You are on page 1of 22

Philosophy of Science Association

Causality: Production and Propagation Author(s): Wesley C. Salmon Source: PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association, Vol. 1980, Volume Two: Symposia and Invited Papers (1980), pp. 49-69 Published by: The University of Chicago Press on behalf of the Philosophy of Science Association Stable URL: Accessed: 28/02/2010 23:06
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at 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 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. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact

The University of Chicago Press and Philosophy of Science Association are collaborating with JSTOR to digitize, preserve and extend access to PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association.



and Propagation1


C. Salmon




A standard of causality has been around at least since the picture time of Hume. The general idea is that we have two (or more) distinct events which bear some sort of cause-effect relation to one another. There has, of course, been considerable the nacontroversy regarding ture of both the relation and the relata. It has sometimes been mainfor instance, that facts or propositions than events) tained, (rather are the sorts of entities which can constitute the relata. It has long been disputed whether individual events or only classes of events can sustain cause-effect relations. The relation itself has sometimes been taken to be that of sufficient sometimes condition, condition, necessary or perhaps a combination of the two.2 Some authors have even proposed that certain sorts of statistical relations constitute recausal lations.3 It is my conviction that this standard in all of its wellview, known variations, is profoundly and that a radically difmistaken, ferent notion should be developed. I shall not attempt to mount arguments against the standard I shall a conception;4 instead, present rather different for purposes of comparison. I hope that the approach alternative will stand on its own merits. 1. Two Basic Concepts

There are, I believe, two fundamental causal which need to concepts be explicated, and if that can be achieved, we will be in a position to deal with the problems of causality in general. The two basic conare production and propagation, and both are familiar to common cepts sense. When we say that the blow of a hammer drives a nail, we mean that the impact produces of the nail into the wood. When penetration we say that a horse pulls a cart, we mean that the force exerted by the horse the motion of the cart. When we say that produces lightning starts a forest fire we mean that the electrical discharge produces When we say that a person's ignition. embarrassment was due to a

PSA 1980, Copyright

Volume 2, pp. 49-69 1981 by the Philosophy Q




the pulse would be an event. qualify deny that as a causal characterized Russell to what Bertrand is similar process of as the persistence be regarded line "A causal line: may always a or what not. are while are represented events time diagrams. I believe. a photon. I shall is a process." which persist are waves and material through tant processes objects will at rest even a material use terms. p. much greater and in many cases. rather cesses. legitimacy briefly duration and temporal extent both spatial mechanics." the shot heard farmers and time can have significant at one place show. transmitted upper atmosphere Signals from a broadcasting in our home. which we had earlier in our lives affect our current is transmitted of these past events By means of memory. to try to reduce tation 2. airplane to our location on the ground. constitute by a pulse from of light.50 thoughtless psychological frequently remark we mean that discomfort. 459). with a window would count A baseball colliding by lines. at other influence places upon what happens time and influence can be propagated because causal through possible are intiand causal causal propagation production Although space. as basic than events rather is to take processes causality I of processes. A sneeze would be a process. a table. of this the scientific approach. Such in everyday contexts. as an event. represented would from the bat to the window. magnitude. the landscape of a cloud moving across Although what I mean by a as causal all processes processes. imporAmong the physically considerable (1948. the baseball. the influence to the present. qualify of causality a theory to develop Before attempting are taken as fundamental. any tempmately one to the other. what happens amples This is and times. than events. processes by points. any rigorous remarks. A sonic of a jet boom makes us aware of the passage in the air is propagated a disturbance from the overhead. The main and make some very informal cite shall examples are relatively is that events and processes between events difference have much greater while in space and time. traveling of of a photocell The activation a process. in which proconsider I should In Newtonian were absolute . resist we should. (or transmission) Experiences behavior. exAs all of these "fired 'round the world. As I shall time. Processes in approaching which I propose One of the fundamental changes entities. something--a of of quality. related to one another. spatial duration. object as a process. perhaps light The shadow is an event. may be constancy given of any but not sudden changes or a gradual change of either. structure. there causal constancy line. traveling. Throughout person. localized temporal processes In spaceextent. station are received by the radio News or music reaches us because waves are propagated electromagnatic In 1775 some Massachusetts from the transmitter to the receiver. a distant star. definition not attempt I shall rather. an inappropriate comment produced of causal occur examples production is equally Causal propagation familiar.

however. This quantity is independent frame of reference. certain however. nor did the duration distance and the temin Newtonian both the spatial mechanics. pp. it was a status physical natural of the special to regard the of relativity consequence theory world as collection of events to one which bear spacetime relations another. The theory of processes types does not mandate an "event relativity ontology". it is done entirely theory with paths of light At any point in space-time. way (developed originally by A. There are. pseudoare incapable of doing so. relativity upon a "process ontology". that we make a distinction Special relativity between what I shall call and pseudo-processes. another however.A. processes Consider a simple that we have a very large cirexample. of relativity Einstein's Both knows. processes which can transpire at arbitrarily velocihigh velocities--at ties that of light. This fact does not violate the vastly exceeding basic relativistic for these are inprinciple. changed special theory the spatial were relativized to distance and the temporal separation of a rigid rod and the duration of a frames of reference. Causal capable are those which are capable of transmitting processes signals. in each for any two events. There is. does demand. greater the velocity of light in a vacuum. struct the Minkowski is cone--a two-sheeted cone whose surface light of all possible which converge generated by the paths upon pulses light that point and the paths of all possible cone) (past light pulses light which could be emitted from that point When all (future cone). them were absolute As everyone between poral magnitudes. approach extended in a natural into account way to general relativity by taking the paths of freely material these falling particles. we can conpulses. "processes" of serving as signals or of transmitting information. causal It is processes a fundamental of that theory that is a first principle light signaZ-that that no signal can be transmitted at a velocity than is. appears legitimate the spacetime structure of the physical world by regarding physical as the basic of physical entities. traveling and time. (see Winnie pulses. Robb) of approaching the special of relativity. showed. We can therefore base special through space this can be Moreover. separation all that. sort if you will--with a spotbuilding--a mounted at its When the light center. it has the same value and every inertial there are good reasons frame of reference.51 The length of a rigid rod did not depend upon a choice of quantities. moving gravitational test are also processes (see GrUnbaum particles 1973. however. are processes. varied from one frame of reference Howtemporal process as Minkowski is an invariant there ever. the entire structure of the spacetime world is determined But light 1977). Suppose cular of super-Astrodome. of a process. spacequantity--the of the time interval between two events. is turned on in the otherlight . These considerations offer for what is sometimes support called an "event ontology". Given two frame of reference. Since for according a fundamental to invariants. events. It therefore to be entirely to approach 735-750). The length to another. light of the light cones are given.

process.52 a spot of light wise darkened it casts If we building. the of a white consists along spot moving regularly any intervention. than a single but that would not conof the spot continuously. is transmitted from the point say. Consider. of light which travels from the spotlight first. pulse. and then off again. process processes of transmitting a mark. If we intervene wall of the building. a pseudo-process is not. capable a pulse to the wall. let us consider rotates. intervention at one it reaches A single remains red until the wall. Suppose. at many We can put red filters itself. transmit messages. of in the absence We have a "process" not be transmitted. causal that the spotlight is mounted on a further. There are a number of ways in which we can as the spotlight a we can place for example. from that in the process it in a way which persists transforms point the light would have remainon. becomes that the spot of light at the wall with the result red filter in the traveling But if we make such a modification red at that point. spotlight white. keep up with If it. This pulse of light. If the light is turned on and set into the spot of light move which it casts rotation. This "process"--the regular to fulfill the conditions Russell used to moving spot of light--seems It is a characterize causal but it is not a causal lines. at a single an intervention stitute point This last spot of us back to the subject brings suggestion no runner could is moving rapidly. whether a causal constitutes that the light process. in front a red filter in the "process". pulse of transin either case it is capable it is modified or not. As of interaction. the "process" at some point. upon the wall will in a highly around the outer wall fashion. pulse from the light travels to the wall. since and can can serve as signals a mark. not be transmitted it will beyond the point spot. process. will be modified make We can. will which. light pulses mitting Clearly. turn the light on for a brief a light moment. therefore. had occurred. at which the red filter soon as the light spot moves beyond the point The mark can be made. on beyond that continue but it will at that point. the . pulse point If to the wall. If we had not intervened. which moves around the wall the spot of light Now. the spot red at other places at an a Zocal intervention but that does not constitute the spotlight. upon the wall. again. mechanism which makes it rotate. of course. paradigm from pseudocausal The basic method for distinguishing processes is A causal is the criterion of mark transmission. traveling is a paradigm of what we mean by a from the spotlight to the wall. of what we mean by a pseudo-process. many interventions places along We could get someone to run around the wall holding one. the a piece of red glass in its path at any point between If we place becomes and the light which was and the wall. intervene to change the spot at some point. light of velocity. in the process isolated point rather but that would involve the wall. ed white its entire from the spotlight during journey a change which we can produce we do intervene at a single place locally We shall of intervention onward. as if no intervention point just in a red lens We can install if we wish. but it become white it will was placed.

rotates at a fixed then it takes the rate. is that the causal transmits pseudo-process. lations of amplitude (AM) or frequency for the pur(FM). loosely. A given whether it be causal or pseudo. set your foot to tapping. it would be physically to make the filter impossible travel fast No material such as the filter. A causal influence transmitted A by sound waves can make your dog come running. reaches about 50. or induce someone to purchase a different brand of soap. to be a rapidly neutron star which sends out a beam of thought rotating radiation. The pulses at the rate of 30 arrive that is the rate at which the neutron star rotates. We can imagine without that. its while the pseudo-process does not. however. Crab nebula which is about 6500 light This pulsar is years away. To make this more vivid. consider an actual which is point example to the rotating There is a pulsar in the quite analogous spotlight. are of transmitting capable and causal marks. per second. There is no upper limit on the speed of pseudo-processes. The electromagnetic equal radiation from the pulsar at the speed of light) takes (which travels 6500 years to traverse the radius of this but the "spot" of circle. If the spotlight rotates once per second. Such processes are the means by which causal influence is in our world. by amplitude of this modifications in the form of moduamong other things--and wave. making any in the spotlight or its rate of rotation. . certain The difference structure. When the beam is directed toward us. then it will be capable of process--is transmitting modifications in that structure. A causal influence transmitted arrow by a flying can pierce an apple on the head of William Tell's son.53 but perhaps a mechanical device could be set up. moves too rapidly. The distinction between which do and those which do not transmit own their processes structures is revealed If a process--a causal by the mark criterion. somewhat that it exhibits a gree of uniformity--we may say. radiation around the circumference of this circle in 1/30th of sweeps a second. influence. can travel at a velocity than that of light. to the distance from the pulsar to the earth. If the spotlight follows. process own structure. or point a television camera aboard a spacecraft toward the rings of Saturn. the outer walls are change At a certain when the radius of the expanded indefinitely. point. may by radio. signals. information. This fact is independent of the size of the building. ther out. a circle drawn with the pulsar at its and with a radius imagine center. Now. enough to keep up. Radio broadcasting transmitting presents a clear The transmitting station sends a carrier wave example. the velocity of the spot exceeds the speed of light. are imposed Processes which transmit their own structure pose of broadcasting. the speed of light As the walls are moved still far(300. which has a certain structure--characterized and frequency. the spot of light will travel around the wall in one second. between a causal and a process I am suggesting. the spot If.000 km/sec). its own structure.000 the spot will be traveling at building kilometers. object. has a certain deprocess. it sends out radiation which we detect later as a pulse. energy. Causal transmitted propagated influences. but no such limigreater tation is placed This can easily be seen as upon the spot on the wall. a fixed amount of time to make one entire circuit around spot of light the wall.

A and B are separated 0975+561 The twin quasars by an angular example. that the propagation or transmission of from one place causal influence and time to another must play a fundain the causal mental role of the world. The point and discussion is Reichenbach's this of the common cause. a who have spent If two friends. come readily discovery. mushrooms they collected to poisonous can be traced and others in the literature. of each other. in such apparent Two quasars of arc. them to chance.7 seconds proximity disthe observed occurrence would be a rather simply given improbable indicates of their Examination of quasars. types with situaand perfect All three forks. hypothesis the two images were probut that two distinct not in fact quasars. together together a striking reveals of their examination close earth. principle that when improbThe principle of the common cause states. roughly. One hypothesis early quite idea how this had the slightest had somehow (no one really quasars Another from a common ancestor. independently copied acute both suffer gastrotogether. pleasant day in the country illnesses we may find that their in the evening. of any two quasars the spectra between to the relations sharp contrast the need to exAstronomers at random. but was unable to find. Conjunctive Forks In order to approach the second causal basic concept. interare three conjunctive. similarity--indeed. be necessary the nature of causal There it will to consider forks. . Thus. tribution equal red spectra are close distances. to attribute recur too frequently able coincidences antecedent. with which we must deal--namely. principle of the conjunctive fork as a device his statistical characterization to elaborate that fundamental causal ?19). is in This situation they are indistinguishable. copied directly from which both of them file a paper in a fraternity for example. Pseudo-processes can can do no It is evident. these objects equal as seen from close as appearing as well in space. are concerned active. production. spectra Moreover. Conto a common causal by reference they can be explained turn in idenin a class If two students sider some familiar examples. by ink someone's marks on a piece of paper heart. I think. types which tions in which a common cause rise to two or more effects gives of departure for are somehow correlated with one another.54 causal influence transmitted one's day or break gladden such things. tical for a common cause-then we search from the other. fine a particularly furnishes scientific considerable significance. that either and if we can rule out the possibility term papers. (1956. shifts. could happen in reality) developed are there that lens effect--that was the gravitational is. have been mentioned Many such examples which has A recent astronomical to mind.5 3. structure As I shall argue beconstitute the causal which connections low. and hence. recognized immediately picked of common in terms of some sort coincidence this astonishing plain twin was that which was entertained cause. distress intestinal and consumed. width of 5. causal processes precisely Hume sought.

the gravitational of the image. A and B do occur independently. it was theorized. C is the emission of radiation in two slightly different directions by a single luminous body.B C) = P(A\C) = P(AIC) > P(A~C) x P(B(C) x P(B~C) of a con- (1) (2) (3) (4) P(A. it is not immediately conditions (1)-(4) obvious. virThis explanation is now.BIC) P(A[C) P(B C) > P(B C) For reasons which will be made clear we shall that below. apply quite straightforwardly Given two effects A and B. elliptical under fortuitously excellent Further observation. In the case of simultaneous the cause C is the common meal which included the illness.B) > P(A) x P(B)7 (5) These relations in concrete situations. as relation the Thus. and which explains the lack of independence between A and B. Reichenbach of the notion ?19) introduced (1956. separate P(X. to the best accepted (Chaffee 1980). product occurrences. given of the common cause as the relationC. In the case of plagiarism. two effects A and B are not independent. P(A.Y) = P(X) x P(Y) of one to the (6) we have considered. would be adequate to produce splitting of my knowledge. in the examples (5) states. the cause C is the presence of the term paper in the file to which both students had access. or by a very large massive black hole.55 effect duced from a single upon the light by body by the gravitational massive an intervening This result might be produced by a object. mushrooms. the fact that the probability of both individuals ill at the same time is greater than the product of the being . In the case of the twin quasar the cause poisonous image. in terms of the following defined four conditions:6 junctive fork. in equation ship among the conditional (1) shows.e. in the case of illness. To say of two events X and Y that they occurred independently another means that they occur with a probability together equal of the probabilities of their i. Although entail P(A. which occur more frequently than together of one another. probabilities Thus.. revealed the presence of a galaxy has subsequently which conditions. the occurrence However. viewing galaxy. tually universally by the experts In an attempt to characterize the structure of such examples common causes. there they would if they were statistically independent is some prior event C which is a cause of A and is also a cause of B. stipulate none of the probabilities in these relations is equal to zero occurring or one.

Re- P(A.8 Then. according A and B statisthe common cause C makes each of the two effects (1). of the example-features which has no other bearing upon the essential of this of the tosses that half pair of dice are made with the namely. potential of the common severe illness. which will us make a further simplify stipulation. also screens the common cause I should of the conjunctive the nature clear To make quite fork. Suppose ships is an that 6 on top. equation (1) entails (7) provided P(AIC)J = P(B\A. in this there are no other of common causes that. means to make statistically Thus. on or off This magnet can be turned surface.56 of their individual is explained illnesses probabilities by the common In this we are assuming meal. we can easily equation off A from B. let us assume.9 6 of double and the probability die is 1/2. this In Reichenbach's says that C screens terminology. is designated the event are tossed. and half electromagnet and the switch. but the arithmetic. We off. that the fact that one person example. with a concealed of the common cause _ an instance C. the relationto exhibit contrived like to use an example deliberately towe have a pair of dice which are rolled involved. for the fact except electroon which they are thrown has a powerful the table addition. in the presence useful There is another to that. if the is on. off. irrelevant By applying tically of it entails that the absence see that (2). is afflicted does not have any direct influence causal upon the illness of the other.C). situation. are made with it turned turned on. independent in as the relationship among the conditional another. and (2). which controls of random device some sort might imagine . calling according way to look at the multiplication (1) equations theorem. magnet is off when the dice these is turned when the electromagnet assume that. as standard dice behave exactly 6 is of getting double and the probability die is 1/6. since positive of C than in the absence of C. 1/36. probabilities Relations assert that C is a (3) and (4) simply (2) states. to the same argument to one another. In that each one has a tiny magnet embedded in it. in the absence gastro-intestinal of one when C obtains--A and B are also cause C--that is. it is considered as C. if the second event dice standard These dice are like of type B.C) 0. is an event that most.BIC) we see that. shows that C screens A similar argument to equation irrelevant. P(BIC) = P(A|C) x P(BIA. seen that conditions It is easily (1)-(4) is 1/4. magnet embedded in its when the electromagnet If the dice are rolled switch. let us assume for the sake of simplicity Moreover. 6 with either of getting Let are fulfilled. equation the probability of each is greater cause of A and B. with either the chance is turned If the electromagnet on. 6 upperwith side die comes to rest of the type A. off To screen off B from A. with side die comes to rest If the first gether. (8) off A from B. Let us further 6 of getting The probability dice.

57 We can readily see condition. 6 is the arithmetical of double average bability which equals of 6 on one die were independent If the occurrence 5/36. Nevertheless. The detosses the probability between the not because of any physical interaction arises. Reichenbach disthree situations: tinguished E E A OB A B A B C C (i) (ii) (iii) . one may win a prize. anywhere along Reichenbach I believe--that forks claimed--correctly. the other was using that the unhappy coincidence source. as explicitly mentioned in the mushroom Likewise. The same consideration to the earlier. satisfies relation (5). of and this arises from a clustering independent. the overall probability be 1/3 x 1/3 = 1/9 ~ 5/36. in addition to relations (1)-(4). they are. pendency on but because of special conditions which obtain dice. in the twin quasar supply Similarly. of whether that the overall regardless probability is 1/3. on causes. background certain of the tosses. the example as of course it must. The coincidence a resulted from the fact that common set of background conditions a common food obtained. arise out of a given so also may we find a common effect common cause. proof 1/4 and 1/36. independent and they are statistically if it is on. When the two students each copy from a paper contrived. but that overall independent nonindeed. poisoning upon the illness of the other. on It may initially seem counterintuitive to say that the results the two dice are statistically if the electromagnet is off. this which realizes equi-probability of 6 on each die. non-independence of sixes which is due simply to the fact that in a subset of the class all of 6 is enhanced for each die. results resulting by getting two dice which add up to seven. might have been avoided. the overall the electromagnet is on or off. physical the process and that by which is produced by which one of the papers the other is produced--in if either had been aware that student fact. In addition. from two distinct For example. from which both ate. conjunctive an important Just as we can have two effects which possess asymmetry. applies less cases. example. in a fraternity between there is no direct interaction file. 6 would of 6 occurring of double on the other. the two images are formed by two separate radiation which processes come from a common source. but which do not directly with interact each other the line. they are not independent. namely. Thus. the illness of one friend had no effect example.

m. a man demonstrates. type fork. out any common effect from A and B conjointly. makes a cup of about 9:00 a. and (iii) events A and B have both a common cause C and a common effect E. and C. and settles coffee. are an essential connections causal part of the causal fork cannot of a conjunctive C at the vertex out them. improbable that such coincidences are explained only in terms of common causes. from the cause is transmitted fluence and withfork. and on these at 8:00 a. promptly a has arrived and prepared his secretary somewhat earlier mornings he is met at on just these fresh Moreover. two events (ii) arising A and B which.58 to two separate rise A and B. circumdental I doubt that. are said to explain it follows forks otherwise coincidences. the common cause C' is. closed Reichenbach's was that situathesis asymmetry of type tions never (ii) forks. if we consider shows up on that his associate made when he arrives (A) and the fact then it might be to be explained. Now. at his office who usually arrives On some occasions. point forks. he arrives very same however. is an common cause The actual either of A or of B. B. the coincicommon cause than E which explains C. C with B.. mornings. day before B. down to read the morning paper. there must also be a could form a conjunctive Here.10 A. while (i) and (ii) (iii) on both ends. he always catches noted that on such mornings In takes the 8:00 a. C with A and another causal connecting connecting process inthe mechanisms constitute These causal by which causal processes These to each of the effects. the event . open or closed always by a fourth of the same temporal that in situations direction.m. that morning (B) as the coincidence the 7:00 a. appointment entirely of A. works at a different who normally his office by one of his associates is already the fact that the coffee location. even in these occurrence fork.m. of course. by his secretary. C cannot be conbut obviously fork satisfying a conjunctive (1)-(4). B. and it is C rather of A and B. represent conjunctive conjunctive which are open are always forks to the and never open to the future Since the statistical relations which are found in conjunctive past. a cause sidered made the a telephone event different C'. Reichenbach allowed E A and B along with their common effect the two events (iii). and E can form a conjunctive stances. bus (C).m. relations to suppose that the statistical It would be a mistake common are sufficient to characterize in conditions (1)-(4) given as an exof correlated in their role as explanations causes effects. He characterized situations is as "open forks". C in a conin which the event the cases to distinguish In order from those in which common cause a bona fide fork constitutes junctive must be a suitable that there let us add the condition it does not. that an even stronger I shall not try to argue it here--namely.. clearly ample. and C form it is plausible this enough to suppose example. I believe claim is warnever common effects. that conjuncranted--though in tive whether event. in the absence of a common cause C. in fact. namely. he usually on other while mornings that A. pot of coffee. jointly a produce common effect a combination of (i) and (ii) in which the E. with(i) a common cause C giving effects. bus (C).11 Consider due to Ellis Crasnow.

only when two causal processes . going is also evident that A. different. both move on as if no such intersection had ever occurred. tain. the intersection of their results in a change in the motion of paths each which would not have occurred if they had not collided. executes We may reasonably assume that the probability tries the the the 8-ball into the pocket is 1/2 if the player going of the cue ball into the pocket and that the probability shot. This will for example. is in terms of forks. the shot that the cue ball will the pocket that bability go into given far has been attempted and that the 8-ball has dropped into the other corner 1). interactive forks violate condition (1) in the definition of conjunctive forks. In the case of the two pool balls. background there is a direct interaction--a the two causal collision--between proin portions cesses which consist of the histories of the two balls. If the paths of two airplanes. when both processes are pseudohappen. In some cases. however. forks common cause A and B C does not statistically screen the two effects from one another. way to look at interactive intersections of processes. Forks Interactive There is another. flying tions at different on a clear altitudes one another. B. In discussing out the conjunctive I took some pains to point fork. pocket (appx. shadows have passed the intersection. A relative novice a shot lie the 8-ball. their respective states of motion are altered. of the balls. the cue ball and fork. by such cases be called interactive the Since (see Salmon 1978). For this I have suggested that forks which are exemplified reason. one of the far corner which is intended to put the 8-ball into pockets. in situations in which separate and disthat forks of that sort occur tinct which do not directly arise out of special interact. The best I believe. cross shadows on the ground may coincide But as soon as the momentarily. another.59 qualify 4. of the pool balls. for the 8-ball in a "scratch". resulting for the cue ball into the one corner let B stand dropping pocket. and C do not about 1/2. two prospatio-temporal cesses without in modification may intersect producing any lasting either. when the player lision between and the 8-ball the cue ball which occurs of the shot. for the colinto corner and let C stand the other dropping pocket. corner Let A stand pocket. that the cue Given that the shot is attempted the probability will ball fall into is not equal to the prothe pocket 1/2) (appx. In the example conditions. It is immediately for C does not screen A and B from one constitute a conjunctive fork. the day. type of common cause which basically used to define the conjunctive the statistical conditions violates a simple Two pool balls. Energy and momentum are transferred from one to the other. processes. Such modifications I shall mainoccur. in different direcprocesses. Consider example. attempts upon a pool table. if the 8-ball into but given the positions falls one is almost to go into the other far corner the cue ball certain pocket. however. as a common cause. If either or both of intersect.

which occur In many cases--and and/or perhaps all--energy momentum transfer and the correlations the modificabetween occurs. for practical photon purthe inat rest. type collision light rays normally pass right one another without effect one of them. analogous that the equality of the conjunctive (1) in the definition fork. tions are direct of the respective laws. through upon either any lasting The fact that two intersecting are both causal is a necessary processes but not sufficient of the production condition of lasting in changes them. has produced action is another of red glass and a piece is a process. of two processes. in many cases. and if that modification persists. Then. the momentum change Similarly. process. Causal processes mentioned causal concepts in processes and changes is propagated.12 conservation consequences This is nicely of an energetic illustrated by the Compton scattering off of an electron which can be considered. are produced by causal The distinction basic notions. are the at the outset. no such mutual modiprocesses fication occurs. these between the close relationship in terms was formulated and pseudo-processes between causal processes in a A mark is a modification of mark transmission. foregoing by the interaction I shall with say the the two fundamental at least I have now characterized. cations say that idea a causal This is the basic intersection constitutes interaction. of white light . it is entirely for two causal However. given that the change other process. influence means by which causal to see We are now in a position interactions. of the intersection for the event stand consisting in the in one and B for a modification Let A stand for a modification to equation we find a relation other. partially. with other occur when they intersect in processes Modifications proof intersection. except is replaced by an inequality: P(A.13 modificorrelated and they undergo When two processes intersect. any subsequent the extremely occurrence of a particle-particle Barring improbable between two photons. if the modifications cesses. in energy The difference between poses. I shall the which persist after the intersection. after there is some correlation between the changes in them. energy the photon is exactly compensated by the momentum change in the electron. a causal in each interaction is process of the produced sort. persist beyond the point and the interinteraction a causal constitutes then the intersection For example.BIC) > P(AIC) x P(BIC) (9) Moreover. initially hv and the scattered hv' is equal to the kinetic photon coming photon in of the recoiling electron. When two causal intersect and suffer modifications processes lasting the intersection. a pulse marks which are transmitted. Let C causal behind what I want to take as a fundamental concept. of the criterion the mark is transmitted.60 the intersecting are pseudo-processes. possible processes to intersect without modification in either.

5. comes to this but it gallery. pulse process. interaction with another accomplished by means of a causal process. go through two holes. transmitting the intersection.. values. we may use the same probability Indeed. processes reason to believe the world abounded in causal that every processes and causal interactions were any human agents there to long before perform experiments. we may often in producing take an active role a mark in order Although to ascertain whether a process is causal (or for some other purpose). are causal in a manner which or pseudo. modified) processes intersection with the other. We live in a world which is full of processes (causal way. in the for each has been marked (i. If we let A stand for a fragment the bull's-eye of target fragment A. that we have installed between the knife-edge and the concrete. We may. setting P(AIC. Relations Between Conjunctive and Interactive Forks that we have a shooting with a number of targets.61 two processes If these if the light intersect--i. turn the presentation around in the following or pseudo). and C for the severing by the we have an interactive fork quite to the example knife-edge. the operation of marking is a process Indeed. If two processes intersect does qualify as a causal then we may conclude that both interaction. no challenge to her. and each fragment will strike its with respective bull's-eye .e. occurs which does not qualify as an interwe can draw no conclusion as to whether the processes involved action. Annie Oakley. each process capable a mark generated since each one has transmitted in marks. Although absorbing into the surrounding environenergy ly acquired may soon be dissipated the glass retains some of the added energy for some time beyond ment. therefore. B for a fragment striking striking the bull's-eye of target of the bullet B..A) P(A C) = P(BIC) = 1/2. If an intersection not. the red glass--then the light becomes and remains pulse goes through the filter an increase in energy as a result of red.B) tical off obviously fails. imagine a steel with two holes in it. analogous of the pool balls.e. = P(B|C. the actual moment of interaction. screening We might. of shows itself Thus. while undergoes which impinges some of the light the newupon it. with one another. to make the situation a hardened steel in such a posiis installed interesting. and these intersections processes undergo frequent Some of these others intersections constitute causal do interactions. while Statis=1. If the shot at the knifetargets plate then the two fragments of the bullet will the edge is good. it should be obvious that human agency no essential plays part in the characterization of causal We have or causal interactions. knife-edge tion that a direct hit on the knife-edge will sever the bullet in a way which makes one fragment hit the bull's-eye of target A while the other hits the bull's-eye of target B. and each process the mark beyond transmits the point of intersection. for she can invariably hit the bull'spresents at which she aims. eye of any target So. Suppose gallery The famous sharpshooter. To make the situation however. are causal. consider another event C*.

Light fredo not furnish Filters red. to turn red light quencies to it is best of irreversible of the physics processes. which was described asymmetry seen by considerThis is easily the same sort of temporal asymmetry. later. makes This example its of the other target. Our ordinary language causal If. is reversed. sides on the two characteristics has different and which comes earlier We do not try to say which part of the process a mark is To erase The same is true when we speak of marking. symmetric function. of the processes which are the states prior To avoid begging states. and C* will interaction between to the physical which is subsequent to a situation has been of the bullet. with temporal is infused causal asymmetry. we should say that two processes of the intersection. provide processes asymmetry. quite and the conjunctive direct which characterizes interactions. background between of contrast There is a further conjunctive point important a kind of temporal forks forks. two billiard between collision ing a simple of states C precedes if a collision in reverse. this type can occur in which C can occur then a collision motion A and B in the two balls. except like of motion states just and causal interactions Causal the collision. absorb filters are irreversible. their Then. we may say.62 of the two Let C* be the event to 1. determined that we have already the words suggest their interaction. and interactive possess Conjunctive do not exhibit forks Interactive above. direction. form a conjunctive B. that forks fulfill furnish conjunctive asymmetry. finding fragment bability the interactive between I believe. virtually equal probability holes. so that they some frequencies. that would have no effect ments with a powerful upon the prolaser. the distinction fork. do not. A. the details and to take terms. in which P(A|C) the cases it is advisable and P(BIC) do to with conjunctive In dealing to our attention restrict . Perfect Forks and interactive forks. conditions. the bullet its way independently and each is going cut into two separate pieces. temporal which are the subsequent and each of the processes intersect. vivid. a temporal is to presuppose erasing of interactions kinds that certain we know on other grounds course. one of the fragto vaporize decide Even if we should of the other. or a mark. of motion that the direction A and B. missing into white transform light we have gone into But until into white. it to basic in applying concepts. the parts By the time we get to C*. be careful but we should of as a result are modified we say that two processes for example. temporal 6. respective through fragments going because C* refers That happens fork. and to the interaction. precede for temporal a basis in and of themselves. questions. physical under special which characterizes arising processes independent fork. to speak of imposing of imposing reverse the exact temporal of In many cases. of A collision balls. in temporally interactions symmetric of causal think as which are furnished processes connections by causal the causal do not interactions and causal Causal processes connections.

interactive Thefact that relations (1)-(4). fied does not constitute a sufficient basis for making a judgment about the temporal of the fork. forks also forks in the limit. The Causal In everyday Structure life. Perfect forks need to be distinguished from the other two types mainly to guard against this of confusion. that the perfect fork is a limiting case of a conseparate grounds. Given that shooting gallery makes perfect her skill in her case). gastro-intestinal The main point to be made concerning is that. the results her trials to claim interactive forks. we when we talk . of the World about cause-effect relations. longer fork. orientation on Only if we can establish. which are used in the characterization of conjunctive are satisforks. source possible 7. until the moment that of she has perfected her technique. than an interactive) can we conclude that the (rather junctive fork. It is. lates relation (9). that she reSuppose turns to the special time after time. from the statistical alone whether the fork should be relationships considered or conjunctive. limiting direct interaction between the processes to the leading respectively two cases of severe distress. Consider Conjunctive perfect yield the example of illness due to consumption of poisonous mushrooms.63 of not assume either is that the relation the extreme values zero or one. Consider the Annie Oakley example once more. we high correlation can ask no more. It would be absurd exemplified when she achieves no the splitting of the bullet that. The essence of the interactive fork is to achieve a conjunctive if the correlation between two results. ill. she improves (at least practice until she can invariably in the manner which rehit the knife-edge in the two fragments sults their respective Up finding bull's-eyes. It therefore seems best to treat this case as a third special type of fork--the perfect fork. equation If probability values were normalized to some (9) must be violated. value other than 1. is perfect.BIC) = P(AIC) x P(BIC) = 1 is a conjunctive case of either a special case of equation or (1) an interactive and it vio- (10) a limiting may represent even though (10) fork. The main reason P(A. however. If we assume--what is by no means always the case--that anyone who consumes a significant amount of the mushroom in question is certain to become violently then we have another instance of a perfect fork. event at the vertex is a common cause rather than a common effect. forks when the perfect take on the limiting it is impossible to tell probabilities values. an arithmetical accident that when perfection while the inequality (1) is fulfilled occurs. a causal as a constitutes but must now be regarded interaction. one might say. Even when these values there no is still obtain. perfection. that result would not obtain.

which in consisting charges to emission of light turn leads from a filament. Homicide by shooting still another An interaction between a gun and a provides example. on common causes difference between There is a striking conjunctive on the other.64 think not necessarily of situations in (though typically invariably) which one event the cause) is linked to another event (which we call the effect) Each of the (which we call by means of a causal process. a switch makes the light In this an interaction case. and which do not interof one another which are physically independent set of backout of some special arise with each other. physical cases--by most. a vacuum and a material wave propagating moving withparticle through interforces out any net external Barring any causal upon it. (or its counterpart The correof collisions. governed--in . Conjunctive and order. in this two events which stands relation is an interaction two between We say. I believe. act directly conditions that such special The fact background ground conditions. it will be recalled. space For another the window (an interactive we say that fork). of structure in the production a vital role tive forks--play two or more processes. example. actions by Maxis governed and the material well's by Newton's particle equations Causal in relativity law of motion first theory). that the window (or more) intersecting processes. are in terms of conjunccharacterized common causes--as 3. between a switching mechanism and an electrical circuit leads to a process of a motion of electric in some wires. which effects of a correlation is the source exist among the various of the common causal backin the absence would be utterly improbable ground. In the conjunctive fork. processes. was broken by boys playing In this baseball. for example. a bullet from the gun to the vic(a causal process) cartridge propels then interacts with the body of the victim. other times causal a basis kinds and various processes three for understanding are the means by which structure from one spacetime of region and order the universe are to in terms of interactive as explicated 2. Causal processes or transmitted propagated and places. sorts are typified interactions by various are involved in the processes which occur the changes between lations conserfundamental if not all. acting wave is governed the electromagnetic in both cases. processes) by causal propagated forks. of causality: fundamental aspects 1. Causal interactions. and interactions the one hand and causal processes seem to be governed interactions and causal Causal processes by basic Conforks. turning go on. ball and a collision of the ball with (a causal through process). of a bat with a ball collision of the (an interactive the motion fork). in structure the means by which modifications constitute (which are produced. tim. in ways which do not apply to conjunctive laws of nature an electromagnetic of causal two paradigms sider namely. is a there situation. where the bullet of characterization The foregoing of causal forks provides.

in such condition the quasar are situated that and the elliptic galaxy from arises a way that directions light coming to us from two different of its a source which radiates from extended portions quite uniformly surface. It should that. the probability of type. junctive chap. conjunctive processes Recall forks conditions.14 are 5 molecules of way--a improbable each type. paragraph. processes conjunctive in the preceding laws of nature However. the right-hand that are 2 there compartment. example. Suppose that we have a box with two compartments connected by a window which can be opened or closed. (1956. depends upon an antecedent state of the world which displays de facto orderliness. The box contains numbers of nitrogen equal and oxygen molecules. What would be a just prior if attributed coincidence to chance is explained hopelessly improbable on the supposition that of each of straightforwardly separate supplies the gases is available The explanation beforehand. Suppose each type. Although to suppose that aZl fundamental it seems plausible interphysical can be regarded actions as exemplifications of the interactive fork. Avogadro's number of molecules--forget it! In a case of this sort we would conclude without hesitation the system that had been prepared by closing the window which separates the two compartments. have said about entropy and the second chap. a non-lawful fact that two members of the same class happen to have access to the same file of term papers. If the box the same sort contains 50 molecules of segregation would each 2 If the box contains 10 -. the chance of finding all of the N2 molecules in the left cules are molecules in of compartment and all of the 02 molecules in the right is a bit less of be than 1/1000--fairly improbable.65 I am not prepared to argue the case in detail. this point articulates in his "hypothesis of the manner in which the branch new sorts It . VIII) law of thermodynamics. In the mushroom poisoning it is a non-lawful that the two participants fact example. sup together out of a common pot. and all of the (N2) (02) N2 molecules are in the left-hand while all of the 02 molecompartment. Consider the simplest sort of example. and GrUnbaum (1973. -4 = 1/16 that of 2 in just that probability they would be segregated somewhat If there coincidence. The window is open. There is a close what has just been said about conbetween parallel and what philosophers forks like Reichenbach III) (1956. apply as sketched in contrast and causal to causal interactions. hardly require and causal forks involve causal the interactions. Reichenbach generalized structure" ?16). been opened to our examination of the box. depend crucially upon de facto background it is In the plagiarism some of the examples mentioned above. vation laws. common causes are not nearly as closely tied to the laws Conjunctive to the extent mention that of nature. If they are distributed there is a randomly. and by filling each with its The window must have compartment respective separately gas. In the twin quasar it is a de facto example.extremely improbable.

process two distinct in then we. relations to same time. strictly speakconjunctive of the common cause. while interaction exBasically. Given the fundamentality and ubiquity of entropy considerthe foregoing that the conjunctive ations. simply tion of two or more processes which produces in each of lasting changes them--then we have simultaneity. we are implicitly which involve statistical generalmaking assertions and it seems to me. It seems to might be simultaneous me that the foregoing discussion a reasonable of resolution provides this If we are talking about the typical cause-effect controversy. they are produced the exploitation of the available in the immediate enthrough energy vironment. propagation lapse hibits the relation of simultaneity. it should be noted. do not states) (highly in isolated but rather. In the thermodynamic we say that low entropy states ordered context. it is the intersection with the red filter which produces the red light. which I characterized above in terms of a causal situation. which makes possible both the kind of order we find in systems in states of low entropy and the kind of order which we get from conjunctive we must ask the cosmologist how forks. that order can emerge except from de facto order. An individual a causal fluence. Another traditional has centered of whether dispute upon the question statements about causal to individual relations or events. answer. characterize Thus. we are dealing with a causal interaction--an intersechowever. of the white Thus. since each process intersects the other at the same time. we have used statistical and interactive forks.66 of order arise from pre-existing states of order. and the light pulse red at the very time of its passage becomes the filter. can sustain connection between process an individual cause and an individual Statements about such effect. they hold properly only with respect I believe. fork also parallel suggests has basic If we wonder about the original physical significance.are dealing with cases interactions. in many instances. Concluding Remarks There has been considerable since Hume's time regarding controversy the question of whether causes must precede their or whether effects. Throughout this discussion of causality. emerge spontaneously systems. have both particular izations. causes and effects with each other. suppose prior 8. stitute the causal A causal connections between cause and effect. pertain whether to classes of events. source of order in the world. like the principle when we invoke something ing. proincess is an individual and such entities transmit causal entity. type. I have argued that causal conprocesses. relations need not be construed as disguised At the generalizations. and why the universe evolved into a state which is characterized by vast of available It does not seem plausible to supplies energy. Causal relations. joining which the cause must precede the effect. the foregoing furnishes a straightforward account Again. I have laid particular . light through involves of time. general aspects. for causal over a propagation finite time interval is an essential of cases feature of this If.

Some are Salmon In Salmon I have attempted to provide a detailed of (1977) analysis the notion of transmission or propagation of causal influence by causal and a justification for the claim that processes. This 8 C2. has. SES-7809146. If is other demonstrated potential and the by Reichenbach common causes corresponding the magnet exist (1956. (1980) given in for and systematic survey a survey (1980). will the question and see what a very different causation'. qualify The variables in the probability A. an excellent historical by the National Science of for (1974) approaches. p. of statistical approaches. B. a 'rope of the phrase into introduced discussions upon this subject. if I understand events nections between intervening by interpolating non-contiguous the same quesThis approach events. form obtain. rather than events 1977). whether it would not be possible through as of an event within the framework processes ontology. by considering for supposing that this of events. we can will die C3 . if another the chain.. program could not be carried source of difficulty for bother to do so. along we look to processes to provide connections the causal (see Salmon in my opinion. is urged The reader to compare the illuminating account of causal . I see no reason series continuous but I would be inclined to ask through. One important why we should conto account for causal him. 160-161). Notes This material Foundation under See the See Mackie various Salmon is based upon work supported Grant No. event to the causal influence is transmitted from one intervening I believe. pp. The difficulty is circumvented.. aspect wear. seemed only to raise precisely for one had to ask how tions connections between about causal events. upon the role of causal It might be asked "event of the so-called abandonment ontology". relations in one We are havior of that assuming the other die. 320). as statistical selves are understood frequencies. upon processes Focusing As John Venn said in enormous value. is that he tried Hume.67 and I have even suggested the stress processes. to carry the same analysis. heuristic (if not systematic) so often 'chain of causation'. does not affect the bea partition C1. "Substitute for the time honoured 1866." (Venn 1866. they legitimately as causal connections. C which appear expressions are taken by Reichenbach to denote and the probabilities themclasses.

irreducibly fork even in principle. was incorrect. to difference it However. be doubled.68 asymmetry symposium. Humphreys in his contribution to this Paul 1I had previously Humphreys kindly attributed pointed this view erroneous out that my attribution role of energy to Reichenbach. able. the example. and momentum trans- in Salmon (1978) of Compton scattering As explained the example has not analyzof being and thus. but fer 12For a valuable in causality. the advantage statistical. . as a perfect (discussed below). 1Strictly. N2 in the each of the probabilities for a distribution consisting right would be just as mentioned of all in this should example and all 02 in the left segregation that a factor as of remarkable is a form of obvious that considered 2 makes no real in the text. discussion see Fair of the (1979). offered by Paul W.

An Inquiry Concerning of the Scientific and Addresses Explanation. Association Philosophical (Reprinted C. Dordrecht Mackie. of Chance. Hans Reichenbach: (1979). 243(5): and the Flow of of a Gravitational Fair. Its Berkeley and Los Human Knowledge.L. The Logic (1866)." Proceedings in American 51: 683-705." Scientific David." ---------------. Theory of Space-Time. Press. Clarendon Hans. 219-250. Philosophical Venn. (1956). Quarterly "Probabilistic 61: 50-74. Causality. (1977). Salmon. Salmon. (ed.)." Erkenntnis (1973).69 References Chaffee. (1974). Macmillan London and Cambridge: Winnie. (1980). Grunbaum. of Causal Influence. Theory Scope and Limits. Press.." Foundations of Space-Time Theories. Reidel Dordrecht Co. Co. Empiricist. Clark and John Stachel. University Angeles: Russell. "Why Ask 'Why?' . . (1979). Lens. Pages 134-205. Publishing Pages 403-425. "Causation Energy. (1978)." Pacific John. Wesley Logical and Boston: D. Minneapolis: University Press. and Co. Simon and Schuster. Studies in the (Minnesota of Science. (1948). (1980). J. "The Causal In (1977). Frederic H.) Edited Philosophy by John Earman. Reidel Publishing the Universe. Reichenbach. New York: Wesley Philosophy C. American "The Discovery 70-88. of Minnesota Glymour. Vol. Philosophical and Boston: The Cement of of Space Problems D. of of The Direction California of Time. Oxford: and Time. Science "An 'At-At' 44: 215-224. John. 14: Jr. VIII. Bertrand.). Adolf. ---------------. 2nd ed.