03. black - why cannot an effect precede its cause

8/12/2019 03. BLACK - Why Cannot an Effect Precede Its Cause

1/11

Why Cannot an Effect Precede Its Cause?Author(s): Max BlackReviewed work(s):Source: Analysis, Vol. 16, No. 3 (Jan., 1956), pp. 49-58Published by: Oxford University Press on behalf of The Analysis CommitteeStable URL: http://www.jstor.org/stable/3326929 .Accessed: 04/01/2012 08:28

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

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 formsof scholarship. For more information about JSTOR, please contact [email protected].

Oxford University Press and The Analysis Committee are collaborating with JSTOR to digitize, preserve andextend access to Analysis.

http://www.jstor.org
http://www.jstor.org/action/showPublisher?publisherCode=ouphttp://www.jstor.org/action/showPublisher?publisherCode=anacomhttp://www.jstor.org/stable/3326929?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3326929?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=anacomhttp://www.jstor.org/action/showPublisher?publisherCode=oup


2/11

ANALYSIS 16 3. JANUARY 1956

The publishers and the Committee of manage-ment of Analysis record with deep regret theirsense of the grave loss which the philosophicalworld has suffered n the death of Miss MargaretMacdonald, the editor of Analysis.

WHY CANNOT AN EFFECT PRECEDE JTS CAUSE?

By MAX BLACK

SUPPOSE a child were to ask, "Why must Monday alwayscome before Tuesday, nstead of the other way round? Whycan't Tuesday sometimes be the day before Monday or a change?"Such an extraordinary uestion would be most difficult o answer

in a sensible way. But once we had recovered from our initialstupefaction, we might make a first attempt as follows : " Thereis no ' must' about it at all. The first day of the week is simplycalled 'Monday' and the second day 'Tuesday'. In just thesame way one child is called 'Jack' and another 'Mary '. Thatis how the names are used. But if you wanted a change, youcould all the first day ' Tuesday and the second day ' Monday .And then Tuesday would come before Monday. Of course, ifyou did make this change, other people would probably notknow what

dayof the week

youwere

talkingabout. So

perhapsyou had better do as they do, after all "This answer treats the original question, "Why must

Monday always come before Tuesday?" as meaning: "Whatreason is there for using the words 'Monday' and 'Tuesday' insuch a way that the earlier, not the later, day is always called'Monday'? " The questioner s taken to be puzzled about onefeature of the use of certain words; and so the answer is aboutwords, too. This interpretation of the original question pre-cludes a straightforward answer to that question. We haverejected he sentence, " Monday must come before Tuesday ".And instead of giving a reason for the present use of the words,'Monday' and 'Tuesday ', we have said that there is no reasonfor their use.

Sometimes, reasons can be given for particular eatures inthe use of certain words. For instance, if somebody were toask, "Why must a final examination always be given at the end


3/11


4/11

WHY CANNOT AN EFFECT PRECEDE ITS CAUSE? 51

fully, we must bring to light the considerations hat make itappear sensible to the speaker. And this may be enough, as inthe case of the question about the feathers, to show that it isnot a sensible question, after all.

What are the considerations hat make the question, "Whymust Monday always come before Tuesday?", seem sensible tothe questioner? What is the supposed discrepancy that givesrise to the question? I suggest that the regular order in whichthe two days come is being thought of as an unexplained andremarkable act. Lightning precedes hunder s a matter of act,for which an explanation might reasonably e demanded. And

so, perhaps, he child thinks t is a peculiar matter of fact thatMonday always precedes Tuesday, and wants to know why.So long as no explanation s forthcoming, he regular rder

of the two days appears rbitrary. s if there ought to be somehidden factor which, being brought to light, will render heuniformity ntelligible. But this demand (if, indeed it doesunderlie he question) t is impossible o satisfy. It would beabsurd, or example, o say " Monday comes first because hatis the day on which people start he week's work " ; but it also

seemso me that

every nswer of the form"

Monday omes irstbecause X " would be absurd. It makes no sense to try to givethis kind of reason; for it makes no sense to treat a necessarystatement s if it were the expression f an empirical egularity.

I cannot call to mind any philosopher sking why Mondayalways comes before Tuesday. But many philosophers haveasked, "Why must there be just three dimensions f Space?and "Why cannot thought be extended? and "Why is itimpossible o remember he future? and "Why cannot aneffect

precedets cause? ".

All of these questions an be treated n the same way as theMonday-and-Tuesday uestion. Construed s questions aboutthe use of words, aving he form " Why are such and such wordsused in such and such ways? " they look trivial and give rise toequally rivial answers. Again, the necessary ruth n questionmay sometimes be logically inferred from other necessarytruths as we did when we offered kind of proof that Mondayprecedes Tuesday); but this, too, will probably not satisfythose who ask such " deep " questions. Philosophers who arepuzzled by the apparently rbitrary haracter f some necessarytruth wish to know what it is in the world hat underlies henecessary ruth. But this wish cannot be fulfilled. There s nomystery o be unravelled--or, ather, t is absurd o talk of therebeing a mystery.

A man who asks "Why cannot an effect precede ts cause?"has more excuse for perplexity than a man who might ask why


5/11

52 ANALYSIS

Monday comes before Tuesday. The ways in which we usecausal words are so complex and flexible that we cannot answer

all questions about their uses without hesitation and reflection. Itis therefore not unreasonable or a man to wonder whether inexceptional circumstances he labelling of an earlier event asan effect of a later one would necessarily constitute a violationof the current rules for the use of causal words. It has beenseriously suggested that such curious phenomena as thosenowadays called cases of " precognition " provide just this kindof exception to the general rule about the temporal order ofcause and effect.

Let us find out. Since we are to examine a necessary tatement,("An effect cannot precede its cause "), it is of no importancewhether he alleged " parapsychological henomena " do actuallyoccur or not. We have only to investigate whether, if theyshould occur, they would properly be described as cases of aneffect preceding ts cause.

Imagine the following remarkable tate of affairs. First, wehypnotize a certain man-Houdini, say--ask him the question," How will the penny fall? " receive one of the two answers," Heads " or " Tails

",and then

proceedto toss a

penny exactlyone minute later. Repeated trials then establish that Houdini'sanswer, in the circumstances related, always agrees with theoutcome of the subsequent rial. Under hypnosis he can alwaysforetell how the penny will fall.

Such a state of affairs would be more remarkable han anyso-called "precognitive " phenomena already reported in theliterature of psychical research. Yet it is logically possible thatwe should discover such a uniform correlation between answersgiven under hypnosis and the results of subsequent trials withcoins. And we can properly ask, here and now, how such astate of affairs ought to be described.

Let us call the event consisting in the hypnotized subject'sreply, A (for 'answer') and the event consisting in the toss ofthe coin exactly one minute later T (for 'toss '). The questionbefore us is: "In the circumstances described, would it beproper to say that the later event, T, caused, or partially caused,the earlier event, A ?"

Before taking the drastic step of answering this question inthe affirmative, here would be several alternatives to be con-sidered. Surely our natural response in the circumstancesimagined would be to suspect some concealed trickery. Ourfirst thought would be that a confederate of Houdini wasarranging or the penny to fall as predicted. If so, the answer, A,would be a partial cause of T; and we would then have a caseof a cause preceding ts effect as we normally expect. If we are


6/11


7/11

54 ANALYSIS

that the event A (Houdini's answer) is not caused by any priorevent. (For we have already ruled out the possibility that it

had the same cause as T and also the possibility that it had acause independent of T's causes.)The final description of the hypothetical case is, accordingly,

as follows : When Houdini is asked under hypnosis the question" How will the penny fall?" there is no earlier cause for hissaying heads rather than tails. When the penny is tossed aminute later and agrees with Houdini's answer, there is a suffi-cient cause for that outcome of the toss, the cause in questionbeing causally ndependent of the previous answer. Supposing

all this to be the case, the question is, " Is A (the earlier event)properly described as caused by T (the later event)? "There is at least one good reason for saying " No ". In the

circumstances described, we can wait until A has happened andthenprevent . For we assumed hat the causes of T (the way thepenny was thrown, the air currents n the room, and so on) werecausally ndependent of A. So, A having occurred, t is withinour power to prevent T's occurring-by simply not tossing acoin.

Thisbeing so,

a man who insistsupon describing

he circum-stances as a case of effect preceding cause must qualify hisassertion that the character of T causes -A by adding that thisis so only when T occurs. In other words: T causes A if Toccurs, yet A may occur without T occurring at all. So A maysometimes occur without being caused either by an earlier ora later event, and yet on other occasions, a precisely similar A,though again not caused by an earlier event, is now held to becaused by a later one. This would be hard, if not impossible, toreconcile with our present uses of causal terminology.

But there is a still better reason for refusing to call T a causeof A. If T's causal antecedents are independent of A, as wefound ourselves required to stipulate, we can arrange for T todisagree with A. Thus, it would be theoretically possible tolearn to toss the coin so that it came down heads or tails as wepleased; all we need to do is to wait for Houdini's answer,and then arrange or the coin to fall contrary o his prediction.If we can do this, the stipulations or the supposed precognitionare logically impossible of fulfilment, because Houdini's answer

will not always agree with the subsequent trial. On the otherhand, if we find that once Houdini has answered we cannotarrange or the penny to come down as we please, we shall becompelled to say that the causal antecedents of T are not inde-pendent of A. We shall have to say that Houdini's answerexerts a causal influence of an esoteric sort upon the subsequenttoss of the coin. This would be extraordinary enough, to be


8/11


sure. Yet it would not be preposterous, as would be the logicalabsurdity of saying that an effect precedes ts cause.

I have been trying to show that a full description of thesupposed case of " precognition" involves self-contradiction.Anybody who says that a later event is a sufficient condition forthe occurrence of an earlier one, while also saying that the laterevent has anterior causes that are causally independent of theearlier event is contradicting himself. But unless such self-contradictory stipulations are made, there is no temptation tospeak of an effect preceding ts cause.

If I am not mistaken, a similar analysis will apply to all casesof so-called "

teleologicalcausation ".

Supposea man

saysthat

the nest-building behaviour of a bird is caused y the impendingarrival of offspring. He is then committed o saying (i) that thereis no prior reason for the nest-building, no earlier causal factorsthat would explain it. (For otherwise it would be redundant oinvoke a further cause.) (ii) He is committed to saying that ifthe fledglings are hatched the nest must have been built (forotherwise the arrival of offspring would not be a sufficientcondition for the nest-building). But in that case, the naturaland proper way of describing he situation s to say that the nest-building is a partial cause of the arrival of the fledglings, and notvice versa.

I have been arguing that our established uses of the wordscause ' and ' effect leave no possibility open for applying them

in such a way as to describe an earlier event as an effect of a laterone. If I am right, to speak of an effect preceding ts cause is asabsurd as to speak of Tuesday preceding Monday. To see theabsurdity we have only to remind ourselves of some of thefeatures of our uses of causal language. And this is what I havetried to do.In a recent discussion of the same subject (Aristotelian ocietySupplementary olume 28, 1954, pages 27-44), Mr. M. A. E.Dummett agrees after some discussion hat an effect cannotprecede its cause (p. 31), but goes on to say that this conclusiondoes not end the matter. Dummett suggests that in certainsituations we might still want to say that a later event stood toan earlier, not, to be sure, in the relation of causation, ut inwhat Dummett wishes to call " quasi-causation .

Dummett says : " We may observe that the occurrence of anevent of a certain kind is a sufficient condition for the previousoccurrence f an event of another kind; and having observedthis, we might, under certain conditions, offer the occurrence ofthe later event, not indeed as a causal, but as a quasi-causalexplanation f the occurrence f the earlier. There are threesuch conditions which would have to be fulfilled if it were to


9/11

56 ANALYSIS

be reasonable o offer such a quasi-causal xplanation. First, theoccurrence of the earlier event, which was to be explained byreference o that of the later event, would have to be incapable,so far as we could judge, of being (causally) explained by refer-ence to simultaneous or preceding events; there must be nodiscoverable explanation of the earlier event which did not referto the later. Secondly, there would have to be reason forthinking that the two events were not causally connected; i.e.,there must be no discoverable way of representing the earlierevent as a causal antecedent (a remote cause) of the later. Thirdly,we should have to be able to give a satisfactory causal) account

of the occurrence of the later event which contained no referenceto the occurrence of the earlier. If these three conditions werefulfilled, and there really was good evidence of the repeatedconcomitance of the two events, then the quasi-causal onnectionbetween them would be a fact of nature which we could do nomore than observe and record " (pp. 31-32).

It will have been noticed that Dummett has been laying downfour conditions for the possibility of a quasi-causal xplanation-conditions precisely parallel to those that emerged in our dis-cussion of the Houdini

example.And it should be clear

bynow

why Dummett's search for a quasi-causal connection holdingbetween a later event and an earlier one led him to formulatejust these conditions and no others.

I have claimed that the four conditions are mutually incon-sistent. But if Dummett had thought so, he would certainlynot have asserted that cases of " quasi-causation might occur.In fact, he provides ingenious descriptions of cases that he thinkswould be examples of quasi-causation; these descriptions willprovide convenient checks of the accuracy of my own analysis.

Dummett says that all of his conditions would be satisfiedin the following case : "A man is observed regularly o wake upthree minutes before his alarm-clock goes off. He often does notknow when he goes to sleep whether or not the alarm has beenwound, nor for what time it has been set. Whenever the alarmhas been set and wound, but fails to go off because of somemechanical accident, which is later discovered, he always sleepsvery late. One morning he woke up early, when the alarm-clock had not been wound, but an acquaintance, who knew

nothing about this queer phenomenon, came in and, for somequite irrelevant reason, set off the alarm-clock ust three minutesafter the man had woken up." " In such a case," adds Dummett," it would be reasonable o overcome our prejudice against thepossibility of giving a quasi-causal ccount of some happening, andsay that the man wakes up because he alarm-clock s going to gooff, rather han o dismiss he whole thing as a coincidence" p. 32).


10/11


There is no trouble n applying our earlier analysis o thiscase. In order o count as a case of quasi-causal xplanation, nDummett's ense, here must be a satisfactory ausal xplanationof the ringing of the alarm-clock, ontaining " no reference othe earlier vent ", i.e. to the man's waking. We know what hisexplanation would be like-it would involve reference o themechanism f the clock work, the position of the lever that setsthe time, and so on. But this means hat whether he sleeperhas wakened or not, we can still make he alarm-clock ing orprevent t from ringing. We need only wait for the man to fallasleep, eave him undisturbed or three minutes, nd then make

the alarm-clock go off. If we can do this, the ringing of thealarm-clock s proved to be not a sufficient ondition or theman's waking, and one of Dummett's conditions or quasi-causal onnection has been violated. Or again, we can wait forthe man to wake, and immediately destroy the alarm-clock.Then that waking could not be caused by the subsequent ingingof the alarm-clock, ince there will be no alarm-clock o do theringing. But if we find t impossible o make he clock ring, orto prevent t from ringing, n the circumstances escribed, weshall have to revise our

assumptionhat we had a

correct,mechanical, xplanation f the clock's ringing. We shall have toadmit, n fact, hat he sleeper's aking xerts causal nfluence, fa very surprising ort, upon hesubsequent ehaviou'r f the clock.

I am not denying hat an event can be a suficient onditionorthe occurrence f an earlier one (as when the ringing of theclock is a sufficient ondition or its having been wound). Soperhaps by relaxing some of his other stipulations, Dummettmight be able to purge his account of " quasi-causality of self-contradiction. But why use the word 'cause' in such cases,qualified hough t is by the apologetic prefix " quasi- "? If theword ' quasi-cause' s to be preferred o ' symptom' or' indica-tion , say, some close analogy must be presumed with the case ofcausation proper. What this analogy s supposed o be is madeclear by another xample discussed y Dummett.

He says: "Imagine that I find that if I utter the word'Click ' before opening an envelope, that envelope neverturns out to contain a bill; having made his discovery, keepup the practice or several months, and upon investigation anunearth no ordinary reason for my having received no billduring hat period. It would then not be irrational or me toutter the word 'Click ' before opening an envelope n orderthat the letter should not be a bill; it would be superstitious nno stronger sense than that in which belief in causal aws issuperstititous." (pp. 43-44).

So the analogy s supposed o be this, that t is not irrational


11/11

58 ANALYSIS

to produce a later event in order to ensure the occurrence of anearlier event of whose occurrence the later event is a sufficient

condition. Now, certainly, f it made sense to talk of producingthe later event in order to make he earlier event happen, t wouldbe tempting to say that the later event caused he earlier one.For a most important use of causal language is to spotlightevents or states of affairs hat we can control: the search or causesis very commonly a search or means by which to produce desiredconsequences or to prevent undesired consequences.

But when the envelope reaches me, the bill has already eeninserted nto it, or not, as the case may be. To suppose that it is

rational to pronounce the word 'Click ' in Dummett's way isto suppose that what I do now can make a difference o the past.And this is not superstition, but a symptom of logical confusion.If my saying ' Click ' is a sufficient condition for no bill to be inthe envelope, I cannot say that word unless there is no bill there.So, if I want to try to say the word, no harm will be done. Butto suppose that what I say makes any difference s absurd. Itwould be as reasonable to suppose that the man who eats theomelette makes the hen lay the eggs; for certainly cannot eatthe omelette unless the

eggshave been laid. Now even the

most daring thinker would hardly recommend he consumptionof omelettes as a way of inducing hens to have aid their eggs.Indeed, in this form of words, the absurdity of post-datingcauses is flagrant.Cornell University.

DELIBERATIVE QUESTIONS: A CRITICISM

By BERNARD MAYO

LOGICALdifferences between types of question are quite

properly investigated by noting differences between locu-tions which count as answers to the questions. Mr. J. M. O.Wheatley' follows this procedure, but arrives at a faulty analysisof what counts as an answer to a " deliberative question ".This in turn involves him in a highly paradoxical onclusion asto what a decision is: namely, that it is a tendency. I shallsay something about both points.

Wheatley follows Bambrough2 n distinguishing between" inquisitive " questions, which " call for" information, and" deliberative " questions, which call for decisions (in the senseof deciding to, not deciding that). But he argues, against Bam-brough, that decisions are not answers o deliberative questions.For " decisions are non-linguistic and such a question, forinstance, as 'When shall we three meet again?' requires a

1 ANALYSIS, vol. 15, no. 3. 2 The Listener, uly 19, 1951.

03. black - why cannot an effect precede its cause

Documents