Knowledge and Certainty Essays and Lectures by Norman Malcolm =e € Prentice-Hall, Inc. Englewood Cliffs, N. J.© 1063 by Prentice-Hall, Ine, Englewood Clif, 81a nc Ne nef te ao ny 2s roped i any form, iy mire on fhe pb tba of Cones Casog Cad Saou rn oe ned Se ‘eewnce-HALL nvnenvanionAL, me, London rmeneeaate or Aor, 2, tm, Spay PRENTICE-HALL OF CANADA, LtD., Toronto ub oF Javan, 1862, Tokyo mencpsia ne mace, a, Mexico Cly Preface Ti storing ey and lestrs ar cy eomeced in to ways: Ft eta opal pcons sent Ene, porecptn,weieton oats Inorigibiyy post ot Sor” ound, proot“concnp at ae’ nde tthe Bheoryf krowlege—are sie om diferent angles, Second, the iaonce of my teas, fends td soon GE Moore ‘tn Lng gente evident rghout Seven of ey hve been pull rei. They are placed herein chonloia rds “The Vetcaton Agurent Sas publ in Phowphicl Ava edted by Max Mack, Felony Inc, Englewood Chi X10 1 publi tere wou rrdon ft tet, But nw fotos ave been fed “Krowlcge nd Dele! ws pb n Mind, Val 1 IS: No 23 Apu 50 Taye nade changes inthe txt and Have ude fot, "Dirt Reception” appeed in The Po sophia Quatery, Va ly Nov 1 Octcb, 158, The tetas itn reve snd’ nun of Totnes added. “Witgrts PosophalImetuation? fist appeared Tho Php Revo Vol Lil No.4 October 1604 There some coneton Of thie ad thor ae nw fotnots “Knowle of Other Sins Srt appeared The Jounal of Pibophy, VoL LY, Novi Nov 1068. 1s pled ee ith seal any ngs. “Ann's Ontlogla” Argument ist appeared fa 159787‘The Philoophical Review, Vol. LXIX, No. 1, January, 1960, ‘Throo new footnotes have been added, “George Edward Moore” was delivered in different versions 1s a lecture at Cooper Union and at several Scandinavian Universities, Its published here for the fist time in English. A somewhat diferent version of it appeared, in Finnish translation, in Ajatus, Vo. XXIV, 1962. “Tho three lectures on memory were delivered to a colloquium st Princeton University in January, 1962, They have been revised. ‘The fist of them, "Memory and the Past,” was published ln The ‘Monist, Vol. 47:2, December, 1962. The sooond and third are published here for the fst time. My thinking about many points in these lectures was greatly benefited by numerous discussions with Professors Jaakko Hintikka and George Henrik von Wright, ‘while T was a Fulbright Research Fellow at the University of Helsinki during the academic year of 1960-1961, I wish to thank the Editors of Mind, The Philosophical Quarterly, The Philosophical Revitc, The Journal of Philosophy, Ajatus, and The Monist, for permission to reprint the articles previously published in those journal. Noman eazcore Maca, NY, April 1963 Contents The Verification Argument, 1 Knowledge and Beli, 58 Divect Perception, 73 Witigentcin's Philosophical Investigations, 96 Knowledge of Other Minds, 190 / Anssinés Ontological Arguments, 141 Geonge Edward Moore, 163 ‘Memory and the Ps, 187 ‘Three Forms of Memory, 208 A Definition of Factual Memory, 222Knowledge and Certainty‘ification Argument 1 AA sumer of guns hav bean wed by von ilonpties to prove tha ett of to emp scent oily cra, tn ts paper I vsh fo examine oe of these Samet. The agumet fat the bes of my know, Been Sat nor ect by CL Lois han by may ater ier wl guote rm im orders oes stong presentation of he Semone “Se nile engaged tn discussing the statement “A pee of whitepaper no before ni Lew ye the lowing ‘This jgment wil be fae if the presentation is lsory: it will bo {ale if what Usps not really paper fab ft ent relly white bat cal as wate Ti tthe es aha One ato ‘tan. “A in the other mph ie eter sl, fo wich i lr the sat Thr ght of an pe before re") 0 here toy any test ofthe Judgment woul prety erly Ine ‘ale some way of acing—making the te, ay contnuog to ook, te tuming my ees, or grasping to tea, ete~-and woul be deterained fy Going or fling to fad some expected rent t expesience, But in his example, if he result of any singe tot isa expected cone states paral verScation of the fudgient oxy; never one which fs Solute deckive snd theoretical complete This eo becute, wile the jgment, 9 far asf slgnfeant, contains nothing which could not be fred stl asa siglcanc which outrans what any singe 1‘ification Argument 1 A. sues of aegunents ave Den wed by varios phicspbiers to prove tat the th of no empl statment i Ebockly crt, In ts paper I wish to examine one of these trgument The argument fn, othe best of my knowl, boon Sled moe focflly by CL. Lae an by any oer writer, wil quot fom him order to obtain aston presentation othe guent ‘While engaged inducing the statement “A place of whitepaper snow Before mie” Lewis ay the lowing” “This judgment wll be fale ifthe presentation is ills it willbe fale I wt I ei not rally paper fa fis ot realy white bot only hake white. Ths objectve judgment alo ts ove capable of core foboraion. A in the oder sxample [the other example, t0 which Eos ers the statement "Thre ls aight of granite steps before ime") so here toy any test ofthe Judgment woul pretty surly ine ‘lve some way of acing—maing the tet 3 by canting to ook prtarng ny ye or grain a, ean ‘rol be determined By Hing on fing Sse expel eat im egret {in dis ezamply if theres of any singe et is as expected, cone Stes. paral verScation of the Judgment ony; never one whichis thiolatly decisive and theoetealy complete This isso because, while the jdgmene, so arson sgnfeant, contains noting which could not be feted sil as a sgheance which outans what any single 12 ‘nim venmeavion ancexces test or any Hnited set of tots, could exhaust, No mater how folly 1 tay ave investigated this objective fact there wil remain se Ahoretial posse of mistake; there wil be farther consequences Which mast be Uhus and’ so if the judgmeot i true, and bot all of these wil have been determined. The possibilty tat such furher {erty Ht madey might have a negative fenik, conot be alogetier ‘reluded; and tis possiity marks the Judgment as, a the tine i estion, aot fully ered and es han absctly ceil. To quibble ch pol dol not cn, be em Bue we are nor ying to weigh the degree of therein duet whic onsmon-snee praeally afoul take account of, but to arve att ‘create anya of lnowledge This character of Deng fuer testable fad les than theretally certain characterizes every judgment of hjctive facta all times; every fdament that such ad such a real {ng exsts or has certain objectively factual property, or Gat & erin objective event actually occurs, x Ut any ehjectiv state of ‘flats sclly the ease” ‘The same argument is stated more dramatically by Lewis in Mind and the World Order. He says Obviously in the statement “This penny is round” T asert implicit cerything the fire of hich tbuldfoofy the statement. The in plat prediction of oll experience which is eset to Rs truth must Ee contsned tn the orginal Judgment. OSterwse, such experience swotld be elevant. All tat further experience the fire of which ‘would lad to the repudiation ofthe prehension Thsoy oF mi ‘les prot ne aiaeat al Now ssp wen How Tong wil i be possible to ver) im some manner the fact that this "around? What tality of experience would veel i completly Eejond’ the possilty of eceotry recnaderaton? seems to be the fact tint no verfeation woul be absoltely complete that sttverleaton is paral and.a mater of dope. snot the se thatthe simpler statement of objective parclar fact imply ‘sets something bout possible experience Chugh ll future tine that theoretically every abjctive fac Is capable of some version at ny Inter date, and that Bo totally of such experience i aboloaly tnd completely sullen to put our knowledge of soc particle Eeyond all posnty of turtyg out toe in enor? For the purpose of refuting it, Lewis considers the supposition that at a certain time, designated ast, the verification of such a G.4 Lews, An Arty of Knolalge and Valuston (La Sle ‘Opes: Cour Pblaing Ca, 1046) p18 Ted a th Wor Ode (Nox Yorks Cue Sere Ses 188), ‘ne venuricarion. ancoseNT 3 statement as “This penny fs round” could be complete. He con- tines the argument: Now suppose furtec that at some date, we put ourselves io past thn tothe the ovaequences of thf, wich Was aceptl a completely established at And. suppor’ tht thse consoguces {alle appear or are not what the mitre of the sccepted fat Te use? that ase wll there sl e no doubt about the acepted ‘eo Or will wit vas sopposely exabished at tbe mbjes to lube a 2? Ad in the latter case ean we suppose i€ was absolute ‘erlfed af tine 1? Since no sigle experience can be sbrlutely eamated a te ver no ied colton sco of EXpeenoes can sbsofitely guarantee an empiical fact as eran be- Yond the posibty of reaatidertion® ‘Many other philosophers have made use ofthis argument. ‘Camap, fr example, uss it in his paper “Testability and Mean- ing." He says: “Tate for instance the following sentence “There is 2 white sheet of pte per on this tab” In order to ssersin whether i hing paper, we Shy mak eto pl barat then, he st semaine tome donb, we may make seme physical and chemleal experiment Here» wo by t czamine sentences which we ffer form the sem tence in guaslon Tes lnfered sentences are predictions about future heeration Tho staber of sch predictions which we en derive om the sentence given i nfs; and therefore Us sentence can newer be completely veiled, To be sure, in many eases we reach racially sect cerany ater anal ur f patie tan at then we stop experimenting. Dut there i alway the theoreti silty’ of clotting the aves of texcobseratons. Theretore bere "ko complete erfction i parle but oly a process of gradually Incresing confirmation Hr continues: “For such a simple sentence as eg ‘There fs 2 white thing on this table’ the degree of confirmation, after «few observations have been made, wil be so high that we practically hid pp, ost. “Toda Camap, “Testabity and Meaning” Philsophy of Science, I, xv (ithe, 1957) seta ed Cusp Pop nd ‘a ial Star Langue (hoe ts Huis, bs 8 er, infact sewer Langage (Nek Hora, gs 8 Wed ith 2 by A) per Fado of Eid Knee feel getank! Seaman Lenrees eee, ‘ah dro Cin aos OT4 ‘ru veRanCATION ancuxcese ‘cannot help accepting the sentence. But even in this ease there remains stil the theoretical possiblity of denying the sentence."* Before proceeding to an analysis of this argument, which I will ‘all “the Verifcation Argument.” T wish to say something about ‘the nature of the conclusion which itis thought ta prove, Previ- ‘ously I sald that i is thought to prove that no emplecal state- iments are absolutly certain. But this remark is not siliciently clear because of a haziness in the meaning of the expression “empirical statements” Certainly no philosopher who has used the Verication Argument has intended that the argument showld apply to necessary ora priori truths. But there ie a class of state- ‘menis with regard to which philosophers have had iiculy in eciding whether to classify statements of that class as empirical statements; and among these philosophers who have used the Verlcation Argument there would be disagreement and hesita- tion about saying whether the argument applies to statements of that class. Statements of the class in question have heen called “incorrigible propositions” or “basic propositions” or “expressive statements” or “senso statements.” The sentence “Tt seoms to me that [ hear seratching sound at the window,” when used in such a way as not to imply that there i seratching sound atthe window, would express a statoment of this class. The sentence “Ie looks to me as Hf there are two candles on the table,” when used in such a way as not to imply that there are two candles on the table, would express another statement ofthis clas. wil call statements of this clas “incorrigible statements” and henceforth 1 vill use the expression “empirical statement” in such a way that the class of empirial statements will be understood to exclude incorrigible statements as well as necessary truths and nooessary falsehoods. tis in this sense of “empirical statement” that the conclusion of the Verification Argument will be understood to be the propotn that every ep saterent is thun avo lute certain.” ‘The class of empirical statement is, of course, enormous. The following are examples of such statements: “There isan ink botle fon that table” “T see a goat in the garden,” "We were in Lugano last winter,” closed the door a moment ago,” “There is no milk abi, p20 sim veuricanon ancoseNT 5 in the ice box,” “Gottlob Froge was not a Spaniard,” *Michelan- glo designed the dome of St. Peter's,” "Water does not flow up Til” "Chickens are hatched from eggs,” “This man's neck is Droken,” "My wife is angry.” The Verification Argument is thought ta prove that whenever any person has ever asserted that the truth of any one ofthese statements is absolutly certain his assor- tion was false or mistaken, and also to prove that if anyone should, {nthe future, make such an aserton his assertion will be false oF mistaken, Tes to be notod thatthe phrase “I is absolutely certain” is only ‘one of several phrases which are used synonymously in certain contexts. Some of the other synonymous phrases are “Tt has bee completely vericd,” “It has boen established beyond « doubt.” “T have made absolutely certain” "I have conclusively estab- lished,” “T know for certain,” “Te has been proved beyond any possiblity of doubt,” “It is perfectly certain.” The Verlication ‘Argument s thought to prove something with regard to each and every one of these phrascs. It is thought to prove that whenover fnyone applies one of these phrases to any empirical statement the assertion which he thereby makes is false or mistaken or in- corrector unjustified. Its thought to prove, for example, that if ny art historlan has ever made the assertion that it is conclusively tstablished thatthe dome of St, Peter's was designed by Michel- angela, his assertion was false or mistaken or incorrect oF unjust fed; and if any art historian should, at any time in the future, make this assertion, his assertion will be false or mistaken or in ‘correct of unjustified, Te thought to prove that if « physician ‘ho has just examined a man struck down hy a bus should ever assert “I have made absolutely certain that his neck is broken,” ‘what he aserts is wrong or improper or unjustified, no matter how earful is examination has been. Tt is common knowledge that ascttons of this sort are often ‘mistaken and that frequently happens tha someone asserts that hh has made absolutely certain that so-and-so is true, when either so-and-so is not true of else, even if sorand-so is true, he has not ‘made so thorough an investigation as to justify his assertion that hho has made absolutely certain that so-and-so is true. The Ve cation Argument i thought to prove, not simply that many as- setions of this sort are mistaken or unjustified, but tht all such6 sum veuricanox ancencet assertions aro, in all cases, mistaken or unjustified, In short, tis thought to prove that it isnot even possible that anyone should, ‘in any circumstances, make such an assertion without the asser- tion Being false or unjustified or improper or mistaken or incor- In order to state tho Verileation Argument as clearly as pos- sible Twill make use of an example. Let us suppose that a dispute has arisen between a friend and myself as to whether William James used the phrase “the steam of thought” asthe ttle of a ‘chapter in his book The Principles of Peychology, my frend con tending that James did not use that phrase, but did use the phrase “the siream of conscioumess” as the tile of a chapter Whereupon I take from a bookshelf Volume I of James's book, tum the pages until I come to page 224, where Lsee the tile "The Stream of Thought” occurring just under the heading "Chapter 1X" Then I say, “You are wrong. Here isa chapter entitled “The Stream of Thought” He says, “Have you made absolutely cer- ‘in?” I reply, “Yes, Ihave. Here, look for yoursel—” T believe that this example provides a natural usage of the ex- pression “I have made absolotely certain” It is, furthermore, a good example for my purposes because 1 do wish to maintsin that on June 17, 1948, T did make absolutely certain that the pphrase “ihe stream of thought” was, on that day, on page 324 of Iny copy of Volume I of James's The Principles of Peychology, ‘The statement “The phrase ‘the stream of thought” was, on June 17, 1945, on page 224 of my copy of Volume I of James's The Principles of Psychology" 1 will call "8." The Verification Argur ‘ment Is thought to prove that I did not make absolutely certain that Sis true. Let us soe whether it does prove thi, ‘The fist step in the argument consists in saying that $ has “consequences” or “expected results in experience” or that from S one can infer statements which ace “predictions about Future observations.” Tt seems to me that i snot dificult to ee an im- ‘portant thing that is meant by these expressions. If, for ‘someone sad “Just now a cat went into the closet,” and I believed the statement, Tsbould expect that i T were to search about in the "yo velnes, 1890, ‘me venuricanion anceseext 1 ‘lost I should see or hea oF touch a eat, IFT did not believe his statement be might naturally say “I assure you that i you look in the closet you will se a eat.” And iF1 did look and didsoe a cat it ‘would be natural to regard this as confirming his fst statement “Just now a cat went into the closet.” Another way of expressing this matter would be to say that if its true that a eat went into the closet just now, then i follows or isa consequence that iT ‘were to search about in the closet I should see or hear or touch a cat. This is a natural uso of “follow” and of “eonsequence” And it is easy to understand what is meant by seying thatthe conditional statement “IFT were to search about In the closet I should see or Ihear or touch cat” states a “prediction about future observa tions” Henceforth I wall use the word “consequence” to express these relationships. I will say that itis a “consequence” of the statement “Just now a cat went into the closet” chat if I were to searchin the closet I should see or hear or touch a cat; and T will, say that the conditional statement “If T were to search in the closet T should see or hear or touch a eat” expresses a “conse- quence” of the former statement. In this use of “oonsequence” i {sa consequence of the statement S that if now, on June 18, 1945, were to look at page 224 of my copy of Volume 1 of James's Principles 1 should see the phrase “the stream of thought” I should certainly expect to see that phrase if I were to look at that ‘page now, and I should be greatly astonished if T didnot see i ‘Tt may be said that what I have ealled a “consequence” of $ is rot a consequence of $ alone, but of $ conjoined with some ‘other statements IT thought that since yesterday someone had erased from page 224 of James's book the phrase “the stream of ‘thought then I should not expect that if I were to look on that ‘page Ishowld see that phrase. Also I should not expect to se ii ‘lnow that my vision was abnormal or tat the room was so dark ‘that T could not make out printed words. The statement "If 1 ‘were to look on that page now Ishould see tha phrase” expresses consequence, not of § alone, but of the conjunctive statement “The phrase was on that page yesterday, and there is no reason to think that the printing on that page has been altered or has changed since then, and my vision is normal, and the Hight is {good.” I cannot see any objection to saying this. I believe that i {sa natural way of speaking to say that FS is tue then itis a8 sme venuricaniow. ances consequence that i 1 were to look at page 224 now I should soo that phrase, and T wll continue to speak in that way: But it wl bh understood that this consequence of 5 i nota consequence of Salone, but ofS conjoined with the other statements mentioned ‘A dificult question now arises, namely, what kind of statements ae the conditional statements that express consequences of S? Consider the statement “IFT were to Took now at page 224 of my copy of James's book I should soo the phrase the steam of {ought wil call this statement “eof the form “It A then EE" Now a view has been pot forward by Lewis that plies that this consequence ofS should really he expressed in these words "IFW were the cao that it should seem to me that Twas loking at page 224 of James's book then it would be the case that 1 should seem to see the phrase ‘the stream of thought T wil ell UUs statement “K” Lewis holds that in a conditional statement ‘which expreses« consequence of an empirical statement, both A and E (Chat is, both antecedent and consequent) mast be re garded as what I cal “incorrigible statement” and what he cals ‘presive statements."* Les cals the statements which exprest consequences of sn empirical statement “predictions” and so does Camap. Lewis talks about his helioving sich a statement as “There isa piece of paper before me.” He says that this bell Involves numerous prediction, eg, that if he were to fold, ‘would not crack; tat if he were t ty t tear ii would tear casily? A moment later, however, be says that he bas not ex Dresod himelf accurately. He sys Jot wat my Intention to mention predictions whic, Yoogh on orl testi ft oj a le ern Gen ee A hr | Dave fie. Tn the Poe, pen al ou ay bet ws be eee th ek ‘ded & tke placo™il be a vel cjetie event sbait wich, ‘the hypoth 4° must he expres something which, # made foe ibe ial Sh contact ope ‘ents cvemualyof experienc, diet and etally recognisable fs ‘vse mot 4 roliatcbjective evn whose faculty cold have, fel eh re co ct cn of ths udgron Athen rogue toh formated expeive (in Analy of Roouladge and Valuation, p. 18). Tad, p17, ‘VERIFICATION ANCUMENT 9 theoretically, my momentary experience could deceive me. What 1 ‘meant to speak of was cortain expected experiences of the ‘nce ond fecling of paper being folded offs seeming to be tor Ho is saying thatthe statement “If I were to try to tear this, it \wonld tear easly” is not a “prediction” and does not express consequence,” in his use of “prediction” and “consequence,” of his belief that there is paper before him. This passage and the one which T quoted from page 181 show that he would regard the statement “If i seemed to me that I was trying to tear this, then it would seem to mo tat it was tearing easly” asthe sort of statement that is a “prediction” and expresses a “consequence” of his belie and if it were true would partially confirm or verify his belief. Thus itis clear that Lewis's use of “consequence,” a8 this ‘word occurs in his presentation of the Verifiestion Argument, is such that he would say that K isa “consequence” of S, and ¢ is not. There i, of course, an enormous difference between k and c ‘The difference could be expressed in this way: IE I were to Took now at page 224 of James's book and were to se there the phrase “te stream of thought" that would ena that page 224 of James's ‘book does exist and thatthe phrase “the stream af thought” Is on that page. Buti now it were to seem to me that T was looking at, ‘page 224 of James's book and if t were to seem to me that Iwas seeing there the phrase “the stream of thought,” that would not ‘entail that page 224 of James's book exists or that that phrase is ‘on any page of any book, 'No other philosopher who has used the Verfation Argument has, tothe best of my knowledge, expressed himself on this point. Garmap, for example, says that when we are tying to very an ‘empirical statement what we do isto infer frm it statements that are “predictions about future observations”** But he does not say whether thes “predictions” are statements like ¢ or state- ments like k. Sine he does not even allude to the distinction it would be natural to assume that hy “predictions” he means state- rents ike . Whether this isso or not Twill heneforth mesa by “slatements which express consequences of $” statements like ¢ and not lke kT have two reasons for this decison. One is that ad *Phsoply of Selene, HL (1938), «25.10 ‘am verPicKTioN ancosees statements tke kare awkward and unnatural. The ober and more Iportant reason is hat Tar not sre that ther any natural tnge af “confi” or “wen” according to which the discovery that kite would cone or very that $f teoe. Suppose that Tree in doubt ax to whether Sis tue If Tere to bok now at Tutt 224 of Jamods book and ee there the phase th sean of Thought” that would fndod conn thatthe phrase was on that page yesterday, But ft were merely the cae tnt it seemed fo Ine now that twas looking at page 32 of James's ook and that it seomed tome tat Twas soln there hat pre, jn a snge tf te precoding words nt is compatible with its Being the ese that Tsim dreaing now or having an ballcintion and am not Sering ay page or any printing a al then how wool this one fm thatthe printed words “the sream of thought were on page 25 of Jans ok yrterday? I not la ome tat would in the leas on 8, im any natural sense of “conte” Tis However, a difleut point and I do not wish to argu it here 1 Inve that no mportant part of my treatment ofthe Verifation “Argument is afleted by the decison to interpret the samen that express consequences ofS as statement ike c and not ke ke and is open to anyone reading this paper to interpret than inthe other may There, however, fact about the relationship between S and the sntements that cares conejuences of 8 whch sould be ‘leary understood, Ths isthe fat that $ doesnot ental Statement tat expreses a consequence ofS, in whichver of thee two ways on interprets thes statements. For example, $ doe ot ental Lethe statement “S but note” isnt selfcntadictory. It's not self-contmadisory t0 wy “The phrase the seam of thought was on page 224 of James book yesterday bt i ot the cate that if were to lok om that page now I should see it there” Nor ead by the conjunctive atement “The pase “testa of thought waton page 224 of James bok yesterday, and thee i no reason to think thatthe printing on that page has ‘Sangedor boon altered since then, and my vison Isr and tho ligt i good” Eventhough thee is no reason to tik that the pre the stream of thought” has disepeared from dat age sce yesterday ft may have dssppeaced, od ft has then FY at that page now T shall not Setter. It should be "TE VERIFICATION ARGUMENT n ‘even clearer that § does nat entail k. The whole statement “The phrase ‘the stream of thought’ was on page 224 of James's book Yesterday but i 6 not true that if it were the case that it should ‘Seem to me that Iwas looking at that page it would be the case that I should scom to see that phrase” isnot selfcontradictory [Nor isk entailed by the previously mentioned conjunctive state- :ment of which $ is one conjonet. When Caraap says that In order to very an empirical statement we "infer" or “derive” from i statements that are "predictions about future observation,” the words “infer” and “derive” must be understood in a sense in hich to say that one infers or derives q from p isnot to say that p entails g. When Lowis speaks of the “implied consequences” # ff a belie its to be understnod that to say that a belie “implies” certain consequences isnot to say tht it entals the statements that express those consequences. Whenever the word. “conse: quence” occurs in my discussion of the Verification Argument it {s to be understood that an empirical statement does not entil any statement that expresses a conserquence of it. This being wn- derstood, T see no reason for not accepting the Grst step in the Verification Argument. When ii aid that S, or any simile state- iment, has “consequences” of “expected results in experience,” of that from $ ono can infer “predictions about future observations itseems to me that this has a fatty clear meaning and is also true. ‘The second step in the argument consists in saying that the ‘number of “consequences” or “expected results inexperience” oF “predictions about future observations” that can be inferred froma $'is “infinite” or “unlimited.” This step in the argument offers some difficulty, but the following considerations may help to x plain it.I said before that itis a consequence of $ that if T were to look nov at page 224 ofthis book T should see the phrase “the stream of thought” But its also consequence that if I were to Took a second from now I should see that phrase, and if T were to Took two seconds from now I should see it, and three seconds from now, and so on for an indefinite number of seconds. What it means to say that this numberof seconds is indefinite” is that ‘no number of Seconds from now can be spocfied such that, ater mi, ‘an nly of Keouladge ond Valation,p. 1762B ‘me venaricarion Ancoseex Stree eeceetseres See Sie peer ee mis (ees eres eae Pei speninnee mrs ae ee Senn Ges Sears “mE vitunicavion ancuneeyr 13 see it, or if my wife were to look now she might not see it, and so on. This proposition in the Verification Argument seems to me to require very careful examination, and T shall return toi later, ‘The fourth step in the Verifieation Argument consist in saying that if some ofthese “consequences” or “expected results fn ex perience" or “predictions about future observations” thet can be inferred from $ should not turn out as expected or shonld prove to be false, then doubt would be thrown on the truth ofS. Lewis ‘expresses this when he says: “And suppose that these conse- ‘quences fall to appear, or are not what the nature of the accepted fact requires? In that ease, will there stil be no doubt about the cepted fact” This fourth proposition in the argument seems to me to express an important truth but one thet is dificult to state, IFT were to take a good look now at page 224 of James's ‘book and were not to see tho phase “the stream of thought,” should not simply conclude that I was mistaken when I asserted previously that it there, and so dismiss the matter, The trth i that I should at first be too dumbfounded to draw any conclusion! ‘When T had recovered from my astonishment what I should con- clade or whether T should conclude snything would depend en- tirly on the circumstances. Suppose that if I were to look now on that page I should se, or seem to seo, the phrase “the stream of consciousness” occurring as the tile, under the heading "Chapter TX,” instead of the phrase “the stream of thought.” As I said T should at the first moment be enormously astonished and not know whet to say. But suppose that T looked again and again and that I stl saw, or seemed to seo, the phrase “the stream of com Sciousnes"; that everything els in the room and the things seen through the windows appeared to Took the same and to be placed 5 T remembered them to be the moment before I looked at the page; that [did not fet il, dizzy, or queer but perfectly normal, that [had asswed myself that the book teas my’ copy of Volume of James's Principles and that the page was page 224; and that there was no reason to believe thatthe printing on that page had ‘been altered sine the lst time I had looked at it fall ofthese things were to occur, thea T must confess that I should begin to foc! a doubt as to whether the phrase "the stream of thought” ever Mind and the World Order, p 822 sue venricarion anceseet that numberof seconds had elapsed, T should no longer expect that if were to lok at that page I shoud so that phrase At Some fare tine that page may he destroyed. But fr as long @ ‘tne asi contsues to exist and isnot injured o tame fred with, and provided that during that period my vision re Thain good, T should expect that if a eny moment daring Gat period {mere to takes good look at that page in «goed light T 5Hould se that pase, however Zong that perod of tne shall be. “There & second consideration: Ido not expect merly that if T ere to lok at tat paige now shoal ce that piss, bt also Tenpect that if iy wife ere to lok at that page naw she would fee iat phrase, and that ifthe man who Hives an the foot below ‘tere (o lok that page he would see tht pres, and tht in Sort if anyone ofan indeintety large number of persons of good ison were to take a goad ok now at that page he would see Uhat phrase, What is meant by saying thatthe number of persons in question is “indentely” large that no mamber of persons, however large, canbe speed such that I sould not expect at If any person of this tutmber were to Took at that page he would See that pase, There i third consideration that combina tin of the preceding two and may be sated as fllows: T sould expect that i anyone of dfn Tage mame of pesos Sflgpod vision were to take «goo loka that page ether ns or at any second of an indefinite number of seconds fom now, te would see that phase, provided thatthe page had not been injured o tapered with aid that he Tooked ati in good light Think thatthe statement that the mamber of “consequences oF “predictions” oF“ sus In experonce” that ‘can be ine {Gred from $ is “inte” or “unlimited” or “indefinitely large” tneans what Ihave esprese in thee Bre considerations and {Feat sso Tam propaed to acept te send step inthe Ver cation Argument “The tht step inthe argument consists in saying that any of these “consequences” oF expected results n experience” may not tum outa enpected ard thst any of these “predict abt fo ttre observations” may prove tobe false. What this implies with Fegaid to, for example, ir tnt 1 were to look now at page 234 of Jamess ook I might not see the phrase “the stream of thought or that T were fo lok «second from now T might not sm VERITON ancuates 3 see it, or if my wife were to look now she might not see i, and so on. This proposition in the Verifeation Argument seems to me to roquire very careful examination, and I shall return toi late. ‘The fourth step in the Verlfeation Argument consists in saying that if some of these “consequences” of "expected results in ex: perience” or “predictions about future observations” that ean be inferred from 8 should not turn ont as expected or should prove to be false, then doubt would be thrown on the truth ofS. Lewis expresses this when he says: “And suppose that these conse {quences fall to appear, or are not what the nature of the accepted. fact requires? In that ease, will ther still be no doubt about the accepted fact” This fourth proposition in the argument seems to me to express an important truth but one that Is dlicult to state, IFT were to take a good look now at page 224 of James's ‘book and were not to see the phrase “the stream of thought," I should not simply conclude that I was mistaken when I asserted previously that its there, and so dismiss the matter, The truth is that Ishould a fist be too dumbfounded to draw any conclusion! ‘When T had recovered from my astonishment what T should con- clude or whether Tshoold conclude anything would depend en- tirely on the cieumstances. Suppose that if were to look now on that page I should see, or seem to see, the phrase “the stream of consciousness” occurring as the title, under the heading "Chapter TX” instead of the phrase “the stream of thought.” As T said 1 should at the fist moment be enormously astonished and not ‘now what to say. But suppose that I looked again and again and that [stil saw, oF seemed to see, the phrase “the stream of con- Sciousnes"; that everything els inthe room and the things seen through the windows appeared to look the same and to be placed. 125 I remembered them to be the moment before I looked at the page; that T didnot feel i, dizzy, or queer but perfectly normal; that I had assured myself that the book tas my eopy of Volume 1 of James's Principles and that the page wos page 224; and that there was no reason to believe thatthe printing on that page had been altered since the last time I had looked at it. If all of these things were to occur, then T must confess that Tshoold begin to {feel a doubt as to whether the phrase "the stream of thought” ever Mind end the World Onder, p, 282“ ‘nur venunicarion ancuen was on page 224 of James's hook. What is more important is that this would be a reasonable doubt. When Levis asks the ehetorical ‘question, “Wil there stil be no doubt about the accepted fat?” the means to imply, of course, that there would be a doubt. But, furthermore, he must mean to imply that the doubs that would exist in those etcumstances would be a reasonable doubt, ie, that there would be good grounds for doubting. For ifthe doubt ‘were not a ressonable one then the fact that it existed would be no evidence that that which had been accepted as a fact was not ‘fact. Now when I say that if certain things were to happen 1 should doubt whether the phrase in question ever did appear on the page in question and that this would be a reasonable doubt, am accepting the fourth step of the Verieation Argument. T ‘am accepting the statement that if certain consequences of S should “fil to appear” then I should have good grounds for be- Hein that I was mistaken when I asserted previously that Thad rade absolutely certain tha Si true. I should have good grounds for believing that the phrase “the stream of thought" was not there on page 224 of James’s book, but also for believing that that phrase never had hecn on that page. Ido not see any mistake in this fourth step in the Verification Argument. "There is afith step inthe argument that is alo diffu to stat. 1 is implied, I believe, in these remarks by Lew [Now suppote . that at some date, fe put ourselves fp postion fo meet he conscquencs of tis fact wie was accepted com ately established st fy. And suppore tnt these eonseqaences fall to lppest or ae not wht the nature ofthe aceeptd fat require? In tekcaso wl there sil be no doubt about the cepted fat? Or wil vat was suppondly exalt at fy be subject to doubt at? And in the. ler ease ean. we suppose was absolutly vered at time 4? ‘The last two sentences in this quotation express the proposition ‘which Ihave called "the Sfth step” in the argument. This propo- sition may be stated as fllowss If at any time there are good ‘pounds for believing that a given statement p i false then at no ‘Previous time was it known with certainty that p is true. This ‘proposition is implied by the rhetorical question, “And i the lat- aa Da ‘u vauncanon Anco. » tor case can we suppose i wat abioitely varied a tine 4° The Teter cue” refered tof he tne fat which there fs sp posed to ext a restonable doubt ast whether something thet Wo accepted at fat ft relly a act. What the tetrad {question tinplis that i there i a reasonble doubt at ty sto Sather something a fact then canna have been theese tat att twas ably vere that hat someting ia fact This positon wil perhaps, be eer if expressed Inthe fllowing Sayt The propeation "There aze sow goed ground for doubting St pis tac ena the proposition “At no previous tine Wasi low with absolute certainty at pis tue? Should we accept this propostion which ithe th tp in the Vercation Argument? Let ue sate fr p the statement “Tame asthe ator of An Auta of A Tecate of aman "Naere"Is here any contradiction in sapposing that some person, say a pub, had sade bully cefain in 170 that ume vas the author of An ABstact of A Tvetse of Human Nature, Int that ‘n 1840 some oer perion, sy 2 tora, ad good sounds for believing that Adam Sth, und not Hue, was the Aithor of {2 Ts thre any contradiction in rapposing tat some Peon a ope tne shoul pone body of evidence that con ively established et so-ancio was the case, bu Ua titer tine apoher person should possess nooe of that evidene but Should poses other evidence on the basi of which twas reason tle to doubt dat rand was theca? I caot soe any con teadicton in tis supposition, Conse hs actual ample” Some sompetn Cis cs are aeo deie wheter Pato wa ‘he author ofthe Lesser Hippas. They cn te ground fr sy that he was and grounds fraying that he wasn. The vow that he was the aioe i subject tos seasonal doubt, Dat Plato may have bon the author If he was the aioe tea lato hin Sel ora contemporary, may have kaon with certainty at he was the author’ Now are he fo sateen (a) Someone at Some ten the past knew with beste certainty Unt Pato as “this statment might be though to bog the quetn. I dee cont swith the cinco ft Weetn Argent snot bog th gc tion with reget ny promi nthe argument cold doo cy oe ‘rose the guna conic the Seton Uist no se cas Kew ih ‘Setainty that ay empl atement tm, fs were contained nt ‘rom ten the Vercaon Argent Suc weld beg the questa,uM su VERIFICATION ARGUMENT ‘was on page 224 of James's book. What is more important i that this would be a reasonable doubt. When Lewis asks the rhetorical ‘question, "Will there still be no doubt about the accepted fact?” The means to imply, of couse, that there would be a doubt. But, farthermore, he must mean to imply that the doubt that would ‘exist in thore circumstances would be a reasonable doubt, ie, that there would be good grounds for doubting. For ifthe doubt were not a reasonable one then the fact that it existed would be fo evidence that that which had been accepted asa fact was not ‘fact. Now when I say that if certain things were to happen T should doubt whether the phrase in question ever did appear on the page in question and that this would be a reasonable doubt, Tam accepting the fourth step of the Verifeation Argument. I fam accepting the statement that If cartala consequences of $ should “fall to appear” then I should have good grounds for bes Tieving that Twas mistaken when I asserted previously that I bad ‘made absolutly certain that Si true. I should have good grounds for believing that the phase “the stream of thought” was not there on page 224 of James’s book, but also for believing that that phrase never had boen on thet page. I do not see any mistake in this fourth step inthe Verlleation Argument. “There isa ith step in the argument that alo dificult to state. {eis implied, I believe, in these remarks by Lewis Now spose... that at same date, we put ourselves poston {D moat the conoruences of this fact whieh war accept So ietlyestablsied str And suppose that these consequences fall Eppes. or are nt what dhe autre of the accepted Ect roquces? In Es ea, ere sil be ng doubt abot the scepted fat? Or wil ‘vit was supporedly established ay be subject to doubt a? And Inthe Inter "case can we suppose war absolutly verled at tine “The last two sentences inthis quotation express the proposition whieh T have called “the Bifth stop” in the argument. This propo- sition may be stated as fllows= If at any time there are good grounds for believing that @ given statement p is false then'at no previous time was it known with certainty that p is tue. This proposition is implied by the shetorical question, “And in the Iat- rh a ‘ni ymRUnCATION ancuseeT 5 ter ease can we suppose it was absolutely verified at time 1?" “The “latter caso” referred to isthe time feat which there is ssp= posed to exist a reasonable doubt as to whether something that ‘was accepted at fas a fact is really a fact. What the rhetorical question implies is that if thore is « reasonable doubs at fax to ‘whether something is 2 fact then t eanmot ave been the case that at tit as absolutely verified that that something is «fact. This proposition will, perhaps, be clearer if expressed in the following ‘way: The proposition “There are now good grounds for doubting that p is true” entails the proposition “At no previons time was it known with absolute certainty that pis tue." Should we accept this proposition which isthe fifth stp in the Verification Argument? Let us substitute for p the statement "Hume was the author of An Abstract of A Treatise of Human [Natures there any contradiction in supposing that sme person, ‘ay a publisher, had made absolutely certain in 1740 that Hume ‘was the author of An Abstract of A Treatise of Human Nature, Dut that in 1840 some other person, say a historian, hed good ‘grounds for believing that Adam Smith, and not Hume, was the author of it? Ts there any contradiction in supposing that some person at one time should possess a body of evidence that cone sively established that so-and-so was the case, but that at later time another person should possess none of that evidence but should possess other evidence onthe bass of which it was reason- ble to doubt that so-and-so was the case? I cannot see aay con- tradiction inthis supposition. Consider this actul example: Some ‘competent Greek scholars are unable to decide whether Plato vas the author ofthe Lesser Hippias. They can ete grounds fr saying that he was and grounds fr saying that he was not. The view that the was the author is subject to a reasonable doubt. But Plato ‘may have been the author. he was the author then Plato hizm- self, of @ contemporary, may have known with certainty that he ‘was the author Now are the tvo statements, (a) "Someone at some time inthe past knew with absolute certainty that Plato was This statement might be thought to beg the queton, 16 does coi tite cnn ofthe Veealon opens una be ges te hoe ny pron tg tcl dm on fs rem i te argemeatcotaied the tertian ht no one cam Sa wi feral tht any empires stereo ut th wee setae a Pete then te Veron Argument hell weal eg he oestion16 ‘nm vermravion ancineese the autor of the Lester Hipp” and (5) “Someone ow has god grounds for doubting that Pio was the author of the rE2eer Hippos lcompatl wth one anche? They ar nt at AIL The proponton tat i the conjunction of (a) tad (B) ie ot slaty. Nong sear fo iagie tan tat ‘ould the cae both that Aristotle knew with absolute or tant dat Pato was the autor of that logue and that 2400 oat later «profstor of Grek, nt having the evidence which Aristotle ad ad noticing inthe dale certain features of se tnebaracteitc of ato, shuld have reasonable dot Bt Plato was the author. Thre Is no contradiction whatover in this Suppostion, lg fay be fae f this export e not scKeconiicery, thon (B) doce mot ental that (2) is fab ‘Therefore, the th step inthe Verhation Argument I an or "Why should anyon lint the emo of ing that ho prop ostion “There are now god grounds for believing that p is fale= tata the proprio °AU ne provour te ws own ith Steslt ataiaty that pis tue Believe tht thee somehing wich may explain why ths error shuld be made, What Ihave Inmindis the lloing‘ sme pron were o mae the serton “Aaitote knew with absolute certainty Hat Pato was the autor af the Leser Hippies but Tout, nd with good reason, that isto wae the autor” then his auction woul coat tn odd sturdy. The most onnary wie of he pase “ne wth abo iste cern” euch that knew with abot cat thet piste” ental pi te.” Threoo, he above astern ental the auton “Pato vas the author ofthe Laer Hippie bt ‘lube that he wat” But syne were fo ase “Plo was the tuthor of the Leer Blip” he would imply, by is amerton, that bo eloned hat Plato was the author of the Lene pp By his aseron “Pato was the ator ofthe Les pla but Tube that he whe woud ert tat he dod stething ‘nt alo ply that he dl not dub that something but believed i" The ste abut would be contin in the aston “ik isrlong but | doubt Ia philonopher ha sensed the pear showed of thi sort of aseton he might be edo conclae Moore bas called attention to the pecalary of ts srt of satan Gp, Moore, “Rass “Theory of Deseions’ "The Philosophy of Baad Rus PA. Sehpp,o8 (Evanson: Nowweten, 1044), p40 ‘nm yewricarion Ancor wv At sine would be abs fr anyone to eter “A a previous time someone knew with able cesta that p tue but Abul, and wih good reason, that pts true” that therfore the Propoiion “Ata previous tine somone Frew with abst ce flay that ps Woe and ow srneone doubt and with god reuon that p is te’ selfcontaicty Tt would be a mistake, howevet o dew ths condasin, Al hough # woul be ard for anyone to eset "tang but 1 doabt, and with gd veo, that sang des a low tha ths proposition i ael-cotadtory. Tat it ot sel centacty is sown fon he face tat te popsiton es ‘aio bute dbus and with good reason, hati ang i cleat not seltcntadiory. Sine he inter prposton i ot sel-contadictary, how can the formar one beso need, can tasty Imagine that sou happen both at tas ang nd Sint tthe same tine, had tonne doubt wt wheter was rani. This suposton seer not sel contadicery Tins the proposton “Ws nog but {ont and with good reason, tht ft slang” root selfcontradtry. although it ‘would bean absurdity 1 wee o ast If is propstion ttre selcontadiory then T doubt, bd with good reo, that iU ranog” would ental It soot reining” But ence we soe clay that hs propetion fmt slEcontaitory ten 1 isk al temptation i Belov tt thr ntlinet does a i te moved TU would abo be absurd for anyone to asert “Aisle new wih abate certaity that Plato ws the author af the Laser Tipps iT dube and on good gonna tat he was” Bat does ot low tat he proposton “Aristotle lnew wth abolte certainty tht Pato wat the ator ofthe Les Hip, bat Someme now doubt, td on goed grounds, tht Plate asthe fuori slcontaicy, Would bea great mite thik Sit th statement “Sosene now doit and bm goa gros he Plt ws th store the Lear Mpa ease sate tment “eis fae tat Atl know with acute aay ht Pato was the authirof tho Leer Hippos” Levis mas His risale, I beliee, wen be stuns tat I someting es 2 opted as. fat at ume tb hat a ne hee arose a enon {Die doubt as to wheter thi something was fact thea18 sum venuricanion anceseevt loves thet it was not absolutely verified otf that this something was a fact. With regard to §, which isthe statement, “The phrase “the stream of thought’ was on June 17, 1948, on page 224 of my copy of Volume I of James's The Principles of Prychology.” this fifth step in the Verllation Argument claims the following: If today, June 18, 1948, tose things which T have imagined as hap- pening should happen, so that T should have good reason to be Tieve that § is false, then It would follow logically that I did not make absolutely certain yesterday that § is true. have tried to shovw that that claim is fls. It i true that it would be an absurdity for me, or anyone, to assert “T made abso- Iutely certain on June 17 that Sis tru but today, June 18, T doubt, tnd with good reason, that Sis tue” This assertion would have the same poculir absurdity asthe assertion °S is true but I doubt that Sis true.” But it does ot follow inthe least that che propos!- tion "On June 17 I made absolutely certain that $is true and on Jne 18 F doubted, and with good reason, that $ is true” i slf- ‘contradictory. I¢ does not follow at all tht the proposition “On June 18 1 doubted, and with good reason, that $ is ruc” entails ‘the proposition “Tis false that on June 17 T made absolutely cer- tain that Sis tru.” It ight be objected that if Ihad made abso- Intely certain on June 17 that § i tro, then on June 18 T should remember this fact and, therefore, should not be able to doubt that $15 true. But itis logically possible that I should not re- ‘member on June 18 that on June 17 I bad made certain that S fs tue, These is no logieal contradiction whatever in the supposi- tom that on Jone 18 T should have a reasonable daubt that $ is tue, although on June 17 1 had made absolutely certain that Ss tm. Thus it sooms to me that there is no reason to acoept the fifth step in the Verification Argument and that, infact, itis definite er. ‘Can the Verification Argument be restated in such a way as to ‘avoid this errr? It seems to me that this can be dane by the fol- lowing to measures, The fest measure consists In strengthening the fourth step in the argument. The fourth step isthe proposition ‘hat Uf cectan things were to happen then T'should have good grounds for doubting that S i trie. This step could be strength- ‘ened by substituting the proposition thet if sufficient numberof things were to happen T-should have absolutely coneluioe ‘nu vemmcarion anouxeenr 19 grounds for thinking that {s false. It sooms to me that f the fourth step were strengthened inthis way it would still be a tue proposition. When discussing the fourth step T imagined certain things as happening such that if they were to happen I should have grounds for doubting that $ is true. But let ws imagine that certain additional things should oceur. Let us imagine tht, being astonished and perplexed at seeing the phrase “the stream of con= sciousness” on page 224 of Jamer book, I should ask my wife to Took at that page and that she too should see the phrase “the stream of consciousness.” Let us suppose that I should examine ‘the manuscript of this paper inorder to verify that there was in ft the statement that T had made absolutely certain that the ‘phrase “the stream of thought” was on page 224, but that I should find instead that in the manuscript was the statement that had made absolutely certain that the phrase “the stream af conscious ness” was on page 22. Let us imagine that I should then examine other copies of Volume I of The Principles of Paychology and see that in each of them the title of Chapter IX was “The Stream of Consciousness; that Ushoold find a number of articles written by psychologists and philosophers which quoted from page 224, ‘of James's book and that each of them quoted the phrase “the stream of consciousness” and that not ane quoted the phrase “the steam of thought that my wife should declare sincerely that she had read page 224 of my copy of James's book on June 17, 1945, and that she recalled that the phrase "the stream of com: sciousness” was on that page and not the other phrase; that every- ‘one of several persons who had recently read Chapter IX in other copies should declare sincerely that to his best recollection James hhad used the phrase “the stream of consciousness” in that chapter and had not used the other phrase. If all of these things were to Ihappen then there would be more than good grounds for doubt- ing that 8 1s true; there would be, I should say, absolutely con- closive grounds for saying that is false. In other words, t would bbe absolutely certain thatthe phrase "the steam of thought” was not on page 224 of my oopy’of Jamess book on June 17,1848; and it would be absolutely certnin that my seemingly vivid reoolle- tion of seeing i there was 2 queer delusion. The fourth step of the Verification Argument could, therefore, be reformulated as saying that {fa suficiont number of things were to occur then there2 ‘um VERIFICATION AncoseENE would be sbnlutely conclusive grounds fo saying that $s false "The second measure novel a revsng tho Velcon Arg: ment would be to change the Bh tp by sbstiting er the false propostion "fat ‘y tine thee shouldbe a rewonable donb at iste then ao previos tne did anyone make ab Solute certain that Sst,” ru proposition at ny tine there shouldbe sbultely conclave grounds that Stale then ata provou tne did anyone make absltly certain that $ fs uve! The later proponon i eal tue Ie a plan aut rn rdnay cco the expressions made abet cor can that pe” and “There a abmltly coche rods that pt fae” aze ned log eppecilon to ons sete Te would be a conadton to ay "Ther are sboltly conc sive grounds hat pi ao but made sboltely can al it tue” The stement “T made absolutely ceraln Uap tr” ental is rc and the statement “There are absolutely Conchsive grande tatp sfc etal fae” Thus the Sttoment “Thee ate absalately condasve gromnde that pis fase bt made sbechtely certain het pls toc ental the bo teaicon “ps fale and pts te” Therefore the propotion “if atany tine there shoud be sbultly conchsive grounds that Si fae then to previous tine did snyoue make abo. Tatty cet tat 5 trate tology Te the Verieaton Argument wore reves by strengthening the fourh step and changing the Bh sep inthe way that T have suggested, then the argument would prove (provided that there ‘wee no oer eorin the argument, wich, 1 Gunk, there) that on Jone 17,108, id not make absoltly certain that 8 tee Buti the angument were reviled in th ay then cou ot be ‘ud to prove the genratprooston hat mo empl statement can bo ealusivly eile as toe, for st obviows reson. Tae revised form the aguneat would contain ana premise the popostion “I at any tine thre are sbslstely” consive {otis hut $a then atm previows dine did anyone make Sally certain tht ig rue” Even fhe srgunent cold be sed a vad prof that Iwas misaken when Tasertd tat tnd indo abltely certain that Si rei could not be wed as ald prof that no empire statement canbe condisively tube a te, Beene oe poi of he engument lls ex “TE vusuncarion ancuneeNT a the supposition tata parla statement, the statment “Sis {ate"can bo concave estaba a eT station with regard to the Veriesion.Arguneat ts, tecor, ag follows ‘Wien tated nts ng fry a pent by Lew he th sep inthe argunet algal ero, When the argument se wed na to avd ths eno, hen cannot bead to ally prove the proposton that no epi statement can be cnc ively elite attr, which ste conduson tt twas tn tended to prove 2 1 wish o pot oat another enor inthe Vaieation Argent, san error that T believe to be of very considerable philosophical Jmportance. The whole argument may be stated as follows: 1. § has consequences. TL The number of consequences of $ is infinite, IIL. The consequences of § may fil to occur. TV. If some of the consequences of $ were to fall to occur, then there would be a reasonable doubt that 8 is true 'V. If at any time there should he a reasonable doubt that is true thea at no previous time did anyone make absolutely certain that § is tris, Conclusion: No one did make absolutely certain that $ is tove. 1 it cll this the “orginal” Verifcation Argument. Before T try show the second erin the argent {wih fo make some remarks he fst place, Ihave stated the argent af tap led nly othe statement S. But Is atndod of course, to Bo 8 perfectly general argument. Thee could be vbsitated for $ ‘ny other empiri statement IE argument sated iSound, then by subst ay ther emplial statment for 8 we ould ‘obtain pall, sound argument In is sense the Vericaton Argument were sound, woud prove, wit espet to any et Pinca seent whatever het no oe id mae ebstely cer fain that that statement Is te dod, it would prove tha no ‘ne can mak absolutely ctain that that statement is te, bo2 ‘Tm VERIFICATION AnoUREENE cause the argument applies at any time whatever. Tn the second place, [have tried to show that premise V is false. I suggested that this false step could be eliminated by the adoption ofthese ‘ovo measures: Firs, to substitute for IV the proposition, which Lil ell “IVa.” “If a suicient nomber of the consequences of were to fail fo occur, then it would be absolutely conclusive that $ is false"; second, to substitute for V the proposition, which 1 wil eal "Vo." “If at any time it should be absolutely conclusive that $ is false then at no previous time did anyone make abso- Intely certain that Sis true” Is seams tome that both IVa and Va are true, I these substitutions are made then the whole argument may be restated as follows 1. $ has consequences. TL The consequences of S are infinite n number. TIL. Tho consoquences of § may fail to occur. IVa. Ifa suficient umber of the consequences of $ were to fail to occur then it would be absolutely conclusive that § is false, Va. If at any time it should be absolutely conclusive that $ is falbe then at no previons time did anyone make absolutely certain that $ is ue. Conclusion: No one did make absolutely certain that $ is ‘This second statement ofthe argument I will call the “revised” Verification Argument. The rovised argument ie diferent fromm the original angument in two respects. Firs, premise V of the original argument is false but the corresponding premise of the revised argument is true. Second, premise TVa of the revised argument contains the assumption that an empirical statement °S is false” «can be conclusively established as true. If we substituted for S in the revised argument any other empirical statement p then 1Va would contain the assumption that it can be conclusively established thatthe contradictory ofp is true. No premise of the original argument, however, contains the assumption that any ‘empirical statement can be conclusively established as true. This, Aiference between the two arguments might be expressed in this sue venuricaTion ancostent 3 ‘way: I all ofits premises were true the original argument would ‘rove, with regard to any empirical statement that was substi- fated for S, that that statement cannot be conclusively estab- lished; and no premise ofthe argument assumes that any empiri- cil statement can be conclusively established; therefore it could ‘properly be sid that i all ofits premises were true the original [argument would prove the general proposition that no empirical Satement can be conclusively established. If all ofits premises were true the revised argument would also prove, with rogard to any empirical statement that was substituted for 8, that that statement cannot be conclusively established; but since one prem= {se assumes that an empirical statement can be conclusively estab- lished, i would be wrong to say thatthe revised argument, i premises were all tre, would prove the general propesition that ‘ho empirical statement can be conclusively established. No matter ‘what statement was substituted for , premise 1Va would assume that the conteadictory ofthat statement can be conclusively estab- lished, The revised argument has this pecullar logical character, that if al of is premises were true it would prove, with regard to ‘any empirical statement, that that statement eannot be completely ‘verified, but it would not prove that no empirical statement can bbe completely vered.* 11 was taken nhs meet of th rv agent Peleg Cea ited ou tome tata propane! he Vast Ager de ot rt cau cgay tet cna secon ost coca cashed He ight ele yal rez to Fev tao Berl ae (yap, 8) et exsist ‘mpl utement can be conlsivelyeublbed tae or fa, then Eat ner ofthe comaqrne” of 8 west alt xe wld SODA emcee ake Wi tao he sre ‘tpument well sal be deo dAed woul at be the oe ABPE prin ee pemet aanescatgely hal soe ep ‘ten be cm sais Sing be fea Sep atten oan bf enocet yea i ss ‘My ech of rene TI ot alfected by this amendment Wes enema Sey aly ay ay © Bad sggee othr way of serge Spe Hee Ca The Pllc Rao, LX, No Ju 1962, BT). He pps to ‘ie aon ching aw mbit SV the flog en cee eae ely tine tae ld be cena Hote at Sue thee a ne pti Ue Ud syone make key esPr ‘ma venuriexTION anceaeENT “The rovisod argument, however, does seem to prove, with ‘gard to my statement 8, that no one can make absolutely cet that $ is true Fit does prove this then i follows that 1 was mis- taken when I asserted that on June 1, 1948 T made absolutely ccarain thet Sis true. Ihave previously accepted premises I, I, 1Va and Va. This leaves premise II. [believe that IH contains ‘serious mistake. T expressed THT in this way: “The consequences FS may fal to occur” Lewis makes nse of Hin the flowing passage: ‘No matter how full I may have ivestigatd this objective fact, here icin ome Unreal posi of mela eve al be Forth consequenoe wich must be thus and 30 ff the Judgment is fu and oo these wave os dees. The ety SE th cet ade mgt re opti raat dhe ine HP gucston, ot ful yenbed andl than senatly ‘When Levis sys, “The possiblity that such further toss, if made, right have a negative result, cannot be altogether prechided,” he is asserting the proposition which T have called “IIL” Carp, in discussing the statement “There is a white thing on this tabi, ‘ays that “the degree of confirmation, ater a few observations hhave been made, wil be so high that we practically cannot help accepting the sentence” When he adds, "But even in this case there remains sil the theoretical possiblity of denying the sen- tence”#" he is, T believe, making use of IE Ti unlikely that he is mercly saying that i is possible that someone should deny the statement to be true, beeause the fact that someone had denied the statement to be true would in no way tend to show thatthe {hat 6 tne, prosded that the eotonee panned ott nade all the fidence poset pt" ibid pe 306) At fry I can see, Frnt ht fn sing tat hs ermoltion of Vian cca apnea of what ‘itendod by pepanents of the Verbeaton Argan” (El, p 308), ae bet to sgt fr sying that he rovuedavfument & dedacely all Cio, pF). Putrnt is verano tetargament Soe et even 2P- een caring ht Da ato ii) Erte eve argument New fea ich have been added to the egal tet of the pee iuly bleed csp, srr enlsed in rakes, rae Ra of Relay an Welton, 18 stay at Meaning Pifovophy of Seience, 1 (1866), 428 ‘me venmexmo ancussest 25 statement cannot be completely verified. Ite ikely that what he is asserting Is that there “remains sil the theoretical possbiity” that some of the statements, which are “predictions about future observations” and which etm be inferred from the statement in ‘question, should turn ont to be flee If this ie a correct interpee- tation of his remarks then he is asserting 1 ‘In order that we shall be clear sbout the meaning of HI let us remind ourselves of what are some of the “consequences” or “expected results in experience” or “predictions about future observations" that can be inferred from S. If $ is true then one consequence is that if I were to look now at page 224 of James's ‘book I should see the phrase “the stream of thought” Another consequence is that if T were to look again two seconds later T should again seo it; another is that if my wife were to look Second later at that page she wonld see that phrase; and so on. ‘What proposition IL says is that i is posible that some or all of these things should fall to ocenr, that i s posible, for example, that iT were to look at page 224 now I should not see that phrase. Although it may appear obvious to some philosophers what ‘meant by saying that such a thing is “posible,” it does not appear tall obvious to me, and I wish to serutinize proposition Il, In order for the Verifeation Argument to be a valid deductive ‘argument (Le. an argument in which the premises entall the con clusion) TIT must be understood in such a way that implies the following proposition, which T wil eall “Hla”: “Tt is not certain that the consequences of $ will occu.” Why must IH be under- stood in this way? For this reason, that someone might agroe that It Is possible that some or al the consequences of $ should fail to oceur but at the same time maintain that itis crtain that they will occur. I might age, fr example, tha itis possible, in some sense of “possible” that if T were to look at page 224 of James's book now T should not see the phrase “the steam of thought,” ‘but maintain, nevertheless, chat it certain that i I were to lok at that page now I should see that phrase. The assertion of proposition Ila Is, therefore, a required step in the argument. IC TIT is understood in such a way thet i implies Hla then the revised argument is a valid deductive argumeat. This wil be seen if wo substitute Ila for I and write down the whole argument, ‘whieh T will eall the “fished” Verieation Argument26 sume venricanion AncoseEN 1. S has consequences IL. The consequences of S are infinite in number. ia, It i not certain that the consequences of $ will occur. Va. Ifa sullcient number of the consequences of § should {ail to occur then it would be absolutely conclusive that § is false, ‘Va. If at any time it should be absolutely conclusive that S fs false then at no previous time did anyone make absolutely certain that § fs true Conclusion: No one did make absolutely certain that $ is 1 secs tome that the conclusion doo fellow logy fom the promise To pt the sngument ore ly. If ithe case that were certain things to happen then would be concasvely stable that § ie fa, nd ts ot certain that hove hinge ‘el not happen, tent fllws that noone has made etn tat 5 stave, have accepted premies 1, 1Va and Va. havead- Intel thatthe conjnedon composed of here pemies and of Tia ental th conclusion, Therefor, iT were to accopt Ila I shouldbe agrocing tat the finished egument is sound and tat ‘contin x pved should veto sd that Twas ie taken when I srtd that I bad made absltely certain tht Ss tuo. do ot se, however any god reaon for accepting la. I believe that it has seemed obviow to the proponets of the Verication Argument both that IT is trac and iat ine lis le, Iwish to show that hse mike 1 wih to show that I may be ndrstood in several senses, that in sme ofthese Sense TT tr and hat sore sala, and ha nly the ‘Sens in which i fle doe nly Hila Tespresiedproposton It in the words “The consequences of $ may Salto aca” Lod have exposed Il aver ober ‘vay: T could ave expensed it by sayingIt spose that the Cousquence of $ wl not coca” or by syag “Tho come ‘Tunes of § might Sot oc” or by sying "Perk he conse: {fue of will not oscar” or by saying “Tt maybe thatthe CSnicqoences of § wil not coat” or by stying “The conse cect of 8 could fll oxen” The word ny” "possble= som venmcano Ancor 2 “igh” “perhaps may be” “ould” “ean” are related to one another in sua way Zt fr any statement that ees oe of these word tere may be skated an oqlvalnt atenent that user anther of them. But whichever one of ts class af atements we employ to expres I eanng wl be open fo fever! eet iret Sevens Tt ot consider some of the diferent tings which might be sant by the propotion “lt is pose that Ue consequences 1 8 wll ot ogo” One thing wich ight be meant st fhe Salement "The consqsenes 68 wil at ocr ont selene traditory. Prequel in plilospbial Asse and sometines inorlnary corse whoh ite eld ts poule that wands wil happen’ of "Sova tay happen” what e mean that the statement “Soandao wil happen? i not selfcontradictory. 1 for example, {were to say atk post that begining ‘torow the temperatures of pia jets wl vary wh ay colors o that it sponte atin one aut the deck on which Tam wring wil vanish from sight one thing which sould mean tat the statements “Beploning tomctow the temper tures of phys! chjects wil vay wh et olor” and “a one ntl wh gh aro ttt. ‘When Ti interpretd in hs way wht says le that {he ent hn the contador of ty statement which expreses a conequonce of fix seltconradicory. I ay, for example that the statment II yee to look now at page 294 of Jans Took Tsbuld nt se the phrase ‘the steam of tooght™ oot sslteontadictory. This nterpetaton of IT cal “y~ ‘Another thing which might be meant by II stato state amet that sates th grounds for heling that any consequence of Swill occur entas hat wll oor, Let re make ths ener Tyan esmple If someone were atk me why Iam se that Twre took now at page 24 of Jars bok {shoul soothe piss th steam of thought night reply se there ye. {Ey and hers pent ble te pgs sce tr been ateed nce then ad my vison foal and he git ir good” Bot what ofr as eons or grounds for ying that Twereto lk at that page nw I chould se tht pat, dos wat nail Ha I oro to ok at Ut page ow sould se tat ls, In the see of pon” wh is edi ly tis pose28 ‘nu venunicaTI0N ancuatene sible thatthe following statement is true: “T saw the phrase on that page yesterday and there is no reason to believe that the printing on that page bas changed or been altered since then and ‘ay vision i norial and the ight is good, but if T were to look ‘on that page now I should not see that phrase.” This statement ix not self-contradictory. I have called the following statement, “c “IE were to look now at page 224 of James's book I should see the phrase ‘the stream of thought.” c expresses one of the con- Sequences of $. Let us call the statement that T have fast used to state the reasons for holding ¢ to be true, fi.” The proposition "R but not c” is not self-contradictory. In other words, H does ‘ot entall Te is possible, inthe sense of “possible” that i used fn TM, that ¢ is false even though R is true. It is possible, in this sens, that tho phase “the stream of thought” has vanished from page 224 even though there is no reason to think that thas and if it had venished T should not see it there. It is not self- contradictory to suppose that it has vanished although there is to reason to think that it has. It seems to me, that with regard to any statement p that expresses a consequence ofS, i isthe ease ‘that iti not entailed by any statement 4 that states the grounds for saying that p is true. This is one natural Interpretation of the ‘meaning of HK T will call “II,” It mast not be supposed that Th, and I, are equivalent. With regard to, for example, Wl, says thatthe phrase “not c” or "eis fale” isnot slf-contradictory. ‘What 11 says, with regard to 6, is that the statement "R but not "isnot self-contredictory. Il says thatthe negative of isnot, by itself, selfcontradictory. Il, says that the conjunction of F and the negative ofc is not self-contradictry, or, in other words, that does not entail e. Il, and Ill, are entirely diferent propo- sitions and both of them seem to me to be clearly trie, The con junetion of I, and Ily is, I believe, what would ordinarily be ‘meant by the statement “It is logically possible that the eonse- jnences ofS will not occur.” ‘Fvih now to pot ot other ues of “posse” and the oor- relative words that are of quite a different kind than the two 20 far mentioned. When it is suid in ordinary life that “I is posible that so and so wall happen,” what i very frequently meant is that there i some reason #0 belive that so-ands0 will happen. Sup- pose that my wife were to say, “t's possible that Mr, Jones will ‘rm vanaricanion anceneext 2 come to see us this evening” 161 were to ask “Why do you think sof” she might naturally eply “He said to me this moming that hhe would come if he did not have to work” In this example the reply consists in stating a piece of evidence in favor of saying that he will come. Suppose thats friend, who looks to be in pet= fect health, should say "I may bo extremely ill tomorrow.” To my. question "Why do you say that?” he might reply "Beeeuse 1 ate scallops for lunch and they have always made me very ill” Here ‘agnin the reply offers a reason, some grounds for saying that he ‘will be il This usage of “possible,” “may,” “might.” and thelr correlatives is enormously frequent in ordinary discourse, and i is strikingly different from the usages noted in IIT, and 1M. When ‘my friend replied to my question “Why do you say that You may be ill tomorsow?” his reply did not consist in pointing ont the Tog cal truth that the statement “T shall be ill tomorrow” isnot sell contradictory; nor did it consist in pointing out the logical truth that no statement which expressed grovnds for saying that he will ot be il tomorrow would ental that he will not be il tomorrow, It did consist in giving some reason, or evidence, or grounds, for believing that he will bell tomorrow. My question "Why do you think so?" or “Why do you say that?” would be naturally under- stood as e roquest forthe reason for believing that he will be il tomorrow. It would be quite absurd for him to reply to my ques- tion by saying, “Because the statement ‘ shall be ill tomorrow is not self-contradictory” or by saying, “Because it does follow logically from the fact that my health i excellent and that I feel perfectly well that I shall not be il tomorrow." If he were to zeply Jn this way it would be rogarded as joke. If 2 Greek schaa ‘were to remark “It's possible that Plato was not the author of The Republic” we should esk “Why do you say so?" and it would be only a joke if he were to reply “The statement ‘Plato was not the author of The Republi’ is no s-contraditory” or to reply “The evidence we have for saying that Plato was the author does not ‘entail that he was the author” We should naturally interpret his fist remark to mean “There is evidence that Pato was not the author of The Republic’ and our question “Why do you say so?” ‘would be naturally understood as a roquest for him to say what the evidence was. His reply gave no evidence. He filed to show that itis possible that Plato was not the author of The Republic,30 ‘me VERIFICATION AncuseeN Jn the senso of “possible” which was appropriate to the context. ‘A radical difference between the ws of “posible” that T am now deseribing and its uses in Mf and Ty consists in the fact that the kind of “possiblity” now being deseribod admits of degree, whereas those other kinds do not. The ordinary expres- sions “There is some possibility,” “It i barely possible,” “There {s a slight possibilty.” “There is a considerable possibilty.” “There i a greater possibility that so-and-so than that such and. such,” “I s very posible,” “There is a strong possibilty.” all e- Jong to the wse of “possible” now being described and not to its tes in Th, and ID. IF the man who says “It is possible that 1 shall be ill tomorrow” supports his statement by saying that he had just eaten seallops and scallops had always made him i, then he could have correctly expressed his statement by the words ‘It Is vory possible that I shall be il tomorrow” or “There is @ strong possibility that I shall he il tomorrow.” But if he spports his statement by saying that “I shall be ill tomorrow" is not self- contradictory, oF by saying that the fact that he is in excellent health does not entail that he will not be ill tomorrow, then he ‘ould not have comectly expressed his statement by the words “Tes very possble that I shall be ill tomorrow” or "There ie a considerable possibility that I shall be ill tomorrow.” The expres- sions “There is some possibility.” “There is a considerable posi- Dility;” “There is 2 greater possibilty that so-and-so than that ‘such and such” mean roughly the same as the expressions "There is some evidence,” "There isa fir amount of evidence,” “There {is more ovidence that so-and-so than that sch and such” The ex- pressions of both types are expressions of degree. Ifthe man who faye "Tt is possible that Plato vas not the author of The Republic” ‘means thatthe statement “Plato was not the author of The Repub Tic” isnot selfcontradictory, then he isnot using “possible” in a sense that admits of degree. There ean be more or Tess evidence for a statement, the reasons for believing it ean be more or less strong, but a stztement cannot be more ot less self-contradictory. ‘The statement “Thero isa slight possibility tha the Smiths are in Paris this week but a greater possibility that they are in Rome” ‘fustates a use of “possibility” which is totally diferent fom logical posabilty. The statement obviously docs not mean that “The Smiths are in Rome” is less slf-contradictory than “The ‘um: venmeaTioN ARGUMENT a Smiths are in Peis” 1t does not make sense to say that one state- sent is "less" self-oontradictory than another. And it obviously does not mean that the evidence as to the whereabouts of the Smiths entails “The Smiths are in Rome” more than it entails "The Smiths are in Pais.” It does not make sense to say “p entails ¢ nore than it entails” The statement obviously means that there |s some reason to think that the Smiths are in Pais but greater reason to think that they are in Rome. hope that I have made it sufficiently clear that there isa com- ‘mon use of the word “posible” and of the comrlative words, c= cording to which the statement “It is possible that so-and-so ‘means “There is some reason to believe that so-and-s0.” When proposition TIT is interpreted in this sense it is equivalent to the proposition “There is some reason to believe that the conse quences of $ will not cour.” I wall ell this interpretation of 11, “Mh,” ‘There are other common uses of “possible” closely analogous to ts use in TT, Suppose thatthe members of a committee are #2 meet together. All of the committee, save one, tum up at the ap- pointed tine and place. omenne asks “Does X. (the missing mem bber) know that there is a meeting?" Inquity reveals that no one recalls having notified K. of the meeting It is also pointed out that, although an announcement of the meeting appeared in the local newspaper, K. frequently does not read the newspaper. A rember ofthe committee sums up the situation by saying, “Then iis possible K does not know about the meeting.” his later statement means that there és no reason to think that K. does know about the meeting; and this seems to me to be very com ‘mon usage ofthe word “possible” and ite corelatives. The propo- sition “Tt is possible that the consoquences of $ will not dee; ‘which i proposition IM, i interpreted in the sense just described, ‘would mean “There no reason to think thatthe consequences of 5 will occur” This interpretation of I Twill ell “TT,” Suppose thatthe question arses as to whether M. was faa cer- tain theater at the time when a murder was commited there, It known that be left a bat only fifteen minutes before and that it ‘would be extremely dificult for eny man to go from the bar to the theater in fifteen minutes. The situation might be summed up by saying “Its unlikely that M. was atthe theater at the time of2 {THE VERIFICATION. ARGUMENT {ve murder bu ti posible that he was” This statement means that theres god atom to belove that M, was no the ater tet ht the fs mot bntely conve It a wey ome ton ise of “pre to say that although there ate erg sounds fr believing that sored the cs, alll possible rat srando isnot the cn, whee thie ie equate ying tint allhngh the grounds fr saying that sande letras ate ‘rong ty as nt sbcltlyconcltive.Il worenkerpreted in ths way wood mean “The grounds fr saying Sat the Consequences of § wil ox ae aot saratlycondhives The interpretation of 1 wl eal Tevet to show tht sentences of the ot “Its posse that soandan It may be tat soanden”‘ethape soand-oe have several diferent se. The wey sme sents fy ferent contests, ie diferent mexninge 1 donot Know that tere toe mo al thr was of hoe sates, bt if See are I ane tot hinkof thr The thd premise ofthe Veriteation Agu ir expred by sentence of thi sre When itis id Tt & pont bie thatthe consequence of $ wil tocar te question ie, therefre, which of thexe diferent ways the sentence being std When Levis says "Ee posblty tat such farther ets ifr, might have a tepntvesenit camot be ogee pe: clude” bs statement equivalent to It possible that free est if rnd, wl have negative ress” Hv are we ode and the ane Of poem this portant prem ofthe ag She in what ene afte posible at so-and oni pes Sha Tere co lok now a page 224 of Jame book f shld tt ace the pase th eras of thought A proponent of te Vaccton “Argument cannot reply that tte pole in te ordinary mening of" pole” Ther sot sac tig the ondary mesbing ofthat phrase. A sentence ofthe srl is posbe that span” docs aot have jst one mening that iG sae in al contort. Te oly core pon os i 0 cx tine each ofthe several ferent inept of th tid premise in order tose wheter thre xy Interpetation of it Sich wil make the “reviled” Veriton Argument sound Seana iti the propotion “ti ponsla thatthe consequences of the sateen Sl not ect Ieee propston “tnt J ‘mr venrieanio ancrcest a ‘known thatthe consequences of $ will cur.” I pointed out previ= ‘ously that inorder for the revised argument to be a valid deduc~ tive argument TIT must be understood in such a way that i im- plies Ila. In order for the revised argument to be a sound argu- ‘ment it must alo be the case that IIIs true, (I am using the pphrase “a sound argument” in such a way that « deductive argur iment i a sound argument if and only if te both the ease that it is a valid deductive argument and that all of its premises are true.) Is there en intepretaion of TI] in accordanct with which [Mf implies Ila and TM is true? The following are the diferent in terpretations of I which arose from the description ofthe several meanings of sentences of the sort "Iti possble that sorand-s0" Uh, The statement “The conequence of $ will ot oc” tse contrdtory. ik, No statement p whlch expreses «conequance of $i ental by ay satrientq whch sates he grounds for al ing that pte ove It Ther rome reson to bev tat the eonsequences of $ will ot oor i. There reson to think that the consequences of Sail ose Til The grounds for holding thatthe consequences of $ wil tat aso conies With regard to these propositions T think thatthe following is the case: If, and Hp are true. But neither of them implies II, Mg, TH, and TH, each implies Ia. But each of them is fale, [wish t0 defend these statements, and I will do so by discussing each ofthese interpretations of HJ. (IML) Teis clearly not selfcontradictory to say either that all or that some of the consequences of $ will not occu. It is possble, in one sense of “Tt is possible” that they will nat occur. But it ‘doesnot follow in the least that i is not absolutely certain thet ‘they will occur. Here isa source of philosophical confusion, With regard to any contingent statement, ii the case that "pis false” js not self-contradictory. A natural way to express this logical ‘ruth about p is to say “Ie fs possible that pis fale.” This pro-ou ‘ru, venuricavion anceneen vides the temptation to say “Since it i possible that pi false, Uherefre its not certain that pis rue” But ths i @ confusion, From the sense of "Iti posible that p is false” in which this, rans that "p is false” isnot self-contreditory it doesnot fellow titer that there is some reason to believe that p is false, or that there is no reason to believe that p i tru, o that the reason for holding that pis tre fs not conclusive. The fact that itis possible that pis false, in this sense, has nothing to do with the question of whether p ts false. In this sense of "Tt is possible that p is false” iis not self-contadictory to say “Ie is certain that pis true although itis posstble that p is false" Tn the senses of “It is pos- ‘ble that pis false” that are expressed by IIT, and ITI itis self- ‘onttadictory to say “Te is certain that pis true although it is pos- ‘sible that p is false.” In the sense of It is posible that pis flso” that is expressed by [ly it [snot se contradictory to say “Tis ‘certain that p is true although itis posible that pis false,” for the reason that T gave in discussing proposition V. But in the sense of II, to say “It is possible that & false” isto say some thing that counts againe! saying “I is certain that p is tue.” Tis easy to be misled by thse diferent uses of “Tt s posible” and to conclude that from the fact that tis possible that p is false, when this means that "p is fal” is not sol-contradictory, that therefore i isnot certain that p is true. But in the use of Tt is possi- ble that is expressed by Il, Itis possible that pis false” only tells, ‘us what kind of statement p is Teonly tells ws that p isa contin- ent statement and not a necessary truth ora neoossary falschood. It tells us nothing about the stato of the evidence with respect to 1p. In the uses of “Tt is posible” that are expressod by Ils, IT, nd Ty, the statement "Te is possible that p is false” does tell us Something about the state of the evidence. It tells us that ere is some evidence for believing that pis false, or that there is no evidence for believing that p is true or that the evidence for p, although strong, isnot conclusive. Ta theso latter uses the state- nent “It possible that pis flee” says something agains ts be- ing cortsin that p is true. The statment “p is false’ is not self contradictory” says nothing whatever against its belng certain that p i true That statement is neutral with regard to the ques: tion of whether p is trac or of whether its certain that p is true. ‘To say that“ I false’ not sel-contradictory” entails "Tis not sw venuricavion Ancessexr 6 ‘certain that p is tra” amounts to saying that “pis e contingent statement” entails "I is not certain that pis true” But to say the latter would be to say something flee. I isnot self-contradictory to ay “There are many contingent statements which [know with certainty to be true.” ¢ isthe statement “If T were to look now at ‘page 294 of Jamer’s book I should see the phrase ‘the stream of thought?” c fs a contingent statement, which entails that “c is {alee is not sol-contraditory. Its correct to express this logieal fact about ¢ by saying “Itis posible that i Lwere to look nov at page 224 should not see the phrase ‘the stream of thought?” But although this statement expresteea truth itis not a trath which is ‘even relecant tothe question of whether itis cortain that ifT were to Took now at that page I should see that phrase (I11,) Previously I suid that the grounds for saying that ¢ is trie are expressed hy the statement “I saw the phrase when T looked there yesterday, there is no reason to believe that the printing on that page is changed or boon altered since then, my Yision ie normal, and the light is good.” I said that this statement FRedoes not entail c, A natural way of expressing ths fact about the logical relationship of to ¢ 1s to say “Even though Ris true itis possible that ¢ is false” This expresses the fact that the inference from R to ¢ is not @ deductive or demonstrative in- ference. Bu it provides another great source of philosophical con- fusion, There 1s temptation to conclude from the fact that i i, possible, in this sense, that c i false even though R is tre that, therefor, i¢ is not certain that ¢ is true. It docs not, however, follow from the fact that R does not entail ¢ either that it wot certain that c is true or that F doesnot stato the grounds on the basis of which itis certain that o is true, The temptation arises from the fact that there are soveral uses of “It is possible that p fs fale” and that frequently these words mean that thore is some reason to believe that p is falso, or that there is no reason to he> eve that p is true, or that itis not absolutely conclusive that p fs true, When one stys “Although R is true ¢ may bo falc” and cepresses by this the fact that “Hand not c” isnot sel-contradic= is easy to be misled by the vaeety of uses of “possible” ray be" into supposing that one has said something that ‘counts against its being certain that is true. But the statement6 ‘nme vemnicarion sncuxemer that R does not entail ¢, Le. that "Rand not eis not selon tradictory, says nothing that counts either for or against its being certain that cls true. Whether itis certain that i tue depends "upon the state of the evidence with regard to e. The statement ‘does not entail c” says no more about the state of the evidence with regard to ¢ than does the statement “eis fale not sel- “contradictory.” Both statements ae irelevant to that mater, The statement “The fact that Ris true makes it absolutely certain that © Is true” is in no way contradicted by the statement "R does not ‘ental o” The two statements are perfectly compatible with ome another. One statement describes the evidence concerning ¢. The other describes « logical relationship between Cand c of which fone could be aware even though one knew nothing whatover about the state of the evidence concerning ¢ The fact that R * (rank (“Phonphlnl Cesta" The Popa Revi, LXXL, 00) ages thee Ho ys it th eon tha de ok fetal i eviouly shat the evince fee Bat sr Feat sso ‘due tt irlovant tothe quoton of how much evsenee thre Wf and in ths ese hs nthing tp do ath the tate of the ese fr Tn 'What ste i abot the evidence fr o kn mapping that he ‘ua ie Prakfut would appear tbe thing at des wat ena then 6 ao on tate tg. He gon on nde, however, at be pte ha ys nha mn ep omy he mr (bid "p S10) Bet He sage tat U have oflrod“orpaagy te” expert foc my "iow (id pe 308). Ph ely apport XH Rew fo gies Ear. ‘plete fp which we sol ay that stg perelly ‘tas or “eoelsv,”and yet whch 1 len ath the Ean lve px‘ he tng in quest, In test thi go onto mene ‘Sim, fr crm he fc ef the pried mur "the ety of hog oe lee a of Tansy fo fs hon pec ean been tte a lags Gh bok has exit” (en 38). Tie I peraay cei andi ty peat bios that tee ew eal she, Ca her tapi te tae pray coin Fay GFraskfoe i the ator oft artic ented “Polo! Cer laity bt my ontence tat be wrote dace a8 etal that He wate) ‘The tin ofthe example Seto Bing gut th at at oe ef he le ea ete ca end” hey Sc Si he ti ere fo th pais ne 1h ropostons Ths iter questions gency freletan. (it mato ‘ray stlvant basse erence nepal propotin asad to ‘gaan, eg,“ spent tal of 1S at the pcr’) Why i op ‘halo ti? Why re we strongly ntinad oink tat eats Ita? Party because af the amby of Pay (0m nom inn fo {nk ects the verb “cnn Incurportes one tot of the ver allows td in odin spec we have anole more fegeeal oe of hn which Te VeFICATION AncUMENT a does not entail ¢ provides no ground for doubting that c Is true. Its a mistake to suppose that because it is possible that is false ‘even though F is trae, inthe sense of IT, that therefore “Te i ‘possible that c i fale,” where these latter words imply that it i rot quite certain that ¢ Is true. What I have said about c applies ‘equally to every other statement which expresses a consequence of. (1M) Ils sod Ht provide no bass whatever for accepting Iie Te sane hing cast besa of ly If there some son for bleving tht any consequence ofS wl ot cc, his counts In favor of balding that ts not cetain tht it wl oc. Bat Titre? I thre Sorrento or rondo evidence for hiking that any consequence vf Swill nt ont for taking, for exe tomple that were to ook now at page 224 of Janet ook Should ot se there the phrase “the stream of though? There i toe hee Lt ws ee what olde eon ne ing tat any consequence of § wl not ecco. I some person total vison had areflly looked fr that pase on ist page a {exw minstes ago and had not found it theres then that would Bea reat ail powerfal one, for inking tat wert ook no T should not set there, Or my copy of Janets book pesesed the pela sharacteate hat somctines the pring on De ges undervent spontaneous changes, that printed words were Eien replaced by diferent printed words without external ase then Bat would hes rato for doubting a fT ee fo Took now at that pag Tsou se that phrase. But there is no ‘enon nk at ty person ha Toke fr that pias and has tot found it thee or tink at my copy of Joe bok dos ste that preliar career. Teaight be objected that albogh ther no reson think tint thee ngs te tue neers hey may be tren which of the sveral sense of nay be iit ht these things may be true Inthe sont of Uk and ly then does 8 fle ‘my rach thing a “Tk does nt follow fom the evidence you ofr that pe meen ht the erence aot god eo bt eatcse, Jos Sottnke metal tht p tn our pomp taking sat pica ‘aly, say Far fo confection ner of Yllows (The {opie ef cera and etalnent are sce a spec connote, inp sor of Winger Insole, pp. S117)3 ‘me vinunicariow ancuneer that itis not certain that they are false, It cannot be said that they ‘may be true in the sense that there is some reason to think that they are tru, for we are supposing it to be admitted that there no reason. May they be tue In the sense that there is no reson to ‘think that they are false, or in the sense thatthe ground for say ing they are false are not conclusive? But there {reason to think that they are false, There is reason to think that no one has tried and failed to find that phrase on that page. The reason fs that I did make certain that the phrase was there yesterday, and if those printed words were there yesterday then itis perfetly certain that they have been there as long asthe book has existed. This is not only a reason but is what would ordinarily be regarded as conclusive reason for saying that no ane of normal vision who has ‘atefully looked for that phrase on that page in good light has failed to see it there. The reason ie obvious for saying that my copy of James’ book does not have the characteristic that it print undergoes spontaneous changes. I have read millions of printed words on many thousands of printed pages. | have not encountered single instance of a printed word vanishing from a page or being replaced by another printed word, suddenly and ‘without external enue. Nor have I heard of any other person who hhad such an encounter. There is overwhelming evidence that printed words do not behave in that way. It is just as conclusive as the evidence that houses do not tum into Howers—that isto say, absolutely conclusive evidence, Tt cannot be maintained that there is any particular evidence for thinking that the consequences of $ will not occu, It might be Ihc, however, that there isa general reason for doubting whether ‘they’ will occur. The reason is that the consequences Of seme statements do fail to oocur. It might be argued that, since some- times people are disappointed in expecting the consequences of a certain statement to ovcur, therefore I may be disappointed in ‘expecting the consequences of $ to occur, This would be similar ‘to arguing that since people are sometines mistaken when they declare a statement to be true therefore I may be mistaken when, declare $ to be true, or that sinoe people sometimes suffer form hallucinations therefore T may have been suffering from an hal ination when T thought that Iwas making certain thet $¥s tre ‘There fs undoubtedly a temptation to argue this way. The fol sm venmiexmtoss ancinsmse 9 lowing semarks by Rowell are but one example of “Lames Year voces which ote peopl dono ea intend feet them wth abnormally ace aig, we ok them up. Ba if we ‘ometines hea setenors which hae ot proceeded from body, sys hist hy be he cst erhape au agiation Tas conjured up al the things at we think oles awe sid fo us? Hore Raval is arguing that since sometimes people in ‘jue vlos, therefore never case when one has 4 VS” One tay Dave imagined the vole emnot undertake fo extmine in {hs pera ofthe sores ofthe tempation tare nts ay They le nome seriou dicale serounding the piosphia Geto, “low do I iow that I nm not dreuing or having sn Batlacnaon” To vette the woul lod way rm the Veriton Argument T do wan fo pit ot hat thi sat fang on the fae of cote Invalid To argue that shee peopl sometimes make Iistkes therefore {may Be mistaken when sty that 5s tes ike aguing that France Bacon ay no have eon an Engh mn boat ste me re not Paghmen or ht Bismarck my ot have been x ste brome sme men ae ot ates, tr tat Tay be lid because seme menace blind, Tis ¢ iraventy of cect easoning, There re some ccunstaness in wich reasoning of at sr scope If he doo tthe {ijpningofce fou and we ae wondering wha the mtn thee is dng sd someone say tat sly te siting at ie desk ne of might roply “He ay nt be tng at Hs desk ease smth ste on the for In these teumstances Sheet at somctines he son the Boor doe count gaat ay that tis surely the cae that he siting at is Jeske Bat we were to open the door and se im sing at hie desk then 1 would be mur for anyone sy “He may aot be stg this deck becuse somes hess onthe Bor Tis sort of fetson ing is scceplable in those circumstances where one has nt yet Jnvetigatel the question tru, where on no ina posi fo now the anvver whee one ca onl mae conjure thers fran unexamined cain the clot od someon sss tht te wonden, we ight reply It may be ttl becae some chal = Dersnnd Rs, An Ouline of Pionphy (W. W. Norton & Company, we, 1827), 80 “rim veniricaviON AnGUMENT are metal” But onco we have looked at, fl and scratched Same splitrs rn thon would e ony amaig to say“ to hel Das se a ae ea Here» Ope eatontng that appropiate in some ccmstances but 9! i a “Sreumsanoes. There ight be Greustance in which ie would be reasonable to say Limay be having an allicinatlon because people do have ballacnations” otto say "T may have iegined a bea a voice becase sometimes t do imagine ht T bar votes but itis en ert suppose tht ise a eoable ng {say inal reumstanes. Tse remarks apply to what we were comadeting asa general reason fr doubling Hat the comoguences of wil oat The eagstion Yas th the consequences of S may fall to occur Intinwe the consequences of some iatement 4 fall 9 oor that since sometines people snake fale statements ad are ds: fppotted when they expect their consequences fo ace, that tkeefore when T assed Tay have made a fae statement turd ney be dcappointed in expcting ts consequences to ocr ‘There re cecunstances in which Rr ily easozaleto tenper the coufdence wth which T ase sme, by reminding 29 Self tat sometimes other people and mysl! make eronous teen ta ital spe esse a Sil creumstances, to supose, for example, ht is reasonable che tats some Tam uae, hse when Say that Lam more than ton yar amy be mistaken and tat ibm quite crtain that Cam ore thas ten. Ths he gg ton tn tere sa ger enon fr Delving that te ste ments that express Consuence of § are fas, and that Bo ‘avon shut fome statements hat people expect fo het rh cut tobe fle, completely in errand presents nobhog mare than a caste of good resoning, Te would be «caricature of ued reasoning if «member ofa sity of Grek scholars ere {Eder oth solty tat hae reson to belove that lato tras nt the author of The Republic and when ake for he rea Son were to reply Ut people offen believe propston which areal Teonlude tht Isis fase, Te asserts that the consequences of 5 may fll to cur inthe sue that heeft some roo for thinking that they wl not occur. But ther io reson tall, ‘rir verurreaTion aNGUMENT a neither any particular eason nor any general reason, for thinking that any ofthe consequences of $ will not occur. (IIL) Propestion 1, which expresses another common usage of "Its possible,” says that there 4s no reason to beliove that the consequences of $ will occur II, is false because thee is «very _good reason fr saying thatthe consequences ofS will oceur. The reason Is that Sis true. What better reason could there be? Two objections might be made to this, Fist, se might be sad that no empirical statement, p, is evidence for another empirical state- rent, q, unless p entails q: § docs not entail sny statement that ‘eapresses a consequence of § (ag, $ does not ental c); therefore the fact that $is tue is no evidence that any statement fs true ‘that expresses a consequence of 8. This objection, however, ea not be made use of by a proponent ofthe Verification Argument. ‘The fourth premise In that argument says that if some of the ‘consequences of S were not to occur then there would be some reason to think that $ is fase. But the contradictory ofa statement ‘which expresses a consequence of § does not ental that Sis false (eg, "eis false” does not entall "Sis false"). Therefore, one step inthe Verification Argument assumes what is clearly corect, that 1 statement, p, can be evidence fora statement, q, even thovgh p ‘doesnot ental g. Th second objection that might be made is that if S is true then there is good reason to think that the conse= ‘quences ofS will occur, but that iti not absolutly certain that § {3 tue. If person made this objection it would be necessary to ask him what his reason is for saying that ifs not absolutely ce tain that § i true. Is his reason that he has looked at that page of James's bock and failed to find that phrase there? In other words {she saying that thee ate particular grounds for thinking that ft is not certain that $i true? But there are no such grounds, Is his reason the general philosophical propesition that no empirical statement i absolutely certain? But that isthe very proposition ‘thatthe Verlfcation Argument is mesnt to prove and s0 tht prop ‘sition cannot be used as a step In the agement, There is no way in which it can be consistently upheld, within the context of the Verfcation Argument, that there is no reason to believe that the consequences of $ will occur. TI cannot be eccepted as an inter- pretation of proposition II in the argument, beeause not only Is2 ‘om vinunearion ancuxeer IL, fase but also its use as a premise would lead either to an inconsistency orto a elrcular argument. (111) Te is unlikely that any philosopher who has used the Verification Argument would wish to maintain either that there i some reason to belive thatthe consequences of § will not occu ‘or that ther is no reason to believe that they wil occur. But une ‘oubtedly he would wish to maintain that the grounds for saying that they will occur are not absolutely conclusive, The statements that express consequences of § are empirical staements and the Veaifcation Argument is intended to prove that the grounds for ‘no empirical statement are absolutely conclusive, If the conclusion of the argument is true then Il, is true, and it would be incon- sistent to accept the conclusion an not to accept Il. But we are now regarding IIIs as a premise in the argument intended to prove that conclusion. Within the context of the Verlfcation Argu nent the proposition that isthe conclusion of it cannot be offered in support of premise TM, because the argument would then be circular. What isto be offered in support of Il? ‘The Verification Argument is subject to serious logical dif cally. It eannot be a valid deductive argument unless it contains the premise that itis not certain that the consequences of § will, ‘occur. The fact that this premise is required is obscured by the ambiguity of proposition III, which is the proposition “It is pos: sible thatthe consequences of S will not oceu.” The meaning of IIT is open to several interpretations. Only if TI s interpreted in such a way that i implies Ia is the argument valid Ils is one natural interpretation of IL, and 1, implies Ia. In fat, Ig and Mla are logically equivalent propositions. Hy entails Illa and Mla entails 11, But Ila (or Hg) isa proposition which requires proof. Proposition Ila is extremely similar to the proposition ‘hich isthe general conclusion of the Verification Argument. The ccnelusion says something about every member of the entire elass ‘of empirical statoments—it says thatthe tuth of not one of thene statements is eompletely certain. Proposition Ta says the same thing about every member ofw certain subclass of empirical sate. ‘ments, namely, the class of conditional statements which express ‘consequences ofS, $ was but one example of an empirial state: nt, picked at random, and could be replaced by any other empirical statement, Whatever statement may be substituted for ‘me vinreaTion ancustest 8 5, proposition Me woul sy that it i not cera Bat he cond Hol statements Gat express consequences of tht Satement are tru. In fect, therefore propotion Illa sys thatthe uth of tot one ofan enormous cst of sateentsnamay al ond tional statements which express consequences ofan) empl statements completely cra, This seeping ad paral Slim rust be fst ae mah ne don th gener co clason of the Veeaton Angiment Every ove of usin odinary Ie requently makes ater of flowing sot “I a. July certain tht if you look trough theo boca you wil see a canoe on the lake “We know for cvtain tat yo pour Ghat acid tno ths sclition you wil sec red pret fn,” "Ie perecthy certain that if you were to touch Ut wie ou would reave ashok” That sto ay everyone of ws equally asserts of seme conditional, empirical statement that is uh fntirly certain and Beyond ration. Shall be sid thatthe Condo, empiteal stents ubich express consequences of 5, or of any statement stitute for Save not cerily te becuse no conditonal, empl statement ty cera’ te? Wnt is the justition fr the Iter proposition? What the Iustcaton for saying that evry tne ayone at made an ase. ton ofthe prcaling srt hi station has ben feo mista nuns? Teresa gap nthe Veriton Axgunent andthe Vercaion Armament ital cannot be sed to fl that gap. What i oH? Some other pilsopical argument? Hare praduced an ag sent tht it were snd, would prove that fin nt cetata tat ‘ny of the conditional statment that express consqunces of § te tue But Hume argument could ot be wed bys proponent of the Verification Argument to prove promise Ila Hues agus iment was ltendd to prove that no inferences about mates af fact ae founded on reasoning” le meant that there ean be mo reason to accept any Inference abot mates of fetta ere can be no reasonable inferences about mates of fact, But prem ise IV af the original Vrifestion Argument ase tat iaome of the consequnens of $ werent to occur then there would be Feason to thik that 5 flue, That premise imps that Ure can {wid Hane, Ax Eagaty Concerning Humon Undentondng, tH“ ‘nun venmicarion ancexcest ‘be reasonable inferences about matters of fact, Thus Humes argi= ‘ment is incompatible with the Verifeation Argument, Perhaps there is some ether philosophical argument that could be offered in support of TM; but until i has been presented swe cannot dltermine whether itis sound or whether itis compatible withthe Verification Argument “The Verifcation Argument docs not stand on its own fet. Proposition Ila, a required premise, makes a claim which is of the some nature as the general conchusion of the argument and only slightly less grandiose. The philosophers who have used the argument have tended to tacitly assume Ila. They have not clearly sen that Ila needs to be set down asa premise and to be supported. The explanation for this, believe is that these philos- cophers have been confused by the variety of uses of the phrases “Te is possible,” “It may be,” and their equivalents. When they hhave sald that “Ii possible that further tests will have a negative result” or that "The predictions about future observations may prove tobe false” they have thought that they were saying some- {hing that i so obviously tre that it does not require support and that shoves that itis not certain that those further “test” will have fa “positive resul” oF that those “predictions” will prove to be ‘rue. The fact is, however, that although there aro natural inter- pretations of IT according to which IIT i obvionsy true, none of those interpretations show that Ta is true; and although there are natural interpretations of III which, if true, would show that Mla is tre, there is no reason to think that TI is trae im any of thoee interpretations. ‘The result of this confusion is that Ila, aldhough a required premise, fs an unsupported premise. "There is one passage in his exposition ofthe Verification Argue ‘ment in which iti clear that Lewis i asorting a proposition eor- responding to TI, and, therefore, to Ila. He is disousing his “belief that there isa pie of paper before him. He says: ‘Aud my belief must imply as probable, anything the ftir of which Thou acept as tending to ise th belt, Also Hi the case {hat mich frre sontngencie imped by te bled reno uch at {ale of them ean he abslutey pecaded ia the Ugh of prior pirical coroborations of what eid. However improbable Scan Winkabe tat sich ltr tests could have oopaive ren “igh hf ete ul inp pte vn of ch ee tern the evidence to date doesnot imply this as more than proba, ‘Tm VERIFICATION AnoUMAERE 6 even though the diferene of this probabibty from theoretical Svety alld e's sight that pay oul be oh a ett ert we om fn Srey at the ‘when we have fle ite sire ofan objective fact persed but later ‘rcumstance has shocked ts out of our surance and obliged us to retractor modify or bail * ‘When he says “The evidence to date does not imply this as sore than probable” (“this refers to “a postive result of such later tests") it clear that Lewis is asserting thatthe evidence for any statement that exprestes a consequence of his belief is not absolutely conclusive. (Ifthe evidence fora statement isnot absolutly conclusive then i follows that it not certain thatthe statement is true, fe, IMT, entails TMla,) What is his reason for saying that thee is no absolutly conclusive evidence that lter tests" will not have a “negative result”? think that part of his reason lies in the statement "However improbable, st remains ‘hinkabl that such later tests could have a negative result” Tis clear that he i using “thnkable” as equivalent to “conceivable.” "The phrase “It is conceivable” is used in ordinary language in cezacly the same way as are the phrases “Tt posible” and “Tt ray be.” The expression “Its conceivable that so-and-so” is open to the samme variety of interpretations as isthe expresion “Tt ft possble that so-and-so.” How shall we understand the statement However improbable, it remains conceivable that ater tests wil have a negative result? If t means that itis not self-contradictory to suppose that later “test” wil have a “negative result” or that the evidence for saying that later “test” will have a “postive result” does not endl that they wil, then this statement is rue; but it provides no ground for denying that the evidence & abso- Intely conclusive that later “test” wil have a “positive result.” If the statement means that there is some evidence that later “tests” will have a “negative result” then the statement fs false. With rogard to 6, i not true that there i some evidence that if 1 ‘were to Took at page 224 of James's ook should not see the phrase “the stream of thought." The statement docs not mean that there is no evidence that later “ess” will havea “postive result; for Lewis clealy holds that € may be probable or even highly An Analyst of Knead Valuation, p. 1786 ‘nur venuriexvion anconees probable tht later “tests” will have a “postive result.” The only thing left fr the statement to mean, so far as I can see is that the fevidence, although strong, is not absolutely conclusive that later “eats” will havea "postive result.” But ifthe statement “However Improbable, it remains conceivable that later tests will have a negative result” has this meaning then it provides no Justification at all forthe statement that “the evidence to date doesnot imply ‘8s more than probable that later tests will have a postive result.” ‘The two staternonts are then identical in meaning and the former statement can provide no justieation for the assertion of the Tatter statement. Both statements are equally in noed of support. believe that there ig something elso in the paragraph just quoted from Lewis that he may have regarded as supporting his aim that “the evidence to date does not imply this as more thao, probable, even though the difference of this probability from theoretical certainty should be so sight that practically it would be foolish to hesitate about it” He continues: “Indeed we could bbe too deprecatory about this diference: if we interrogate ox ‘perience we shall ind plenty of occasions when we have felt quite ute of an objective fact perceived but later circumstance has shocked us out of our assurance and obliged us to retract oF ‘modify our belief.” In terms of my statement S, 1 understand Lewis to be saying the following: Its n0 more than probable that the consequences of S will occur; but it may be so highly probable that thore fs no “practical difereneo” between this high probability and “thooretical certainty.” It may be so highly probs fable that it would be foolish to hesitate over this difference and to feel any doubt thatthe consequences ofS will occu But then, he wars, perhaps we are deprocating this dilference too much. (CIndeed we could be too deprecatory about this difference”) 1 ‘understand him to be saying that we shoold remember that it i not certain thatthe consequences of Swill cur and that perhaps ‘we should hesitate a litle, ic, feel a sight doubt that they will ‘occur. Why? Recause there have been numerous oocaions when ‘we felt sure of something and then dseoverod later that we wore mistaken. If understand Lewis correctly, he is using the later fact both to reinforce his claim that itis not conclusive that the ‘consequence of § will occur and as « ground for suggesting that perhaps it would be reasonable to feel a slight doubt that they ‘ru vEnicanion Ancescent a will occur. But if he is doing this then he fe making a mistake that [ mentioned in my discussion of proposition IMs. That mistake consists in thinking tha theo i a general reason for doubting any ppaticular statement that we believe to be tric, the reason being that it has frequently happened that what we believed to be tre fumed out to be false. Iam not entirely conident that Lewis is arguing in that way; but if be is then enough was said in our discussion of Iy to show that this alleged general reason for doubts no good reason at all for doubting thatthe consequences of S will occar and that to argue in this way is to commit a travesty of coreect thinking, ‘The passoge that Ihave just quoted contains the clearest aser- ‘tion of proposition IM, that T ean find in Lewis’ writing or in the ‘writing of any ether propanent of the Verifcation Argument. In this passage no good grounds are offerod in defense of If, and ‘the assrtion of it seems to obtain its plausibility from the ambi gully of the expression “It is conceivable” ("thinkxble"), which hus the same ambiguity as the expression “It is possible” Almost snyone who reflects on these matters wil, indeed, fel an inclina- tion to say that Illy is trve. What is the source of this stiong inclination? T believe that it lies exactly in that ambiguity. Con- sider ¢, which isthe statement “If I were to look now at page 224 fof James's book T should see there the phrase ‘the stream of thought,” and which exprosses a consequence of S. One feels compelled to say that i is posible tht ¢ is false. And this & cormect, It possible that ci false im the sense that “ois false” is not sef-contraditory and én the sense that the grounds fr afi ing ¢ do not ental ¢. Now fooling assured that the statement "It is possible that « is false” is undeniably true, one wants to con= clude “Therefore i {snot certain that ¢ i tre,” And from the latter statement one correctly concludes “Therefore the grounds for arming c are not conclusive.” Reasoning in this way leads ‘one to accept IMy. But this reasoning is fallacious, The error lies in the step from “It is possible that ¢ is fale" to "Therefore it i ‘ot certain that ¢ i tue” Tn the senses of “It is possible” in which ‘tis undeniably tue tht its possible that cis ale, the fact that it is possible is izelovant to the question of whether ot not i i certain that ¢ is true. The fact that, In those senses, it i possible that cis false is entirely compatible with the fact thatthe grounds8 ‘mmr venurrcavion ancuneen for alfrning © ane perfectly conclave and that i is perfectly certain that cis tro. The grounds T should give for affing ¢ tre that saw the phrase “the steam of thought” when I Toke At page 221 of James's book yesterday and that there no reason to alive thatthe printing om the page has changed or been tered since then, and that my ison Is normal and hat the light is geod. Those grounds would be accepted as absolutly tonclasive by everyone in ordinary ie, in what way 80 hey fai to bo conve? Te wil be sid “It posible that you had am hallucination yestrday and did not sx the page of book tall” As Tid Bolo, there reconnected with this statement problems of great imporance which cannot be stoded tn this pape. 1 il iit injec to thse remark, The meaing ofthe sateen s 8 that there ie some reson to think that had an hllcinationyester thy. The philosopher who makes this statement doesnot tnd toda hat hy vite of «parler knowledge of me and of my Gireamstances yesterday ht has evidence that I fered fem a fallcination, This statement is intended to make the general claim that every tne anyone hes believed tthe did peeve a certain thing ts pon that he did not prev that thing at All and that be bed an hallucoation insted Furthermore this Statement doesnot cai merely that whenever anyone has be: Trove that he percived a certain thing tis posible that be wa Ining an haicination 1s intended to lim that i aio pow ie tht he wes dreaming or tat he bad an opted aston, ‘Shor that be aufred from an error of some sort. Te pio: sophical statement “Whenever snyone has mado a peceptal judgment iis pon that he was suring from hallucination” I guise way of claiming Whenever anyone hes ade «per ‘opus adgmont itis possble Gat his judgment wai eon” GF dainng "to posh that every perceptual statement fate Now is it posible that very perceptual statement is fle fn any sense of I is pol” Irom wich fellows at not ovttn tht any percptnalsiaterent i tre? Lat ut review the thes of “Ils posable that we have described. Any pecepeal Statment may be false inthe sense that the cntadny of ay ‘eveoptalsaemont isnt selfcontradctorys butt doesnt foe “ME vexmexTIOn ancuseese 0 low that isnot corti that aay perceptual statement I tue (it, Tes wae, lee, that the Gideace that ons could afer in behalf of my peroepal statement doesnot ent! Cat the Stntament Is Bus but, agua, docs nt follow thats wot evtain that any preoptaal statement i tue (I). Tt cate be ‘untied that wih respect o each perceptual statement here some prictlr evidence tht that statement i fle thre to evince at all hat my statement at Ls a page of book Yesterday Is false (I). To argue tat shoe ume perceptual Statement ar fale therefore ot eran that au partear propa atemen 8 te unsound reasoning (le) It would Eetaurd to contend that thre no reason te Scep any pe ‘ceptual statement (IUI,). Nothing ria to he meant bythe statement "I is possible that every preptalatteent false excep the cla tthe founds for accepting any peropalstateneat are never Com thisive (I) As Tai, Tbelov thatthe grounds which one Could offer in behalf of any percoptnal sateen do ot ental Sha te satmont ste I dr not low in he lease however, tht he ol ar mat pee com pe enormously goo ground for acxping my peeptaalsatement that T caw the phrase Whe stain of thought page 224 af James's bok yesterday. The best way to show that owe grounds ssot ove woo at sme eee or yg tat id not se that pase yesterday. But no philosopher pre pero t do Bison, te pili da hat th rounds are not eonlsive dos ot rest on eoldnce. On What lcs it ret? Ona confusion, I eleve Ons ince to argue “tts not conclusive that ht percep atment ita Denese ‘tis possible tha xfs." But examination of thi statment show's thatthe words "Te posse tat tl fale do nat mean that hres oddone dat ts fle They anean tats ogaly pose Dati fab. But the fac that logialy pase that fe doesnot tend to show fn any way snot cre ite he octizton to contend ha is possible tat ion is hallucnatory ross, in pt at Yeast, upon the cae com fason which Bsa he root of the Verilestion Argument «cn fosion over th usige ofthe expression “tis possbles One ean50 sue vERIICATION AncusEENE construct an argument intended to prove that T did not have an hallucination yesterday, ‘sembles, in an important respect, the Verfeaton Argument. This ‘argument may be stated as follows: Tf certain things were to happen there would be good reason to ‘believe that Thad an hallucination yesterday. Tis posible that those things will happen ‘Therefore, iti not certain that T didnot have an hallucination yesterday. . ‘The second premise of this argument corresponds to premise I of the Verifeation Argument. In order that the conclusion should follow, this second premise must be understood in such a tvay that it implies the proposition “Te is not certain that those ‘things will ot happen.” I contend that there is no natural Inter- pretation of thie premise in which it is both the ease that the premise is true and that it implies that proposition. have tried to show that there is no sense ofthe expression “It is poste,” and the correlative expression, In which the state- ‘ment Tt is posible that the consequences of $ will ot occur” Doth is true and implies the statement “It isnot certain that the consequences of $ will occur.” To show this isto expose the most important error in the Verieation Argument. The Verifeation “Argument is « very tempting argument. From the propositions that § has an infinite number of consequences and that itis possible that these consequences will not occur and that if a fuficient number of them did not occur it would be conclusive ‘that Sis flee and that if were conclusive that Sis false then no ‘one previously made certain that $ is trac, it wems to follow that Tid not make certain yesterday that 8 is true. Tho proposition that itis posible that these consoquences will not occu is the premise of central importance. When one fist meets the argu- Inent one feels tha thie premise cannot be questioned. It seems 0 obviously trae that thre is scarcely need to state it: Ths ep= perently tnvullerable premise conceals @ serious fallacy. This premise must be understood in such a way that it implies that it {snot certain thatthe consequences of $ will occur. Anyone who ‘undertakes to examine carefully the several ordinary usages of "Te is possible” should soe that inthe usages expressed by Ill and Tl, this premise doesnot imply in the least that iis not certain {mE VERIICATION ANCUMENE 5 that the consequences of $ will occur. He should see that in the usages expressed by IN, and TI, this premise is clearly false. He should soo that in the usige expressed by III tis premise stands in need of support and that the proponents of the Verification Argument have offered nothing valid in support of i and that if t were to be supported by philosophical argument it could not, without cireularty, be supported by the Verieation Argument itself. The porsuasiveness of the Verfeation Argument arses fom the failure to distinguish soveral usages of “Ie Is possible” that ‘occur in different contexts in ordinary discourse. The result of this failure is that inthe philosophical context of the argument fone tres to make that phrase stiddle several different ordinary ‘sages all at once. Tn the usages expressed by Illy and Illy the proposition “Tt is possible that the consequences of $ will not ‘occur is an obvious logical truth. In the usages expressed by II, Mil, and My this proposition expresses a doubt, implies an un. Certainty. Through neglecting t0 distinguish these two sets of "usages one is led to think both thatthe proposition “It is possible that the consequences of $ will not oceut is an obvions truth and ‘that it implios that i€ is not certain that the consequences of S will occ ‘The proponents ofthe Verlleation Argument have emphasized their proposition that the consequences of an empirical statement are tnfinite in number. They have exerted themselves mainly in snguing for that premise of their argument, while they have said hardly anything at all about proposition Mia. If, however, fa i true then it does not matter, in a sense, whether IT is tr oF no. 1S has only one consequence and if that consequence is such that ft failed to occur $ would be refuted and if is not certain thatthe consequence will ccur, then i follows both that ifs not certain that $ 1s true and that I did not make certain yesterday that 8 i true. It willbe replied, of course, that if § ad only one ‘consequence then we could put that one consequence to the test. If 6, for example, express the only consequence of § then we could find out whether ¢ is true by my performing the action of ooking now at page 224 of James's book. If we knew that c is ‘uue and fe expressed the only consequence of 8, then we should kKoow with certainty that $s true. But $ has not just one or soveral consequences, but an infinite number. We eannot put aa2 ‘rm vexunicavion ancencest infinite number of consequences to the test. Therefore we cannot now with certainty that Sis rue. ‘This argument makes an important assumption. The assump: tion is that Teannot know that any consequence will ce. I can lknow that it de occurting end, perhaps, that it has occurred, but ‘not tha twill occur T assumes that T eannot know that true ‘until [perform the action of looking at page 224 of James's book, This assumption ts identical with proposition Wa. Why should ‘we accept this assumption? The philosophers who use the Ver tation Argument have given us no reason at al. This assumption ‘goes against our ordinary way’ of thinking and speaking. T should Say, for example, that itis certain that if I were to look now at page 224 of my copy of Jamess book I should see there the phrase “the stream of thought.” My grounds for saying this sre that I saw the phrase there yesterday, that there Is no reason to think thatthe printing on that page has changed or been altered since then, that my vision is normal, and that the light is good. ‘These are not merely “very good” grounds; they would ordinaely bbe regarded as absolutely conchisve. What grounds do those philosophers have for saying that tis nat certain that if I were to Took naw at that page T should seo that phase? None at all “There fs nothing whatever which prevents me from knowing now that ¢ is true. do not hace to perform the act of looking now in onder to know that sf T did perform it now I should sce that phrase. I should also say that {ts certain that if my wife were to Jook at that page now she would see that phrase and that i i cettan that ff my neighbor were to Took now he would see it and. +0.0n for an indefinite number of persons. IFT can know now that (is true I can also know now that any number of other stato= ‘ments, which express consequences ofS, are true That this num- baer of statements is finite or unlimited or indefinitely large docs rot prevent me from knowing that they are all true, I cannot perform an infinite number of actions of looking; but i dose not follow in the least that I cannot know what the results wonld be 4 any of an iaflaite number of possible actions of looking were performed. With regard to any one of an infinite numberof state- ‘nents which express consequences ofS, T can give grounds for saying that it is certain that that statement is true and the grounds fare what would ordinarily be regarded as perfectly conclusive, ‘cm vatunicaTION ARUN 53 ‘The philosophers who uso the Verlcation Argument have. put their emphasis in the wrong place. The ertieal step in the argu ‘ment is not the proposition that an empirical statement has an infinite number of consequences; i i the unjustified assumption that i cannot be certain that those consequences will ooo. 3 ‘Our attention has been concentrated on the fallacies contained {in propositions IIT and V of the original argument. There are, however, other erzors involved i the thinking that surrounds the ‘argument. One of these errors consists in a misunderstanding of the ordinary usage of expressions such as “verify,” “establish,” “make certain” “find out The proponents of the argument say that fT want to find out whether @ certain proposition i true T make a few “tests” or “observations.” These few tests may be ‘enough “Yor practical purposes” but, they say, Tcan go on making tests forever, "But there is always the theoretical possibilty of continuing the series of tes-abservations. Therefore here also no ‘complete verification i possible but only a process of gradually Increasing confirmation” ‘Let ws take an example. Suppose that I think that Paradise Lowe begins with the wards “OF Man's frst disobedience,” but that T ‘am not sure and wish to verify it. T take from the shelf a book ‘entitled Miton's Poetical Works. I turn tothe fst page of verse and under the heading Paradise Lost, Book I, soo that the rst four words of the fst line of verse aze “OF Man's fst disobedi= ence.” It would ordinarily be said that I had verified it The proponents ofthe Verification Argument would say that Ted not Fcompletely” verified it. They would say that I had not even, “completely” verified the fact thatthe st four words of verse on the page before me aze the words "OF Man's fst disobedience” ‘What shall Ido to further verify this latter fact? Shall 1 look ‘again? Suppose that T do and that Tee the same thing. Shall 1 ‘sk someone ele to look? Suppose that be looks and that he socs the same thing. According to this philosophical theory itis still not “completely” verified, How shall I further verify it? Would i aoap, “Testubty and Meanog." Phlowphy of Stone, 1H (1806),2 ‘mi venunIcATION ancoxeese Infinite number of consequences to the test, Therefore we cannot know with certainty that Sis tue "This argument makes an important assumption. The assump- tion is that T cannot know that any eonsegquence will ecu. T can know that i i occurring and, perhaps, that i har occurred, ut not that i il cen. It assumes that I cannot know that cis true tun T perform the aetion of looking at page 224 of James book. ‘This assumption is idontical with proposition Wa. Why should wwe acoept this assumption? The philosophers who use the Verii- tation Argument have given us no reason at all, This assumption ‘gocs against our ordinary ways of thinking and speaking, T should say, for example, that iti certain that if T were to look nov at [page 224 of my copy of James's book I should sew there the [phrase “the stream of thought” My grounds for saying this sre ‘hat T saw the phrase there yesterday, that there is no reason to think that the printing on that page has changed or been altered since then, that my vision is normal, and that the light is good. ‘These are not merely "very good” grounds; they would ordinary bbe rogarded as absolutly conclusive. What grounds do those philosophers have for saying that fis not certain that i [were to Took now at that page I should see that phase? None at lll ‘There is nothing whatever which prevents me from knowing now that cis true, Ido not have to perform the act of looking now in ‘order to know that if T did perform it now I should woe that phrase. T should slzo say that itis certain that sf my wife were to fook at that page now she would see that phrase and that i is certain that # my neighbor were to Took now he would se it and 50 on for an indefinite number of persons. IF I can know now that fs true T ean also know now that any number of other state ments, which express consequences ofS, are true. That this num- ber of statements i infinite or unlimited ar indebaitely large does rot prevent me from knowing that they are all trae. I cannot perform an infinite number of actions of looking; but it does not follow in the least that T eannot know what the results would be {f any of an infinite number of possible actions of looking were performed. With regard to any one of an Infinite numberof state= ‘ments which express consequences of §, I can give grounds for saying that it certain that tat statement Is true and the grounds ‘are what would ordinarily be regarded as perfectly conclusive ‘ime VERIFICATION ARGUMENT 5 ‘The philosophers who use the Verification Argument have put their emphasis inthe wrong place. The eriticl step in the argue rent is not the proposition that an empirical stitement has an Inflate number of consequences; i isthe unjustied assumption ‘that it cannot be certain that those consequences wil cur. 3 Our attention has been concentrated on the fallacies contained {in propositions TIT and V of the original argument. There are, however, ther errors involved in the thinking that surounds the argument. One of these errors consists In a misunderstanding of the ordinary usage of expressions much as “verify,” “establish,” “make certain” “ind out” The proponents of the argument say that if T want to find out whether a certain proposition is true T make a few “tests” or “observations.” These few tests may be ‘enough “or practical purposes” bu, they say, Tean go on making tests forever. "But there is always the thoorctical possibilty of continuing the series of testabservations. Therefore hero also no ‘complete verification is possible but only a process of gradually ‘increasing confirmation”= Lotus take an example, Suppose that I think that Paradise Lost Dogine with the words “OF Man's first disobedience,” but that 1 ‘am not sure and wish to verify i. T take from the shelf « book entitled Milton’ Poetical Works I turn to the fist page of verse ‘and under the heading Paradise Lost, Book 1,1 soo thatthe fast, four words ofthe fst line of verse are “OF Man's fst disobedi- fence” It would ordinarily be said that I had verified i. The ‘proponents ofthe Verification Argument would say that I had not pletely” verifed it. They would say that T had not even “completly” verified the fact thatthe Sst four words of verse on the page before me are the words “OF Man's first disobedience.” What shall I do to further verify this latter fact? Shall T Took gain? Suppose that T da and that I see the same thing. Shall T ‘sk someone elee to Took? Suppose that he looks and that he sees the same thing. According to this pillosophical theory It is stil not “completely” verified. How shall I further verify it? Would it ‘Cama, “Testability and Meang” Phosphy of Science, (188),au ‘Tun VERIFICATION snounteN be “further verification” iT wore to look again and again at this page and have mare and more other people look again and again? [Not at alll We should not describe it ol Having looked once are fully, if T then continued to look at the page we should not say that I was “urther verifying” or “tying to further verify” that the fist four words of verse on that page are "OF Man's fst dis- obedience.” Carnap declares that although it might be foolish or lmpractical to continue “the series of tstobservations sill one could do so “theoretically.” He implies that no matter ‘hat the circumstances we should describe certain actions as “further ver- fying” or “further confirming” this fact. That isa mistake. Suppose that T continued to look steadily atthe page and someone won- ‘dered why I was behaving in that way. If someone else were to say “He is uying to further verify that those are the first four words,” this would be an absurd and humorous romark. And this description would be equally absurd if my actions consisted in showing the hook to one person after another. In those clrcum- stances there is nothing which we should call “further vera tion.” To suppose that the “process of verification” can continue ‘without end” is simply to ignore the ordinary usage of the word “verify.” Its false tha “there fs always the theoretical possiblity of continuing the series of tesobservations” It is possible that should continue 10 look at the page. It is not possible that 1 should continue the verification ofthat fact Because, in those cle- ‘cumstances, we should not describe anything as “further verfca- Hon” of it. The verification eames to an end, ‘Camap would say that the statoment “The Best four words of verse on this page ate ‘OF Man's first disobedience’” ie nat “co pletely” verifled because “there remains stil the theoretieal posi- bility of denying the sontence."*" What does he mean by “there remains stl the theoretical possibility of denying the sentence"? ‘Does he mean that itis logically possible that someone should deny that statement? This i true, but irelevant tothe question of ‘whether it has been established thatthe statement i true. Docs he ‘mean that the contradictory of the ahove statement is not self- contradictory? This is aso true and also iselevant. Does he mean, that there is some reason for thinking thatthe statement is fale, tha, 428 ‘me yenuncavion anuneenr 5% or that there is no reason for thinking t true? But there is the best of reasons for saying itis true, namely, that I looked carefully at the page « moment ago and saw that those were the frst four ‘words; and there is no reason whatever for saying that iis fale Does he mean thatthe fact that I looked atthe page and saw that those were the fist four words of verse does not “completely” ‘establish thatthe statement is tric? In what way does It fil to ‘establish it "completely"? Shall we repeat that i does not “com pletely” establish i because “there remains stil the theoretical possibilty of denying the statement”? But this is circular reason ing. Carnap's statement “There remains still the theaetical possi Diy of denying the sentence” embodies the sume confusion that ssuounds premise IIL of the Verifeation Argument, the confi son produced by the failure to dstingush the several diferent ‘usages ofthe expression “Its possible.” ‘Some philosophers have thought tha, when it i sid in ordi= nary disoourse that it is absolutely certain that so-and-so, what this means is that tis practically certain that so-and-so This 8 cloarly a mistake. The ordinary usage of “practically cartain” is ‘quite diferent from the ordinary usage of “sbsolutely certain” Tt tially certain” norally means "It is almost certain.” To is practically certain that so-and-so implies that it is ‘not absolutely certain, “It is practically certain that p is tue implies that i is reasonable to have a slight doubt that p is true snd implies thatthe evidence tht pie tre isnot absoltely con= ‘lusive. “It is absolutely certain that p is true” implies, on the ‘eantrary, thatthe evidence that p is tre is absolutely conclusive ‘and implies that in the light of the evidence it would be wnreason able tohave the slightest doubt that pis tue. Lewis and Camap do not, of course, make the mistake of ‘identifying absolute certainty with practical certainty. They make 4 different mistake. They identify absolute certainty with “theo- retical certainty.” Lewis, for example, uses the expressions “abso- Intely certain’ and “theoretically certain” fterchangeably.*" Both the and Carnap say that the tith of an empirical statement ean bo practically certain but no "theoretially certain.” How are they using the expression “theoretical certainty"? What state of afairs, op. An Ana of Knowlege and Valuation, 180,56 ‘ne veranicaTIOn ancien if could be realized, would they call “theoretical certainty”? In ‘what circumstances, supposing that such circumstances could exist, would it be “theoretically cortain” that a given statement is ‘rue? The answer is clear from the context oftheir arguments It ‘would be “theoretically certain” that a given statement is true only if an infinite number of “tests” of “acts of verification” had been ‘performed. It is, of course, a contradiction to say that an infinite ‘umber of “test” or acts of any srt have been performed by any~ ‘one, [tis not that itis merely impossible in practice for anyone to perform an infinite number of acs. It is impossible in theory. ‘Therlore these philosophers misuse the expression “theoretically certain.” What they call "thooretical certainty” cannot be attained ‘even in theory. But this misusage of an expression fein itself of slight importance. What is very important is that they identity ‘what they mean by "theoretically certain” with what is ordinarily meant by “absolutely certal.” IE this identification were eoeroct then the ordinary mesning of “absolutely certain” would be con- twadictory. The proposition that itis absolutely cartain that a given statement is true would ental the proposition that someone bad performed an infisite number of acts. Therefore, it would bi ‘a contradiction to say, for example, "I is absolutely certain that Socrates had a wif.” Statements of this cort are often false, or they are often unjustified on th strength of the evidence at hand But to say that auch statements are one and all self-contadictory 1s perfectly absurd. A philosophical theory that has such a conse~ ‘quence is pny false 2 (Praha very propely ras the question ay whtcetainty, and the ighest degre of cosy x He cools tat I have ot “ver empl ‘espealy and recly to sp whet carainty sand what fo gene eit ‘Se ("Philsuophiel Cartaing.” ‘The Philsophicd Review, LXXI, 307). To pot do tht nthe present cay, but Ido tee In" Knowledge and boc Uy to dune what I il "the song nse af Sm and E ‘als the cinerea © ir wht various Pitopbors heve ha a mind ‘tes thoy lve spolen of "see" “perfect” or “neta” cera (ee p70) tin tote et Diet Beep Ce p28) ‘rant Mina tus to sharctarae the noton of ean Dy desing {ngecton between «pesnsrgung usttement a Sta cade bing Svaling tke the sks anced wt tn aerent there neing began by vein to do 0” (The Phswophiea! Review, op ep. S1). [He gnron fo ay at“ satoat cate only if Reval be soca for anyone posting tho evidepoesvaable frit roar the ssemect ‘cot (ii) Holly he charters the hight depen of coal, ‘nmr venmicarion ancuxeene 3 hich all “pubsophia eta,” a follows “A ttennt eno the Bes gre o cert aly is appated by erence wk ost fs silos to mak the greaten posite peel on the arth of Sfatement Fern peso to roged semen beng taint hs ‘eee, be st be wing to mk angling oa Re tule eae p33) Beg ginete oeeeee te a ly ones es rear at SE aoa lon mimeo {bios sunt hat whtaever ie Gotan that stateaent tw Phopaall ues tccenieaniers eal iota coteaner cieaaltal rs see nEhGe Gea, Cre ee aes fenpiea, proc ope) wih respect to whi t wea be rao Grecia abi ote Gartnrencrate i aa eran appa ik tear amet oem tea Sotgal Cornmeal aaa ree Seen ace ies hag acer teen Se ery Be oy onime encore ore Speaaat coy een at casein etectioe Statice sat ge sae ian Ree ireagtaga san a bury. Some of these kinds are dceced tn “Knowodge and Beli” oclagy aah tana dane moe Tees eee ee Bae 1058), pp 7-00 ny PresKnowledge and Belief « fe must recognize that when we know some: thing we alther do or by reflecting, can know that our condition is one of knowing that thing, while when we believe something, ‘wo eithor do ar can know that our condition is one of belicving and not of knowing: so that we cannot mistake belie for knowl ae ‘This remark is worthy of ivestigetion. Can I dis cover in myself whether I know something or merely belive i ‘Let us bogin by studying the ordinary usage of “know” and “believe.” Suppose, for example, that several of us intend to go for a walk and that you propose that we walk in Cascadila Gorge. I protest that T shoold like to walk beside a flowing stream and that at this season the gorge is probably dry. ‘Consider the following cases (2) You say “I bliove that it won't be dry although have no particular reason for thinking so." If we went to the ‘gorge and found a flowing stream we should not say that you new that there would be water but that you thought so and were right, (2) You say “I beliove tht it won't be dry because “HL A, Paced, Knowledge and Perception (Orford; The Clarendon Pres 1860), 88 8 ENOWLEDOE Ano meter 50 {trained only three days ago and usually water flows in the gorge for at least that long after a rain.” If we found water we should he ‘inclined to say that you knew that there would be water. It would bbe quite natural for you to say “I knew that i woulda’ be dry"; and we should tolerate your remark. This case difers fom the previous one in that here you had a reason, (8) You say “T know that it won't he dry” and give the same reason as in (2). If we found water we should have very litle hesitation in saying that you knew. Not only had you a reason, Dut you sid “I know” instead of “I believe.” It may seem to tus that the latter should not make a dliferenco—but it does, (4) You say “I know tha it won't be dry” and give a steonger reason, og, “T saw a lot of watar fowing in the gorge when I passed it this moming” If we went and found. water, there ‘would be no hesitation at all in saying that you knew. If, for ‘example, we later met someone who sail "Weren't you surprised to see water in the gorge this afternoon?" you would reply "No. [knew that there would be wator; Thad been there earlier in the day.” We should havo no objection to this statement. (5) Everything happens as in (4), except that wpon going to the gorge we find it to be dry. We should not sty that you knew, Dut that you belicved that there would be water. And this is true even though you declared that you new, and even though your evidence was the same as it was ia ease (4) in which you did now. Tish to make some comments on the usage of “kno,” knew," “believe,” and “believed,” as illustrated in the preceding cases (a) Whether we should say thet you knew, depends in part on whether you had grounds for your assertion and on the strength of those grounds, There would certainly be less hesitation to say ‘that you knew in case (4) than in case (3), and this ean be due only to the difference in the strength of the grounds (b) Whether we should say that you knew, depends i part on how confident you were, In caso (2), i you had sald “It rained only three days ago and usually water fows in the gorge for at Teast that long after a rain; but, of course, I don't fel absolutly sure that there will be water,” thea we should not have said that ‘you knew that there would be water. If you lack confidence thato NOWLENGE AND RELIRE ‘pis true then others do not say that you know that p is true, fven though they Know that p is true. Being confident is a neces sary condition for knowing. (c) Prichard says that if we reflect we cannot mistake belief {for knowledge. In case (4) you knew that there would be water, ‘and in case (5) you merely believed it. Was there any way that you could have discovered by relletion, in ease (5), that you ‘id not know? Tt would have been useless to have reconsidered your grounds for saying that there would be water, beeanse in fase (4), where you did now, your grounds were identical, ‘They could be at fault in (5) only if they were at fault in (4), and they were not at fault in (4). Cases (4) and (5) differ in ‘only one respeet—namely, that in one case you did subsequently find water and inthe other you did not. Prichard says thet we can determine by reflection whether we know something or merely believe it, But where, in these cases, isthe material that reflection would strike upon? There is none. ‘There is only one way that Prichard could defend his position. He would have to say that incase (4) you did not know that there would be water. And its obvious thet he would have said this. But this is false. It s an enormously common usage of lan= quage to say, in commenting upon just such an incident as (4), Tle knew thatthe gorge would be dry because he had soen water flowing there that morning” Tt isa usage that all of ws are fam Jar with, We so employ “know” and “knew” every day of our lives. We do not think of our usage as being loose or incorrect —and itis not. As philosophers we may be staprsed to observe that it con be that the knowledge that p is true should difer from the belie that p is true only in the respect that in one case pis tee and in the other false, But that is the fat ‘There is en argument that one Is inlined to use asa proof that ‘you did not know that there would be water. The argument is the following: It could have turnod out that you found no water; if ithad s0 turned ont you would have been mistaken in saying that you would find water; therefore you could have been mis: takes but if you could have been mistaken then you did not now. Now it certainly could have turned out thatthe gorge was quite ‘ry when you went there, even though you savr lots of water xxOWLEDCE AND BELIEE, a owing through it only a few hours before. This does not show, however, that you did not know that there would be water. What it shows is that although you knew you could have been mis: taken? This would seem to be a contradictory result; but It is not, Tt seems so because our minds are fied upon another usage Of “know” and “lnew"; one in which “It could have tured ont that T was mistaken,” implies “I didnot know." ‘When is “know” usod in this sense? I believe that Prichard tases it Jn this sense when he says that when we go through the ‘proof of the proposition thatthe angles of triangle are equal to two right angles we know that the proposition is tue (p. $9). He says that if we put to ourselves the question: Is our eoniton ‘one of knowing this, of iit only one of being convinced of i? ‘then “We can only answer ‘Whatever may be our state on other fccasions, here we are knowing this” And this statement §s an ‘expression of our knowing that we are knowing; for we do not believe that we are knowing this, we know that we are” (p. 89). “He goes on to say that if someone were to object that we might bbe making a mistake “because for all we know we exn later on. discover some fact which is incompatible with a triangle’ having angles that are equal to two right angles, we can answer that we now that there can be no such fact, for in knowing that tr angle must have such angles we also know that nothing can exist ‘which is incompatible with this Fae” (p. 90). ‘tis easy to imagine a non-philosophical context in which st ‘would have been natural for Prichard to have sad “T know that the angles of a trlangle are equal to two right angles” Suppose that « young man just beginning the study of geometry was in doubt as to whether that proposition is trie, and had even con- structed an ingenious argiment that appeared to prove it false, ‘Suppose that Prichard was unable to find any error in the argu ‘ment. He might have said to the young man: "There must be an Some reer ses to Baye thought tht Twas denying be tit “T that ent that That as ny neti, my word donot hn tat plication If hal sl “ldough you knew you ote Imistaon” I shoud have denied the above ental nd aio Soult fie minged "Enew-" The difrense Hetwosn th stg ed weak sens “know and “how isa tat th eaten as forthe stg ltt forth wea ee blr for bh fff tap hc oe es and di a) now tht poe xvownmcr AND mr error ini, Low that the angles of a telangle are equal to two right angles” ‘When Prichard says that “nothing can exst which is incom- patible with” the truth of that proposition, i he prophesying that ‘no one will ever have the ingenuity to construct a flaless-lokin ‘argument against i? I believe not. When Prichard says that “we now (and implies that he knows) that the proposition is true land knot that nothing can exist that is incompatible with its being tre, he i not making eny prediction as to what the future svi bring in the way of arguments or measurements. On the ‘contrary, he is asserting that nothing thatthe future might bring ‘ould ever count as evidence against the proposition. He is ime plying thet he would not call anything “evidenes” against it He Js using “lnow” in what I shall call its “strong” sense, “Know” is ‘used in thie sense when a person's statement “I know that p is true" implies that the person who makes the statement would look upon nothing whatever as evidence that p is false. It must not be assumed that whenever “know” is used in con- nection with mathematical propositions itis used in the strong sense, A great many people have heard of various theorems of geometry, eg, the Pythagorean. These theorems are a part of “common knowledge.” If a schoolboy doing his geometry assign- ‘ment fet a doubt about the Pythagorean theorem, and said to an ‘adult "Are you sure that it is trae?” the latter might reply “Yes, know that Its” He might make this reply eventhough he could ‘ot give proof of itand even though he had never gone through ‘proof of It If subsequently he was presented with a “demon- stration” thatthe theorem is false, or If varions persons reputed to have a knowledge of geometry soberly assured him that it is false, he might be bled with doubt or even be convinced that he ‘vas mistaken, When he said "Yes, I know that i trae,” he did not pledge himself to old to the theorem through thick and thin. He did not absolutely exclude the possibility that something could prove it to be false. I shall say that he used “know” in the “weak” renee, (Consider another example from mathematics of the difference between the strong and weak senses of “Inow." I have just now rapidly calculated that 92 times 16 is 1472. If Thad done this in xsownzoon aN mete 6 the commerce of daily life where a practical problem was at stake, ‘and if someone had asked “Are you sure that 92 x 16 = L472" might have answered "I know that iis 1 have just now ealou- latod it” Bur also T night have answered “T knovr that iis; but will caleulate it again to make sure” And here my language points to a distinction, I say that I Know that 92% 16= 1472. Yet Tam willing to confirm that &, there is something that T should call “making sue”; and, likewise, there is something that should call “Boding out that itis falso.” IFT were to do this calculation again and obtain the revue that 92 x 16 = 1872, and 41 were to carefully check this latter calculation without finding any error, I should be disposed to say that I was proviously mis: taken when I declared that 92 x 16= 1472, Thus when I say that I know that 92 x 16 = 1472, Tallow for the possiblity of a ‘refuation; and so Tam using "know" in its weak sense. "Now consider propositions like 2+ 2= 4 and 7-+5= 12. Ikis hard to think of circumstances in which i would be natural for sme to say that I know that 2+ 2=4, bocause no one ever ques tons it. Lat us try to suppose, however, that someone whose {intelligence I respect aagues that certain developments in arith retic have shown that 2+ 2 does not equal 4. He writes out a proof of this in which I can find no flaw. Suppose that his de- Ieanor showed me that he was in earnest. Suppose that several persons of normal intelligence became persuaded that his proof ‘was correct and that 2 +2 does not qual 4. What would be my reaction? I should say “I cant see what is wrong with your proof; bbut it és wrong, beeause T know that 2+ 2 4." Here I should bbe using “know” ints strong sense. I should not admit thet any argument or any future development in mathematics could show that it is false that 2+ 2= 4 "The propositions 2+2= 4 and 92x 16= 1472 do not have the same stats, There con be a demonstration that 2+ 2= 4. [But a demonstration would be for me (and for any average per- son) only a curious exereise a sort of game. We have no serious interest in proving that proposition? It does not need a proof. It +See ad philosophers have taken an tc proving tat 2 PETA Regs Lala: Nee Binoy on tho Undoranding "Be 1, chof XNOWLEDGE AND RELIEF stands without oe, and would not alfa proof went against ‘The cate it diferent with the proposition tht G2 x 18 = M72 We take an Interest in the demonstration (calculation) because that proposition depends upon its demonstration. A clealation say lead mo to rject it 8 fale. But 2} 24 doesnot depend os demonstration. 1 des ot depend on anything! And ih the calculation that proves that 92> 16 1472, there are steps that do not depend on any calculation (eg, 2x 6= 1% 5+ 54 0=1) ‘There i correspondence betwen this dualism inthe logical status of mathemateal propositions a the two senses of now” When T wie “know in the weak sense Tam prepared to Teta investigation (demonstration, calolation) determine wheter the something that T claim to know ie tric or false. When T ase “koow" in the strong senso Iam not prepared to lok upon any- things an Hncetzation T do not concede that anything what socver could prove mo mistaken; Ido not regard the matter ax ‘open to any question; I do nok admit that my proposition could tem out toe flo, that any future investigation could refute It corcast doubt on ‘We have been considering the strong senso of “know” in its pplication to mathematical propositions. Does i have applic. tion anywhere in the realm of empirical propostions—for ex. mpl, to propositions that assert of imply that certain physial {ings ext? Descartes said that we have a “moral assurance” of the truth of some ofthe ater propositions but that we lack “metaphysical certanty."® Locke said thatthe pereption ofthe csstence of physical things fs not "so certnin at our intuitive ig ps the Renato of rte es 0; Thay bre we 1 thew tt cia be deduced from eran rian po determine what reins and ale of ference ee regi nthe dou, Tir rest [ect boon i the ont ofthe dolcton, Compare thse remsey abest te stra Spi of "knew" with same of te emt ht itis Enka the ed ‘eo rm Fran, duty amination a ion th Seton that depends the certainty and ven of all our ow, wh coin ay oe Ba eos ft inane evn rogue a gust, Eck, ay 3 Wy Sse. ~ NOWLEDGE AND BELIEE 6 Jnowlege, or the deductions of our reason” although “it is assurance that deserves the name of knowledge.” * Some philoso- phers have held that when we make judgments of perception such as that there are peonies In the garden, cows in the field, for dishes in the eypbosrd, we are “taking for granted” that the ponies, cows, and dishes exit, but not knowing it in the “strict” sense. Others have held that ali empirical propositions, including judgments of perception, aro merely hypotheses” The thought Dohind this exaggerated mode of expression is that any empirical proposition whatever could be refuted by future experience— that i, it could turnout tobe false. Are these philosophers right? ‘Consider the following propositions (1) The sun fe about ninety million mils from tho earth (Gi) There is a heart in my body. (i) Here is an inkebotdle In various circumstances I should be willing to assert of each of ‘these propositions that I know it to be true. Yet they differ stik- ingly. This I see when, with each, I try to imagine the possiblity that itis false. (i) If'in ordinary conversation someone said to me “The sun is about twenty million miles from the earth, isnt it?” I should reply “No, it i about ninety million miles from us." If he suid “Think that you are confusing the sun with Polaris” I should reply, "I know that ninety milion miles i roughly the sun's dis- tance from the earth.” I might inite him to veefy the figure in fan encyclopedia, A thd person who overheard our conversation ‘could quite correctly report that I know the distance to the sun, ‘whereas the other man did not. But this knowledge of mine is lite better than hearsay. I have seen that figure mentioned in 1 few books. [know nothing about the observations and ealeula- ‘ons that led astronomers to accept it. If tomorrow a group of ‘eminent astronomers announced that a grost error had been made and that the correct figure is twenty million miles, T should not “Yok, an kh 3. sea sein ema a ae 8 le ee SE int wt Bor Pubes’ fe ioo {eNOWLENGE AND MELIEY insist that they were wrong. Tt would surprise me that such en enormous mistake could have been made. But I should no longer hae willing to say that T now that ninety milion is the correct figure. Although I should now claim that T know the distance to be about ninety milion miles, itis easy for me to envisage the possibilty that some future investigation will prove this to be false, Gi) Suppose that after a routine medical examination the ex sted doctor reports to me thatthe X-ray photographs show that Thave no heart I should tll him to get a new machine, I should bbe inclined to say thatthe fact that I have « heat is one of the fw things that T can count on as absolutely certain, T ean feel {tbeat, [know ifs there, Furthermore, how could my blood ein late i€T didat have one? Suppose that later on T suffer a chest, injury and undergo a surgical operation. Aftrwards the aston- ‘shed surgeons solemnly declare that they searched my chest cavity and found no heart, and that they made incisions and looked shout in other likely places but found it not. They are convinced that Tam without a heart. They. are unable to under- stand hovr circulation ean occur or what accounts for the thump- ing in my chost. But they are in agreement and obviously sincere, and they have clear photographs of my interior spaces. What ‘would be my attitude? Would it be to insist that they were all ristaken? T think not. I believe that I should eventually accept their testinony and the evidence of the photographs. I should consider to be false what I now regard as an absolute certainty. (ill) Suppose that as T write this paper someone in the next room were to call out to me “I cant find an ink-botle; is there ‘one in the house?” I should reply “Here is an inkcbotle” If he said in a doubtful tone “Are you sure? T Tooked there before.” should reply "Yes, Tow there is; come and get it” Now could it turn out to be false that there isan ink-bottle directly in front of me on this desk? Many philosophers have thought so. They would say that many things could happen of sich a nature that if they did happen it would be proved that I fam deceived, [agree that many extraordinary things could hap: pon in the sense that there is no logical absurdity in the suppos tion, Ie could happen that when Inet reach for this ink-botte ‘my band should seem to pass through it and T should not feel NOWLEDGH AND mex or the contact of any object. 1t could happen that in the next mo- ‘meat the ink-hottle will suddenly vanish from sight; or that 1 should find myself under a tree in the garden with no ink-bottle about; or that one or more persons should enter this room and Aeclaze with apparent sincerity tht they see no inkcbotle on this desk; or that a photograph taken now of the top of the desk should clearly show all ofthe objects on it except the ink-botle, Having admitted that these things could happen am I com: pelled to admit thet if they did happen then it would be proved that there is no inkcbotle here now? Not a alll T could sty that ‘when my hand seemed to pass through the iak-bottle I should then be suffering from hallucination; that ifthe inkcbotle sud= enly vanished it would have miraculously ceased to exist; that the other persons were conspiring to drive me med, or were them- selves victins of remarkable concareent hallucinations; that the ‘camera possessed some strange flaw or that there was trickery in developing the nogative I admit that inthe nest moment I could ‘ind mysell under a tee or in the bathtub, Bt thief not to admit ‘that it could be revealed in the next moment that Tam now ‘dreaming. For what T admit i that T might be énstantancoushy ‘transported to the garden, but not that in the next moment 1 ‘might wake up in the garden. There is nothing that could happen to me in the next moment that I should call “waking up"; and therefore nothing that could happen to me in the next moment ‘would be accepted by me now as proof that I now dream. "Not only do I not have to admit that those extraordinary oeeur- renoes woul be evidence that there i no ink-botle here; the fact fs that Ido not admit it. There is nothing whatever that could happen in the next moment or the nest year that would by me be called evidence that there is not an ink-botle here nov. No *1My voit somewhat dferet here fey wht io “The Vere Sati Aree Tce am cna wth ig oie ZEN te tf ro he tinh the = ‘nas rae der fel hat ern ern tht the things fn si ‘i ct apps Flly T held threat fe pred een at they Wil sot pee. Her, {arta dingrecing wih ay of thse poly, bk ‘ss aug cfr po hat iy tte oe {ning could loppam, dace not vague me to ade that they were ‘ppm tnt weld bene Sat hr blero)68 vow AND neta fature experience or investigation could prove to me that Tam mistaken, Therefore T were to say "I know that there is an ink- bottle here” I should be using "know" in the strong sense. Tt will appear to some that I have adapted an unreasonable attitude toward that statement. There is, however, nothing wn- reasonable about it. Te seems so because one thinks thatthe state- ‘ment that here is an ink-bottle must have the same status as the Statements that the sun is ninety milion miles away and that T Ihave a heart and thet there wil be water in the gorge this after- noon. But this a prejudice. Tn saying that I should regard nothing as evidence that there {sno ink-bottle here now, Iam not predicting what I should do if vacious astonishing things happened. If other members of my family entered this room and, while looking at the top of this desk, declared with apparent sincerity that they soe no ink-botle, might fall into a swoon or become mad, I might even come t0 believe that there is not and has not been an ink-bottle here, 1 ‘cannot foretell with certainty how I should react. But if iis not ‘prediction, what is the meaning of my assertion that I should regard nothing as evidence that there is no ink-botle here? ‘That assertion desribes my present attitude towards the state- ment that here isan ink-botle. Te does not prophesy what my attitude would be if various things happened. My present atti- tude toward that statement is radically diferent from my present altitude toward those other statements (eg that Tave a heart) * 1 do now admit that certain future occurrences would disprove the later. Whereas no imaginable future occurrence would be ‘considered by me now as proving that there is not an ink-botle haere “These remarks are not meant to be autobfographical. They are ‘meant to throw light on the common concepts of evidenes, proof, and disproof. Every one of us upon innsmerable occasions of * (he word “attode” int veny stfactory, but ennne hi of a thera tat pl tet, By “ny site” Fr, be T should sy ond think vats things ware to happen. yy pase ‘ueinde Than see shal uae tno Toe ge thos fatate tine Sos thingsaceually dal happen a tine, Te tio tinction that shows that ny dept of Sy peeet thud” W na prophecy] xvowusncr AND ReLIn? « daily life takes this same aide towards various statements ‘about physeal things, eg, that here isa torn page, that this dish is broken, that the thermometer reads 70, that no rug ison the oor. Furthermore, the concepts of proof, dsproof, doubt, and conjecture require us to take this attitude. In order for It to be porible that any statements about physical things should turn ‘ut to be fal it necessary that some statements about physical things cannot turn ont 0 be fale ‘This will be made clear i we ask ourselves the question, When do we say that something turned out to be fase? When do we tse those words? Someone asks you fora dollar. You say “There {sone in this drawer” You open the drawer and look, but it is perfectly empty. Your statement turned out to be false. This ean be ssid because you discovered an empty drawer. Tt could not bbe anid if t were only probable that the drawer is empty or were still open to question, Would it make sense to say “I bad better take sue that it is empty; perhaps there is «dollar init aftr all?” Sometimes; but not always. Not if the drawer les open before your eyes, That remark i the prelude to a search, What search ‘an there be when the emptiness of the drawer confronts you? In certain circumstances there is nothing that you would call “making sure” that the drawer is empty; and likewise nothing that you would call “ts turning out tobe fale” that the drawer is empty. You made sure that the drawer is empty. One statement about physical things turned out to be false only because you ‘made sure of another statement about physial things. The two ‘concepts cannot exist apart. Therefore it i impossible that every ftatement about physieal things could turnout to be false. Tin a certain important respect some a prior statements and some empirical statements possess the same logical character. ‘The statements that 5 x 525 and that hore is an ink-botle, both lie beyond the reach of doubt. On both, my Judgment and reasoning rests, IF you could somehow undermine my confidence in either, you would not teach me caution. You would fil my ‘mind with chaca I could not even make conjectures if you took faway those Bxed points of certainty; just as @ man cannot fy to ‘limb whore body has no support A conjecture implies an un- ‘ersanding of what certainty would be. If it is not a certainty that 5x 5~ 25 and thet here & an ink-bottle, then T do not un-0 XENOWLEDCE AND RELIE derstand what it i, You cannot make me doubt elther of these statements or teat them as hypotheses, You cannot persuade me that future experience could refute them. With both of them ifs perfectly unintelligible to me to speak of a “possbiity” that they are false. Ths sto say that T know both of them to be true, ia the strong sense of “know.” And I am inclined to think thet the strong sense of “know” is what various philosophers have had In mind when they have spoken of “perfect,” “metaphysical.” or “strict certaimty.” It will be thought that T have confusod a statement about my “sensations” or my “sense-data,” or about the way something looks or appears to me, with a statement about physica things. It will be thought that the things that Thave said about the stato- ment “Hore isan ink-botle” could be trae only if that statement |s interpreted to mean something like "There appears to me to be an inkcbotl here,” interpeeted so as not to assert or ime ply that any physical thing exists, I wish to make it lear that my Statement “Here is an ink-botle” is not to e interpreted in that ‘way. It would be utterly fantastic for me in my present eircum- stances to say “There appears to me to be an in-hole here.” 1 someone were to call me on the telephone and say that he urgently needed an ink-bottlo ¥ should invite him to come here and get this one. If he said thet it was extremely urgent that he should obtain one immediately and that he could not aford to ‘waste time going to a place where there might not be one, T should tell him that iis an absolute certainty that there is one hore, that nothing could be more certain, that i i something T absolutely guarantee, But if my statement “There is an ink-botte Dero” were a statement about my “sensations” or “sensedata,” ot fit meant that there appears to me to be an ink-botle here or that something here looks to me like an inkcbotl, and i that 5, Desay, for example, appreny took shi riteon for something’ boing “eptelyeeen® Wht he cull not ane fn the least groan Boob ms al fla ge ee cme lane fase tot ce dul fo gouras nga lend dose an de a ‘neal tere ot Shwe a cele fe cate Sep pttAL Sang few W) Asta (iy md) ‘id of “ttutive knowlege tat one onnatdmagin a pater sony, fed hat favor no soom for estan, dour stato” Eo, BE WV ch 230 . a NOWLEDGR AND BEE a all that I meant by it—then I should react quite differently to his urgent request. 1 should say that there is probebly an ink- Dott here but that I could not guarantee it, and that if he needs fone very desperately and at once then he had better look else- ‘where. In short, wish to make it clear that my statement “Here fs an ink-hottle” is strictly about physical things and not about “sensations” “sense-daa,” or “appearances.” ™ ‘Let us go back to Prichard’s remark that we ean determine by roflection whether we know something or merely beliove it. Prich- far would think that "knowledge in the weak sense” is mere be- lief and not knowledge. This is wrong. But if we let ourselves speak this way, we ean then see some justification for Prichard romark. For then he would be aserting, among other things, that wwe can determine by reletion whether we know something in the strong sense or in the weak sense. Ths isnot literally true; however, there ie this truth in st—that rellection can make us realize that we are using “I know iin the strong (or weak) sense {nna particular cae. Prichard says that relletion can show us that “our condition is one of knowing” a certain thing, or instead that “our condition is one of believing and not of knowing” that thing do not understand what could be meant here by “our condi ton." The way I should put itis that reflection on what we should think if certain things were to happen may make us vealizo that ‘we should (or should not) eall those things “proof” or “evidence” that what we claim to know is not so. I have tried to show that the distinction hetween strong and weak knowledge does not run parallel to the distinction between a prior! and empirical know- ‘edge but cuts across it, ie, these two kinds of knowledge may be distinguished within a prior! knowledge and within empirical Tenowledge, TRetiction can make me resliz that Iam using “know” in the strong sense; but can reflection show me that T know something in the strong sense (or in the weak]? It isnot easy to state the (The rmainor of the cmy & newly weiten. Th ods consion| yas wongy sted The rea seed tote flowing exchange be {roca Redd Taylor aad sy i respect to te oigial pepe yr, SSCS on Rove an say Ml Jone 166), hen, Sant nd Rt Aa, SV, ch 18, ay, "eden evowLEnce AND BEL logical facts here, On the one hand, s I make an assertion of the form “I know that p” it does not fellow that p, whether or not Tam using “know” inthe strong sense, fT have sald to someone jutside my room “OF course, I know thet Freddie is in here” find Tam speaking in the strong sense, it does not follow that Freddie is where [ claim he i. This logical fact would not be altered even if I realized that I was using, “know” in the strong, sense. My reflection on what I shoald say i... cannot show ime that I know something. From the fact that [should not call ‘anything “evidence” that Freddie is not here, it does not follow that he i hore; therefore it does not follow that I know he is here, (On the other hand, in an actual case of my using “now” in the strong sense, I cannot envisage a possibility that what T say to be true should turn out to be not tue. IF 1 were speaking of ‘ancther person's assertion about something, 1 could think both that he i using “know” inthe strong sense and thet nonetheless what he claims he knows to be s0 might turn out to be not so. But in my own case I cannot have this conjunction of thoughts, and this is @ logical and not a psychological fac. When T say that I now something to be so, using “know” in the strong sense, it is unintelligible to me (although perhaps not to others) to sup- pose that anything could prove that itis not so and, therefore, that Ido not know i* “Tigi the bast summary Team gv of what wrong and ight in ‘om lan tt one cad dete by stn whee one Lav ‘eg ly ble A dt he es the ay wee re ‘Soveraton wi Witgetse ef and rough acount he lle 0 fan my tong iter Mem (Sew York Orford Unversity Pres 1958), pp 87-98, Jako Hina provides Unecote etnent of the topic of “arving tal te Knowa wl specie ‘etree to char’ in Se he Knowledge end Balu (ithaca: Caroll Usiomty Prom, 1082) 8 Direct Perception 1 Tost tise te cin nition of Pro- fessor G. B. Moore to consciously bold two apparenty cont dictry views about perception. One i th view that “we do constantly se directly pat ofthe surfaces of pyskal objets * Tho other isthe yew that we cannot directly see any physical surface. Moore saya “Lam strongly ci’ to tke both ofthese incompatible views. Lam completely pucled about the mater, ou only wish T coud see any way of eling it I bleve that fm examination ofthis coat in Moore's thought wil ld an Insight nto a major problem ofthe philosophy of perception. Moore «highly intersting way of explaining sown philosophical wea the expersion “cl ee” He al attention to tha nda use ofthe word “see” n which a person sho hasan afterdimage when his eyes are closed ay be sald to Soe the aferimage. This we of “st,” Moore ells “rectly se He thins tht there ar analogous es of hea” “el “taste” Dively seing, directly sneling ete, are varices of “dicly apprehending® He so uses the expression “sense datum that “4 Reply to My Cries” The Phlcaophy of GB. Moor, PA. Sei ‘ek, (Byamtone Neeson, 1042), pr 6465 a “bu p 659 2™ mace vencerio’ from the fact that something is directly apprchendod it follows that it is a sense-datum. Thus an after-image isa sense-datum, ‘Moore also says that “a toothache which you feel is necessary a ‘rense-datumn"# ‘Moore expreses his inclination to think that no physical sar face can be directly apprehended as fllows: “I am inclined to think that itis as Impossible that. . - anything whatever which 4s divecly apprehended, any sense-datum, that's, should exist, ‘unpereeived, es it i that a headache should exist unfet, If this Js $0, t would follow at onee, that no sensedatum can be iden- tical with any physical surface [Moare is confident thatthe sup- position that a physieal surface might exist unperceived is free from contradiction] which is the same thing as to say that no pysieal surface can be directly apprehended tht it isa contra fiction to say that any is."* He offers, so far as T know, no sort of argument or explanation In support of his opposite clination to think that physieal surfaces can be directly appre Dended. Philosophers discussing problems of perception have fequently used the expressions “direct perception,” “direct apprehension,” “immediate awareness,” and thelr cognates. But these expressions Ihave no familie use in ordinary life and those philosophers ust- ally have given either no explanation, or an obscure ane, of thelr intended philosophical use, Moore's directions as to his use of “dirty seeing” (the "visual variety” of “directly apprehending”) are the most lucid that exist in the literature of perception.” To seo an aftersimage with eyes closed is to directly se its and to see ‘anything, x, in that same sense of “see” in which one may see an after-image with eyes closed, is to directly see x" ‘Unfortunately, when wo begin to think about the notion of see- ing an aftorimage, we find that i is fall of plilosophical difi- calties. These may prevent us from correctly tinderstanding the dF. 8. orien’ stony dd, pp. 620631, ‘Moar’ eleronce tg Macbaihe woe af te sentence “Is thie « dagger arhth seo Before ef" a another example of See being el to ea, ory har ey at nt ep fe th afersnage; for ao abou jut how Sct i spose to ‘hve been wsing his sentence and thr so way of Song ot ‘pms vencernioN: % Important conceptions of "socing directly” and “immediately ap- prehending,” that this notion is meant to elucidate. By studying the uso of statements about atersimages, I intend to examine the concept of seeing an after-image, and fo campare it with the eon- cept of sooing a physical reality. I should say that it does not appear to me that the distinction between seeing afteimages with open eyes and soeing them with closed eyes hes any impor tance for this inguiry. My main alm isto eaefy the philosophical cemeept of direct perception. A secondary aim isto provide 8 con- clusive proof that a physical surface cannot be “directly seen,” in Moore's explicit senso ofthis expression. [Let us begin by noting a remark that Moore made a few years ‘before—namely, that “there is an absurdity’ in supposing that any one of the aftersimages which I saw could also have been seen by any’ one ese: In suppesing that two diferent people can ‘ever sce the very same afterimage."" Agreeing that this Is an absurdity, I want to know what sort of absurdity iti, Suppose that two persons, A and B, performed the following experiment: Buch of them gazed steadily ata piece of white paper eut in the shape of a fourpointed star and placed on a black ground, and then tumed his eyes toa shoot of white paper, and beth of them Tooked at the very same white sheet. Suppose that each of them sw a gray patch on the white ground, the gray patch having roughly the Shape of a four-pointed stat. Suppose that each of them traced with « finger the location of his afte-image on the white ground, and each taced the very same path. Should we say that the after-image that A saw is the very same one that B saw; or should we say that A's after-image was exactly similar to 1s and was seen in the sme place, but that A's was numerically diferent from B's? What further experiment could we make ia frder to answer this question? What favestigation shonld we carry out? To whet exterion should we appeal? The answer is that we do not have the faintest idea as to what would be an in vestigation ofthis matter Tt snot the case tht we have a concep tion of an investigation but are prevented by practical obstacles from caning i out. Rather, the word “investigation” has no 11 Prat of Eel Wo” Prosi of he Bt Adomy, XV,6 Dpmecr pencerion ‘usage in this connection. We do not understand the sentence "It ‘urmed out that A and B saw numerseally the same image.” Nor do we understand the sentence “I turned out that A's image was ‘not numerically identical with B's image” Neither sentence, to be size, expresses a contradiction. Nonetheless, nether sentence has ‘any mesning, in the sense that no one has the least notion a to What the circumstances should be in which we should say the ‘one sentence as oppated to the other, One way to state this fact, ‘would be to say that the concepts of numerical identity and m= merical diferenee do not apply to after images. If Ihave painted two circles of the same color, area, and shape ‘on the wall ofa room, and have placed A and B together in the room, with the instruction to deseribe what they soe, then I Standing outside the room and hearing their identical descrip tions, may be in doubt as to whether the area that A deseribes Js exactly similar to but not identical with the area that B de- scribes, or whether it is identical with it This doubt I can quickly resolve by entering the room and observing that A and B do oF {donot look and point atthe very same part ofthe wall. I ean have 1 doubt as to whether the hat that A wears in the moraing is ‘exactly similar to the ove that B wears in the afternoon. T can ‘discover that itis oF is not. Having discovered that itis T ean ‘wonder whether tis the very same hat. Perhaps I eannot find out, because A and B share an apartment to which no other person i ‘over admitted. But there is something that you and I would call nding out"—eg, watching from the closet and observing whether the very same hat that is on 4's head when he returns from his moraing walk is transferred to Bs head. The concept of ‘numerical identity applies to hats and paintod cirles but not to ser-images, ‘Moore's remark that there is an “absurdity” in supposing that two people can see the very same after-image must not be under stood, if taken as true, as implying that they see diferent after- images, but ar implying that questions of identity or difference simply do-not belong to our discourse about after images, Tis a6, tabrurd to suppose thatthe after-image that I soe e not identical ‘withthe one that you sce as to suppose that it A’ afterimage ‘may be diferent from B's in the sense of eing unlike it, But when Duct rencernioss 7 ‘we are speaking of aftar-images, sis unlike y" doesnot entail “x ‘is numerically diferent from y."* If from the fact that A has an afterimage and B also has one we “conclude” that “at least two after-images exist,” this would ‘only be a bizarre way of saying that each of two poople sees aa after-image. To “infer” from the fact that Ihave seen alter-images ‘three times In my life, that “at least three after-images have ex Jsted” would be an eccentric way of announcing that om at least three oceasions someone hes seen an after-image. It would be idle to raise a further question as to whether the aterimage seen at one of those times Was or was not numerically identical with one seen at another time. Consider another question about after-images, Would it be pos sible for A to have made a mistake about the gray star, e.g, £0 have got the number of points wrong, there being actualy five ne stead of four? What would determine tht this was or was not so? Would Bs report that he, B, saw four points, be evidence that A saw four? One could not think so exeept by making the sense- less supposition that A and B saw the very same thing. T say “senseless” advsodly, because itis a supposition that one can nd sore deny than an ‘tis tempting to suppose that A might look more carefully at the gray patch and discover for himself that it had five points in stead of four. But ean he look atthe gray patch? If one can look at something thea one can also Took aay from i, And if I ean, look away from something, that implies that # ean continue to ‘exist when I no longer see it. Would it have meaning to thinle that A's afterimage existed alter he saw it? We may note that the sentences “A hasan afterimage” and “A sees an afterimage” are synonymous in their ordinary usage and that, therefore it ‘would be a contradiction to say that A has an after-image that Ihe does not see. Nor could it be sald thet someone other than A. “(This remark males me wef, One cataily bas 2 srng fe ‘insti 6 deduce numeral diferene fem qualitative dMeroce, Teed ethane ter to ace ta er ngewe lt {hing}. Bort the ease of afertmage, 2 and y ave altatively te ‘me, ti makes no see to sake a frhorquston ae 1a her they ae ‘mimeraly the ssn]n mec vuncernion has or does not have that “very same” after-image of A’s when ‘A docs not see it; for questions of numerical identity cannot be raised with regard to alter-images, "That there is no sense in supposing that my afterimage might ‘exist when T do not see it will soom obvious to many. It is con- nected with a point that will not seem ebvious—namely, that there is no senso in saying that I might look away from my after- {image and, therefore, no sense in saying that T look at it If T ean not look at it then 1 eannot look at it again, nor examine t more carefully. “The reason that one cannot look away from an afterdmage i not that it will follow one’s line of sight. It might oF might not IE upon turing one’s headone found that one no longer saw the after mage, one could allthis “Tooking away” from the after- mage, But it would not mean the same as looking aveay from a minor image. One could raise an intelligible question as to ‘whether « micror image continaed to exist while one's head was ‘tumed-— intelligible inthe sense that one can conceive of evidence for or against? There is no parallel intelligible question concern- tng the after-image, We have no idea what should be considered ts evidence that my afterimage did, or did not, exist after I looked aseay from it “Looking away” from an after-image may feel like looking away from a shadow or a reflection, The difference between the to i onceptual, not a diference in experience. We are dealing with ‘questions that cannot be settled by producing an afterimage and ‘noting what happens. ‘Can one fail to notice certain features of one’s afterimage? It {s an ordinary use of language to say “I didn’ notice” in response to various questions put by another about one’s recent after- Image, e.g, "Was ited inthe center?” But this uso ofthat expres- son i quite unlike its ase when itis said in reply tothe question, “Did the man whom you saw in the garage yesterday went shoes?™ In the later caso the reply, “I didnt notice,” suggests a need for further inquiry. In the former case it ene inguiy. Tf Tsay “T didet notice whether it was od in the center,” that tut of cpane 4 miro nage i refction It wold not be « mizor image fonts fae Sppearad” in ter when wus nt meted Taleo tea wat we shoul ell ‘bmn pencerioN » statement is properly regarded as belonging to my description of my afterimage. Its equivalent to saying that my afterimage ‘was in that respect indefinite. 1 the desceiption of an afterimage is fragmentary it doos not follow that the desripton is incom plete. Descriptions of after-images difer in their logical status de- pending upon thelr degree of contiguity in time with the after: Images they describe. If several days after seeing an afterimage T declare tht 1 did not notice its color a the center, one ean sup pose that T did notice but have forgotten. This supposition has sense because there could have been a previous report of mine that supplied this detil. But ifn a description that is contempo- raneous or nearly contemporaneous with the afte-4mage T declare that did not notice some detail, then iis absurd to think that T have forgotten. That description is not “more likely” to be ac- curate than the description that I give from memory @ month later. Ibis not “better evidence.” Its your criterion for determin ing the character of my afterimage. "Thus there is such a thing ar not noticing a feature of one's after-imago; but itis not faling to notice. If it happens that while Thave my afterimage I report that “I eannot tell” whether i i red in the conte, or immediately after say that “I did not notice” ‘whether i was, then it would be foolish for you to argue that the center of it either was red or was not red, and to conclude that I hha failed to notice which t was, What would be your authority for saying that either it was or was not red in the center? Are you. appealing to an a priori principle? But what reason is there to think that whatever isan a prior principle of our discourse about physical surfaces is also an a priori principle of our discourse bout afterimages?“L did not notice whether it was red in the center” has the same finality, in an afterimage report, as “It was shaped lke a star.” fT do not knovr what its color inthe center ‘was this “not knowing” fs not ignorance ‘You find ont about my afterimage by hearing that report of ‘mine that is contemporaneous or nearly contemporaneous with the afterimage. Ido not find out about my present or just past after-image, inthis way or in any other. could ind out about ita year later, by consulting a record of my original report. Bat the ‘ecuracy ofthat original report cannot be questioned by either of0 smwer vencerrion us, It is tempting to think that we have the concept of is being in error, but that you cannot determine whether it i, and only 1 can, This is entirely wrong. Suppose that while I have an after- image [report that itis starshaped, Then, while I stil have i, ‘ean I observe it and compare my report with it? How ean T ob: serve it? “Observing” it could not even imply looking ati, as I have previously pointed out. Nomally, by “observing” and also by “comparing,” we mean something that can be done accurately ‘of inacourately. IF I declare "Yes, i is star-shaped,” what would it mean to say that T have accurately observed it? Tan ty to observe more accurately the shape of « shadow by changing my positon so as to obtain a better view of i. What would it mean to {get a better eew of my aftersmage? Nothing at all. I cannot even ty to “observe” an after-image “more accurately.” Nor would my ssoond report “Yes, it is starshaped” confirm my fst report that ‘it was star-shaped. For if it would, then the report “No, it isn't weshapes!” would disconfirm the fst report. But tits not how Wwe should interpret such a sequence of reports. Normally, the latter sequence would be understood as meaning that previously [saw a starshaped afterimage but no longer do. If my second report was "No, it ut star-shaped and my aftermage has not changed.” we should not take even this as refuting the frst report Instead, we should rogard that whole sequence of discourse as tunimtellgible, In this sense the subsequent report can disredit ‘the previous report. ut it doesnot show it to be inaccurate, ean see a star-shaped afterdmage and a star-shaped shadow. In either case I can count the number of pots. Is it the same sense of “count”? Suppose that, with eyes open, T see an after- {mage on surface of white paper. Suppose thet I determine tht, it has seven points by counting in succession from one to seven, pointing my fingers at a diferent point of the afterimage each {imo Futter a numeral. Might not I count the points ofa shadow inthe same way? It is truc that my movements and utterances would be the same In both cases; and there might he no diferenes in my experience —but the logic of the two cass is very diferent. It is sensible for me, and for you, to wonder whether, unwittingly, Thad placed my finger twice on one point ofthe shadow. There might be various proofs that Thad done so and that, therefore, my count was a meer rERcEPRON 8 riscount, Could there ever be a reason to think that {had mis- counted the points of my afterimage? Tt appears so. IT were to exclaim “Teas six points, not seven; I miscounted,” then would not you have reason to say that I had miscounted-—the reason, being my own assertion? do not doubt that it would be a natural occurrence for a per- son reporting on his after-image to eorrect himself in this way, ror do I doubt that anyone else would naturally conclude that the former had miscounted. 1 do not doubt, therefor, that there is a natural uso of “miscount” in this connection, But what te that use? What is the status of my assertion that T miscounted? Could Ibe mistaken about that; isi perhaps the ease that seven is the right count after all? The question tras off into emptiness. ‘We are trying to lay upon this notion of “miscounting” « burden ‘that it cannot support. My assertion that I miscounted has for you ‘the same sort of unquestionable validity as does my assertion that the after-image ¢ star-shaped or thatthe pain in my shoulders, dull pain, All ofthese assertions might be called “self-confiem- ng.” implying by this that really they have no confirmation TET report that the number of pages in a manuscript fy seven, T may say later, “Tmiscounted; six is the correet number.” You can discover that, contrary to my belief, my former count was not a riscount. You cannot dscover, “contrary to my belie” that my first count of the points of my afterimage was not a mlscount, My assertion that it was a miscount is aot responsible to any Investigation, not even my own, For shall T count again? Does not “counting again” pose a difeulty similar to that of “looking again”? Counting involves shifting one's attention successively along the series of items counted. One might ask: “Is it possible that the constitution of an afterdmage is fected by alterations in the degree of one's attention to its parts?” The question is sly, in the sense thet it points nowhere. We do not know what to look for to answer it. “Counting again” does not havea clear sense with after-images. This shows that the notion of miseounting that applies to after.images is only a shadow ofthe notion of miscount- {ng that applies to physical realities, I lacks that body of implica tions that belongs to the later. ‘There could be other reasons for thinking that a man's deetara- ‘ton that his aftedmage is seven-pointedrevuts from a miscount,Co mace vencerno Suppose that he commonly miscounts physical realitios such as ‘eggs and onions. This might suggest the possibilty that his count ing with respect to the after-images is in error. Or suppose that many others have been subjected to the sume physical stimula ‘tom and that every one of thom has reported sooing a six-pointed tfterimage. This might suggest that itis probable that he mis- ‘counted, But if he sinoerely reassures us that seven isthe correct ‘number, that decisively sefutes our conjectures, These “possbili- ties” and “probabilities” collapse before his confident reasserton that he counted correctly. This brings oxt the respect in which the latter ascrtion is “self-confirming” It shows also how flimsy isthe possibility” that he miscounted, There is a possibilty that he made a mistake only insofar as there is a possibility that he will any that he made a mistake, The possbily that he is mistaken bout his after-image is only a weak imitation of that fllledged sense in which itis possible that he made a mistake in counting the forks and spoons, For the latter has lear meaning inspite of huis sincere denial that he miscounted. With regard to anyone's description of his own after-image, there is no clear difference in meaning between his assertion that he made a mistake and his ‘assertion that he expressed himself incorrectly ‘A report on ones ov after-image is an odd sort of discourse, in ‘which each statement is unimpeachable, unless amended or ean- ‘celled by a subsequent statement in that report And there can be ‘only a minimum of correcting, beyond which we should hesitate ‘tw call ita genuine roport on an afterimage. Suppore that a man traces upon the wall the outline of an aftr- ge that be sees agains it, and traces only six points although Ihe says that he aves seven. Do we have evidence that he mis- ‘counted? If be sees seven points then he isnot tracing correctly tnd if he is tracing correctly thon he does not see seven points. He says bath that his afterimage has seven potnts and that he has traced it aceurately. Which statement is true? We can shrug our shoulders at thie confit in his deseription—and nothing more! “Any preference for one statement over the other would be entirely arbitrary. If someone's deseription of a physical reality Is self- contradictory, it is sensible to speak of trying to find out whether this one or that one of the contradicting statements i trae and this makes i meaningful to assert “Either this one or that one DMECT PERCEPTION 83 imust be true” When the contactor stataments belong, to an afterimage description the latter assertion is dewid of meaning forthatdeserptin i our etrion of what the ftrinage ike and consulting it exhasts vr Inu. It self omtudictry then defeat sin away tha sel contaditory description of physical realty docs not We are prevented fom drawing ony conelision {for example thatthe man misounted) expt the enchasion that we donot Know what to ake ofthis piece of di: ours that started ont as «description ofan triage. "wil mark the diference between afteriagedevrptions and reports and descriptions and rept of phys rales, by sy dng tnt the former are incrrgla® and tho ter are no 1 watt ware, however, agai some misundertandings tat may be caused by the ose of thi technical term, When say tht the former are “incorrigible” I do ot imply, ofcourse, at man cannot correct his own afternage description by telling us that Imispoke (og tht he sds reddish cc bein the center” wien he should have saa Dish ctl red inthe cone) or by amending or widening a previous statment in is dst tin, Nor do imply that snyting whatover tht i offered aa description ofan atesmage should be accepted without que tion, If were selecontraictory wo shold mot regard ty 8 perfectly good, altongh undsual, description of an aftrsmage If-we found tht someone conta missed certin color ade tive im his description of plysial reat, then we should hot Accept a fac vale his atersitnage descriptions conning those jective there was eny language tall whose ae in elation to physi reales he hed wot mastered, then we soul disallow Hiroe oft nan alleged aftersmage report Furtbemere there a sense n whi «person canbe mistaken even in reporing that he sees sn afternage, An afterimage 1) LAs xities“ncnliy” a eS ond Sensis (New Yost Ono Univer Prey Ie 1002), pp. si&. He tachog 2 pene of wow ome ld by A.J Ay ad eat eat dat Iie hw uth any af what hehe ger ng hepsi fake hee i en oy te ke hh Ie inepable once uted, of being mabecqunty amend retraced” {kd ps Lid) snag so, bot that i fot howe expo te noon tocol, (A dss ceteed om ae natin keer bn AV [ontes PhzvophialImctigetons ae pp. 1i0"tad 17)ot mcr eacernos (called an “after senso” by some pychologlst) "th impre- Slonot vv scnton rtloed ater tho eternal ase ith {ravat (Oxford Engloh Dion). aman In oom lok at the window though which the ight pours my be hat when fe dts gant th da il of cles hs ee, wl ce tn image of te wind wi ls bight panes and darker sabes {orth right an dark may be reve) He wl en be sxing faltering the conte to look at the window and to sce ites dant as befor, and if be as oot previously Ikeda an ven bightr area at thre nite sor scoma tim to be ty decease in the intensity of the window gt, then he cat Tes an afterinnge became tere has bee no esston ot ‘lininten of sensations. One cannot ace an afterinage In ost tou crcumstagoc, ts being mater of defision an nt an Crperinental face Aman sight not koow that cota spectl ‘Dhara must be fllledbeore Is croct o sy that ses Staller tage: Or be might know this ut thnk that they were Feed when they were nt ln ethr ease bi set that he fo an aferinage could bes oe aero fe” It fT pewon, Ar placed In crcameancos so markedly Inappropriate feb srg an aftersinage tnt would he mise of guage to oy Aon an aer-inge” were to sy “Tse an aera Shon would be ite comet oy that Aniston tether A a sn repel a te of the expeson “feria” a0 provouly ose cone and if be were pace in czeumstano such at srould nt bea mus of language to sy “A sro an aernage Thanks seca alterna, oul bo ete senslas foros to mppos that A vas mistan. Ths estore of aerdnage Stemens Surply ferent ten fom statements soot purprtd physal sealife. 1A bad long ago lurve to wo Cnty the words ot sce aces” and had never show sy tendency to misuse them, and if his ceatances were such that I would pot be a ese flange ty "A sos Suen” then UPA reported" se lanes” would stil be sensble for us a sippor, Unt Als mide, We coal spon, for example, hat STEN appear to be fumes on the orzo wil turnout toe owing colin th sky ‘ Thos in saying tat fringe reports ae “acomigble I do pmwcr vencerion 85 not deny that there fea sense in which someone can be mistaken ‘when he asserts that he seos an alterimage. He oan properly be said to be mistaken if would be a misuse of language to say that he secs an afterimage, But there is also sense in which & person's report that he sees an after-image cannot be mistaken; nd itis this sense that T intend when T say that his report is incorrigible If he has correctly used the words “T see an after- mage” on numerous aecasions and has never misused them, and if his circumstances are such that it would not be a misuse of language to say of him that he sees an after-image (e.g, he has just looked steadily at bright light and then closed his eyes) then there cannot bea question of his being i error when be says “Tce an after-image." There cannot be a question of whether he “takes” something to be an afterimage that is really not one. ‘There can be a question as to whether what he says is true; but it {is identical with the question of whether he is bbing, Not only is ther a sense in which a person cannot be mistaken {in reporting that he sees en afterimage; there is also a sense in which he cannot be mistaken as to the color, shape, or other sensible characteristics of his after-image. To be sure, if n his di course about physical realities he constantly misused the word “blue,” this would be a sufficient condition for saying that he Aeseribes his afterimage inooreectly when he says to “Tt i blue.” Furthermore, itis @ sufficient condition for saying that 4 part of his description is mistaken thet, while he still ces the afterimage, he honestly says that i s mistaken; but here there is tno clear distinction in meaning between “making a mistake” and “expressing oneself incorrectly” If, however, his use ofthe word. “blue” in his talk about physica realities Is perfectly normal, end 'f he has not only said bit earnestly reamed that his after- Jimage is blue, then the supposition that he might be mistaken as to the color of it is entirely without meaning, This difersre- markably from a ease in which he is ealled upon to deseribe the color and shape of a part of the surface of « wall. Perhaps be is too far from itor the light is poor or itis partly concealed by & curtain, It is plain enough what steps would be in order for removing those possible causes of errr. In contrast, think how Tadierous it would be to suggest that he ought to come closer to hhisafter-image or to look at it in a better light in order to improve86 mec vincerrioN his view of it orto suggest that perhaps part of itis concealed ‘rom hin. 1 will formulate the difference between after-image descriptions and descriptions of physical realities by saying that the former, like the later, can embody errors of use and of ex pression, bot that, walike the latter, they cannot embody errors of perception. u Let ns bring the procoding observations to bear on the question of whether physic realities can be “directly perceived” or “di- rectly apprehended,” in the sense that Moore explicitly gave to those expressions. Tt is easily seen that they eannot be. Tt Js quite impossible, for example, that anyone should “direcly sec” a physical surface. To dirctiy see’ something is to see {tin “What sense of see" in which one may see an afterimage. he later use of “ses” cannot be set forth except by deserib- {ng the logical features of those statements In which it occurs, ise, reports and descriptions of afterimages, The two statements “The ploce of cloth that T see it round, bluish and sed in the center” and “The after-image that I see is round, bluish and red in the center” differ in 4 most important feature end can, with propriety, be said to ilistrate different senses of “se.” There Js.a kind of error—namely, an arr of pereeption—in regard to ‘which the suggestion is meaningful that perhaps Tam In zor In ‘thinking that Tzeally see a cloth ofthat deseeiption; and in regard to which the suggestion is not meaningful that perhaps I am in ‘error in “tinklng’ that I soe an afterimage of that description, "Therefore it sa consequence of Moore's explanation of his own, use of “directly apprehend” that physical realities cannot be directly apprehended. My argument to prove tis isa justiieation et ted oy hi sd may De mae pn St eer soba i ‘Ths idence that “wee” is used in ee ese ae lsu the ee ene eae Sees FTEs cae Stop dc al diene Cie oer et he a ie Se See area a ee aah Se eee Pe a ere ee Sees Dinecr rmncernion s ‘of Moore's inclination to think that they cannot be, Indeed, it shows that “inclination” is too modest a word. "But why does Moore have an inclination to think that physical realities can be divctly apprehended? What isthe nature ofthe ‘confusion from which this contrary inclination springs? I have an hypothesis to offer; bt I wish to emphasize that itis extremely tentaice Tet us fist note an important feature ofthe use ofthe expres- sions “direct apprehension,” “direct perception,” “immediate awareness,” and ther eognates, that has prevailed among philor ‘ophers. The latter have commonly thought of disect perception as something that is not open to the posit of error. Berkeley, discussing errors of perception (eg, mistaking a equate tower, con ata distance to be round), remarked of the perton in eror “His mistake les not in what he perceives immediately and at present (it being @ manifest contradiction to suppose that he Should ere i respect of that), but im the wrong judgment he makes concerning the ideas he apprehiends to be connected with those immediately perceived, . 2" Here Berkeley says that i {sa contradiction to suppose that a man could be mistaken about what he perceives immediatly; and this remark may be taken as stipulating « nocesary condition for the use of the expressions “immediate perception” or “direct perception” We should further note that Moore says that "diectly see" is “the visual variety of what Berkeley called ‘direct perception’ Thus Moore thinks that his use of “direet perception” conforms to Berkeley's, ‘Temay appear that notions other than the impossibilty of error have been included in the standard philosophical conception of direct perception. For example, it has been implid that if one directly perceives something one cannot doubs that one perceives it2"T do not believe, however, that this i relly a diferent idea. tis unlikely that i is intended as « psychologieal observation, ‘More probably t expresses the thought that, i a common use of “doubt,” 2 person cannot have « doubt that he perceives some: thing wnless it makes sense to suppose that he is mistaken in * Third Dlg een Hye and Penne. Reply, po 029. iad x Ao, opin sd La Mh Ta, wa88 puwcr reacernios thinking that he perceives it, and that whenever this does not make sense nether does & “doubt” Tis said also that perception is immediate only ft contains no {inference What shall be the criterion for determining whether a given perception “contains an element of inference"? An appeal to the ordinary use of “inference” will not be very helpful, be- ‘cause the philosophical use is different. The fllowing quotation {rom J. S. Mill fs an example of the latter: 1 aff, for example, that Chea «muri voles, This would passim common language, or a dect perception All, however, which realy poten thst Tear sud That the sound x a vole, st tat vole tho voice ofa man, are ot perceptions but inferences Fatima my toe oa har i orig If any propockion concerning & matter of fact would commont ‘818 to known bythe dct testimony of the nse th sly ‘would beso The tath, however ifr oterwie Tory i cra louredsurace or rather Ifa the Kind of val senations which Sr urally produced by «colored surface; and from these ss marks, Eon to bes iy fie xp | cnet fy other, T might have fied. sensations presieely similar when 2 The warn there might fave gon Some oer pan na resembling him in appeurance ag atthe distance and withthe degree tf ston wich bestowed to be mistaken for him. aight have San anep, and dreamed that T saw him or fo a nervous ste of border, wich bnoogit his image before me n'a waking halhcia tion Tn al these mares, many ave boen Tet belive tat they Sw persons ‘well known to them, who were deal or far dant. If Sty Af these suppostions had een try the aflmation that T a0 Iny Brother would have been emoneats but whatever was mater of {Ret perception, namely the visual sematons, woul have been real The tfsrence only would have Been Ml grounded; T shold ave Aserbed those sensation to & wrong eases In Mils philosophical terminology, « perception involved “infer cence” and, therefore, was not “direct,” if it "might have been Some philosophers who meke litle or no use of the phase “arect perception” express the same idea by speaking of “the siven” in perception, Commonly they contrast that which is 2g th th oes pie mtn wih hy ot poe sgl: ey oak ern” (eee, Be Dg) 25 Sytem of Logie, Hk WV, ch 1s 8 (Sow Yorks Longmans 00, The, 1888). Dmecr vuncernos 0 “given” with that which isthe result of “interpretation” For ample, CI Lewis says “Perocptnal knowledge has two aspects oF phrases; the givenness of something given, and the interpretation ‘whieh, in the light of past experienee, we pit upon it” ” What is the criterion for distinguishing these two aspects? The following Subtract from a pereeptual experience “all that conceivably could be mistaken; the remainder is the given content of the experience..." thin, ia short, that “impossibility of error” is the main feature of the standard philosophical conception of direct perception. I take the liberty of making this explit by constructing a definition of “direct perception” in torms of it, as follows: “A directly per- ceives x ifand only if A's aserton that he perceives x could not ‘be mistaken; and A directly perceives that x has the property F, if and only if A's assertion that he perocives that x is F could not bbe mistaken,” T wil later point out a respect in which the defini tion is ambiguous, ‘The next step in putting forward my hypothesis about Moore's ‘contrary inclination will be to vemind you of Moore's disposition to assert, in the course ofa philosophical argument, that not only ‘does he perceive some physical object (eg, a hand ora tree) but, furthermore, that he could not be mistaken about it. Consider the following remarks: 1am aot, therefor, sid to sy tat I do now perceive Et tat 4s a doo, sod that that is = Soger some philosophers seem to zo 1 have denied ht we ever So i tk mc hinge a then, fea ehers no ony hat we ever ow tht bt lp nt they are ever tru Ad ii fac, I ever do know wih «thing of is Tee le lf tna hat Te pee ah ver unis I bath Isaw and i te, BOC eens tome su {ifn refutation of such views aa these, simply to ol ot eases In weich ove do know such tage.‘ ters you know realy linger thee 16 no doubt abot eT inow 2d you ll nw #0 ‘And recall his assertion, at one point in his British Academy "An Anais of Knslage ond Valuation rvniing Che) p38 ind pa * oot Plnpha Stir (New York Harcourt, race & Worl, Ie, waa)y pp bara Sill, Is Open Courtea pmcr rencernon lecture, that he knew, a few moments before, that he perceived ‘wo ands: ow absurd i would be to suggest that I didnot know i, but on believed It and tint perbaps fers not the eae! You might at well ‘suggest tht T do not Snow that Tam now standing up ane talking — {het pesaps afer all Tn not, and that is not que certain that T an “Moore thinks that sometimes, but not always, whea he says “I 00 doot” he knows that there isa door there. believe that he ‘ses “know” this connection, in what Thave called it “strong” sense?" I think that what he is tying to expres whea he sys in Tis emphatic way, sucha thing as I now that that i a door,” is that nothing would be eoidence that there is no doo ther, that it could not furn out that there was none, that he could not be ‘shown to be in error. And these remarks would not reveal an tide of prejudice or dogmatism. They would not be a denial that enything whatever could happen Instead, they would reveal an important feature of his (and our) ordinary use ofthe expres- sions “evidence,” “turning out to be fale,” “shown to be tn eror” namely, that inthe particular casein question, there is nothing ‘hat Moore would callevideneo” that ther Is no door befare him; ‘hat he does not understand what #t would be like fr t to “tara ‘out that there is no door, nor what discovery could “show” When he says “there is no doubt about it” he is not making the psychological assertion that he and others fee! no doubt, but the logical assertion that for him (and others similacy placed) doubt that there is 2 door there would be a senseless thing, because it would have no reference to any inquiry. Whenever Moote is placed in circumstances that are the best posible for ‘examining the soe of object that a door fs, then he cannot be in doubt as to whether ts door, in that sense of “being in doubt” that implies that he who isin doubt has some conception of what {slacking and what to look for. ‘Thus Moore is disposed to hold that sometimes when he and others perceive a door or a toe, they Know that there is @ door for a tree before them. The sense in which they know this might "Poot of an Eten Woks" Proceedings o he Brith Academy, XXV, * Seep 8, ‘pmecr rrnceemion a ‘quite neturally he expressed by saying thet they “could not be tistaken” about it. What this ambiguous phrase would mean i this use of iis not that they have strong or even conclusive evl- dence, but rathor that nothing could show or tend to show that they were mistaken Tn accordance with the definition of “direct perception” that 1 proposed, we may say that Moore's view is that sometimes he and. others directly perecive physical objects and physical surfaces, since he holds that on some occasions he and others cannot be mistaken when they assert that they perceive physical surfaces. (Of course, Ihave not presented, nor could present, anything like conelusice evidence to show that Moores inclination has this ongin, m Leaving aside the question of whether the preceding considera- tions do account, in part at leas, for Moore’ inclination to think ‘hat sometimes physical surfaces are directly perceived, I want to task whether itis blunder to old that sometimes when # person asserts that he percolves some physical reality, itis impossible that he should be mistaken, Clearly it would be a blunder to bold that the later is so in the cory same sense in which it i impossible that he should be mistaken when he assrts that he sees an after-image. The respect in which afterimage reports feannot be in ertor (their “inooregibilty”) was elucidated by tween them and reports about physical that sometimes the latter are incorrigible ‘would be to destroy the whole distinction 2 =n ey ae “pte He hat a pg St ea ee Saute Li Dees ata te ays SEIS hagas Se eee es SE ESE i ete ar gy So Shem cra ne dg ici a ne nie ge fe alate Eee scraeccel cv ev a ls SE SY nda tapes boa SE eee Pune a Pd Sant Pe miracle SRL eho anchsaurad a dion92 Dpmncr vrncernos Isit tenable poston to costed that report of py eal tua, alough eagle sometines cannot be in etr? Tis i we tan mae cut a senge of “cannot bein ero” and “ould nt be titan that ater than te sense of hve ples in which they apply to fringe statement, yet a sea tn which they Sonstiges apply to statements sboutpyial realities, Inde tally fe do hyo shall have revealed an smbigity ny Afton of “rect perception” and shal hiv, as aro, poe idl to senses for is expression ec cer nc eet of pista object statement) tying tose, St, the respok wildy tcanot be in nor and ead the apt a whith ti, nevertices, not incwgible, Suppo that you and are seard Ing fora msing doer pate 1 ink that Lape it undr the bed. Tats pretty dark unde ore 1 an precy well understand the suggestion tat perhaps what Le sot plat, ut oly a ‘hil or perhaps rug Then 1 fetch tout ad mak sre tat itise plate Ifyou inqued rr ator room, wheter Isc 4 dlnnerplat | Could spy that Ida, Now Hf you came up. acd Saving al cose iow off ld “Patbape tt x pla but ny &shadov, or shaybe aug” then I should ot understond yovr remark T do understand te normal se of your sentence iy fating to uodstand in thi ete ot ike my fling to nde the sentnce"Doakeyehiberoste smth whi ha to normal use Indeed, te coneguone of my falar wih the normal use of your extence that {shouldbe Bevidred by Your soying it in th ease, Your omar wold srk me 3 fool ii"expesion of doubt Best, nthe svaton, nether of us Ins any conception of what woul bea Deter proaf tha isa plate Wo dd aves moment lore, wen twas under the bed Then, 1 unesood the posit ht T might be mistaken; now, “erage ort a te mig nv tr fecha teats rearreaeec Lo Sy aon nl etn de aes cle Feep eater “arigbie® 1 ewe inh Pla bee Sy ene Ae ie lt cen eae Be dtaants dy fs petite chee ts ki Sekirei tales i tt ha ee SECS el ene ct EMS beled ate “eal mg tt {5 coe hy the edary epesin “co mt be saben) mec rencernox 3 do not. It will be thought that at least T must conceive the possibilty that I have an hallucination, But this isnot so. The nation might be such that at that time I eould conceive of no future happening that would show, or tend to show, that I was having an hallucination, Does this mean thet the statement “I see « plate” that I might utter a that first moment when I saw beyond question that what held was a plate, would he incorrigible in those circum- stances? ™ Not if we use the term “incorrigible,” as T do, to in clude the whole logic of afterimage statements. For it is not Aiffeult to see several pots of difference, I will mention two. Fin of all, any person who does not see the plate in my hand ‘can understand the supposition that peshape my statement that 1 see a plate is in error: for example, he can understand that he right look at my hand and see nothing in it at all, i, he can understand the supposition (although 1 cannot) that T may be having an hallucination. In contrast: if Tanmounce that I see an after-image, neither T nor anyone else attaches meaning to the suggestion that I may be making an error, other than sn error of use oF of expression, In th second place, to any other person who does not have a good view of the plate and who voices « doubt that I really see one, Teould say, "Twill prove it to you; came here tnd look at it yoursell.” A similar invitation to one who doubts that [see an afterimage would be without meaning, tis noteworthy that it is only in a qualified sense tha the con- cept of a possible exror of perception fils to apply to the percep tual satement, ie, only the person who utors the statement (and any other person who has an equally privileged view of the ob- ject) cannot conceive that he is making an error of perception; ‘nd all others can. Whereas it isin an unqualified sense that the concept of a possible error of perception fails to apply to the afterimage statement, 2, no one at all ean conceive that the person wh uters itis making an error of perception Ww Let us return to our stating point—namely, Moore's opposite ‘inclinations. So long as we adhere strictly «9 Moores explicit [Now Austin's remahs quoted i Ue immediatly preceding fstote)94 mac pencerio directions as to his intendod use ofthe expression “direct percep Lion” (actually of its syonym “dltect apprehension”) then must bbe admitted that its an error to think that physical realities can ever he directly perceived. But we remarked that inthe history of theorizing about perception an important condition forthe se of the phrase “direct perception” has boca the idea that one could not he mistaken about what he dieectly perceives, This phrase, “could not be mistaken,” i natural and apt phrase for surnmariz- ing the peculiar loge of afterimage statements, Is also a natural snd apt phrase for summarizing th partly similar and also widely ‘different logie of peroeptual statements attered in circumstances that are the best possible for pereciving the physical realities claimed to be perceived. If the illuence of this natural phrase tended to control one’s use of the philosophical expression “direct perception," one would be led to confuse two, resembling but distinct, concepts of direct perception. One woul be disposed to hold that both afterimages and physical realities can be diretly perceived; and might parly see and partly fal to se that difer- ‘ent ideas of “direct perception” are involved. It is my surmise that ‘this isthe explanation of Moore's being tom between opposing {inclinations ‘There is one difculty with this explanation which I must men- ‘on briefly. If it were correct one would expect Moore to be in- clined to hold that bodies are sometimes directly seen, and not solely surfoces of bodies, But Moore maintains that bodies (at Teast, opaque ones) cannot be dict seen. “You certainly do not see your hand in that same sense in which you se that part of ts ‘surface which is turned toward you; and it only the later sense of 'sod—that in which you are seeing a particular part ofits sur- ace, and not seeing other parts—which can possibly be identical with tht sense of ‘seein which you see an afte-image with closed. ‘ees."= In his short paper "Visual Sense-Data,"#* Moore says that in the case ofa transparent body, “ike a drop of clear wate,” it seems possble that one ean “see the whole abject at once” but [he rcnainder of this sey & nowy writen] site ey a us i ih Pow nthe bce (New Yorks 1 Mcrlan Company, 1057), pp. 206-1 i p38. ‘pmoct rencerriow: 95 that when you see an opaque body you do not see “all parts ofits surface no ts inside." It is evident that, for Moore, « necessary ‘condition of seeing something directly is that you should see every Ditof ist oncel Is worth noting thet this condition i met when ‘you see an afterdmage. IF we added this condition to the condi- ‘om expressed in my definition of the standard philosophical con- ception of direct perception,® we should obtain the following definition of “directly seo": “A directly sees x, if and only if A ‘could not be mistaken in asserting that he sees and there is not ‘abit of + which he docs not se." The second condition could not De satisted in the ease of an opaque body. There could be circum stances in which both conditions were sated in respect to part ‘of the surface of a body. In such a ease Moore would be inclined fo say tht that part of the surface was deectly seen. Why should {8 be merely an inclination, and why should he have a contrary inclination? My suggestion i that he ft, without being distinctly aware of, the ambiguity of the first condition. The language of this condition ean express either the notion of “incoeigibility” which dacs not apply to the seeing of physical surlacss, ar the notion of “knowing in the strong sense” that one sees something, ‘which does apply to the seeing of physical surfaces. Whether oF not T have soceeded in diagnosing Moore's hestations, this tmmbiguity seems to me ta be an important souree of confusion for the philosophy of pereeption®® I wor noting tht Moar, hs “Visual Sense-Data” op. et. 210.11), publabed four yours tier the preset say, Bally cy 0 clus which 1 hold toe comes Hat ona cannot "dived" So 3 to esac o's bodyWittgenstein’s Philosophical Investigations" Ein Buch ist ein Spiegel; wenn ein Affe hinein- {uckt, so hana freilich kein Apostel herausschen. Ac attempt to summsite the Inestigetions would be setter succenfl torn.” Wiagensatncompresed his tights to the pont wher farther eomprosion i impose ‘What ts noadad stat ny be led sed the connections tween them traced ot Aly ist rection to the book wil be to rgard H asa poling collacion of relocton that ate scones dividaly bilan, ut potest no ity present 0 ‘nom of dean tach tho nity i thar, bat ext be per ceived without strenuous exertion. Within the scope of evew {he connetedase can et be brought ou thi by concentra ing on some single tople—in spite ofthe fact tat there are no separate top, foreach ofthe inventions i te bok ee. Cronos apa abd sgn with every other on the following T center my stention on Witenes reatment ofthe problems Af how language i elated to loner experencer—toscnains, "Ludwig Witgentsn, Philosophical Instigate, Corman and Engh facing pages Me by G. EM Atcnbe (Now Yorks The Maca ages Te by ‘Company, 189). , ‘wrrromnsrEin’s puuLosontucst, wwvasricaTioNs OT feelings, and moods. This Is one ofthe main inguities ofthe book and perhaps the most diffcult to understand, 1 am sulleseatly aware of the fact that my presentation of this subjeet will eer tainly fil to portray the subtlety, elegance, and force of Wittzen- stoin's thinking and will probably, in addition, contain positive mistakes. References to Part I will be by paragraph numbers, eg. (207), and to Part II by page numbers, eg. (p. 207). Quotations will be placed within double quotation marks, riste language. Let ws see something of how Wittgenstein at tacks what he eas" ten of privat language.” By 4a Jngoage is meant one that not merely is not but cannot be under stone by anyone other than the speaker. The reson fortis that the words of dis language are sspposed to "refer to wiat ean only tbe Known to the person speaking, to his immediate private sensi tion (248). What i supposed is that T“assocate words with Sensations and use these names in descriptions” (230). 1 x my attention on a sensation abd establish « conacton between a ‘word and the sensation (258). is worth mentioning tha the conception that itis pose and even necesary for one to have m Private language 1 not centre Rather isthe view that comes most naturally to any tne who philosophies onthe subject of the relation of words to txperences. The iden of private language Ys presupposed by very progam of tnfering or constructing the ‘extemal world and oor minds’ I is contained inthe plilosophy of Descartes nd in dhe theory of Klas of easel Beith empitcsm, ae well 4 in recent and contemporary phenomenal and sense-dtum hoary. At hotom iis the dea that there i ony a contingent and not an esenal connection between a steation and its outward cxprssion—an ide that sppeal fous all Such thoughts as eso are tpical expressions of the Kea ofa private Innguage: that T Jznow only from my on case ora the word pa’ means (298, 205); tht Tcan only Belive that someone ese iin pain, bt { Joe it if Cam (300); hat another person eannot have my pains (25); that Tan ondertake to al hs (pointing inward) pai fn the futare (269); that when Tsay ‘am in pa’ Tam at any rate justified before must (230).95 warrcesren’s rimosonmucaL nvEsTIcATIONS In order to appreciate the depth and power of Wittgenstein's asseult upon this idea you must partly be its captive. You must feel the strong grip of It. The passionate intensity of Wittgen= stei's treatment ofits due tothe fact that he lets this idea take ‘possesion of im, drawing out of himself the thoughts and im- ‘gery by which it expressed and defended—and then subject- ‘ng those thoughts and pictures to fercest scrutiny. What is wat- ‘wen down represents both 2 logical investigation and a great phi- Tsophar’ struggle with his own thoughts, The logical investiga- tion will be understood only by those who duplicate the struggle in themselves, (One consequence to be drawn from the view that I know on from my own ease whet, say, tckling’ means i that "Tknovr only what T call that, not what anyone else does” (347). T have not learned what ‘ticking’ means, I have only called something by that name. Perhaps others use the name diferently. This is a regrettable diffculty; but, one may think, the woed will still work for meas name, provided that I apply consistently to a certain ‘sensation, Bot how about ‘Sensation? Don't I know only from my ‘own case what that word means? Perhaps what [calla “sensation” ‘others call by another name? It wall not hep, says Wittgenstein, to say that although it may be that what I have isnot what others calla "sensation," at least I have something. For don't I know only from my own case what “having something” is? Perhaps my use of thase words is contrary to common use. In trying to explain how T gave ‘tickling’ its meaning, T discover that I do not have the right to use any of the relevant words of our common Ian- ‘uage. “So in the end when one is doing philosophy one gets to the point where one would Ike just to emit an inarticulate sound” (281). Let us suppose that I did fix my attention on a pain as I pro- ‘nounced the word ‘pain’ to myself. T think that thereby Testabe lished a connection between the word and the sensation, But I «didnot establish a connection if subsequently I applied that word to sensations other than pain or to things other than sensations, eg, emotions. My private definition was a success only i it led me to use the word correctly in the future. In the present case, ‘correctly’ would mean ‘consistntly with my own definition’ for the question of whether my-use agrees with that of others has ‘WrrroexsrEn’s PHILOSOPHICAL, INVISTIOATIONS 0) ‘bon given up as bud jo. Now how i to be decided whether have wed the word consstenty? What wil be the dflrence ‘toe iy having used consistently ands seeming to me that have? Or bas this distinction vanished? “Whatever going fo scm right te is right. And tat only means that bere se can't talk abot ight” (238) fhe distinction between correct and ‘Soom coset has dnppeared, thon so has the concept comet Wfolows thatthe ler of my private language are only inpres sons of rules (250. My impression that I follow a rule docs not conti that fallow the rule, ules there canbe something that ‘el prove my impression comet. And the something camet be {nother impression for this wuld “a if smeone were to buy fever copies ofthe morning paper to asrre himself that what It sald was trac” (265). "The proof that T am following a rale ost. appeal to someting Sndependent of my impresion that fm. If in the nature ofthe ease here canot be sch an appeal, ‘hon my private language doesnot have rls, fr the concept of tru rete that there be a dference between es allowing 4 Tule’ and He is under the inpreson that he is following role'—jst a the concept of understanding a word veges that there a dference between He understands ths wort and ‘He think hat be understand ths wort (ef 259). “Even i T cannot prove and cannot know that Cam comecthy folowing the rls of my private language might be said til ‘may be thet Tam. eas maning sy that The supp tiom makes sense: you and T andertond i Wittgenstein es reply to this (34553). We ae inclined to think that we know wat it means to sy Te Bie lock om the sun or "This eon: etal deat-mte alls to himself inwardly ina ysa! language fF The stove is in pun’ These sentenes produce pcre in oar ‘minds, and fe coms to ws tat te pictures ell ws how to appli them--tat tell wat we have to lok for, wat we hve fo don order to determine wheter what is pctred isthe cse. But we make a mistake in thinking thatthe picture cone in Stuff the instructions as to how we a to apply it Think ofthe picture of blindness «dae inthe solo inthe head of the Bind man (424). There x nothing wrong with a a piture "But tohat its application?” What shall count fr or ant Ita belng ‘sid Ut this that mani in, that he pete spies to hi?100 warren ‘The ploture doesnt say. If you think that you understand the sentence I follow the rule that thle sto be called “pan” (a rule of your private language), what you have perhaps is a picture of yourself checking off various felings of yours as elther being ts ‘or not. The picture appears to solve the problem of how you Aetermine whether you have done the ‘checking’ right. Actually it ‘ocart give you even a hint in that direction; no maee then the picture of blindnest provides so much asa hint of how it isto be Aletermined that tis or that man is blind (348359, 422-495, p 184) ‘One will be inclined to say here that one ean simply remember this sensation and by remembering it will know that one is mak {ng a consistent application ofits name, But wall it also be possible to have a false memory impression? On the privatelanguage hy- pothesis, what would show that your memory impression is fase or trie? Another memory impression? Would this imply that ‘memory is court from which there is no appeal? But, as & matter of fact, that is not our concept of memory. Imagine that you were supposed to paint particular eslour °C,” ‘wich war the cour Wat appeared when the hemes! substances X 1d Y combined Suppose tat the colour strck you as bigter ‘on one day than on andthe would you not sometines say: “T must ie wrong, the colour i certainly the same ae yesterday’? This shows that we donot always reort to what memory tells asthe verdict ofthe highest cour of appeal [56] ‘Thevo is, indeed, such a thing as checking one memory against ‘another, eT check my recollection of the time of departure of train by calling up a memory image of how a page ofthe time- table looked—but “this process has got to produce a memory ‘which is actually corect Ifthe mental image of the timetable ould not tuelf be tested for corectness, how could i confirm the correctness of he rst memory?” (265). TT have a language that is rally private (Le, it sa logial Smpossiblity that anyone else should understand itor should have any basis for knowing whether fam using a particular name con- sistenty), my asertion that my memory tells me 0 and so wil be “utterly empty. “My memory’ will not even mean—my memory Impression. Por by a memory Impression we understand some- ‘thing that is either accurate or inaccurate; whereas there would wrromsrans rumosomucaL isvisnoaions 101 not be, inthe private language, any conception of what would establish a memory impression as correc, any conception of what ‘comet’ would mean here ‘The sume. Ono wants to say, Surely there can't be a difculty in sowing whether a feeling of mine sor isnt the some as the feel- ing I now have. I will cll this feeling “pun” and will thereafter call the same thing “pain” whenever it occurs. What could be ‘esier than to follow that rule?” To understand Wittgenstein reply to this attractive proposal we must come closer to his treat ment of rales und of whet it sto follow a rule. (Here he forges a remarkably illuminating connection between the philosophy of psychology and the philosophy of mathematics.) Consider his ex- ample of the pupil wha has been taught to write down a cardinal rnumber sees of the form ‘0, n, a, $n. . 2 at an order of the orm “+n: so that atthe order "+ he waites down the series of natural numbers (185). He has successfully done exercises and tests up to the mimber 1,000, We then ask him to costinee the series “+2’ beyond 1,000; and he writes 1,000, 1,004, 1,008, 1 012. ‘We tell him that this is wrong. His instructive replys, "But 1 ‘went on inthe same way” (185). There was nathing in the previ= ‘us explanations, examples and exereises that made it impossible for im to regard that es the continuation ofthe series. Repeating those examples and explanations won't help him. One must sey to him, in effect, “That dent what we call going on in the same way.’ Te is a fact, and a fact of the kind whose importance ‘Wittgenstein constantly stresses, thet i fs natural for human be- Ings to continue the series fa the manner 1,002, 1,004, 1008, given the previous training, But that is merely what it isa fact of ‘buman nature. ‘One is inclined to retort, “OF course he can misunderstand the fustruction and misunderstand the order "+2; but if he «nder- ‘stands he must go on in the right way.” And here one has the ‘dea that “The understanding itself isn sate which i the source ‘of the comect use” (146)—that the correct continuation of the series the right application of the rule of formula, speings from ‘one’s understanding of the rule, But the question of whether one ‘understands the rule cannot be divorced from the question of ‘whether one will go on in that one particular way that we call100 swrrroensre’s rumtosormncas, wvesTicATIONS ‘The ploture doesn't say. If you think that you understand the sentence I follow the rile that thie tobe called “pain” (a rule of your private language), what you have pethaps is a picture of yourself checking off various felings of yours as either being this ‘or not. The picture appears t0 solve the problem of how you determine whether you have done the ‘checking’ right. Actually it ‘docant give you even a hiat in that direction; no mate than the picture of blindness provides so much as a hint of how it isto be ‘etermined that this of that man is blind (348359, 422-426, p. 184), ‘One will be inlined to sey here that one can simply remember this sensation and by remembering it will know that one fs mak {ng a consistent application ofits name, But will it also be possible to have a false memory impression? On the privatelanguage hy- pothesis, what would show that your memory impression is fase or trie? Anothor memory impression? Would this imply that ‘memory is court from which there is no appeal? Bu, a8 & matter (of fact, that isnot our concept of memory. Imagine tht you wee supposed t0 paint s particular elour “C" “HIS? techn that ested ben he Shem sans {ind T combined —Soppore that the colbur track you as brighter fon day han on attr, would you ot sti ay I st song elt cette ha etn he that do nt alae ret fhe memory tele ws the vert af the highest cout of appeal [90] ‘Theze fs, indeed, such a thing as checking one memory against another, 2, I cheek my recolletion ofthe tine of departure of a train by calling up a memory image of how # page ofthe tine- table Tooked—but “this process has got to prodice « memory Which is actually correct. Ifthe mental image of the time-table ‘onld no itself be tested for correctness, how could i coaim the comrectnes of the rst memory?” (265). TT have a language that is really private (Le, iti logial {impossibility that anyone else shoold understand i or should have any basis for lnowing whether Iam using a particular name con- sistently), my assertion that my memory tell me so and 0 wall be tutenly empty. “My memory’ will nt even mean—my memory {mpression. For by a memory Impression we understand some- ‘thing that is either aceurate or inaccurate; whereas there would ‘wrrroeNsrEN’s emmosormiest, INvESMEATIONS 101 ot be, in the private language, any conception of what would ‘establish a memory impression as corret, any conception of what “eartect would mean here ‘The some, One wants to say, Suzcly there can't be dilleaty in knowing whether a feeling of mine is or ft the same as the fe {ng Tntvw bave, Iwill call this feling “pain” and will thereafter call the same thing “pain” whenever i occurs. What could. be feasier than to follow that rule? To understand Wittgenstein's reply to this attractive proposal we must come closer to his treat- ‘ment of rules and of what itis to follow a rule. (Here he forges @ remarkably ihuminating connection between the philosophy of psychology and the philosophy of mathematics.) Consider his exe “ample of the pupil who has been taught to wrlte down a cardinal ‘umber series of the form “o, m, 2n,2n . . 7 at an order of the form “+1 so that at the order “+I' he wrltes down the series of natural numbers (185). He has successfully done exercises and tests up to the number 1,000. We then ask him to continue the series +2 beyond 1,000; and he waites 1000, 1,04, 1,008, 1,012. We tell him that this is wrong. His instructive reply is, “But I ‘went on inthe same way” (155). Theze ws noting in the previ- ‘us explanations, examples and exerelss that made it impossible for him to regard that as the continuation of the series. Repeating those examples and explanations won't belp him. One must sty to him, in eect, “That ist what we call going on in the some way. Tt Is a fact, and a fact of the kind whose importance Wittgenstein constantly stresses, that i natural for human be- {ngs to coatinue the sees inthe manner 1,002, 1,004, 1,06, given the previous taining. But that is merely what it isa fact of human nature, ‘One is inclined to retort, ‘OF course he can misunderstand the instraction and misunderstand the order "+; but he under stands ithe must go on in the right way.’ And here one has the fdea that "The understanding itself isa state which isthe source fof the correct use” (146)—that the correct continuation of the series, the right application ofthe rule or formula, springs from fonc!s understanding of the rule. But the question of whether one tunderstands the rule cannot be divoread from the question of ‘whether one will go on in that one particular way that we call,