Causation by Disconnection Jonathan Schaffer Philosophy of Science, Vol. 67, No. 2. (Jun., 2000), pp. 285-300. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28200006%2967%3A2%3C285%3ACBD%3E2.0.CO%3B2-Z Philosophy of Science is currently published by The University of Chicago Press. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/ucpress.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org. http://www.jstor.org Sat Sep 22 20:38:24 2007 http://links.jstor.org/sici?sici=0031-8248%28200006%2967%3A2%3C285%3ACBD%3E2.0.CO%3B2-Z http://www.jstor.org/about/terms.html http://www.jstor.org/journals/ucpress.html Causation by Disconnection' Jonathan Schaffert: Department of Philosophy. University of Houston The physical and/or intrinsic connection approach to causation has become prominent in the recent literature, with Salmon, Dowe, Menzies, and Armstrong among its leading proponents. I show that there is a type of causation, causation by disconnection, with no physical or intrinsic connection between cause and effect. Only Hume-style condi- tions approaches and hybrid conditions-connections approaches appear t o be able t o handle causation by disconnection. Some Hume-style, extrinsic, absence-relating, nec- essary andlor sufficient condition component of the causal relation proves to be needed. I. Introduction. It is widely believed that causation requires a co~znection from cause to effect, such as an energy flow. But there are many ways to wire a causal mechanism. One way is to have the cause connect to the effect, but another is to have the cause disconnect what was blocking the effect. The detonator button is pressed and the bomb explodes. Surely this is causation (this is not a trick case: no backup detonators or anything like that). If you do not already consider the causation here obvious, I offer the following intuition-buttressing considerations. Counterfactually, had the button not been pressed then the bomb would not have exploded. Statistically, bomb explosions will universally follow button pressings in such circumstances. The explanation for why the bomb exploded will surely include the button pressing. Knowing of the pressing will license a prediction of the explosion, and knowing of the explosion will license a retrodiction to the pressing. Button pressings in such circumstances con- stitute an effective strategy for explosion-hungry agents. The button 'Received November 1999; revised February 2000 +Send requests for reprints to the author, Department of Philosophy, University of Massachusetts, Bartlett Hall, Amherst, M A 01003. :Thanks to Phil Dowe, Jim Hawthorne, and David Lewis, and especially to Ned Hall for bringing my attention to this type of case. Pli~losopliy of Sclence, 67 (June 2000) pp. 285-300 0031-82481200016702-001l92.00 Copyright 2000 by tlie Philosophy of Science Ascociation. All rightc r e c e r ~ e d 286 JONATHAN SCHAFFER presser will be held morally responsible for the consequences of the explo- sion. Here all the central connotations of causation are in full force. But I have not said how the detonator is wired. Perhaps pressing the button generates an electrical current which connects to the bomb and makes it explode:' Button not pressed Button pressed pressing 0-X explosion pressing @-3 explosion Or perhaps pressing the button disconnects an electrical current that was inhibiting an independent source from triggering the explosion: Button not pressed Button pressed pressing OLpO explosion pressing BL7 @ explosion source '"ri"ibitOr a i"ibitO'source "/@ Either way is causation full force.2 See the pattern and you'll find it everywhere, even in the most paradig- matic causal sequences. The killer gets angry (C), pulls the trigger (Dl), fires a bullet through the victim's heart (D2), and the victim dies (E). Here is a paradigm of causation, witlz causation by disconnection at every step. Working backwards, surely the firing of the bullet through the victim's heart (D2) is a cause of his death (E). But heart piercings cause death only by disconnection. The brain is kept alive by an influx of oxygenated blood, and heart piercings cause death by disconnecting this influx, allowing ox- ygen starvation to run its course: 1. Diagram conventions: filled circles doubly represent neurons that fire and events that occur. unfilled circles doubly represent neurons that d o not fire and events that d o not occur, arrows represent stimulatory connections. and lines ending in squares represent inhibitory connections. If two neurons are connected by a stimulatory connection and the first fires, then the second will fire unless inhibited. 2. Terminological note: Since the pressing prevents the inhibitor from preventing the source, disconnections may also be called (following Ned Hall) "double preventions." But beware: I use "prevention" (following Phil Dowe) t o denote a relation that occurs but once when the button is pressed, viz., 287 CAUSATION BY DISCONNECTION Heart not pierced Heart pierced piercing os?.geI! death starvation brain cell brain cello oxygen use oxygen use The Center for Disease Control (which is fully aware of the wiring) iden- tifies heart disease as "the leading cause of death" in the United States. At the next step backward, surely the killer's pulling the trigger ( D l ) is a cause of the firing of the bullet through the victim's heart (D2). But trigger pullings only cause bullet firings by disconnection. The spring is kept coiled by the sear, and trigger pullings cause bullet firings by discon- necting the sear, allowing the spring to uncoil (propelling the striker onto the powder, compressioil of which produces the explosion which propels the bullet): striker n spring Diagrammatically: Trigger not pulled Tripper pulled pulling 0firing pulling Q @ firing spring spring @ cocked cocked Even the National Rifle Association, which insists that "guns don't kill people, people kill people" to blame the shooter rather than the weapon, concedes thereby that shooters of guns can cause death. Finally back to the first step, surely the killer's getting angry (C) is a cause of his pulling the trigger ( D l ) . But nerve signals only cause muscle contractions (such as that of the trigger finger) by disconnection. Muscle 288 JONATHAN SCHAFFER fibers, according to the generally accepted sliding filament theory, contract by myosin-actin binding, which is blocked by the presence of tropomyosin on the binding sites of the actin. Contraction occurs when the electrical impulse from the nerve triggers a calcium cascade through the fiber. The calcium binds to troponin (attached to each tropomyosin molecule) and the troponin-calcium con~plex disconnects the tropomyosin from the bind- ing sites: KxxxnxX actin- m~tropomyosin on grooves between \\\\\ ///// double helixes of actin, with troponin m myosln.t m attached to each tropomyosin molecule. (Myosin proteins work like coiled springs, set to grab the actin when re- leased). Diagrammatically: Nerve signal not fired Nerve signal fired firing C: contraction firing @ contraction tropo- ~-2'1, actin-myosin actin-myosin myosin binding myosin myosln @' myosin @' Philosophers such as R. C. Collingwood (1940), Douglas Gasking (1955), G. H. von Wright (1975), and Huw Price (1991) have alleged that causa- tion is an anthropomorphic concept, inextricably entangled with the idea of human agency. While I would maintain that an objective concept of causation can be isolated, surely this concept must still apply to voluntary human actions, which are due to muscle contractions. See that disconnections are both ubiquitous and paradigmatically causal and you'll find them in even the most theoretically salient instances of causation. When Saul Kripke (1972) speaks of the reference of names as transmitted via causal chains, it should be obvious that his claim does not turn on just how the printing press is wired. When Brian Skyrms (1980) speaks of rational decisions as those that maximize the expected utility of their effects, it should be obvious that this calculation is indifferent as to whether the effects are achieved by connection or disconnection. When Alvin Goldman speaks of perception as a causal relation (1977), he ex- plicitly notes that one can perceive black holes without any energy coming from them (rather the black hole disconnects light that would otherwise be visible), and Michael Tye notes that: "This difficulty is not peculiar to 289 CAUSATION BY DISCONNECTION astronomical contexts. It seems to me that perfectly black objects which are not too small or too distant may be seen with the naked eye provided that they are located against light backgrounds" (1982, 324). Causation by disconnection is causation full force, involving all the central coililotations of causation such as counterfactuals, statistics, ex- planation, inference, agency, and responsibility, iilvolved in even the most paradigmatic cases of causation including all instances of human action, and involved in the most theoretically salient cases of causation relevailt to the analyses of reference, decision, and perception. 1 cannot see a stronger argument that anything is causal.' Causation by disconnection is different than causation by connection (and liable to be overlooked if one concentrates on colliding billiard balls) but no less causal thereby. There is more than one way to wire a causal mechanism. 2. Physical Connections. David Hume glossed our naive conception of causatioil as that of necessary connection. While Hume thought the con- nection not in the objects but projected by the mind, a number of subse- quent philosophers have addressed the Humean problematic by seeking a coililection in the objects via physical processes such as energy flows. Causation by disconilection refutes this program. The causal connec- tion of which Hume speaks may well be objective, but is more abstract than simple physical connection. There are three intimately related research programs running today that take physical connection to be at least necessary (if not sufficient as well) for causation. I will be extremely brief here, since causation by discon- nection refutes not the details of these various programs, but their under- lying thought. First, there is the program of Jerrold Aronson, David Fair, and Hector- Neri Castaneda, on which property transfer, or more specifically energy transfer, is taken to be at least necessary for causation. For Aronson, causation is transference of a quantity (velocity, momentum, kinetic en- ergy, and heat are given as examples, though no definite class of relevant quantities is specified) by intervening contact action of cause into effect, so that, " 'A' in 'A causes B' refers to an object that successfully transfers one of its quantities to the effect object." (1971, 422). For Fair, "A causes B iff there are physical redescriptioils of A and B as some manifestation 3. Cases of disconnection have floated through the literature for some time now, al- though their full force has perhaps yet to be appreciated. F o r example, Douglas Ehring (1986, 251) speaks of causation by elimination of transfer in the case of turning off a light switch; but then Ehring (1997) requires physical connection between cause and effect. with n o mention of disconnections. 290 JONATHAN SCHAFFER of energy or momentum or refer to objects manifesting these that is trans- ferred (flows), at least in part, from the A-objects to the B-objects" (1979, 236).4 And for Castaneda (1984) causation requires both transfer (of 'caus- ity', empirically identified with energy) and Humean constant contiguous conjunction. A second physical connection program is that of Bertrand Russell, Wesley Salmon, and Phil Dowe, on which processes are taken to be basic entities, the causal ones of which must be specified. For Russell (1948) the specification is in terms of spatiotemporally continuous qualitative persis- tencies. For Salmon (who notes that Russell's specification applies to joint effects of a common cause such as a moving spot of light projected from a rotating beacon) mark trailsmittability is added, where, "A mark is an alteration to a characteristic that occurs in a single local intersection" and "A mark is transmitted over an interval when it appears at each spacetime point of that interval, in the absence of interactions" (1998, 250). Thus a causal process, for Salmon, is a spatiotemporally coiltinuous persistence that is capable of transmitting a mark, of propagating structure. And for Dowe (1992, 1995) a causal process is the worldline of an enduring con- served-quantity-bearing ~ b j e c t . ~ A third physical connection program is that of J. L. Mackie and Doug- las Ehring, on which a causal mechanism is postulated as: "a process which underlies a regular sequence and each phase in which exhibits qual- itative as well as spatio-temporal continuity" (Mackie 1974, 222). Mackie finds in this ilotioil certain senses in which causes necessitate their effects, especially that of the 'intimate tie' between cause and effect. Ehring (1997) develops this thought by specifying the relevant qualitative continuity as trope identity-through-time, and the intimate tie found therein as intrinsic relation. All three research programs are deeply interrelated, and owe their dis- tinctive flavors as much to historical pedigree as to pl~ilosopl~ical differ- ence. All understand physical connections as persistencies, and differ only in what is said to persist: unspecified for Russell and Mackie, properties for Aronson and tropes for Ehring, energy for Fair and Castaneda, struc- ture for Salmon, and objects iilstailtiatiilg coilserved quantities for Dowe. It is obvious that these property transfer/causal process/underlying mechanism programs have no room for causation by disconnection, since 4. According to Wladyslaw Krajewski (1997), energy flow accounts of causation trace back to Robert Mayer (the discoverer of the law of conservation of energy) in 1842, and have been advocated by such scientific luminaries as Helmholtz, Planck, and Lor- entz. 5. Salmon (1994, 1998) now embraces Dowe's shift from capacities to conserved quan- tities, and now speaks of causal processes as those that transmit conserved quantities. CAUSATION BY DISCONNECTION 29 1 causation by disconnection involves no persistence line between discon- nector and effect, but rather the severing of one, as the wiring diagram makes manifest: No line from disconnector to effect disconnector C! 35 effect 'b /,' inhibitor a--~-? / source d" Thus for Aronson, Fair, and Castaneda there will be no property trans- ferred from, e.g., the nonbeating heart to the oxygen-starving brain, for Russell, Salmon, and Dowe there will be no causal process between them, and for Mackie and Ehring there will be no persistence mechanism tying these events together. Not all ways of wiring a causal mechanism involve persistence. Those connection theorists alive to the problem of disconnection have hitherto bitten the bullet. Here is Aronson: Consider a weight that is attached to a stretched spring. At a certain time, the catch that holds the spring taut is released, and the weight begins immediately to accelerate. One might be tempted to say that the release of the catch was the cause of the weight's acceleration. If so, then what did the release of the catch transfer to the weight? Noth- ing, of course (197 1, 425). Such a dismissal might be tolerable were it limited to the isolated case of launching a weight by a spring, but once it is seen that spring-release-like mechanisms are ubiquitous in nature, present throughout the most intui- tively paradigmatic as well as the most theoretically salient causal se- quences, with all the conceptual connotations of causation in full force, such a dismissal becomes absurd. No theory so dismissive is close to de- serving the name "causation." 3. Intrinsic Relations. Hume's empiricist reconception of causation is that of regularity of sequence in the objects together with felt necessity in the mind: constant contiguous conjunction with conditioned conceptual con- nection. Hume notes, as a counterintuitive consequence of his reconcep- tion, that it is "drawn from circumstances foreign to the cause" (1975, 362), and some subsequent philosophers have rejected the Humean recon- ception by seeking a notion of causation not "drawn from circumstances foreign" but rather drawn from intrinsic relations. 292 JONATHAN SCHAFFER According to Peter Menzies, causation is a theoretical notion, func- tioilally defined as the intrinsic relation that typically induces probabilistic counterfactual dependence between actual, distinct events. A relation is intrinsic in the iilteilded sense (intrinsic to its pairs) iff whenever (a, b) and (a', b ' ) are duplicate pairs, either both or neither of the pairs stand in the relation, where (a, b) and (a', b ' ) are duplicate pairs iff a and a' have the same natural properties, so d o b and b', and the natural relations between a and b are the same as between a' and b'. Meilzies asserts that: The distinctive mark of our intuitive coilception of causation . . . is that it takes causal relatioils to be determined by the natural properties of the relata and the natural relations holding between them, taken in isolatioil from everything else happening in the world. (1996, 100) According to David Armstrong, causation is a primitive notion, em- pirically identified with the instantiation of the second-order ilecessitatioil universal N (Armstrong's lawmaker) between states of affairs: F a and G b are causally related in virtue of an iilstantiatioil of N(F,G). Armstroilg notes with satisfaction that, "[Elach instantiation of a universal is com- plete in itself, so the law will be present conzpletely in each instantiation. So where singular causation is the instantiation of such a law it will be a completely intrinsic relation" (1999, 184).6 The idea of causatioil as an intrinsic relation may be regarded as a generalizatioil of the idea of causation as a pllysical connection. As Men- zies puts it, "[Tlhere is an intrinsic relation connecting the event of my throwing the stone and the window's breaking, a process consistiilg in the transfer of energy-momentum from cause to effect." (1996, 105) For Arm- strong energy flows could be regarded as a this-worldly instance of lawful sequence. And for Ehring requiring the physical connectioil of trope per- sistence is a way to ensure intrin~icness.~ It is obvious that intrinsic relation programs, like physical connections, have no room for causation by disconnection, since causation by discon- nection turns on the extrinsic circumstailce of a preexisting connection to be severed. For example, the nerve signal and fiber contraction are causally 6. F o r Armstrong, the states of affairs related by causation contain first-order universals (the F of Fa). I will argue in the next section that causation by disconnection requires absence causation. Since Armstrong rejects negative first-order universals, mediation by absences represents an additional conflict (other than extrinsicness) between what Armstrong calls 'causation' and any relation that covers disconnections. 7. According to Alexander Rueger, neither energy-momentum flow nor conserved quantity transmission are in fact intrinsic as scientifically defined, since both depend on global spacetime structure. Rueger concludes that "the localist intuition, though pre- sumably part of the 'folk theory of causation', is just not correct and has t o be given up" (1998, 37). CAUSATION BY D I S C O N ~ E C T I O N 293 related only due to the extrinsic circumstailces of tropomyosin being pres- ent, since a perfect duplicate of the signal and the contraction could exist without any causal relation if the fibers in question were missing the tro- pomyosin that the nerve signal is normally needed to disconnect. In that case there would be nothing for the signal to accomplish: Tropomvosin absent (no causation) T r o ~ o m v o s i n present (causation) firing contraction @L7@ tropo- tropo- actin-myosin myosin 2 binding / myosin @ myosin @ Here the 'foreign circumstance' of the presence or absence of the inhibitor changes whether C and E are causally related. Not all ways of wiring a causal mechanism involve intrinsic relations. I admit that there is something intuitive about intrinsicness, but deny that this intuition can withstand philosophical scrutiny. Intrinsicness seems an isolated intuition, with no clear relation to the other central connotations of causation (such as counterfactuals, regularities. explana- tion, inference, responsibility, and agency). Causation by disconnection pulls these apart. It reveals that we must choose between, on the one hand, a conception of causation that is tied into the holding of the central con- notations of causation, that applies to such paradigmatic causes as heart attacks, gun firings, and human actions, and that supports a causal anal- ysis of reference, decision, and perception, inter alia. versus, on the other hand, a conception of causation that is intrinsic. Thus I think intrinsicness, like the principle of determinism, the law of universal causation, and the homology of cause to effect, belongs among the platitudes which a sys- tematic conception of causation should reject. A limited intrinsicness thesis may still hold for a component of the causal relation. I think causation is best understood in terms of a hybrid condition- connection account (C causes E iff an actual E-connection depends on C; see $ 5 ) . I think the connection component is best understood (Armstrong- style) in terms of sequences of events subsumed under antecedent and consequent of the fundamental dynamic laws.8 Now, holding the laws fixed (which may themselves be extrinsically determined by the systematic regularities), the connection component of causation is intrinsic to the extent that the laws project intrinsic properties. Perhaps this more modest 8. See my "Processes as Lawful Sequences" (unpublished b) for further discussion. 294 JONATHAN SCHAFFER sense of the intrinsicness-given-our-laws of the connection component of causation is all the folk ever had in mind. 4. Conditions and Absences. What constraints are imposed on a theory of causation adequate to understand how the many ways of wiring a causal mechanism are all instances of c a u ~ a t i o n ? ~ I suggest that the constraints imposed by disconnections are that an adequate theory of causation must involve necessary and/or sufficient conditiorzs, and must relate absences. The main alternative to physicallintrinsic connection approaches to causation is the Hume-style, generalist, abstract, extrinsic, necessary and/ or sufficient conditions approach. Conditions approaches (such as Hume's regularity and David Lewis's counterfactual dependence approaches) have no difficulty with disconnections. In whatever sense causes are supposed to be necessary and/or sufficient for their effects, the disconnector will be necessary and/or sufficient for what it releases. For example, the death will counterfactually depend o n the heart piercing. l o But what of the spatiotemporal continuity of causation? Perhaps spa- tiotemporal continuity is not metaphysically necessary. Perhaps it is not even physically necessary in the quantum domain. But surely it holds in cases of heart failure, gun firings, and muscle contractions: n o magical or quantum connections are a t issue. What are the intermediaries between, e.g., heart piercing and brain death? Answer: the relevant intermediaries are absences: the heart piercing causes an absence of oxygenated blood traveling from the right ventricle, through the relevant arteries, to the brain, which absence causes an ab- sence of oxygen resupply to the brain cells, which absence causes oxygen starvation. That is to say that the continuous sequence is: Heart pierced starvation 9. I assume (a) that the ways of wiring a causal mechanism have something substantive in common in virtue of which both are causations, and (b) that something informative can be said about this commonality. Thus I assume in what remains that causation is a univocal nonprimitive. 10. Probabilistic and agential approaches also count as conditions approaches. All these approaches, once extended to the indeterministic case, converge on understanding cau- sation as some form of probabilitj,-mising relation. In general, the disconnector will raise the probability of what it releases. 295 CAUSATION BY DISCONNECTION where the second link is a line of absences. Pure coilditioils approaches need absences to recognize the spatiotemporal continuity of discoilnective causation. Some philosophers find absence causation unproblematic while others find it abhorrent. I think this clash of intuitions is a manifestation of the clash between conditioils and connections coilceptioils of causation: ab- sences are perfectly suitable as conditions but not as connectors. Thus Lewis (a conditions theorist) takes it as intuitively obvious that when Fred omits the precautioils he causes the accident (1986, 189-193), while Arm- strong (a connections theorist) intuits: "[O]missions and so forth are not part of the real driving force in nature. Every causal situation develops as it does as a result of the presence of positive factors alone" (1999, 177). So far this clash seems stalemated. Intuitively we want to say that the gardener's failure to water the flowers (absence) caused them to die," but to deny that your, my, or the queen of England's failure to water the flowers caused them to die. Those like Lewis who speak of absences as causal may point to the holding of the central connotatioils of causation between the gardener's failure and the flowers' death, while those like Armstrong who deny that absences are causal may point to the prolifer- ation of counterintuitive causes thereby coui~tenanced.'~ I think the defender of absences as causal already has the better of it. A pragmatic explanation of why it is infelicitous to speak of the queen's failure as a cause is readily available: since I never presumed that the queen would deign to water my flowers, to speak of this absence is to impart 110 information not already supposed. And since both the conditions and connectioils theorists must (counterintuitively) count my birth as a cause of my coughing now, both sides are already committed to the pragmatic machinery involved. In any case, since disconnections show that absences mediate the cau- sation in such paradigmatic cases as heart attacks, gun shots, and human actions, and call mediate the causation in such theoretically salient cases as are relevant to the analysis of reference, decision, and perception, these considerations should definitely settle the issue over absences. Thus con- ditions approaches which relate absences (as was already thematic for the 1 1 . See Hart and Honore 1959 for numerous examples of causation by omission rec- ognized in the law, such as "failure to deliver to a manufacturer on time a piece of machinery, which he has ordered, may be held the cause of the loss of those profits which would have been made by its use" (1959, 56). 12. The most intuitive theory would count the gardener's failure but not the queen's as a cause. But it is doubtful that there is any ontological basis for such a discrimination. It would seem that, for any theory of causation with aspirations t o objectivity, the only options with respect to absences are all or none. 296 JONATHAN SCHAFFER conditions theorist) provide a unified and informative account of what the many of ways of wiring a causal mechanism have in common. 5. Hybrid Approaches. Other than the necessary and/or sufficient condi- tions approach, the only other approach to causation I know of which is adequate to disconnections is a hybrid conditions-connectioils approach (which explicitly iilvolves coilditioils and relates absences). It will emerge, however, that major revision of the extant hybrids is required. Fair concludes on intuitive grounds that absences are causal, and rec- ogilizes that his energy flow account cannot accomn~odate them, since absences "cannot be the sources or sinks of actual energy-momentum" (1979, 246). He responds by suggesting a hybrid counterfactual-energy approach: (1) Connective causation: C causes E iff (a) C and E are actual, distinct events and (b) C is physically (/intrinsically) coililected to E. (2) Causation by prevention: C causes -E iff (a) C occurs and E does not, and (b) C is a coililective cause (by 1) of an E' incompatible with E. ( 3 ) Causation h j omission: -C causes E iff (a) C does not occur and E does, and (b) had C occurred, it would have connectively caused (by 1) an E' illcompatible with E. (4) Cc~usmtiorzh j omission ofprevention: -C causes -E iff (a) neither C nor E occurs, and (b) had C occurred, it would have connec- tively caused (by 1) E.13 It might seem that the hybrid theorist can trace the conditions theorist's approach to spatiotemporal contiiluity and maintain, for example, that heart piercing (C) causes an absence of oxygen resupply to the brain cells (-D) by (2), that the absence of oxygen resupply to the brain cells (-D) causes brain death (E) by (3), so that heart piercing (C) causes brain death (E). But the hybrid approach does not chain. Heart piercing causing brain death is of form (I), and the hybrid approach equates causatioil of form (1) with connection. Solution: understand conditions (1)-(4) as defiiliilg direct causation, and understand causation generally in terms of chains of direct causation^.'^ 13. Dowe (I 999) refines Fair's hybrid approach, analyzing preventions and omissionsvia counterfactuals about connection. Yet Dowe insists on a pure (actual) connections ap- proach t o causation, and so relegates his refinement to ersatz status. I think this is a highly unstable position. Disconnections show that what Dowe dismisses as ersatz causation* is far more deserving of the title "causation" than the pure connections approach. 14. Fair clearly intends the hybrid relations to be closed under transitivity (1979, 248). Chaining can also be accomplished by requiring the spatiotemporally continuous hold- ing of the hybrid relations between C and E, rather than the mere transitive closure thereof. Either works here. 297 CAUSATION BY DISCONNECTION Thus disconnections can be understood to be prevention-omission chains. But chaining infinitely exacerbates the hybrid theorist's disunity prob- lem. Fair has already split causation into four relations, and now with chaining when C is said to cause E all that is so far specified is some arbitrarily complex combination of direct causations. We have not really understood why (1)-(4) are all causations in the first place, and so are in no position at all to maintain that all n-length chains thereof (e.g., (1)-(2)- (4)-(3)-(2)- . . . , (4)-(3)-(2)-(4)- . . . , etc.) are all causations.'* Solution: the hybrid theorist, to regain unity, must (a) unify absences and presences into a single form, and then (b) generalize to a common hybrid relation across the now formally unified (1)-(4). Thus chaining would entail simple repetition rather than arbitrary complexity. (If neither this nor any other hybrid-unifying strategy can be implemented, then disconnections will have the upshot that only a pure conditions approach is adequate.) In order to unify absences and presences into a single form as per (a), the hybrid theorist looks to have three main approaches. The first is to take the causal relata as abstract facts, since that there is oxygen resupply to the brain and that there is no oxygen resupply to the brain are equally of factive form. But it is hard to see how there could be any connection to such an abstraction. The second is to take the causal relata as concrete unstructured events and interpret the event described as "an absence of oxygen resupply" to refer to a present proxy that stands in the incompat- ibility relation to this absence. But it is hard to see how the conditions could hold for such a proxy, since, for instance, the proxy has the wrong counterfactual behavior: if the proxy for the fridge holding no beer is its being full of sausages, then it does not follow that had the fridge not been full of sausages then it would have held beer.16 The third main approach, which seems best, is to take the causal relata as concrete structured events on which the causal relata are ordered (Property, Region) pairs with neg- ative properties allowed so as to include absences." These pairs have the right counterfactual behavior, since the supposition that the (absence of 15. Lewis (1998) suggests understanding the hybrid relations as direct causation, raises the disunity objection, and concludes that while the hybrid approach might well specify a supervenience base for causal relations, it is far too disjointed to be regarded as a n analysis of the causal concept. 16. Perhaps this is salvageable, since it is generally recognized that there is a t least some context-dependence in the assessment of the counterfactual value of -C (Lewis 1986, 6. Thus it might be maintained that referring to the proxy via the absence-description creates a context in which the supposition that the proxy does not occur is best inter- preted as the supposition that the absence in the description is present. 17. This approach is inspired both by Jaegwon Kim's understanding of the causal relata as (Object, Property, Time) triples, as well as by D. C. Williams's understanding of the causal relata as tropes. 298 JONATHAN SCNAFFER beer, fridge region) event does not occur is the supposition of a beerful fridge. Call the disjunction of a positive relata or a positive proxy for a negative relata the nznnifestation of the relata in question. Then the rele- vant connection for these pairs is the connection to the manifestation of the effect. Here then is a suggestion as per (b) for framing a unified hybrid relation for the (Property, Region) pairs that formally unify absent and present relata: C directly causes E iff (a) C and E are actual, distinct events, and (b) had C not occurred, then an (actual) connection to the E-manifesta- tion would not have occurred. In short, the thought is that an (actual) E-connection depends on C. Con- nective causation (CIE), previously understood in terms of C being con- nected to E, is now understood in terms of connection dependence. In other words, if E would have occurred at all without C it would have been via a nonactual connection (such as via a preempted backup). Causation by prevention (CI-E), previously understood in terms of C being a con- nective cause of an E' incompatible with E, is likewise understood on the counterfactual: had C not occurred, then if E' (the E-manifestation) would have occurred at all it would have been via a different connection. The reader may compare the understandings of causation by omission (-CIE) and by omission of prevention (-C/-E).I8 One consequence of this unification is that disconnections can again be understood directly oust as with the conditions approach), since, e.g., the connection to brain death depends directly on the heart piercing. So we can kick away the ladder of direct causation and reach: C causes E iff (a) C and E are actual, distinct events, and (b) an E-process depends on C. (Where an E-process is an actual connection to an E-manifestation.) Both the conditions and (suitably refined) hybrid conditions-connec- tions approaches seem equally adequate to causation by disconnection. I believe, however, that when we turn to problem cases such as preemption the hybrid approach will perform significantly better. Moreover, the hy- brid approach can explain what is right in the connections approach: cau- sation does turn out to be based in actual connections, just not directly. 18. There are differences between the disunified relations reported above and their unified generalization developed here. In the CIE case for instance, the unified gener- alization entails that if C is a n iu~e.~~~:sentirri~~rrr.tof the connection (a speck of dust on the bullet), then C is not a cause of E, since that connection would, in essence, still occur without C, while the disunified relations must count such an inessential C as a connec- tive cause of E. Further confirmation for the unified generalization. CAUSATION BY DISCONNECTION 299 Rather we must take a further step of abstraction and speak of conditions for connections. I leave a full comparison of these approaches to another day.'" I know of no other approach to causation able to explain how the many ways of wiring a causal mechanism are all instances of causation. Since both the conditions and hybrid approaches involve conditions and relate absences, I conclude that causation by disconnection yields a powerful argument for the involvement ol' Hume-style, extrinsic, absence-relating, necessary and/or sufficient conditions in any adequate theory of causation. Armstrong. David (19991, "The Open D o o r : Counterfactual versus Singularist Theories of Causation", in Howard Sanltey (ed.), Cau.satioi~ trnil Lnw.s of ~ V n t l ~ r e . Dordrecht: Klu- wer. 175-186. Aronson, Jerrold (1971). "On the Grammar of Cause", Sjrrtlzese 22: 414--430. Castaneda, Hector-Neri (19841, "Causes, Causity, and Energy", in Peter French. Theodore Uehling Jr.. and Howard Wettstein (eds.), ~Micill,e.i.tStudies in P h i l o s o ~ ~ h j L'ohirrte IX Causatioi~ trntl Cau.i.nl Theories. Minneapolis: University of Minnesota Press. 17-28. Collingwood, R. G. (1940). An E.i..i.nj or2 Metci~~h,vsic.i.. Oxford: Clarendon Press. Dowe, Phil (1992). "Wesley Salmon's Process Theory of Causality and the Conserved Quan- tity Theory". Phi1osoph.v qf Science 59: 195-216. . (1995). "Causality and Conserved Quantities: A Reply t o Salmon", Philosophy qJ Science 62: 321-333. . 119991. "Good Connections: Causation and Causal Processes". in I-ioward Sankev (ed.), ~trlistrtiorz arzd Laws of Nuttire. Dordrecht: Kluwer, 247-264. Ehring, Douglas (19861, "The Transference Theory of Causation", Sjnthe.i.e 67: 249-258. . (1997). Cnu.i.ntion trizil Persister~ce. New York: Oxford University Press. Fair, David (1979). "Causation and the Flow of Energy". Erl~erzntrzi~14: 219-250. Gasking, Douglas (1955). "Causation and Recipes". Mirzd 64: 4 7 9 4 8 6 . Goldman, Alvin (1977). "Perceptual Objects", Sjrzt1ze.i.e 35: 257-284. Hall, Ned (2000), "Two Concepts of Causation". Jouriinl of Plziio.i.oplzj (forthcoming). Hart, H . L. A. and A . M . Honore (1959). Cau.siition in the L n i ~ ' .Oxford: Clarendon Press. Hurne, David (1975). An Enquiry Corzcernirzg Hnrrznrr I~rzder.i.tnrzdirzg, edited by L . A. Selby- Bigge and P. H . Nidditch. Oxford: Clarendon Press. Kim, Jaegwon (19931, "Events as Property Exemplifications", in Sl~j~eri~ei~ierzcetrrzci Mirztl: Selected Plzilosoplzictrl E,c.i.try.i.. Cambridge: Cambridge University Press. 33-52. Krajewski, ~ l a d ~ s l a w (19971, ..Energetic, 1nformationa1,-and Triggering Causes", Ericenizt- rzii. 47: 193-202. Kripke. Saul (1972). Naming anil ,Vecessit,v. Cambridge. MA: Harvard University Press. Lewis. David (1986). "Causation", in Philo.sophictri Ptrpers, vol. 2. New York: Oxford Uni- versity Press: 159-21 3. . (19981, "Void and Object", preprinted by the Department of Philosophy, University of Melbourne. Mackie, J. L. (19741, The Cenzent qf'tlze CTniver.se: A Study qf'Cc~irsntion. Oxford: Clarendon Press. Menzies, Peter (1996). "Probabilistic Causation and the Pre-emption Problem", mi nil 105: 85-1 17. 19. See my "Causes a s Probability-Raisers of Processes" (unpublished) f o r a r g u m e n t s favoring t h e hybrid conditions-connections a p p r o a c h developed here over t h e condi- tions a p p r o a c h . 300 JONATHAN SCHAFFER Price. Huw (1991), "Agency and Probabilistic Causality", Britirh Jonr.ntrl,for the Plzilosopli,~ of Science 42: 157-176. Rueger, Alexander (1998), "Local Theories of Causation and the A Posteriori Identification of the Causal Relation". Erlcenntnis 48: 25-38. Russell. Bertrand (1948), Hnrnan Kno\c,lecige: Its Scope and Limit.7. New York: Simon and Schuster. Salmon. Wesley (1994), "Causality without Counterfactuals", P h i l o s o ~ ~ h yof Science 61: 297-312. . (1997), "Causality and Explanation: A Reply to Two Critiques", Philo.soph,v qf Science 64: 4 6 1 4 7 7 . . (1998), Ccrn.rality and E x ~ ~ l a n n t i o n . New York: Oxford University Press. Schaffer, Jonathan (unpublished), "Causes as Probability-Raisers of Processes". . (unpublished b), "Processes as Lawful Sequences". Skyrms, Brian (1980), Cazl.sal Necessity. New Haven: Yale University Press. Tye, Michael (1982), "A Causal Analysis of Seeing", Philo.rophy and Phenomerzologiciil Re- search 42: 31 1-325. Von Wright. G. H . (1975). Causality and Determinism. New York: Columbia University Press. Williams. D. C. (1953), "On the Elements of Being", Revieit, qf'1ML.tcrpl1ysic.r 7: 3-18. You have printed the following article: Causation by Disconnection Jonathan Schaffer Philosophy of Science, Vol. 67, No. 2. (Jun., 2000), pp. 285-300. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28200006%2967%3A2%3C285%3ACBD%3E2.0.CO%3B2-Z This article references the following linked citations. If you are trying to access articles from an off-campus location, you may be required to first logon via your library web site to access JSTOR. Please visit your library's website or contact a librarian to learn about options for remote access to JSTOR. [Footnotes] 5 Causality without Counterfactuals Wesley C. Salmon Philosophy of Science, Vol. 61, No. 2. (Jun., 1994), pp. 297-312. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28199406%2961%3A2%3C297%3ACWC%3E2.0.CO%3B2-S References Wesley Salmon's Process Theory of Causality and the Conserved Quantity Theory Phil Dowe Philosophy of Science, Vol. 59, No. 2. (Jun., 1992), pp. 195-216. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28199206%2959%3A2%3C195%3AWSPTOC%3E2.0.CO%3B2-D Causality and Conserved Quantities: A Reply to Salmon Phil Dowe Philosophy of Science, Vol. 62, No. 2. (Jun., 1995), pp. 321-333. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28199506%2962%3A2%3C321%3ACACQAR%3E2.0.CO%3B2-M http://www.jstor.org LINKED CITATIONS - Page 1 of 2 - NOTE: The reference numbering from the original has been maintained in this citation list. http://links.jstor.org/sici?sici=0031-8248%28200006%2967%3A2%3C285%3ACBD%3E2.0.CO%3B2-Z&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8248%28199406%2961%3A2%3C297%3ACWC%3E2.0.CO%3B2-S&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8248%28199206%2959%3A2%3C195%3AWSPTOC%3E2.0.CO%3B2-D&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8248%28199506%2962%3A2%3C321%3ACACQAR%3E2.0.CO%3B2-M&origin=JSTOR-pdf Causation and Recipes Douglas Gasking Mind, New Series, Vol. 64, No. 256. (Oct., 1955), pp. 479-487. Stable URL: http://links.jstor.org/sici?sici=0026-4423%28195510%292%3A64%3A256%3C479%3ACAR%3E2.0.CO%3B2-S Probabilistic Causation and the Pre-emption Problem Peter Menzies Mind, New Series, Vol. 105, No. 417. (Jan., 1996), pp. 85-117. Stable URL: http://links.jstor.org/sici?sici=0026-4423%28199601%292%3A105%3A417%3C85%3APCATPP%3E2.0.CO%3B2-M Agency and Probabilistic Causality Huw Price The British Journal for the Philosophy of Science, Vol. 42, No. 2. (Jun., 1991), pp. 157-176. Stable URL: http://links.jstor.org/sici?sici=0007-0882%28199106%2942%3A2%3C157%3AAAPC%3E2.0.CO%3B2-Z Causality without Counterfactuals Wesley C. Salmon Philosophy of Science, Vol. 61, No. 2. (Jun., 1994), pp. 297-312. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28199406%2961%3A2%3C297%3ACWC%3E2.0.CO%3B2-S Causality and Explanation: A Reply to Two Critiques Wesley C. Salmon Philosophy of Science, Vol. 64, No. 3. (Sep., 1997), pp. 461-477. Stable URL: http://links.jstor.org/sici?sici=0031-8248%28199709%2964%3A3%3C461%3ACAEART%3E2.0.CO%3B2-E A Causal Analysis of Seeing Michael Tye Philosophy and Phenomenological Research, Vol. 42, No. 3. (Mar., 1982), pp. 311-325. Stable URL: http://links.jstor.org/sici?sici=0031-8205%28198203%2942%3A3%3C311%3AACAOS%3E2.0.CO%3B2-T http://www.jstor.org LINKED CITATIONS - Page 2 of 2 - NOTE: The reference numbering from the original has been maintained in this citation list. http://links.jstor.org/sici?sici=0026-4423%28195510%292%3A64%3A256%3C479%3ACAR%3E2.0.CO%3B2-S&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0026-4423%28199601%292%3A105%3A417%3C85%3APCATPP%3E2.0.CO%3B2-M&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0007-0882%28199106%2942%3A2%3C157%3AAAPC%3E2.0.CO%3B2-Z&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8248%28199406%2961%3A2%3C297%3ACWC%3E2.0.CO%3B2-S&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8248%28199709%2964%3A3%3C461%3ACAEART%3E2.0.CO%3B2-E&origin=JSTOR-pdf http://links.jstor.org/sici?sici=0031-8205%28198203%2942%3A3%3C311%3AACAOS%3E2.0.CO%3B2-T&origin=JSTOR-pdf