Principia Mathematica Pdf English Download

CHAPTER II THE THEORY OF LOGICAL TYPES Tno theory of logical types, to be explained in the present Chapter, recomrnended itself to us in the first instance by its ability to solve certain contradictions,of which the one best known to mathematiciansis Burali-X'orti's concerning the greatest ordinal. But the theory in question is not wholly dependentupon this indirect recomnrendation:it has alsoacertain consonance rvith comuron sensewhich makes it inherently credible. In what follows, we shall therefore first set forth the theory on its own accbunt, and then apply it to the solution of the contradictions. I. The Vicious-Circle Pdnaipla An analysis of the paradoxesto be avoided shorvsthat they all result from rr,certain kind of vicious circle*. The vicious circles in question arise lrom supposingthat a collection of objects may contain memberswhich can only be rlefinedby meansof the collectionas a whole. Thus, for example,the collection of propositions will be supposed to contain a proposition stating that ' all prop<:tion=' total.'=' call=' this=' 'vicious-circle=' principle,'=' be,'.rrs,,it,=' enablesus=' vicious=' circles=' involved=' assurnptionof=' rll,.git,irnatetotalities.=' arguments=' condemnedby=' vicious-circle=' '=' scc=' lsst=' section=' present=' chapter.=' cf.=' also=' [l=' poincarc,=' ler=' mathcoetiques=' et=' f=' rr=' logirllro,'reauede=' m€taphysiqueet=' ile=' morale,=' mai=' 1906'=' p.='>
38
II(TRODUClION
[cner.
principle will be called.,vicious-circle fallacies,' Such irgumente, in certain ci*umstances, may lead to contradictions, but it olten halpens that the conto_which , they lead are in fact true, though ihe argt,ments are 1l1sio.ns fallacious. Take, for example, the law of excluded ..=iddl', in the form..all propositions are true or false.' rf from this law we argue that, because the Iaw of excluded middle is a proposition, therefore the l# of excluded middle is tr'e or false, we incur a vicious-circle fallacy, 'AI propositions,' must be in some way linrited before it becomes a legitimate toialily, and any limitation which makes it legitimate must make any statement about the totality fall outside the totality. similarly, the imaginary seeptic, who asserts that he knows nothing, and is refuted by being asked if he knows that be knows nothing, has aseerted nonsense, and has been fallaciously refuted by an argument which involves a vicious-circle fallacy. rn order that the ..'pti't assertion may become significant, it is necessary to place some limitation ugon the things of which he is asserting his ignorance, becausethe things of which it is possible to be ignorant form an illegitimate totality. But as soon uul a suitable limitation has been placed by him upon the collection of propositions of which he is asserting his ignorance, the proposition that he is ignorant of every member of this collection must nob itself be one of the collection. rlence any significant scepticism is not open to the above form of refutation. The paradoxes of symbolic logic concern various sorts of objects: propo_ . . sitions, classes,cardinal and ordinal numbers, etc. All these so.ts or oLlects, as we shall show,represent illegitimate totalities, and are therefore capable of giving rise to vicious-circle fallacies. But by means of tbe theory (to be explained in Chapter III) rvhich reduces statements that are verballv concerned with classesand relations to statements that are concernedwith propositional functions, the paradoxes are reduced to such as are concerned with propositions and propositional functions. The paradoxes that concern propositions are only indirectly relevant to mathematics, while those that more nearly concem the mathematician are all concerned with propositional functions' we shall therefore proceed at once to tbe consideration of propositional functions. II. The Nature of propositional Xurctions. a 'propositional function' *S something which contains a .B-y variable a, and expresses a propotitr,on -..n as soon * , u't' is assigned to a. That is to say, it differs from a proposition sorely by the fact that it is ambiguous: it contains a variable of which the value is irnassigned. rt agrees with the ordinary functions of mathematics in the fact of containin! an unassignedvariable; where it differs is in the fact that the values oithe functionare propositions.Thus e.g,,,n is a man'or,,sins:1,,isa propo_ sitional function. we shall find that it is possible to incnr a vicious-ciicle
rrl
pBoFogrrrolrar, rrrucr.rolvs
gg
fallacy at the very outsOt,by admitting as possibleargumentg to a propositional function terms which presuppose the function ThiJform of the felliv is verv instructive, and ite avoidance leads, ao we ehail see, to tbe hierarchy of typur. The question as to the nature of a function* is by no means an easy one. It would oeem, however, that the essential characteristic of a functlon is ambigu'ity. Take, for erample, the law of identity in the form,,A is .r1,'which is the form in whieh it is usually enunciated. It ie plain that, regarded psychologically,we have here a eingle judgment. But what are we to say of the object of the judgment ? we are not judgrng that socrates ie socratee,
ambiguity bhat constitutes the eesenceof a function. When we speak of .,{a,', rvhereo is not specified,we mean one value of the function, but not a definite one. We may expressthis by saying that,,,6s', ambiguouslyitenotes $a, fi, $c, etc., where +', +b, $c, etn.,are the varjoue values of ..fa.' When we say that 'fua' anrbiguously denotes eo, Qb, $c, etc.,rve mean lhat '6a' means one of the objects 6a, #b, $c, etn., though not a definite one, but an undetermined one. rt follorvs that'Qa' only hL a well-defined. meaning (well-defined, that is to say, except iu so far as it is ofits essenceto
was definite, while conversely,as we have just seen,the function cannot be definite until its values are definite, This is a particular case,but perhapethe most fundamental case, of the vicious-circle principle. a function is what ambiguously denotce gome one of a certain totality, namely the valnes of the function I hence this totality cannot contain any rnembers which involve the function, since, if it did, it would contain memberr involving the totality, which, by the vicious-circle principle, no totality can do. It will be seen that, according to the above account, the values of a function are presupposed by the function, not vice verga. It is sufficiently obvious,in any particula,r case,that a value of a function does not p.u.oppo.L the function. Tbus for example the proposition 'socrates is human' eai be perfectly apprehended without regarding it as a value of the function,,a is human.' It is true that, conversely,a function can be apprehendedwithout 'wheatheworil'function'isueclintheeequel,,iprclnsitionalfunction,'isalwrysmeent. Otber lunotions will not be in quostion in the preeelt Chopter.
40
INTAODUOTION
[cuer.
its being recessaryto apprehend its values severally and individually. If this were not the case,no function could be apprehended at all, since the number of values (true ard false) of a function is necessarily infinite and there are necessarily possible arguments with which we are unacquainted. What is necessaryis not that the values should be given individually and extensionally, but that the totality of the values should be given intensionally, so that, concerning any assignedobject, it is at least theoretically determinate whether or not the said object is a value ofthe function. ft is necessary practically to distinguish the function itself from an undetermined value of the function. Wo may regard the firnction itself as that which ambiguously denotes,while an undetermined value of the function is that which is ambiguously denoted. If the undetermined value is written ' $u,' we will write the function itself 'f2.' (Any other letter may be used in place ofo.) Thus rve should say 'Qn is a proposition,' but '{d is a propositional function.' When we eay ' $n is a proposition,' we mean to state something which is true for every possiblevalue of o, though we do not decide what value o is to have. We are making an ambiguous statement about any value of the function. But wben we ay ' $k is a function,' we are not making an ambiguous statement. It would be more correct to say that we are making a statement about an ambiguity, taking the view that a function is an ambiguity. The function itsel{ {?, is the singlething which ambiguouslydenotes its many values; while {e, where c is not specified,is one of the denoted objects, with the ambiguity belonging to the manner of denoting. We have seen that, in accordancewith the vicious-circle principle, the values of a function cannot contain terms only definable in terms of the function. Now given a function $h,lhe values for the firnction* are all propositions of the form {r. It follows that there must be no propositions, of the form {a, in which u has a value which involves OA. Gf this were the case, the values of the function would not all be determinate until the function was determinate. whereaswe found that the function is not determinate unless its values are previously determinate.) flence there must be no such thing as the value for {d with the argument {6, or with any argument which involves fd. That is to say, the symbol '+ (06)' must not express a proposition, as 'fa' doesif {a is a value for {6. In fact '{ ({6)' must be a symbol which doesnot expressanything: we may therefore say that it is not significant. Thus given any function {0, thep are arguments with rvhich the function has no value, as well as argumerits with which it has a value. We will call the arf;uments with which f6 has a value 'possible values of a.' We will say that gh is 'significant with the argument a' when $h has a value with the argument ,r. * W e e b g l l sp e ski n th i sCb e p ter of t' taluealor ,,valueaoJqx ,' m eani ngi aer c h C4'sndof oaee the ssme thing, namely Oa, Qb, ec, etc. Tho tlietinction of pbraseology serves to svoial embig[it;r sherc evera,l vrisbles are ooncemeil, eepeoially when oae of them is e function.
II]
PossIBLEARGuMENTS FoR FuNoTIoNs
41
When it is said that e-9.' + @2' is meaningless,and therefore neither true nor false, it is necessaryto avoid a misrrnderstanding. If 'Q ($2)' werc interpreted as meaning 'the value for f2 with the argument Q2 ie trrc,' t,hat would be not meaningless,but false. It is false for the same reason for which 'the King of France is bald' is false,namely becausethere is nosuch f,hing as 'the value for {2 with the argument $2.' Bfi when, with some trgunrent a, we assert fa,we are not meaning to assert 'the value for ffwith the algument a is true'; we are meaning to assert the actual proposition rvhich a'sthe value for {0 with the argument o. Thus for example if {b is '2 is a nran,' rf (Socrates)will be 'Socrates is a man,' not 'the value for the function 'h is a man,' with the argurnent Socrates, is true.' Thus in accordance with our principle fhat '$$2)' is meaningless, we cannot logitimately deny 'the function 'd is a man' is a man,' because this is nonsense, but rve can legitimately deny 'the value for the function'6 is a rnan' with the argument '6 is a man' is true,' not on the ground that the value in question is false, but on the ground that there is no such value for the function. We will denote by the symbol '(a).Q*' the proposition 'fz always*,' i, e. th.e proposition which asserts al,l, the values for fd. This proposition involvesthe function $h,not merely an ambiguousvalue of the function. The rwsertion of fe, where a is unspecified, is a different assertion from the one which asserts all values for $6, for the former is an ambiguous assertion, whereasthe latter is in no senseambiguous. It will be observedthat,,(n).$n' doesnot assert'fa with all valuesofa,' because,as we have seen,there must Lrcvaluesof a wiih which '{a' is meaningless.What is assertedby,, (u).5n' is all propositionswhich are values for Qh; hence it is only with such values of z as make '{a' significant, i.a. with all possibleargtments, that {a is asserted when we assert '(c). {o.' Thus a convonientway to read '(*). Q*' is 'Sa is bruewith all possiblevalues of er.' This is, however,a less accurate reading than 'Sa always,' beeausethe notion of truth is not part of the content of what is judged. When we judge 'all men are mortal,' we judge truly, but the notion of truth is not necessarily in our minds, any more than it need be when rvejudge 'Socratesis rnortal.' III.
Def,nition and, SystematicAmbiguity of Truth and, Fal,sehood^
Since '(o). far' involves the function Sk, it must, according to our principle, be impossible as an argument to {. That is to say, the symbol 'Q{@). {aJ' must be meaningless.This principle would seem,at first sight, lo have certain exceptions. Take, for example, the function 'B is false,' and r:rrnsiderthe propositioa '(p).p is false' This should be a proposition rrsserting all propositions of the form ' gt is false.' Such a proposition, we ' We use 'alwaye' as meaning 'in all cases,' not t'at ell times.' will me&n ' in Bomec&ses.'
SimilarlJr ,,8ometime8'
42
IIEM)DUOTTON
[cuer.
should be inclined to say, must be false, beopuse 'p is false' is not always true. flence we should be led to the proposition ' {(p) . p is false} is false,' i.e. we ehould be led to a proposition in which '(p).p is false' is the argument tn the function 'f is false,' which we had declared to be impossible. Now it, will be eeen that '(p).p is false,' in the above, purports to be a proposition about all propositions,and that, by the general form of the viciouscircle principle, there must be no propositions about all propositions. Neverthelegs,it seemsplain that, given any function, there is a proposition (true or false) asserting all its values. Hence we are led to the conclueion that 'p is false' and 'g is false' must not always be the values, with the arguments p and q, for a single function 'f is false.' This, however, is only possible ifthe word 'false' really ha'emany different meanings, appropriate to propositions ofdifferent kinds. That the wonds'true' and 'false' have many different meanings,according to the kind of proposition to which they are applied, is not difficult to see. Let us take any function Qfu,and,let fo be one of its values. Let us call the sort of truth which is applicable to {a ' f,rst truth.' (This is not to assume that this would be first truth in another context: it is merely to indicate that it is the first sort of truth in our context.) Consider now the proposition @). Qa. If this has truth of the sort appropriate to it, that will mean that every value sz has 'ffret truth.' Thus if we call the sorb of trutb that is appropriate to (c), $u'second, truth,' we rnay define '[(u).Qa] has second truth' as mearring'every value for f? has first truth,' i.e. '(a). (So has first truth).' Sinrilarly,if we denoteby '(gc) . $u' the proposition'fa sometimes,' i.e.aawe moy less accurately expressit, 'fa with somevalue of a,' we find that (ga). fc has secondtruth if there is an o with which fa has first truth; thus we may define '{(S'). {eJ has secondtruth' as meaning 'some value for f0 has first truth,' i.e.'(ga). (fa has first truth).' Similar renarks apply to faleehood. Thus '{(a). fa} has second falsehood'will mean 'some value for $b has first falsehood,' i.e. '(ga). (fr has firrt falsehood),' while has second falsehood' will mean 'all values for f0 have first '{(9').{a} falsehood,'i.e. '(u). (fo has flrst falsehood).' Thus the sort of falsehoodthat can belong to a general proposition is different from the sort that can belong to a particular proposition. Applying these considerations tl the proposition '(p). p is false,' we see that the kind of falsehood in question must be specified. I{, for exaurple, first falsehood is meant, the function 'p has fimt falsehood' is only significant when p is the eort of proposition which has first falsehood or 6rst truth. Elence '(p).p is false' will be replaced by a statement which is equivalent to 'all propositions having either ffrst truth or first falsehood have first falsehood.' This proposition hae secozd falsehood, and is not
rr]
TBIxTEAND FArsEEooD
r possible argunrent to the funcbion 'p bas frsf npparent exception to the principle that 'Ql@).{a}' dieappears,
43 falsehood.' Thus the must be meanirgless
Similar considerationswill enable us to deal with 'not-10' and wit'h'p or q.' It might seem as if these were functions in rvhich any proposition might nppear aa argument. But this is due to a systematic ambiguity in the meanings of 'not' and 'or,' by which they adapt themselves to propositions of any order. To explain fully how this occurs, it wil be well to begin with a definition of the simplest kind.of h'uth and,falsehood. The universe eoneists of objects having various qualities and standing in various relatione. Some of the objects which occur in the universe are complex. When an object is complex, it, consists of interrelated parts. Let rueconsider a complex object composedof two parts a and b standing to each other in the relation .8. The complex object 'a-in-the-relation-.R-to-b' may be capable of being perceiued; when perceived, it is perceived as one object. Attention may show that it is complex I we lhen juilge that a antl b sta,ndin t'he relation ,ll. Such a judgment, being derived from perception by mere attention, may be called a 'judgment of perception.' This judgment of perception, considered as an actual occurrence, is a relation of four terms, namelyo and 0 and .B and the percipient. The perception, on the contrary, is n relation of two terms, namely 'a-in-the-relation--B-to-b,' and the percipient. Since an object of perception cannot be nothing, we cannot perceive 'a-in-therelation-R-to-0' unless o is in the relation .E to b. Hence a judgment of perception, according to the above definition, rrust be true. Thig does not rnean that, in a judgment which aryears to us to be one of perception, we Rre Bure of not being in error, since we may err in thinking that our judgment has really been derived merely by analysis of what was perceived. But if our judgment has been oo derived, it muet be true.iln fact, we may define truth, where such judgments are concerned,as consisiing in the fact that there is rt, complexcorresptundingto the discursive tboughtwhich is the judgment. That is, when we judge ' a has the relation n b b,' our judgment is said to be true when there is a complex 'a-in-the-relation-.R-to-b,' and is said to be fal,se when this is not the case. This is a definition of truth and falsehood in relation to judgments of this kind. It will be seen that, according to tbe above account, ajudgment does not hlve a single object, namely the proposition, but has several interrelated objects. That is to say, the relation which constitutes judgment is not a relation of two terms, namely the judging mind and the proposition, but is a rclation of sevgral terms, namely the mind and what are called the constituents of the proposition. That is, when we judge (say) 'this is red,' what occurs is a relation of three terms, the mind, and ' this,' and red. On the other hand, when we perceiae'the rednessof this,' tbere is a relation of two terms, namely
44
INTRODUOTTON
[cner.
the mind_and complex object ,.the rednessof this.,, When a judgment -the occurs, there is a certain cornplex entity, composed of the ,rrirrd tl' objects of the judgmenl. When it e joa'gment is true, in the'ni case of :.'.i:1' the kind of judgments we have been considlrin'g, there is a corresponding complex of the objectsof the iudgment, alone, F;alsehood, ir, ..gn.d to oo. present class ofjudgments, consists in the absenceof a correspondirg co-ple* composed.of the objects alone. ft follows from the above theoiy that a 'proposition,'inthesense_inrvhichapropositionissupposed tobeiheobject of a judgment, is a false abstraction,b'u-o.' ajodgment has severalobjJcrs, not one. It is the severalness of the objects in judgment qas opposed to perception)which has led peopleto speak of thougirt ri,,discorsiue',i though they do not appear to have realized ciearly what was meant by this epithet :ts of a single judgment, it follows that rnsein whir:h this is distinguished from ) entrty at all. That is to say, the phrase ; we call an .,incomplete,,symbol*; it
tt'.l;; ',X'l1',ii; to acquire a complere meaning. rh;n'r;:T,'.'T'J#flf
circumstancethat judgment in itself suppliesa sufficient supplement,roi ,t r, Judgment in itself rnakes no aerbal addition to the p.opo.ilion. Thus ,,the proposition 'socrates is human' uses ,.socrate. i, ho,ouo,'io , *'y rrtri'n requires a supplement of some kind befole it acquires a complete .;r';;, but when Ijudge 'socrates is human,' the meaning i. iir','i'r judging, and-we no longer have an incornplete symbol rhe'o-pt'tiJLy fact that p.opositions are'incompletesymbols'is irnportant phirosophicalry, and is relevait points in symbolic logic. 'i 'rt'i., The judgments we have been dealing with hitherto are such as are of the sale- fgrm as judgments perception, i.e.their subjects are always particular 'f and definite. But there are many j udgments which are not of this form. such are 'all men are mortal,' .,I met a riran,,,..somemen ,.e G.euks.,, Bu[.' dealing with suchjudgments, we will introduce some technical terms. We wilf give the name of ,,a complet:,,to any such object as .,cr in the re_ lation 8 to D' or,,a having.the quality q,,, or,,L and b anrt c standing in the reJation S.'-Rroadly speaking, a cornpliu is anything *hi'h u.'o..'io ih' universe and is n-ot simple. We will- call a juigmeit elementary when iu merely asserts such things as ' o hfs the reration o b,' ,, a rrr. trr'qurriiy, q -81 or 'a and b and c stand in the relation s.' Then an erementarylodg*'nt'i. true when there is a corresponding complex, and false when there-is io .o.rusponding complex. But take now such a proposition as ..all men are mortal.,, Here the judgment doesnot correspondto one compLex,but to many,mamely,,socrates * Se Chapter III.
IIJ
oENERAL
JIIDGIIENTS
4E
rr.t necessaryto know what rnen there are. we must admit, therefore, as a r';rdicallynew kind ofjudgment, s'ch generalassertionsas ,,all men are mortal.,,
It is evident (as explained above) that the definition of truth is different in the caseofgeneral judgments from what it was in the caseofelementarv .irrdgments. Let us call the meaning of truth which we gave for elementar! .judgnenbs'elementary truth.' Then rvhen we assert that it is true that ail lncn are mortal, we shall mean that all judgments of the form ,,a is mortal,' where a is a man, have elementary truth. we may define this as ,,truth of l,he secondorder' or 'second-order truth.' Then if we express the proposition 'tll rnen are mortal'in the form '(u) . u is mortal, where a is a man,,, nrrd call this judgment p, then 'p ie true' must be taken to mean ,,p has second-ordertruth,' which in turn means '(r).', is mortal'has elementarytruth, wherea is a man.,'
rs :Lre men, but may have any value witb which ,,,a is a man' implies .a is rrr.rbrr,l'issignifcant,i.e. either true or false. such a propositionis called a ' lir.rn:rlimplication.' The advantageof this form is thai ttre valueswhich the r^ri;rble may take are gi'en bythe function towhich it, is the argument: the
46
IT.ITRODUCTIoN
[cner.
values which the variable may take are all those with which the function is significant. We use the symbol '(r).Qa' to express the general judgment which assertsall judgments of the form '{2.' Then the judgment 'all men are mortal' is equivalent to '(*).', is a man' implies 'a is a mortal,' z'.e.(in virtue of the definition of implication) to '(u) . n is not a man or u is mortal.' As we have just seen, the r4eaning of tnrth which is applicable to this proposition is nob the same as the meaning of truth which is applicableto'oisa man' or to 'a is rnortal.' And generally,in any judgment (a). fa, the sense inwhich this judgment is or maybe true is not the same as that inwhich {o is or may be true. If fz is an elementaryjudgment, it is true when it points fo a correspondingcornplex. But (.2). fo does not point to a single correspondingcomplex:the correspondingcomplexesare as numerousas the possible values of c. It fbllows from the above that such a proposition as 'all the judgments made by Epimenides are true ' rvill only be prima facie capable of truth if all hisjudgments are of the sameorder. If they are ofvarying orders,ofwhich the nth is the highest, lr'e may make n assertionsof the form 'all the judgments of order ror made by Epimenides are true,' rvhere nz has all values up to n. But no suchjudgment can include itself in its own scope,since such a judgment is always ofhigher order than thejudgments to which it refers. Let us cousider next what is meant by the negation of a proposition of We observe,to begi,nwith, that '{n i4 some cases,'or the form '(u).Qd' 'fo somebimes,'is a judgment which is on a par with 'fa in all cases,'or '{z always.' The judgment '{a sometimes'is true if one or more valuesol .r exist {brwhich {z is true. Wewill expressthe proposition'fa sometirnes' by the notation '(go).Qr,' where 'g' stands for 'there exists,' and the whole symbol may be read 'there exists an a such that $n.' We take the two kinds of judgment expressed by '(') . Qn' and.' (S*) . Q*' as primitive ideas. We also take as a primitive idea the negation of an el'ementaryproposition. We can then define the negationsof (a). $u and (qr) . fz. The negation of any propositionp will be denotedby the symbol'-p.' Then the negation of (a) . Qo will be d'ef'ned,as meaning
'(e*).- d','
as meaning ' (u) . Qu.' Thus, and the negation of (ga) . {a rvill be d,efi'ned, in the traditional language of formal logic, the negation of a universal affirmative is to be defined as the particular negative, and the negation of the particular affirmative is to be defined as the universal negative.' Hence the meaning of negation for such propositions is different from the meaning of negation for elementary propositions.
II.l
SYSTEMATI9 AMBreurry
47
An analogousexplanation will apply to disjunction. Consider the statenrcnt 'either p, or On always,' We will denote the disjunction of two lrropositionsp, q by 'p v g.' Tben our statement is 'p. v . ('). Qr.' We will supposethai p is an elementary proposition, and thai fa is always au elemenl,lry proposition. We take the disjunction of two elementary propositions as rr lrrinribive idea, and we wish to ilef,ne the disjunction ' 1 t . v . ( n ) .g a |' 'Ihis may be definedas '(c) . pv Qa,'i.e.'either p is true, or fa is alwaystrue' is to mean 'p or $a' is always true.' Similarly we will define ' p . v . ( g u ) .9 x' rrs meaning '(gs).pvQn,' i,.e.we define 'eitherp is true or there is an a lirr which {o is true' as meaning 'there is an n fot which eitherp or fr is true.' Similarly we can define a disjunction of two universal propositions: ' @ ) . Q u , v . ( y ) . {y ' w i l l b e d e f i n e d a s m e a n i n g '( n ,y) .g a vg y,' i .e . 'oither {rr is always true or rlry is always true' is to mean '$a or !ry'is tlways true.' By this method we obtain definitions of disjunctions contriining propositionsof the form (r).4' or (gr). fc in terms of disjunctions of elementarypropositions; but the meaning of 'disjunction' is not the same for propositionsof the forms (c) .Q',(gr).#r,u it was for elementary propositions. Similar expllnations could be given for implication and conjunction, but this is unnecessary,since these can be defined in terms of negation and disjunction. IY. Why a Giuen Xunction c'eclwiresArgwments of a Certain Type. The considerations so far adduced in fbvour of the view that a functiorr cannot significantly have as argument anything defined in terms of the function itself have been more or less indirpct. But a direct consideration of the kinds of functions which have functions as argume4ts and the kinds of functions which have arguments other than functions will show, if we are not mistaken, that not only is it impossible for a function f2 to have itself or anything derived from it as argument, but t,hat, if is another function ^12 strch that there are arguments a with which both 'fa' and,'{ra' are sigrrificant, then ^lr2 and anything derived from it cannot significantly be nrgument to $2. This arises frorn the fact' lhat a function is essentially an ambiguity, and that, if it is to occur in a definite proposition, it must occur in such a way that the ambiguity has disappeared, and a wholly unambiguous statement has resulted. A few illustrations will rnake this clear. Thus '(c) . fa,' which we have already considered,is a function of {6; as soon ns {6 is assigned,we have a definite proposition, wholly free from ambiguity. tsut it is obvious that we cannot substitute for the function something which is not a function: '(o).6*' means'{e in all cases,'and depends for its significance upon the fact that there are 'cases' of $a, i.e. upon the
48
INTBODUCTION
[cuer,
ambiguity which is characteristic of a function. This instance illustr.ates the fact that, when a function can occur significantly as argument,something which is not a function cannot occur significantly as argument. But conversely, when something which is not u function can occur significantly as argnment,a function cannot occur significantly. Take, e.g.,,a is a man,' and consider 'f0 i. a lnan.' Here there is nothing to eliminate the ambiguity which constitutes f6; there is thus nothing definite which is said to be a rnan. A function, in fact, is not a definite object, which could be or not be a manl it is a mere ambiguity awaiting determination, and in order that it may occur significantly it rnust receive the necessarydetermination, which it obviously does not receive if it is merely substituted for somethingdeterminate in a proposition*, This argument does not, however, apply directly as against such a statement as.,{(,z).fr} is a, man.' Common senselvould pronounce such a statenrent to be meaningless,but it cannot be condemnedon the ground of ambiguity in its subject. We need hereanewobjection,namelythefollowing: Apropositionisnotasingleentity, but a relation of several; hence a statement in which a proposition appears as subject will only be significant if it can be reduced to a statement about the terms which appear in the proposibion. A proposition,like such phrases as'the so-and-so,'where grammaticallyit appearsas subject,must be broken np into its constituentsif we are to find the true subject or subjectsf. But in such a statement as 'p is a man,' wherc p is a proposition, this is not possible. Hence '[(a). {o} is a man' is meaningless. Y, The H,i,erarchyof Functions and, Propositions. We are thus led to the conclusion,both from the vicious-circleprinciple and from direct inspection, that the functions to rvhich a given object a can be an argument are incapable of being arguments to each other, and that they have no term in comnronwith the functions to which they can be arguments. We are thus led to construcb a hierarchy. Beginning with o and the other terms which can be arguments to the same functions to which o can be argument, we come next to functions to which a is a possible argument,and then to functions to rvhich such functions are possible arguments, and so on. But the hierarchy which has to be constructed is not so simple as nright at first appear. The functions which can take o as argurnent form an illegitimate totality, and themselvesrequire division into a hierarchy of functions. This is easily seenas follows. Let f ({2, r) be a function of the two variablesf2 and ra. Then if, keeping a fixed fofthe moment, we assert this rvith all possible values of f, lve obtain a proposition: @) . f @?, a) .
* Note thet st&tements concerning the significance of a phrase coutaining ,,C, ' conceru the synbol, 'Q2,' autl iherefore do not fall uncler the rule that the elimination of the functional ambiguity is ueceesary to signiticance. Significanoe is a property of eigas. Cf, pp. 40,41, f Cl. Chapter III.
II]
TEE ErERAxcEy oB FurircrroNs
49
Ilcre, if c is variable, we h,avea function of a; but as this function inv,rlves n totality of values of $2*, it cannot itself be one of the values included in the totality, by the vicious-circle principle. It follows that the totality of values of {2 concerned in (f) . f (62, o) is not the totality of all functions in ivhich ,r can occur as argument, and thab there is no such totality as that of all functions in which o can occur a^sargument. It follows from the above bhat a function in which f2 appears as argument requires that'Q2' should not stand for any funcbionwhich is capable of a given argument, but must be restricted in such a way that none of the functions which are possiblevalues of ,'{2' should involve any referenceto the totality of such functions. Let us take as an illustration the definition of identity. We might attempt to define .,c is identical with y' as meaning 'whatever is true of n is true of y,' i.e. 'fra alwaysimplies {gr.' But here, since we are concerned to assert all values of '+a implies fy' regarded as a function of f, we shall be compelled to impose upon f some limitation which will prevent us from including amongvaluesof { valuesin which ,,all possibie values of f ' are referred to. Thus for example ,,e is identical rvith a,, is a function of r1 hence,if it is a legitimate value of $in',ga always implies Qyi' *e shall be able to infer, by meansof the above definition, that if a is identical with a, and a is identical with y, then y is identical with o. Although the conclusion is sound, the reasoning embodies a vicious-circle fallacy since we have taken '({). {r implies So,' as a possible value of fa, which it cannot be. If, however, we impose any limitation upon f, it may happen, so far as appears at present, that with other values of f we might have Qn true and Sy false, so that our proposed definition of identity would plainly be rvrong. This difficulty is avoided by the 'axiom of reducibility,' to be explained later. I-or the present, it is only mentioned in order to illustrate the necessityand the relevance,qf the hierarchy of functions of a given argument. Let us give the name 'o-Iunctions' to functions that are significant for a given argument o. Then suppose we take any selection of a-functions, and consider the proposition 'a satisfies all the functions belonging to the selection in question.' ff we here replaceo by a variable,we obtain an o-function; but by the vicious-circleprinciple this a-function cannot be a member of our selection,since it refersto the whole of the selection. T,ei the selectionconsisr of all those functions which satisfy/(f2). Then our ney function is @). {f (+2) implies {o}, where n is the argument. It thus appears that, rvhatever selection of a-functions we may make, there will be other o-functions that lie outside our * When we the erplanation simpler to*rito argument must
Bpeok of 'velues oJ 92' it is O, uot z, that is to be assigned. This follows from in the note on p. 40. When the function iiself is the variable, it is possible and 0 rather then 6i, etcept in positions where it is necesearyto emphasize th&t sn be supplieil to eecure signilicanoe,
50
III1TRODUOTIOII
[ouer.
selection Such o-functions, as the above insteuce illustrates, will always arise through taking a functiou oftwo argumenti, {2 and r,and asserting all or some of thevalues resulting from varying f. What is necessary,therefore, in order to avoid vicious-circle fallacies, is to divide our o-functions into 'types,' each ofwhich containsno funcbionswhich refer to the whole of that tyPe. When something is assertedor denied about all possiblevalues or about some (undetermined) possible values of a variable, that, variable is called apparent, after Peano. The presenceof the words all or sonr,e in a proposition indicates the presenceof an apparent variable; but often an apparent variable is really presentwhere languagedoes not at once indicate its presence. Thus for example'/. is mortal' means'there is a time at which r{ rvill die.' Thus a variable time occurs as apparent variable. The clearest instances of propositions not containing apparent variables are such as expressimmediate judgments of perception,such as 'this is red' or 'this is painful,' rvhere'this' is something immediately given. In other judgments, even where at first sight no variable appears to be present, it often happensthat there really is one. Take (say) 'Socrates is human.' To Socrateshimself, the rvord 'Socrates' no doubt stood for an object of whiclhe was imrnediatelyarvare,and the judgrnent 'Socrates is human' contained no appareut variable. But to us, who only know Socrates by description, the word 'Socrates' cannot mean what it rneant to him; it means rather 'the person having such-and-suchproperties,'(say)' the Athenian philosopherwho drank the hemlock.' Norv in all propositionsabout 'the so-and-so'there is an apparent variable, as will be shown in Chapter III. Thus in what we have in mind when we say 'Socrates is human' there is an apparent variable, though there was no apparent variable in the corresponding judgment as made by Socrates,providedwe assumethat there is such a thing as irnmediate awarenessof oneself, Whatever may be the instances of propositions not containing apparent variables,it is obvious that propositional functions whosevalues do not contain apparent variables are the sourceof propositions containing apparent variables, in tbe senseinwhich the function {,? is the sourceof the proposibion(a).5a. For the values for Qh do not contain the apparent variable a, which appears in (o).Sc; if they contain an apparent variable y, this can be similarly eliminated, and so on. This prciCessmust corne to an end, since no proposition which we can apprehend can contain more than a finite number of apparent variables, on the ground that whatever we can apprehend must be of finite complexity. Thus we must arrjve at last at a function of as many variables as there have been stages in reaching it fronr our original proposition, and this function will be such that its values contain no apparent variables. We may call this function the nntriu ofour original proposition and ofanyother
rtl
MArBrcEs
61
arguments prolxrsitions and functions to be obtained by turning some of the t,,,r,irofunction into apparent variables. Thus for example,if we have a matrixlrrrcticrnwhosevalues ate $(a,y), we shall derive from it il. Q@, Y), which is a function of r' @) . Q @, y), which is a function of Y, with all possible values of o and y' k,',yt . + @, y),meaning '{ (c, 9) is true variable, i.e. no variable excepb no real containing proposition is a 'l'hi, i*t
so-and-so,' The first matrices that, occur are those whose values are of the forms
6r,*(r,il'x@'!,2'..),
characterized by the fact that they involve no variables except individuals. Such functions we will eall 'f,rst'ord'er functions'
Again, if a is a given individual, 'f ! a implies { ! a with all possiblevalues of {' is a function of o, but it is not a function of the form { ! o, becauseit involves an (apparent) variable f which is not an individual' Let us give the name ..pruA^ilt' to any first-order function { ! CI. (This use of the word ' predicate'
52
INTRODUCTION
[cEA'.
is only proposed for the purposes of the present discussion.) Then the stabement '{ ! o implies f ! a with all possiblevaluesof 0' be read 'all the -uy about o, but does predicates of a are predicates of a.' This makes a statement not attribute tro a a preilica,ts in the special sensejust defined. Owing to the introduction of the variable first-order function g'1,2,we now have a new set of matricee. Thus '{!o' is a functionwhich contains no apparentvariables,butcontainsthe two real variables$12 ard,o. (It should be observedthat when f is assigned,we may obtain a function whose values do involve individuals as apparent variables,for example if f !a is (y).*(r,y). But so long as { is variable, { ! r contains no apparent variables.) Again, ifa is a definite individual, {!a is a function of the one variable f!2. If a and D are definite individuals, '{ ! o irnplies rlr ! 6' is a function of the two variables+12,h'!2, and so on. We are thus led to a whole set of nery matrices, f (+ t 2), s @ t2, ^1,| 2), F (O ! 2, c), and so on. These matrices contain individuals and first-order functions as arguments, but (like aII matrices) they contain no apparent variables. Any such matrix, if it contains more than one variable, gives rise to new functions of one variable by turning all its arguments except one into apparent variables Thus we obtain the functions whic h is a f unc t ionof { 12. $).5 @t2 , 'hl2 ) , (t). F(Q I 2, o), which is a functionof St2. (0) .i7(0 I 2, a), which is a function of e. We will give the name of second,-ord,er matricesto such matrices as have finst-orderfunctions among their arguments,and have no arguments except first-order functions and individuals (It is not necessarythat they should have individuals among their arguments.) We will give the name of second,order functions to such as either are second-ordermatrices or are derived from such matricesby turning some of the arguments into apparent variables. It will be secn that either an individual or a first-order function may appear as argument to a seconcl-orderfunction. Second-orderfunctions are such ascontain variables which are first-order functions, but contain no other variables except (possibly)individuals. We now have variousnerv classesof functions at our comnrand. In the first
(S).6!e, and so on. (These resuit from assigninga value tof leaving to be assigned.) We will call such functions,,predicative functions o? { first-order functions.'
ill
sEcoND-oRDEBFttNcrroNs
53
l rr t hc eecondplace, we have second-orderfunctions of two argumentg one ,,1'whiuhis a first-orderfunction while the other is an individual. Let us denote rrrrr|r,t,rrrrrrined valuesof such functions by the notation f t($12, u). Ax x,x,rrts c is assigned,weshall have a predicative function of { 12. If our litrrcl,ioncontainsno first-order function as apparent variable, we shall obtain t grnxliurtivefunction of o if we assign a value to S!2. Thus, to take the xirrrlrlcst possible case,ifft ({ 12,a)is f ! a,the assignmentofa valueto { gives rrnrr,prcdicativefunction of c,invirbue of the definitionof 'f !2.' Butif l'l (6 I 2, u) contains a first-order function as apparent variable, the assignment ,rl r vrrlueto $12 gives us a second-orderfunction ofa. In the third place,we have second-orderfunctions of individuals. These n'ill n,llbe derivedfromfunctionsof the form/! (Q12,n) by turning{into an rrplra,rcntvariable. We do not, therefore,need a new notation for them. We have also second-orderfunctions of two first-order functions,or of two nrrt:hfunctions and an ildividual, and so on. We may nolv proceed in exactly the same way to third-order nratrices, rvhich rvill be functions containing second-orderfunctions as arguments,and torrtaining no apparent variables,and no arguments except individuals and lirsb-orderfunctions and second-orderfunctions. Thence we shall proceed,as lxrlbre,to third-order functions; and so we can proceed indefinit'ely. If the highest order of variable occurring in a function, whether as argumerrt or as npparent variable, is a function of the ath order, then the function in which il, occursis of the ?z+ lth order. We do not arrive at functions of an infinite order,becausethe number of arguments and of apparent variables in a function rnust be finite, and thelefore every function must be of a finite order. Since the orders of functions are only defined step by step, there can be no procesg ol 'proceedingto the limit,' and functions of an infinite order cannot occur. We will define a function of one variable as predticat'iuewhen it is of the next order above that of its argumenb,i.e. of the lowest order compatible rvith its havingthatargument. Ifa function has severalarguments,and the highest order of function occurringamongthe argumentsis the nth, rve call the function predicative if it is of the rl + lth order, i.e. again, if it is of the lorvest order compatible with its having the arguments it has. A function of several trguments is predicativeif there is one of its arguments such that, when the obherarguments have values assignedto them, we obtain a predicative function of the one undetermined argument. It is imporbant to observethat all possible functions in the above hierarchy t:anbe obtained by meansof predicative functions and appa.rentvariables. Thus, i$ rve saw, second-orderfunctions of an individual n are of the form ( +) . ' f t ( +t 2 , u ) o t ( s d ) . / t ( $ t 2,n ) o r ( 0 ,9 ) .ft( +t2 ,a l 't'2 ,n ) o r e t'c'., where / is a second-order predicative function. And speaking generally, a
54
INTRODUCTION
[cuer.
non-predicative function ofthe zth order is obtained from a predicative function of the nth order by turning all the arguments of the a - lth order into apparent variables. (Other arguments also may be burned into apparent variables.) Thus we need not introduce as variables any functions except predicative functions. Moreover, to obtain any function of one variable a,lie need not go beyond For thefunction ('t).'f|($|2'rlrI2,n)' predicativefunctionsof ,Tzovariables. where ;f is given, is a function of $12 and a, and is predicative. Thus it is of is of the form the form Fl($12,r), and therefore(+,*).'ft(+12,*12,a) (+) . /' ! (Q t 2, u). Thus speaking generally, by a successionof stepsve find that, if f ! 0 is a predicative function of a sufficiently high order, any assignednonpredicative function of c will be of one of the two forms
(+).r! ($ti, n),(sd).rt ($ri, n), where .F is a predicative function of S ! 0 and r. The nature of the above hierarchy of functions may be restated as follows. A function, as we saw at an earlier stage, presupposesas part of its meaning the totaiity of its values, or, what comes bo the same thing, the totality of its possible arguments. The arguments to a function may be functions or propositions or individuals. (It will be remembered that individuals were defirredas whatever is ncither a propositionnor a function.) For the present we neglect the case in which the argument to a function is a proposition. Consider a function whosc argument, is an individual. This function presupposesthe totality of individuals; bub unless it contains functions as apparent variables, it does not presupposeany totality of functions. If, however, it does contain a function as apparent variable, then ib cannot be defined until some totality of functions has been defined. It follows thai, we must, first define the totality of those functions that have individuals as arguments and contain no functions as apparent variables. These are the predicattiue functions of individuals. Generally, a predicative function of a variable argument is one which involves no totality except that of the possible values of the argument, and those that are presupposedby any one of the possible arguments. Thus a predicative function of a variable argument is any function which can be specified rvithout introducing nerv kinds of variables not necessarily presupposed by the variable which is the argument. A closely analogous treatment can be developed for propositions. Propositionswhich contain no functions and norbpparentvariablesmaybe called el,ementaryprogtositions. Propositions which are not elementary, r'hictr contain no functions, and no apparent variables except individuals, may be called f,rst-ord,er propositions. (It should be observed that no variables except aryo,rent variables can occur in a proposition, since whatever contains a reol variable is a function, not a proposition.) Thus elementary and first-order propositions will be values of first-order functions. (It should be remembered
fll
TEE axrou oF BEDUcrBrrrry
55
t,hrt, o function is not a constituent in orre of its values: thus for erample l,lrofunction '2 is human' is not a constituent of the proposition 'Socratee ix hurnan.') Elementary and first-order propositions presupposeno totality r,xrxrpt(at rnost) the totality of individuals. They are of one or other of the l,ltrrxrforms $ t u ; ( a ) . St a ; ( g a ) . St a , whorr: { ! o is a predicative function of an individual. If follows that, if p rrrprcsentsa variable elementary proposition or a variable first-order propoxition, a function /p is either/({ t r) or f l(a) . Q I u} or J l(go) . { I o}. Thus n lirnction of an elementary or a fimt-order proposition may always be reduced k' t function of a first-order functiorr. It follows that a proposition involving t'lrr:botality of first-order propositions may be reduced to one involving the t,ol,rllity of first-order functions l and this obviously applies equally to higher orrlcrs. The propositional hierarchy can, therefore, be derived from the lirrrotional hierarchy, and we may define a proposition of the nth order as otrrrwhich involves an apparent variable of the n,- lth order in the functional lricrrrchy. The propositional hierarchy is never required in practice, and is orrly relevant for the solution of paradoxes;hence it is unnecessaryto go into lirrt,herdetail as to the types of propositions. YI.
The Ariom of Reilucibilitg.
It rernains to consider the ' axiom of reducibility.' It will be seen that, ru:rxrrding to the above hierarchy, no statement can be made significantly rlrout 'all a-functions,'where a is some given ob.ject. Thus such a notion rrx ' all properties of o,' meaning ' all functions which are true with the rlgument, o,' rvill be illegitimate. We shall have to distinguish the order .l lunction concerned.We can speak of ' all predicative properties of a,' all xr,t:ond-order properties of a,' and so on. (ff a is not an individual, but an ,,bjcct of order n, 'second-ordet properties of :a' will mean 'functions of r)r(lorz*2satisfiedbyo.') But we cannot speak of 'all propertiesof o.' Irr some cases,we can see that some statement will hold of 'all nth-order pnrperties of a,' whatever value n may have. In such cases, no practical lrrrrrnresults from regarding the statement as being about 'all properties of rr.,'provided we remember that it is really a number of statements,and not r single statement which could be regarded as assigning another property to rr, ovcr and above all properties. Such caseswill always involve somesysterrrrt,icambiguity, such as that involved in the meaning of the word 'truth,' rx oxplained above. Owing to this systematicambiguity, it will be possible, xorrrr,tirnes, to combineinto a single verbal statement what are really a number ol rlilfcrent statements, corresponding to different ordert in the hierarchy. 'l'lris is illustrated in the case of the liar, where the statement 'oll .d's xl,rrl,crncnts are false' sbould be broken up into different statementsreferring l,o lrin stabenrents of various orders, and attributing to each the appropriate k irul of falsehood.
56
[culr.
The axiom of reducibility is introduced in order to legitimate a great mass of reasoning,in which, prima facie, we are concernedwith such notions as 'all properties of o'or 'all a-functions,' and in which, neverbheless,it seems scarcely possible to suspect auy substantial error. In order to state the axiom, we must first define what is rneant by ' formal equivalence.' Two functionsf0, rf,? are said to be 'formally equivalent' when,with every possible argument x, $r is equivalent to ^lrr, i.e. $a and,rfrc are either both true or both false. Thus two functionsare formally equivalentwhen they are sabisfied by the sameset of arguments. The axiom of reducibility is the assumption that, given any function {i, there is a formally equivalent predicatiue function, i.e. lhere is a predicative function which is true rvhen {a is true and false when fo is false. In syrnbols,the axiom is: F t (gr/r) : Qn . --n. rl', a. variables, we require a similar axiom, namely: Given any function For two
+@'il'thereis'*;1ffi lit,Tr::l':Ti::,'#'''
In order to explain the purposesof the axiom of reducibility, and the nature of the grounds fol supposingit true, rve shall first illustrate ii by applying it to some particular cases, If we call a pt'ed,icateof an object a predicative funcbionwhich is true of that object,then the prcrlicatesof an object are only someamong its properties. Take for example such a proposition as ' Napoleon had all the qualities that , make a great genenr,l.'Wc rnay interpret this as meaning 'Napoleon had all the predicatesthat, mrke a great general.' Here there is a predicatewhich is an apparent variable. If we put ' f (+ l2)' for ' f I 2 is a predicaterequired in a great general,'our propositionis @) : f@! 2) iurplies { ! (Napoleon). Since this refcrs to a totaliby of predicates, it is not itself a predicate of Napoleon. It b.yno meansfollows,however,that there is not someonepredicate common and peculiar to great generals. In fact, it is certain that there is such a predicate. For the number of great generals is finite, and each of them certainly possessedsome predicate not possessedby any other human being -for exanrplc,the exactinstant ofhis birth. The disjunction ofsuch predicetes will constit,ute a predicate common and peculiar to great generals*. If we call this predicate ,112, the statement we made about Napoleon was equivalent to {r ! (Napoleon). And this eduivalenceholdsequally if we substitute any other individual for Napoleon. Thus we have arrived at a predicatewhich is always equivalent to the property we ascribedto Napoleon, i.e. it belongs to those objects which have this property, and to no others. The axiom of reducibility states that such a predicate always exists, i.e. that any property * When a (finite) set of predicates is given by &ctuol eDumeration, their clisjunction is a predicate, be@us no preilicaCe occurs as eplnrent variable in the diajunction.
il|
THE AXIOM OF REDUCIBILITY
57
.l' trr olrjcct belongs to the same collectiorrof objects as those that possess fi r ) i l to l ) tc ( l i C ate.
W(! fniry next illustrate our principle by its application to id'entit'y. In l,lrin txrrrnection, it hasa certain affinity with Leibniz'sidentity of indiscernibles. I t, is ;rfrrin that, if a and g are identical, and fo is true, then {y is true. Ifere il, rrurrurt rnatter what sort of function Sd may be: the statement must hold firr orrrTfunction. But we cannot say, conversely: 'Il with all values of f, ,fr: inrplies fy, then a a,nd u are identical'; because'all values of {' is iruulrrrissible.If we wish to speak of 'all values of S,' we must confine ,,ursr.lvosto functions of one order. We may confine f to predicates,or to x,rnrrd-orderfunctions,or to functions of any order we please. But we must rrr,r:cssirrily leave out funcbionsof all but one older. Thus rve shall obtain, so 1,.slxrak, a hierarchy of different degrees of identity. We may say ' all the ;rrrrrlicl,tesof r belong lo y,' ' all second-orderproperties of c belong to y,' rrn,l se 61. Each of these staterirents iuplies all its predecessort: for ,,x:rrrrple,if all second-orderproperties of o belong to y, then all predicates .l o bclong Lo y, for to have all the predicatesof z is a second-orderproperty, nrrrlbhisproperty belongsto a. But we cannot,,without the help of an axiom, r lgu(i converselythat if all the predicatesof a belong to u, all the second-orcler pr',rlxrrties of rr must also belong to g. Thus we cannot,,without the help of irrr rrxiorn,be sure that o and g are identical if they have the samepredicates. lr.ilrrriz'sidentityofindiscerniblessuppiiedthisaxiom. It shouldbe observed t,lrnl,by ' indiscernibles' he cannot have meant two objects rvhich agreeas to rrll [heir properties, for one of the properties of o is to be identical with u, rrrrrlbhereforethis property would necessarilybelong to y if t and y agreed rrr rll their properties. Some limitation of the common propertiesnecessary L' rrriLkethings indiscernibleis thereforeimplied by the necessityof an axiorn. liirl purposesof illustration (not of interpreting Leibniz) we may supposethe lr,rnrnonproperties required for indiscernibility to be limited to predicates. 'l'lr.n the identity of indiscernibles will state that if a and y agree as to rrll thcir predicates,they are identical. This can be proved ifwe assumebhe rrri,,rnof reducibility. Eor, in tbat case,every property belongs to the same l.lk'ction of objects as is defined by some predicate. Hence there is some prr.rlicatecolnmon and peculiar to the objects rvhich are identical with o. 'l'lris prcdicatebelongsto o, since c is identical with itself; hence it belongs f,r r7. since y has all the predicates of a; hence y is identical with c. It f,,ll.ws lhat we may d,efinea arrd y as identical when all the predicatesof o l r , , l r r r rtgt i y , 'i . e . w h e n( f ) r f ! n . ).6 1 y. We th e r e fo r ea d o p t th e fo l l o w i n g r l , l i r r i l , i u no f i d e n t i t y *: la= y.::
( f ) ' f ta .) .
$ ! y D f.
' Nol,r'tlrrt in thie alefinitionthe second sign of equatity is to be regarded as combiuiag with ' I rl ' 1,, f(,rm one eymbol; whst is definetl is the sign of equality not followed by the letters ,, Df.'
58
INTRODUOTION
[oner.
But apart from the axiorn of reducibility, or someaxiom equivalent in this connection, we should be compelled to regard identity as indefinable, aud to admit (what seems impossible) that two objects may agree in all their predicates without being identical. The axiom of reducibility is even more essential in the theory of classes. It should be observed,in the first place, that if we assumethe existence of classes,the axiom ofreducibility can be proved. For in that case, given any function {2 of whatever order, there is a class a consisting of just those objects which satisfy f2. Hence '$a' is equivalent to 'a belongs to a.' Bfi, ' n belongs to a ' is a statement containing no apparent variable, and is therefore a predicative function of ,2. flence if we assume the existence of classeg the axiom of reducibility becomes unnecessary. The assumption of the axiorn of reducibility is therefore a smaller assumption than the assumption that there are classes. This latter assumption has hitherto beeu made unhesitatiugly. However,both on the ground of the contradicbions,which require a more comirlicated treatment if classesare assumed,and on the ground that, it is always well to make the smallest assumption requiled for proving our theorems,we prefer to assumethe axiom of reducibility rather than bhe existenceof classes. But in order to explain the nse of the axiom in rlealing with classes,it is necessaryfirst to explain the theory of classes,which is a topic belonging to Chapter III. We therefore postpone to that Chapter the explanationofthe use ofour axiom in dealing with classes. ft is worth while to note that all the purposesserved by the axiom of reducibility nre equallywell servedifwe assumethat there is alwaysa function of the zth order (where n is fixed) which is formally equivalent to f6, whatevermay be tbe order of $h. Here we shall mean by.,a function of the nth order' a function of the zth order relative to the argumentsto f6; thus if these argumentsare absolutelyof the mth order, we assumethe existenceof a function forrnally equivalent to {6 whoseabsolute order is the rl + zr.th.The axiom of reducibility in the form assumedabove takes z:1, but this is not necessaryto the use of the axiom. ft is also unnecessarythat z should be the same for different values of nz; what is necessaryis that roshould be constant so long as aazis constant. What is needed is that, where extensional functions of functions are concerned,we should be able to deal rvith any o-function by means of some formally equivalent funq{ion of a given type, so as to be able to obtain results which would otherwise require the illegitimate notion of 'all a-functions'; butit doesnot matter rvhat the given typeis. It does not appear, however, that the axiom of reducibility is rendered appreciably more plausible by being put in the above more general but more complicated form. The ariom of reducibility is equivalent to the assumption that 'any
ill
TEE AxroM oF REDUcrBrrrTy
59
corrrbinationor disjunction of predicates* is equivalent to a single predicate,' r,a, f,6 61,oassumption that, if we assert that a has all the predicates that xrl,isfy a function (+l 2), there is some one predicate lvhich o will have 'f wlurneverorrr assertion is tnre, and will not have whenever it is false, and ri rnilarly if we assert that r has someone of the predicates that satisfy a function ./ @12). For by meansof this assumption,the orderof a non-predicativefunction rrur be loweredby one; hence,after somefinite number of steps,we shall be able kr gct from any non-predicative function to a formally equivalent predicative lirrrction. It does not seem probable that the above assumption could be nrrbsbitutedfor the axiom ofreducibility in symbolic deductions,since its use wonld require the explicit introduction of the further assumptionthat by a lirrite number ofdownward sbepswe can passfrom any function to a predicative lirrrction,and this assumption'couldnot well be made without, developments l,hlrtare scarcelypossibleat an early stage. But on the above groundsit seems pl:rin that in fact, if the above alternative axiom is true, so is the axiom of rcrlucibility. The converse,which completes the proof of equivalence,is of rnrrrseevident. YII.
Reasonsfor Accepting the Aniom of Red,ucibility. That the axiom of reducibility is self-evidentis a proposition which can lrrr,rrllybe maintained. But in fact self-evidenceis never more than a part of l,lre rcason for accepting an axiom, and is never indispensable. The reason li,r'accepting an axiom, as for accepting any other proposition, is always lrrlgt'lyinductive, namely that nrany propositionswhich are nearly indubitable lirrr be deducedfrom it, and that no equallyplausible way is known by which l,lr,.scpropositionscould be true if the axiom were false,and nothing which is pr',rlrt,blyfalse can be deduced from it. If the axiom is apparently self-evident, llrrrl, only means, practically, that it is nearly indubitable; for things have lrr.r'rrbhoughtto be self-evidentand have yet'turned out, to be false. And if llrrr rxiorn itself is nearly indubitable, that merely adds to the inductive ,,r'irl,'rrce are nearly indnbitable: derived from the fact that its consequences rt,rIr.s not provide new evidenceofa radically different kind. Infallibility is rr.vrrr ;rltainable,and therefore some element,of doubt should always attach I'r ov(iry axiom and to all its consequences.In formal logic, the element of ,l,,rrlrl,is less than in most sciences,but it is noi absent,as appearsfrom the lrrr'1,l,lr:rt the paradoxesfollowed from premisseswhich were not previously [r,,rvn k) require limitations. In the case of the axiom of reducibility, the rrr,lrr,:l,ivc cvidencein its favour is very strong,sincethe reasoningswhich it rr,rrrl p,,r'rrril,s the resultsto which it leads are all such as appear valid. But il, seemsvery improbable that the axiom should turn out to be false, ',lllnrrglr ' I l.rl t.lro oonrbination or disjunction is supposeil to be given iuteusionully. If given erten'r,,r'trlly (i./. by enumeration), no assumption is required; but in tbis case the numbcr ol c r r tr c or nedl m us t be fi ni te. r ,r ,,Il r r r r l ,,r
60
INTRODUCTION
[cuer.
it is by no means improbable that it should be found to be deducible fron, someother more fundamentaland more evident axiom. rt is possiblethat the useof the vicious-circleprinciple,as embodiedin the above hierarchy of types, is more drastic than it need be, and that by a less drastic use the necessity for the axiom might be avoided. Such changes,however,rvould not, render anything falsewhich had been assertedon the basisof the principlesexplained above: they would merely provide easierproofs of the sametheorems. There would seem,therefore,to be but the slenderestground for fearing that the use of the axiom of reducibilitv mav lead us into error. V ttt.'tn' Contrad.ictions. We are now in a position to show how the bheory of types affects bhe solution of the contradictions.whichhave beset mathematical logic. For this purpose,s'e shall begin bv an enumerationof some of the more irnportant and illustrative of these contradictions,and shall then show how they all ernbody vicious-circlefallacies,and are thereforeall avoided by the theory of types. It will be noticed that these paradoxesdo not relate exclusively to the ideas of number and quantity. Accordingly no solution can be adequatewhich seeks to explain them merely as the result of some illegiiimate use of tliese ideas. The solution nust be sought in some such scrutiny of fundarnentallogical ideas as has been attcnrpted in the foregoingpages. (1) The oldest contradiction of the kind in question ts the Eytimenides. Epinrenidesthe Cretnn said that all Cretnns rvereliars, and tll othet statentents made by Cretans u,erecertainly lies. Wtrs this a lie? The simplestform of this contradiction is afforded by thc man who says 'I arn lying'; if he is lying, he is speaking the truth, and vice versa. (2) Let ,rz be the class of all bhoseclassesrvhich are not members of themselves.Then, whatever classc may bc, 'n is a w' is equivalent to ,'a is not an c.' Hence, eiving to a bhe value w, 'w is a w' is equivalent to 'w is not a w.' (3) Let ? be the relation which subsistsbetween trvo relations l? and ,S whenever -E does not have the relation fi to S. Then, whatever relations .E and S may be, '11 has the relation T to S' is equivalent to '-r9does not have the relation -R to S.' llence, giving the value ? to both -ll and ,S, ' ? h a st herelatio n TtoT' is equiv alentt o'?does not hav e t h e r e l a t i o n T to T.' (4) Burali-Forti's contradiction* may be statcd as follows: It can be shownthat every well-orderedserieshas an ordinal number, that the series of ordinals up to nnd including any given ordinal exceedsthe given ordinal by one, and (on certain very natural assumptions)that the seriesof all ordinals (in order of magnit,ude) is well-ordered. It follorvs that the series of.all ' ' UnB questions eui numeri transfiniti,' r r. (189 ? ) . S e e*2 56 .
Rentl,iconti tlel circolo natenntico d,i Palermo, Yol.
rr]
oF coNTBADIorroNs ENUMERATToN
6l
orrlinals has an ordinal number, O say. But in that case the seriee of all ,rdinals including O has the ordinal number O + 1, which must be greater l,h:r,nO. Hence O is not the ordinal number of all ordinals. (5) The number of syllables in the English names of finit€ integers l,cnds to increase as the integers grow larger, and must gradually increase indefinitely, since only a finite number of names can be made with a given linite nurnber of syllables. Hence t,he namesof someintegers must consistof r[ lcasbnineteen syllables,and among thesethere must be a least. Hence ' the lrlwt integer not nalneablein fewer than nineteen syllables' must denote a rlcfirrite integer; in fact, it denotes 111,777. But 'the least integer not rrrrneablein ferver than nineteen syllables' is itself a name consisbingof .ighteen syllables; hence the least integer not nameable in fewer than ninel,r'cnsyllablescan be named in eighteen syllab'les,whioh is a contradiction*. (6) Arnong translinite ordinals some can be defined,while others can not; lirr the total number of possibledefinitions is Not, while the number of translirritc ordinals exceedsNo. Ifence there must be indefinable ordinals, and Bu tth i si sd e fi n e d a s'th e l e a sti n d e fi n a b l e r r r r r o n g t h e s e t h e r e r n u s t b e a l e a st. j. ,'rrlinal,'which is a contradiction (7) Richard'sparadox$is akin to that of the least indefinableordinal. It i' ;w follows: Considerall decimals that cirn be defined by meansof a finite rrrrrnberof words ; lct E be the class of such decimals. Then .E has No termsl lrt,nccits rnemberscan be ordered as the lst, 2nd, 3rd, . ' .. Let ,lf be a number rlt:finc'das follows: If the nth figure in the zth decimal is p, let the nth figrrrc in 1[ be p + 1 (or 0, if p :9). Then y'f is differeni from all the mernbers .1 1/, since,whatever finite value n rnay have, the rzth figure in -ltris different li,rn bhe nthfigure in the rzth of the decimalscomposirrg-E,and therefore-lI ir rliffcrent from the nth decimal. Neverthelesswe have defined Ir in a finite rrrrrnbcrof rvords,and therefore .lVought to be a member of .E Thus il both is rrnrlis not a mernberof Z. In all the above contradictions (which are merely delectionsfrorn an irrrLrlinitenumber) bhereis a common characteristic,which we may describe sclf-rcferenceor reflexiveness. The remark of Epirnenides must include 'rrr rt,s.ll in its orvnscope.If oll classes,provided they are not membersof themru,lr.s,:rre members of tr, this must also apply to w; and'similarly for the ' 'l'lris contradiction was suggesteclto us by Mr G. G. Berry ol the Bodleiau Library. t N,, ir lhc number of finite integets. See *123. I (tl Ki)nig, 'Ueber tlie Grundlagen der Mengonlbhre und das Kontinuumproblem,',lloth. fr r r r ,r /,r rYi , r l . r x r . ( 1905) ; A. C . D i x on, 'O n ' w e l l -o rt l e re d ' & g g re g a t e s , ' P ro c . L o n i l o t t Mq t h . li,rr Hr'rics2, Vol rv. Part r. (1906); aud E. W. Ilobson, 'On the Arithmetic Continuum,' iDid. 'l'1,'.'r,lul,;on ,rlTeredin the last of these papers dependsupon the variation of th€ ' apparatus of ,l.,riilrti,'r.' rntl is thus in outline in agreement with the solutiou adopted here. But it does not tr r r r r l r r l r tr .tl r {)s tatem ent i n the t€x t, i f ' de f i n i t i o n ' i s g i v e n a c o n s t a n t me a n i n g . 1i t il l'.inorrd, ' Les mathematiqueE ct I& logique,' Retue de MatuPhysiqueet de Mot'ale, [fnr lillffi,,.x1u'cirlly rectiotrs Yu. ancl Ir.; also Peano, Seuista ile ilQ,heDratica,Vo]. vrrr. No.5 ( ttx x l ) . l ' I l l l f.
62
INTRODUCTION
[cxer.
analogous relational contrerdiction. In the casesof names and definitions, the paradoxes result from considering non-nameability and indefinability as elements in names and definitions. In the case of Burali-Forti's paradox, the series whoseordinal number causesthe difficulty is the series of all ordinal numbers. In eachcontradictionsomethingis said about oll casesof somekind, and from what is saida new caseseemsto be generated,which both is and is not of the same kind as the casesof which all were concernedin what was said. But this is the characteristicof illegitimate totalities, as we defined them in stating the vicious-circleprinciple. Hence all our contradictionsare illustrations of vicious-circle fallacies. It only remains to show, therefore,that the illegitimate totalities involved are excludedby the hierarchy of types which rve have constructed. (1) When a man says'I am lying,' we may intcrpret his statement as: ' There is a proposition which f am affirming and which is false.' That is to say, he is assertingthe truth of sornevalue of the function 'I assertp, and p is false.' But we sarvthat the word 'false' is ambiguous,and thab, in order to make it unambiguous,rvemust specifyt,heorder of falsehood,or, rvhat comes to the sarnething, the order of the proposition to rvhich falsehoodis ascribed. We sawalso that,ifp is a propositionof the nth order, a propositioninwhich p occursas an apparentvariable is not of the nth order,but of a higher order. Hence the kind of truth or falsehoodwhich can belongto the statement'there is a proposition p which I am affirming and which has falsehoodof the ath order' is truth or falsehoodof a higher order than the nth, Hence the statement of Epimenides does not fall ruithin its own scope,and therefore no contradir:tion emerges. If we regard the statement'I amlying' as a compactway of simnltaneously making all the following statements:'f am assertinga falsepropositionof the first order,'I am assertinga false propositionof the secondorder,'and so on, we find the following curious state of things: As no proposition of the first order is being asserted,the statement 'I am assertinga falseproposition of the first order' is false. This statenrent is of the second order, hence the statement 'I am making a falseFtatement of the secondorder' is true. This is a statement of the third order,and is the only statement of the third order which is being made. Ifence the statement ,'I am making a false statement of the third order' is false. Thus we seefthat the statement .,I arn making a false statement of order 2rt,*l' is false,while the statement,,I arn making a false statement of order 2n' is true. But in this state of things there is no contradiction. (2) In order to solve the contradictionabout the classof classeswhich are not members of themselves,we shall assume,what will be explained in the next Chapter, that a proposition about a class is always to be reducedto a statement about a function which defrnesthe class.i.a, about a function which
Itl
vlcrous-cmoLE FALr,AcrEs
6A
rloesnot satisfyits defining function, and therefore (as will appear more fully irr chapter rrl) is neither a member of itself nor not a member of itself. This 13an jmmld_i1te consequenceof the limitation to the possible arguments to a function rvhich was explained at the beginning of the present cf,apbr. Thus if a is a class,the statement 'a is nob a member of a'ls always meaningless, and there is therefore no sensein the phrase 'the class of those which are not members of themselves.' Hence the contradiction which'lasses results from supposing that there is such a class disappears.
terms of f, and this no function can do, as we saw at the beginning of this chapter. rlence '-E has the relation R to,s' is meaninglesg and the contradiction ceases. (a) The solution of Burali-Forti's contradiction requires some further developments for its solution. At this stage it must suffice to observe i;hal, a series is a relation, and an ordinal number is a class of series. (These statements are justified in the body of the work.) rrence a series of ordinal numbers is a relation between classesof relations, and is of higher type than any of the series which are members of the ordinal numbers in question. Burali-F,orti's 'ordinal number ofall ordinals'must be the ordinal number ofall ordinals of a g'iven type, and must therefore be of higher type than any of these ordinals. Hence it is not one of these ordinals, and ihu.' ir oo contradiction in its being greater than any of them*.
' The solution of Burali-Forti's parador by neans of the theory of types is given in cretail in r256.
64
INTBODUCTION
[cner.
of 'nanres.' It is easy to see that, in virbue of the hierarchy of functions, the theory of types renders a totality of 'names' impossible. We may, in fact, distinguish names of different orders as follows: (o) Elementary narnes rvill be such as ere true 'proper names,' i.a. conventional appellations not involving any description. (b) First-order nameswill be such as involve a description by means of a first-order function; that is to say, if f ! 6 is a firstorder function, 'the term which satisfies $!0' will be a first-order name, though there will not always be an object narnedby this name. (c) Secondorder names will be such as involve a description by means of a second-order function; arnongsuch nameswill be thoseinvolvinga referenceto the totality of first-order names. And so we can proceed through a whole hierarchy. But at no stage canwe give a meaning to the'rvord 'nameable' unlesswe specify the order of namesto be employedI and any name in which the phrase'nameable by namesof order n' occursis necessarilyof a higher order than the ath. Thus the paradox disappears. The solutions of the parzrdoxabout, the least inlefinable ordinal and of Richard's paradox are closely analogous to the auove. The notion of 'definable,'which occurs in both, is nearly the same as 'nanreable,'which occurs in our fifth paradox: 'definable' is whnt 'nameable' becomes when elementary nalr)esare excluded,i.a. 'definable' means 'nameable by a name which is not elementary.' But here there is the same ambiguity as to type a^stherc rvirs bcfore, and the sarne need for the addition of words which specify the type to which the definition is to bclong. And however the type may be specified,'the least ordinal not definableby definitionsof this type' is a definition of a higher type; and in Richard's paradox,when we confineourselves,aswemust,to decirnalsthat have a definition ofa given type, the nunrber 1[, which causesthe paradox,is found to have a definition which belongsto a higher type, and thus not to comewithin the scopeof our previous definitions. An indefinite number of other contradictions,of similar nature to the above seven, can easily be rnanufactured. In all of them, the solution is of the sanrc kind. In all of them, the appearance of contradiction is produced by the presenceof some word which has systematic arnbiguity of type, suclr zts truth, fulsehood, fu,nction, property, class, relation, cardinal, ordinul, unrc, d,ef,nition. Any such word, if its typical ambiguity is overlookcd,will apparentiy generate a totality containing members defined in terms of itself, and will thus give rise to vicious-circle fallacies. In most cases,the conclusionsof arguments which involve vicious-circle fallacies but whereverwe have an illegitimate totality, will not be self-contradictor-v, a little ingenuity rvill enable us to construct a vicious-circlefallacy leading to a contradiction, which disappearsas soon as the typically ambiguous words are rendered typically definite,i.e. are determined as belonging to this or thal, type.
rr|
vlcrous-crRcr,ErArrrAcrEs
6b
'l'hus the appearanceof contradiction is ahvaysdue to the presenceof words r,rrrlxxlyinga concealedtypical ambiguity, and the solution of the apparent r,,rrl,rrdictionlies in bringing the concealedambiguity to light. lrr spite of the contradicbionswhich result from unnoticed typical rurrbiguity, it is not desirable to avoid words and symbols which have l,ypi<:*lambiguity, Such rvords and symbols embrace practically all the rrl,'^s rvith which mathematics and mathematical logic are concerned:the nyrrl,t:r'ticambiguityis the result of a systematicanalogy. That is to say,in all the reasoningsrvhich constitute mathematics and mathematical ^lrrr.st Irgir:,we are using ideas which may reccive any one of an infinite number of rlrllirlcnt typical detenninations,any one of which leavesthe reasoningvalid. 'l'lrrrsby employingtypically ambigrrouswordsand symbols,we a,reable to make ,,rrrrt:hrin of reasoningapplicableto any one of an infinite number of different ,'tscs,which would not be possibleif we were to forego the use of typically rrrrrbiguous words and symbols. Arnong propositions wholly expressedin terms of typically ambiguous rr,l,i'ns practically the only oneswhich may differ,in respectof truth or falseh,xxl,:r,ccording to the typical deternination which they receive,are existencellrr,,r.ms If we assumethat the tot'l number of individuals is ru,then the I.l'r1l1,,rn6'rofclassesofindividualsis2',thetotalnumberofclassesofclasses ,rl irrrlivid.als is 2z',and so on. Irere a maybe either finite or infinite,and in ,,il,ltlr case2' > n. Thus cardinalsg,:eaterthan n but not greater than 2' exist ru nlrPliedto classesofclasses,but not as applierl to classesof individuals, so tlrrtl,rvhatevermay be supposedto be the number of individuals,there will bc . r rrl,cnce-theorems which hold for higher types but not for lower types. Even Ir, r,', however,so long as the n.mber of individuals is not asserted,but is rrr,'r','lyir.ssumed hypothetically,rvemayreplace the type of individuals by any ,t,lr*r' t,ype,provided we make a correspondingchange in all the other types ,'r'r'rrrrirrg in the samecontext. That is, we rnay give the name',relativein,lrvrrlrrals'to the members of an arbitrarily chosen type r, and the narne ' r,,lrrl,iveclassesof individuals' to classesof,,relative individuals,,,and so on. 'l'lrrrxs. long as only hypotheticalsa.e concerned, in which existence-theorems li'r' ,rr. bypeare shown to be implied by exisbence-theorems for another,only r r'lrr/ rr. types are relevant even in existence-theorems.Thi s applies also to case-s rvfr,'rr'l,lrc hypothesis(and therefore the conclusion)is usserted,, provided the ,r'r'r.r'l,i(). holds for any type, however chosen. For example,any type has at l,'rrnl... rncmber; henceany tlae rvhichconsistsof clnsses,ofwhateverorder, lrrrrrrrt,l.rrst two members. But the further pursuit of thesetopics must be left t,' l,lr,,lrrrly of the work.