositions. Thus, it is now true that the dog barked (past tense) and now true that the cat is alarmed (present tense). If this suggestion can be made to work, then we can reject premise (2) of the above argument. For then both cause and effect exist at present.

Unfortunately, there is a problem with Bigelow's suggestion. The past-tense proposition that the dog barked, being both singular and contingent, is in dire need of a truthmaker. There is need of an ontological ground of its truth. The truthmaker principle states (roughly) that true truth-bearers (subject to certain restrictions) require worldly truthmakers. I hope the principle is intuitive; in any case, I won't explain it further here.

Now if presentism is true, i.e., if only temporally present items exist, and The dog barked needs a truthmaker, the question arises as to what that truthmaker could be. A truthmaker obviously must exist to do its job, which implies that it must exist now, given presentism. Alan Rhoda in his paper Presentism, Truthmakers, and God tackles the question whether presentism is consistent with the truthmaker principle. What exercises him are questions about statements about the past. Since many of them require truthmakers, how is that possible given presentism? His solution, roughly, is that the truthmakers of statements about the past are divine memories.

My question for Rhoda is this. How does he understand causation? Is it necessary to bring in God for an adequate account of causation in the natural world? If yes, is there a drift towards occasionalism?

2. The materialist has more than one option, but a very tempting one is to reduce reference to causation. The idea is roughly that the referent of a word is what causes its tokening. Thus 'cat'-tokenings refer to cats because cats cause these tokenings. Suppose I walk into your house and see a cat. I say: "I see you have cat." My use of 'cat' on this occasion -- my tokening of the word-type -- refers to cats because the critter in front of me causes my 'cat'-tokening.

3. Unfortunately, this is only a crude gesture in the direction of a theory. For it is obvious, as Hilary Putnam points out (Renewing Philosophy, p. 37), that pictures of cats cause 'cat'-tokenings, but pictures of cats are not cats. Cats are typically furry and cover their feces. Pictures, however, are rarely furry and never cover their (nonexistent) feces. Yet I might say, seeing a picture of a cat over your fireplace, "I see you like cats." In that sentence, 'cat' is used to refer to cats even though what caused my 'cat'-tokening was not a cat but a non-cat, namely, a picture of a cat.

Indeed, practically anything can cause a 'cat'-tokening: cat feces, cat hair, cat caterwauling. Suppose I see Ann Coulter engaged in a 'cat-fight' with Susan Estrich on one of the shout shows. I say, "They are fighting like cats." Nota Bene: my use of 'cat' in this sentence is literal, but it is not caused by a cat! (My example, not Putnam's).

4. Our problem is this: what determines the reference of a word like 'cat'? What makes this bit of language represent something extralinguistic? 'Cat' is a linguistic item; a cat is not. (French 'philosophers' take note.) In virtue of what does the former target the latter? If both a cat and a pile of cat scat can cause a 'cat'-tokening, then the causal theory of reference is worthless unless it can exclude these extraneous (excremental?) cases.

The problem is that 'cat' refers to something quite specific, cats, whereas 'cat'-tokenings can be caused by practically anything. To solve this problem, a notion of causation must be invoked that is also quite selective. It turns out, however, that this selective notion of causation presupposes intentionality, and so cannot be used in a noncircular account of intentionality. Let me explain.

4. Context-Sensitive versus Context-Independent Concepts of Causation. We often in ordinary English speak of 'the cause' of some event, a myocardial infarction say, even though there are many contributing factors: bad diet, lack of exercise, hypertension, cigarette smoking, high stress job, an episode of snow-shovelling. Which of these will be adjudged 'the cause' is context- and interest-relative. A physician who gets a kick-back from a pharmaceutical concern will point to hypertension, perhaps, so that he can prescribe a massive dose of angiotensin-converting enzyme inhibitor, while the man's wife might say that it was the snow shovelling that did him in.

To take a more extreme example, suppose a man dies in a fire while in bed. The salient cause might be determined to be smoking in bed. No one will say that the flammability of the bedsheets and other room furnishings is the cause of the man's incineration. Nevertheless, had the room and its furnishings not been flammable, the fire would not have occurred! The flammability is not merely a logical, but also a causal, condition of the fire. It is part of the total cause, but no one will consider it salient. The word is from the Latin salire to leap, whence our word 'sally' as when I sally forth to do battle at a chess tournament. A salient cause, then, is one that jumps out at you, grabbing you by your epistemic shorthairs as it were, as opposed to being a mere background condition.

Putnam cites the example of the pressure cooker that exploded (p. 48). No one will say that it was a lack of holes in the pressure cooker's vessel that caused the explosion. A stuck safety valve caused the explosion. Nevertheless, had the vessel been perforated, the contraption would not have exploded!

What these examples show is that there is an ordinary-language use of 'cause' which is context-sensitive and interest-relative and (if I may) point-of-view-ish. A wholly objective view of nature, a Nagelian view from nowhere, would not be able to discriminate the salient from the nonsalient in matters causal. In terms of fundamental physics, the whole state of the world at time t determines its state at subsequent times. At this level, a short-circuit and the current's being on are equally causal in respect of the effect of a fire. Our saying that the short-circuit caused the fire, not the current's being on, simply advertises the fact that for us the latter is the normal and desired state of things, the state we have an interest in maintaining, and that the former is the opposite.

The ordinary notion of cause, then, resting as it does on our interests and desires, presupposes intentional notions. I cannot be interested in or desire something unless I am conscious of it. And I cannot adjudge one state of affairs as normal and the other as abnormal unless I have interets and desires.

5. Recall what the problem was. The materialist needs to explain reference in physicalist terms. He thinks to do so by invoking physical causation. The idea is that the referents of a word W cause W-tokenings. The reference of a word is determined by the causal influence of the word's referents. So it must be cats, and not pictures of cats, or the past behavior or English speakers that causes 'cat'-tokenings. But surely the past behavior of English speakers is part of the total cause of present 'cat'-tokenings. (See Putnam, pp. 48-49.) 'Cat' does not refer to this behavior, however. To exclude the behavior as non-salient requires use of the ordinary interest-relative notion of causation. But this notion, we have seen, presupposes intentionality.

6. The upshot is that the above causal account of representation -- which is close to a theory proposed by Jerry Fodor -- is viciously circular: it presupposes the very notion that it is supposed to be reductively accounting for, namely, intentionality.

I suggested that the essence of idolatry lies in the illicit absolutizing of the relative. A finite good becomes an idol when it is treated as if it were an infinite good, i.e., one capable of satisfying our infinite desire. That our desire is infinite is shown by the fact that it is never satisfied by any finite object or series of finite objects. Not even an infinite series of finite objects could satisfy it since what we really want is not an endless series of finite satisfactions -- say a different black-eyed virgin every night as in popular Islam's depiction of paradise — but a satisfaction in which one could finally rest. "Our hearts are restless until they rest in Thee." (Augustine) What we really want, though we don't know it, is the absolute good which is goodness itself, namely God. This idea is common to Plato, Augustine, Malebranche, and Simone Weil.

Ultimately, all desire is desire for the Absolute. A desire that understood itself would understand this. But our de-luded desire thinks it can find satisfaction in the finite. Therein lies the root of idolatry. The Buddha understood this very well: he saw that desire is infinite in that it desires its own ultimate quenching or extinguishing, its own nibbana, but that finite quenchings are unsatisfactory in that they only exacerbate desire by giving birth to new desires endlessly. No desire is finally sated; each is reborn in a later desire. Thus the enjoyment of fleshpot A does not put an end to lust; the next night or the next morning you are hot for fleshpot B, and so on, back to A or on to C, D, . . . and around and around on the wheel of Samsara. The more you dive into the flesh looking for the ultimate satisfaction, the more frustrated you become. You are looking for Love in all the wrong places.

So Buddha understood the nature of desire as infinite. But since he had convinced himself that there is no Absolute, no Atman, nothing possessing self-nature, he made a drastic move: he preached salvation through the extirpation of desire itself. Desire itself is at the root of suffering, dukkha, not desire for the wrong objects; so the way to salvation is not via redirection of desire upon the right Object, but via an uprooting of desire itself.

In Buddhist terms, we could say that idolatry is the treating of something that is anatta, devoid of self-nature, as if it were atta, possessive of self-nature. Idolatry arises when some finite foreground object, a man or a woman say, is falsely ascribed the power to provide ultimate satisfaction. This sort of delusion is betrayed in practically every love song ever written. Here are some typical lyrics (trivia question: name the song, the singer, the date):

You are my world, you're every move I make
You are my world, you're every breath I take.

There are thousands more lyrics like them, and anyone who has been in love knows that they capture the peculiar madness of the lover, the delectable madness of taking the finite for infinite.

Or will you deny that this is madness, a very deep philosophical and perhaps also religious mistake? I say it is madness whether or not an absolute good exists. Whether or not it exists, reason suggests that we should love the finite as finite, that our love should be ordered to, and commensurate with, its object. Finite love for finite objects, and for all objects if there is no infinite Object.

Now what does this have to do with causation? Malebranche's idea, as I understand it, thanks in part to Brandon Watson's fine post, is that our belief in the efficacy of secondary causes is rooted in idolatry: we idolatrously ascribe to secondary causes productive power when sober Humean analysis -- to put it anachronistically -- reveals nothing but regular succession and neither production nor necessitation. The flame does not produce ignition in the dry paper; what is given is a spatiotemporally contiguous event-sequence that instantiates a regularity. Secondary causes are mere occasional causes, occasions of the exercise of the causality of the only true productive cause, God.

To put God to work in the theory of causation seems wildly extravagant; but it seems less crazy when one faces up to the difficulties involved in productive natural causation. And if sense can be made of occasionalism as a general theory of causation -- as opposed to an ad hoc solution to the interaction problem faced by the substance dualist -- then, as Brandon points out, there emerges a moral dimension to the theory of causation. And that's a surprising result.

Churchland apparently thinks that a substance, to be 'substantial,' must be material. Churchland thereby betrays his inability to conceive of (which is not the same as to imagine) an immaterial substance. (Note that this expression is not an oxymoron like 'immaterial matter.') But let that pass. The issue for now is simply this: How can two things belonging to radically disjoint ontological categories be in causal contact? But here again, Churchland seems to be laboring under a false assumption, namely, that causation must involve contact between cause and effect. But why should we think that this 'billiards ball' model of causation fits every type of causation? Why must we think of causation as itself a physical process whereby a physical magnitude such as energy is transferred from one physical object to another? I discussed this briefly in a previous post and suggested that on a regularity or counterfactual theory of causation there is no difficulty in principle with the notion of a causal relation obtaining between two events that do not make physical contact.

But today I have different fish to fry. I want to see if the materialist's conception of mind-body interaction is free of difficulty. If the materialist view of mind were without diffuclty I would be a materialist, on grounds of parsimony.

Suppose that every token mental state is identical to some token brain state. (Token as opposed to type: If you don't undersdtand this distinction, I will explain it.) This token-token identity will be held both by type-type identity theorists as well as by functionalists who reject type-type identity. Accordingly, my pain sensation after touching a hot stove is strictly identical to (not merely correlated with) a state of my brain. If so, there would seem to be no difficulty in principle with this state's being brought about by other physical states. For then physical-mental causation would just be a special case of physical-physical causation — which we assume to be unproblematic. At first glance, then, it appears that physicalism does not face the sort of interaction objection that substance dualism faces. To put it graphically, there is no 'ontological chasm' to 'jump over.' The mind-body problem gets traded in for the brain-body problem which presents no special philosophical difficulties, as difficult as it may be to work out the scientific details.

But why does the brain state which is (identically!) my pain sensation cause aversive behavior such as the withdrawal of my hand from the stove? Because it has the property of being painful. It is the painfullness of the sensation that is causally relevant. Why did I remove my hand from the stove? Because touching the stove hurt. A causal explanation of my aversive behavior (the yanking back of the hand) cannot merely invoke a brain state such as the firing of C-fibers (or whatever a completed neuoscience would specify): it must make reference to a brain state that has the property of being painful. This is a qualitative property or a quale to use a piece of phlosophical jargon. It is the peculiar first-person feel of the experience, the what-it-is-like of it. This raw feel (to us an old expression of Herbert Feigl) has what Searle calls a "first person ontology": its esse is its percipi. One cannot distinguish its appearing from its being.

Now here is the problem: if the pain sensation is identical to a brain event, then there is no place for the felt painfulness. For the brain event has a third person mode of existence and among the properties of this brain event you will not find anything that has a first person mode of existence. A brain event is purely physical and has only physical properties.

But even if you insist that the phenomenal property of painfulness -- the sensory quale -- is a property of the brain state, this phenomenal property surely plays no causal role in bringing about the aversive behavior. For it is only the physical aspects of the brain state that play a causal role in bringing about the the other physical states that comprise aversive behavior.

The materialist seems to be no better off than the substance dualist. Both face the problem of interaction since common sense strongly suggests that mind acts on body, and body acts on mind. (Parallelists and occasionalists deny this, but they move quite a distance from common sense.) The problem for the dualist is the problem of bridging the gap between two disjoint ontological categories, while the problem for the materialist is explaining the causal relevance of a pain sensation, say, to other physical events when the very painfulness which accounts for the aversive behavior cannot play any causal role if the pain sensation is identical to a brain state.

Although the materialist avoids the dualists' gap problem, he faces a different problem which is just as bad: the problem of explaining how a brain state can bring about aversive behavior when the property of this brain state that ought to be causally relevant, namely, the phenomenal painfulness of it, cannot be causally relevant.

In sum, the materialist solves the gap problem by construing mental-physical interaction as physical-physical interaction, but he can do this only by rendering the very phenomenal properties that first give rise to the mind-body problem causally inefficacious.

Why then the double standard? Why is the substance dualist taxed with a problem of interaction, when the materialist faces an equally difficult problem? My conjecture is that there is an ideological bias in favor of materialism that is at the root of this double standard.

Recommended Reading: Kenneth Einar Himma, "What is a Problem for All is a Problem for None: Substance Dualism, Physicalism, and the Mind-Body Problem," Amer. Phil. Quart. (April 2005), 81-92.

But suppose you are a latter-day Cartesian: you are both a substance dualist and an interactionist: you believe that mind and body are distinct (kinds of) substances and you also believe that some mental events cause physical events and some physical events cause mental events. You will be taxed by some with a supposedly insurmountable difficulty: how can there be mind-body and body-mind causation if mind and body are radically different kinds of substance?

There are benighted souls who think this objection is decisive against Cartesian dualism. They are mistaken.

To show that the objection is not decisive it suffices to set forth a theory of causation that would allow mental events to cause physical events (and vice versa) even if a mental event is construed as an irreducibly mental substance's instantiation at a time of a mental property and a physical event is construed as an irreducibly physical substance's instantiation at a time of a physical property.

Well, a regularity theory of causation would do the trick, would it not? Suppose we say that:

Event-token e1 causes event-token e2 if and only if (i) e1 temporally precedes e2, and (ii) e1 and e2 are tokens of event-types E1 and E2 respectively such that every tokening of E1 is followed by a tokening of E2.

On this Hume-inspired theory (sans the contiguity condition), causation is just regular succession. If this is the correct theory of causation, then there is nothing problematic about mental events causing physical events, and vice versa.

Of course, if you think that causation must involve the transfer of some physical quantity such as energy or momentum, then of course substance-dualist interactionism is out. But there is nothing in the very idea of substance-dualist interactionism to render it incoherent. It all depends on how causation is understood.

Suppose one adopted a counterfactual analysis along the lines of: c causes e =df had c not occurred, e would not have occurred. There is nothing here to rule out substance-dualist interactionism.

BV: Thanks for writing, Dennis. I have been thinking about you in connection with your beloved Pollo alla Marsala. Later, I am going to explain how you can prepare it for yourself so as to end some of the suffering induced by the culinary backwater in which you presently reside. But I digress.

The underlying facts you mention are not in dispute. But I am not sure I understand your argument. If S1 pushes S2 off a building, but S2 survives, then S1 has not caused the death of S2. What you must mean is that if S1 pushes and S2 dies, then S1 caused the death of S2 regardless of the fact that in similar circumstances not every 'pushee' dies. Now I don't deny that there is causation here at the level of event-tokens. But I should think that this causation at the token level is a species of necessitation.

So let's consider the question at the type level. Then your claim is that the event-type pushing causes the event-type dying (subject to all relevant caveats and provisos) even if a tokening of the first event-type merely raises the probability (to a degree <1) of a tokening of the second event-type.

Now what you are saying makes sense and is in accord with the usage of many philosophers nowadays. But for me, causation is a species of necessitation, and the 'action' occurs at the level of tokens, not types. I think you will agree that types as such don't do any causing. It is only the tokening of a type that can serve as a relatum of a causal relation. Now either token t₁ necessitates token t₂ or it does not. It seems to me to make no sense to say that one event-token raises the probability of the occurrence of a second event-token.

But to pursue this further, I would have to hear what your theory of causation is. As you know, there are several theories on offer. To lay one of my cards on the table: I reject regularity theories.

There is also the point that the type smoking and the type lung cancer (or else the disjunctive type: lung cancer or emphysema or . . .) are rather unlike the types mentioned in your two examples. Almost no one survives a ten story fall. (Are you aware of even one documentable case?) But plenty of people smoke all their lives without getting lung cancer. In my family alone, there are perhaps a half dozen examples. One uncle smoked unfiltered Camels.

2. As you might have guessed from my earlier email about liberals and conservatives, I’m not overly happy about the paragraph starting with “[l]iberals and leftists engage in this loose talk for at least two reasons.” Can’t liberals believe as they do because it seems to them, on reflection, the right thing to do, whether or not they can perform a power grab in the process? And the claim that their “loose talk” “aids them in their denial of individual responsibility” seems to me to rest on an implicit false dilemma: surely a liberal can believe in individual responsibility and believe that there are various social structures that are harmful and ought to be censured or restricted in various ways.

BV: That's a good objection, Dennis, and I take it seriously. Let me first say that I write several different kinds of posts. Some are pure philosophy posts. These purport to be rein wissenschaftlich. Others are of a political and social nature, and in them I allow myself to engage in a bit of polemic (from the Gr. polemos, war, strife.) It is a tough question whether there can or even should be pure, polemic-free political philosophy. I am currently thinking about this in connection with study of Carl Schmitt and Leo Strauss, but I have just begun to seriously engage this issue and have no fixed view. In any case, I believe (rightly or wrongly) that the Left must be defeated. If I am right, then some amount of polemic may be not only permissible, but also obligatory.

Did I qualify 'liberals and leftists' with 'many'? I'll have to look. Certainly many libs and lefties are motivated by a desire for power and will say anything to grab it and hold it.

You and I will agree that a theory or policy cannot be refuted by tracing it back to the psychological motivations of its proponents. Theories and policies must be evaluated on their own merits. But once I have refuted a view, then I consider myself justified in explaining how someone could hold such a (false, incoherent, unsupported . . .) view. Why, for example, do MANY Dems oppose SS reform? Their arguments are bad and they have no proposals of their own. (Barbara Boxer would be a good example.) It seems clear to me that they oppose Bush type reforms because they want to retain their power over people, the power that comes from controlling how much money people receive in retirement benefits. If someone maintains a position that obviously benefits him in terms of money, power, status, yet has either no arguments for the position, or demonstrably worthless ones, then it is entirely legitimate to 'psychologize' such a person.

But the psychologizing must come after their arguments have been refuted, not before.

Regarding the false dilemma, you want to say that a liberal can uphold individual responsibility while also holding that certain "social structures" ought to be "censured or restricted." Well, any sane person will agree with that. For example, adults ought to be free to purchase and consume alcoholic beverages, but no one wants them to be made available to children. But you may not be appreciating my point: the liberal tendency is to deny individual responsibility and to displace it onto factors external to the agent. That's not to say that everyone who calls himself a liberal does this. Alan Colmes calls himself a liberal but I'm sure he would come out in favor of individual responsibility. We are talking types and tendencies here. We have gone entirely too far in the liberal direction, as it seems to me, and now for the good of the country we need to make a correction and move back in the conservative direction. I wouldn't be calling myself a conservative if it were not for the extremism of contemporary liberals, an extremism from which I wish to dissociate myself. In some ways, I myself am a classical liberal.

Thanks again for your civil comments, and feel free to counterrespond.