Documente Academic
Documente Profesional
Documente Cultură
I
I
I
I
I
I
I
I
I
Austiia
Italy
u
u
u
u
u
u
u
u
u
Pluto
Venus
Meicuiy
Tis ielation is iefexive on the set Mais, Pluto, Venus, Meicuiy}; it is
also symmetiic. It fails to be tiansitive since diiect aiiows aie missing,
foi instance, fiom Mais to Venus. Te ielation is not asymmetiic oi
antisymmetiic since theie aie paiis of objects such as Mais and Meicuiy
that have aiiows going back and foith between them .
Both ielations discussed so fai aie not equivalence ielations, as they
aie not tiansitive.
Te ielation has some peculiai piopeities: its diagiamis empty. It is
iefexive on the empty set , but on no othei set. It is symmetiic, as theie
is no aiiow foi which theie is not any aiiow in the opposite diiection.
But it is also asymmetiic and antisymmetiic because theie is no aiiow
foi which theie is an aiiow in the opposite diiection. is also tiansitive.
Consequently, is an equivalence ielation on .
Te ielation with the diagiam below is not iefexive on the set with
the two elements Ponte Vecchio and Eifel Towei, because theie is no loop
attached to Eifel Towei.
Eifel Towei
Ponte Vecchio
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
i
In this diagiam, theie aie aiiows fiom Fiance, Italy, England, and
the United Kingdom. Te set containing Fiance, Italy, England and the
Volker Halbach :oo8/:oo
: Sets, Relations, and Arguments :
United Kingdom is called the domain of the function. Te names of the
thiee cities ieceive aiiows; the set of these thiee cities is called the iange
of the function.
uiiii1io :..
(i) Te domain of a function R is the set d. there is an e such that
d, e R}.
(ii) Te range of a function R is the set e. there is a d such that d, e
R}.
(iii) R is a function into the set M if and only if all elements of the range
of the function are in M.
Te elements of the domain seive as inputs oi aiguments of the
function; the elements of the iange aie outputs oi values.
In the above example the set containing Fiance, Italy, England and the
United Kingdom is the domain of the function, while the set with Paiis,
Rome and London as its elements is the iange. Accoiding to (iii) of the
above defnition, the function is a function into the set of all Euiopean
cities, foi instance.
uiiii1io :.,. If d is in the domain of a function R one writes R(d} for
the unique object e such that d, e is in R.
Te ielation containing all paiis d, e such that d has e as a biological
mothei is a function: if d has e as biological mothei and d has f as
biological mothei, then e and f must be identical. Its domain is the set
of all people and animals, its iange the set of all female animals with
ofspiing.
In contiast, the ielation containing all paiis d, e such that d is the
biological mothei of e is not a function: my biothei and I have the same
biological mothei, yet we aie not identical.
Volker Halbach :oo8/:oo
: Sets, Relations, and Arguments :,
:. o-vi:vv vii:1ios
Te ielations I have consideied so fai aie binaiy; they containonly oideied
paiis. Expiessions such as d loves e expiess binaiy ielations; the expies-
sion d loves e expiesses the ielation that contains exactly those oideied
paiis d, e such that d loves e. In contiast, the expiession d piefeis e
ovei f expiesses a teinaiy (3-place) ielation iathei than a binaiy one. In
oidei to deal with teinaiy ielations, oideied tiiples (oi tiiples foi shoit)
aie used. Tiiples aie veiy much like oideied paiis.
A triple d, e, f is identical with a triple g, h, i if and only if they
agree in the frst, second and third component, respectively, that is, if and
only if d = g, e = h and f = i.
,
Ternary relations are sets containing only triples.
Besides oideied paiis and tiiples theie aie also quadiuples and so on.
Tis canbe geneialisedto evenhighei aiities n: an n-tuple d
I
, d
2
, . . . , d
n
In oidei to use a language competently, one must mastei all thiee aspects
of it.
Syntax is conceined with the expiessions of a language baie of theii
meanings. In the syntax of a language it is specifed what the woids oi
sentences of the language aie. In geneial, the giammai of a language
belongs to the syntax of that language, and ofen syntax is identifed with
giammai. In oidei to use the language competently, one must know the
giammai of the language. In paiticulai, one must know how to foim
sentences in the language.
Semanticssemantics may be desciibed as the study of the meanings
of the expiessions of a language. Cleaily, to use a language one must not
only know what the woids and the sentences of the language aie; one
must also know what they mean.
Te expiession Im Mondschein hockt auf den Gibein eine wild
gespenstische Gestalt is a well-foimed Geiman declaiative sentence. In
a syntactic analysis of that sentence one may iemaik that hockt is a veib
in piesent tense, and so on. All this is meiely syntactic infoimation; it
does not tell one anything about the meaning of that sentence. In oidei
to undeistand the sentence, you need infoimation about meaning. Foi
instance, it is a semantic fact of Geiman that im Mondschein means in
the moonlight.
Te tiichotomy was intioduced by Moiiis (:,8).
Volker Halbach :oo8/:oo
: Sets, Relations, and Arguments :,
Te thiid component, piagmatics, will not be studiedheie. Piagmatics
is, ioughly speaking, the study of language in use. Assume John calls
Maiy and asks hei whethei she wants to come along to the cinema. She
ieplies, I am ill. Obviously, John should not expect Maiy to come along,
but the sentence I am ill does not mean the same thing as I dont want
to come along to the cinema; the foimei sentence only says something
about Maiys health. But utteied by Maiy in this paiticulai situation, the
sentence I am ill, spoken by Maiy, conveys the infoimation that she will
not join John. Tus, John needs piagmatics in oidei to undeistand that
Maiy is not going to come along. Puie semantics would not tell him.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Piopositional Logic
In this section I shall intioduce the language of piopositional logic. All
othei foimal languages that will be discussed in this manual aie based on
the language of piopositional logic.
Befoie intioducing the syntax of this language I will biiefy outline a
method foi talking emciently about the expiessions of a language and foi
desciibing the syntax of a language. Te method is by no means specifc
to the language of piopositional logic.
:. : qUo1:1io
By enclosing an expiession in quotation maiks one can talk about that
expiession. Using quotation maiks one can say, foi instance, that A is
the fist lettei of the alphabet and that Gli enigmi sono tie is an Italian
sentence. Te quotation of an expiession is that veiy expiession enclosed
in quotation maiks.
:
Quotation maiks allow one to designate single expiessions. Desciib-
ing the syntax of a language usually makes it necessaiy to talk about a
laige oi infnite numbei of expiessions. Foi instance, one would like to
be able to state that one can constiuct new sentences in English by com-
bining sentences using and (ignoiing capitalisation and punctuation).
: Cappelen and LePoie (Spiing :oo,) piovide an oveiview of the intiicacies of quotation
and of pioposed theoiies. A classical text on quotation is Quine (:o).
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic :,
Logicians would expiess that geneial claim by saying the following:
(:u) If and aie English sentences then and is an
English sentence.
It is iaining is an English sentence. If one takes It is iaining as both
and in the above iule, then the iule says that It is iaining and it is
iaining is also an English sentence (again we ignoie the absence of the
full stop and the missing capitalisation). One can then use It is iaining
and it is iaining as again and It is iaining as to conclude fiom the
iule that also It is iaining and it is iaining and it is iaining is an English
sentence. In this way one can constiuct longei and longei sentences and
theie is no limit to the iteiations.
I think that (:u) is faiily stiaightfoiwaid and should be easy to
undeistand. Teie is, howevei, something puzzling about it as well: the
pait of (:u) claiming that and is an English sentence is decidedly
not about the expiession in quotation maiks. Te letteis and aie
Gieek letteis, and the expiession with as fist symbol, followed by a
space, followed by and and anothei space, followed by , is defnitely
not an English sentence. Only once and aie ieplaced by English
sentences does and become an English sentence.
Te Greek letters used in this way are metavariables or metalinguistic
variables.
Tus, the above iule may also be expiessed in the following way:
AnEnglishsentence followedby and (inspaces) andanothei
oi the same English sentence is also an English sentence.
Tis way of iephiasing (:u) does not iely on quotation maiks but on
talking about expiessions following one anothei. Tis method is peihaps
safei than using (:u) with its quotation maiks and metavaiiables, but it
is also moie cumbeisome when applied to intiicate giammatical iules.
Tus, I will piesent defnitions in the style of (:u) iathei than talking
about expiessions following one anothei.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic :8
Logicians haidly evei use the expiessions of foimal languages in
the way they use the expiessions of theii mothei tongue, but they ofen
talk and wiite about the expiessions of these foimal languages. Since
the expiessions of the foimal languages they aie conceined with difei
fiom expiessions of English, logicians usually diop the quotation maiks.
Instead of saying
(P (QR}} is a sentence of the language of piopositional
logic
they say,
(P (QR}} is a sentence of the language of piopositional
logic.
I will follow this convention and usually diop quotation maiks aiound
the expiessions of foimal languages in this manual. Tis also applies to
expiessions containing metavaiiables.
:. : 1ui sv1:x oi 1ui i:cU:ci oi vvovosi1io:i iocic
Now I can desciibe the syntax of the language L
I
of piopositional logic.
uiiii1io :.: (si1ici ii11ivs). P, Q, R, P
I
, Q
I
, R
I
, P
2
, Q
2
, R
2
, P
3
,
Q
3
, R
3
, and so on are sentence letters.
Using metavaiiables I will defne the notion of a sentence of the lan-
guage L
I
of piopositional logic.
uiiii1io :.: (si1ici oi L
I
).
(i) All sentence letters are sentences of L
I
.
(ii) If and are sentences of L
I
, then , ( }, ( }, ( }
and ( } are sentences of L
I
.
(iii) ^othing else is a sentence of L
I
.
Given what I have said about metavaiiables, (ii) implies that ( }
becomes a sentence of the language of piopositional logic when the Gieek
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic :
letteis and have been ieplaced by sentences of the language of
piopositional logic. Te Gieek letteis and themselves aie not
expiessions of the language L
I
.
As I explained on page :,, I could have foimulated pait (ii) of Defni-
tion :.: without using the metavaiiables and by expiessing (ii) in
the following way:
Te negation symbol followed by a sentence of L
I
is again a
sentence of L
I
. Te symbol ( followed by a sentence of L
I
,
followed by the symbol (oi , , ), followed by a
sentence (not necessaiily distinct fiomthe fist one), followed
by the symbol }, is a sentence of L
I
.
I hope that (ii) is not only shoitei but also easiei to giasp.
Te last clause in Defnition :.: says that only expiessions one can
obtain by using clauses (i) and (ii) aie sentences. Veiy ofen this last clause
is omitted and the clauses (i) and (ii) aie implicitly taken to be the only
means of aiiiving at sentences. At vaiious points in this manual I will
piovide defnitions that aie similai to Defnition :.:. In those cases I will
diop the analogues of clause (iii) foi the sake of simplicity.
Logicians also say the negation of iathei than . In this termi-
nology, is the negation of , and similarly ( } is the conjunction of
and , and is the disjunction of and . Te sentence (P Q},
foi instance, is the negation of (P Q}.
ix:mvii :.,. Te following expiessions aie sentences of the language L
I
:
((P P} Q
436
},
(R (P (P
3
Q
4
}}},
((P Q
4
} P}.
In the next example I show how to piove that the last sentence above
is indeed a sentence of L
I
.
ix:mvii :.. By Defnition :.:(i), P is a sentence of L
I
. Tus, by (ii),
P is also a sentence of L
I
. By (i) again, Q
4
is a sentence of L
I
. By (ii) and
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,o
by what has been said so fai, (PQ
4
} is a sentence, and by (ii) again and
by what has been said so fai, ((P Q
4
} P} is also a sentence of L
I
.
Te symbols , , , , are called connectives. Tey ioughly coi-
iespond to the English expiessions not, and, oi, if . . . , then . . . and if
and only if , iespectively.
name in English symbol alternative
used here symbols
conjunction and ., &
disjunction or +, v
negation it is not the ,
case that
arrow if . . . then ,
(material
implication,
conditional)
double arrow, if and only if ,
(biconditional
material
equivalence)
Te names in biackets and the symbols in the iightmost column aie used
by some othei authois; they will not be used heie.
Te expiessions in the in English column indicate how the connec-
tives aie commonly iead, iathei than theii piecise meanings.
:. , vUiis iov uvovvic vv:cxi1s
A sentence with many biackets can be confusing. Foi convenience I shall
employ ceitain iules foi diopping biackets. Tese iules aie not ievisions
of the defnition of a sentence and they do not foim pait of the omcial
syntax of the language L
I
of piopositional logic. Tese iules aie meie
conventions that allow one to wiite down abbieviations of sentences
instead of the sentences themselves in theii omcial foim.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,:
Most logicians employ at least some of these iules. Foi instance, haidly
anyone wiites (P Q} instead of P Q. Tis, then, is the fist iule:
vv:cxi1ic covi1io :. Te outer brackets may be omitted from a
sentence that is not part of another sentence.
Foi instance, one may wiite P (Q P} instead of (P (Q P}}.
Howevei, this convention does not allow one to diop any biackets fiom
(P (Q P}}, because the sentence (P (Q P}} is wiitten heie as
a pait of the sentence (P (Q P}}.
Heie a waining is in oidei: Te syntactic defnitions in Section :.:
apply to L
2
-sentences but not to theii abbieviations. Foi instance, I have
defned the negation of the sentence as . Now one might think that
P Q is the negation of P Q. Tis is not the case, howevei. P Q is
shoit foi (P Q} accoiding to Convention :; and the negation of (P Q}
is (P Q} and not P Q, which is shoit foi (P Q}.
vv:cxi1ic covi1io :. Te inner set of brackets may be omitted
froma sentence of the form((}}. An analogous convention applies
to .
One may abbieviate ((P Q
2
} P
2
} as (P Q
2
} P
2
by Conven-
tion :, and then one may also diop the iemaining paii of biackets by
Convention :; so P Q
2
P
2
is shoit foi ((P Q
2
} P
2
}.
(( } } may be pait of a laigei sentence. So, using Conventions
: and :, one can abbieviate the sentence
((((P
2
P
3
} Q} R} ((P
2
P
3
} Q}}
with the following expiession:
(P
2
P
3
Q R} (P
2
P
3
Q}.
In aiithmetic one may wiite 3 3 + 4 instead of (3 3} + 4, because
the symbol foi multiplication binds moie stiongly than the symbol +
foi addition. Analogously, theie aie conventions foi oideiing logical
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,:
opeiations: and bind moie stiongly than oi . And just as in the
case of aiithmetic, this yields fuithei conventions foi diopping biackets.
Foi instance, the sentence P QR is in abbieviated foim. In oidei
to decide whethei the sentence is an abbieviation of ((P Q} R} oi
of (P (Q R}}, imagine that the connectives and compete foi
the sentence lettei Q. Since binds moie stiongly than , gains the
uppei hand and (P (Q R}} is the coiiect ieading.
In moie abstiact teims, this iule foi diopping biackets can be ex-
piessed as follows:
vv:cxi1ic covi1io ,. Assume , , and are sentences of L
I
, is
either or , and is either or . Ten, if (( }} or ((} }
occurs as part of the sentence that is to be abbreviated, the inner set of
brackets may be omitted.
Afei diopping the outei biackets fiom ((P Q} R) accoiding
to Convention :, one may shoiten the sentence fuithei to P Q R
since binds moie stiongly than . Similaily (P ((Q P
3
} R}}
may be abbieviated as (P (Q P
3
} R}. In ((P Q} Q}
only the outei biackets can be omitted in viitue of Convention :, so
that (P Q} Q is obtained as an abbieviation. One cannot use
Convention , to obtain P Q Q, because the oiiginal sentence does
not contain ((P Q} Q} as a pait. Intuitively, in P Q Q the
negation symbol iefeis only to P and not to the sentence (P Q} as it
does in the oiiginal sentence.
Teie is no deepei ieason foi the choice of these thiee conventions.
It is nevei incoiiect to use the unabbieviated sentences of L
I
with all
theii biackets iathei than theii abbieviations. In situations wheie the
application of the biacketing conventions can give iise to confusions,
it is bettei not to use them. Also it is peifectly legitimate to apply the
iules selectively. Foi instance, one may apply only Convention :, but not
Convention ,, and wiite (P Q} R foi ((P Q} R}. I will apply
these conventions extensively.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,,
:. 1ui sim:1ics oi vvovosi1io:i iocic
In Section :., I gave a chaiacteiisation of the logical validity of aiguments
in English. In this section I will defne validity foi aiguments in the
language L
I
of piopositional logic. Te infoimal Chaiacteiisation :.,
of validity foi English aiguments will be adapted to the language L
I
of
piopositional logic and theieby be tiansfoimed into a piecise defnition.
In oidei to defne logical validity foi the language L
I
, the notion of
an inteipietation foi the language L
I
needs to be made piecise. Fiist, I
need to say which expiessions can be inteipieted in difeient ways and
which aie always inteipieted in the same way. Te connectives aie logical
symbols, the biackets meiely auxiliaiy symbols; logical and auxiliaiy
symbols cannot be ie-inteipieted (insofai as one can speak of auxiliaiy
symbols being inteipieted at all). Te sentence letteis aie the only non-
logical symbols of L
I
; they can be inteipieted in difeient ways.
Te inteipietations of the sentence letteis will be piovided by so-called
L
I
-stiuctuies. Tese L
I
-stiuctuies need only piovide enough infoimation
to deteimine whethei a sentence is tiue oi false. Now, undei which
conditions is the sentence P Q tiue: If the connective functions
like and in English, then both P and Q must be tiue foi P Q to be
tiue; otheiwise P Q will be false. Similaily, since woiks like not, the
sentence R is tiue if and only if R is false. As coiiesponds to oi, the
sentence P Q is tiue if and only if P oi Q is tiue (oi both aie tiue).
Te aiiow coiiesponds ioughly to the English if . . . then, but the
lattei has a iathei complicated semantics. Te L
I
-sentence P Q is false
if and only if P is tiue and Q is false; otheiwise it is tiue. Te phiase if . . .
then, which coiiesponds to the aiiow, does not always behave like this.
How the aiiow is ielated to if . . . then will be discussed in Section ,.:.
Geneially, in L
I
, the tiuth oi falsity of a sentence of L
I
depends only
on the tiuth oi falsity of the sentence letteis occuiiing in the sentence;
any fuithei infoimation is not ielevant. Teiefoie, L
I
-stiuctuies need
only deteimine the tiuth and falsity of all sentence letteis.
Instead of saying that a sentence is tiue, logicians say that the sentence
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,
has the tiuth-value Tiue. Tis sounds like a philosophically piofound
move since new objects aie iequiied: tiuth-values. Howevei, tiuth-values
aie haidly mysteiious objects. Some moie mathematically minded logi-
cians use the numbei I foi the tiuth-value Tiue and 0 foi the tiuth-value
False. In the end it matteis only that Tiue and False aie distinct. It is
possible, although not veiy customaiy and technically less convenient,
to develop the semantics of L
I
without tiuth-values by saying that a sen-
tence is tiue (oi false) instead of saying that it has the tiuth-value Tiue (oi
False). I shall use the lettei T as a name foi the tiuth-value Tiue and F
foi the tiuth-value False.
Tus an L
I
-stiuctuie piovides inteipietations of all sentence letteis
by assigning to eveiy sentence lettei exactly one tiuth-value, T oi F.
uiiii1io :., (L
I
-s1vUc1Uvi). An L
I
-structure is an assignment of ex-
actly one truth-value (T or F) to every sentence letter of the language L
I
.
:
One may think of an L
I
-stiuctuie as an infnite list that piovides a
value T oi F foi eveiy sentence lettei. Te beginning of such a list could
look like this:
P Q R P
I
Q
I
R
I
P
2
Q
2
R
2
T F F F T F T T F
Staiting fiom the tiuth-values assigned to the sentence letteis by an
L
I
-stiuctuie A, one can woik out the tiuth-values foi sentences con-
taining connectives in the following way. Te shoitest sentences aie the
sentence letteis; theii iespective tiuth-values aie fxed diiectly by the
L
I
-stiuctuie A. Foi instance, P could be assigned the tiuth-value T and
R could be assigned the same tiuth-value. In this case PR would ieceive
the tiuth-value T, too. If P
I
is given the tiuth-value F by A, the sentence
P
I
(P R} gets the tiuth-value T, because P R is tiue; and s being
tiue is sumcient to make a sentence tiue.
: In moie mathematical teims, an L
I
-stiuctuie is a function into the set {T, F} with the
set of all sentence letteis of L
I
as its domain.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,,
Tus, the tiuth-values of the shoitest sentences, that is, of the sentence
letteis, aie fxedby the L
I
-stiuctuie A, andthenthe tiuth-values foi longei
sentences aie deteimined successively by the tiuth-values of the sentences
they aie made up fiom.
I will wiite ||
A
foi the tiuth-value of that is obtained on the basis
of A; it is deteimined by Ain the following way:
uiiii1io:.o (1vU1ui:L
I
-s1vUc1Uvi). Let Abe some L
I
-structure.
Ten | . . . |
A
assigns to every sentence of L
I
either T or F in the following
way.
n
and are L
I
-sentences, is short for
I
, . . . ,
n
} . So the claim of
Example :.:, may be wiitten P Q, Q P.
Fiist I diaw a tiuth table foi the piemisses and the conclusion in the
following way:
P Q P Q Q P
T T T F F T T F T
T F T T T F F F T
F T F T F T T T F
F F F T T F F T F
(:.:)
Now I will check whethei theie is any line in which the entiies in the
main columns of the piemisses all have Ts, while the conclusion has an F.
In the fist line of tiuth-values the fist piemiss ieceives an F, the second
a T, and the conclusion an F. Te second and fouith lines also have Fs
foi one piemiss. Te thiid line has Ts foi both piemisses, but also a T foi
the conclusion. Tus, theie is no line wheie all piemisses ieceive Ts and
the conclusion an F. Teiefoie, the aigument is valid, that is, P follows
fiom P Q, Q} oi, foimally, P Q, Q P.
Foi fnite sets of piemisses, one can ieduce the pioblem of checking
whethei to the pioblem of checking whethei a single sentence is
logically tiue. To do this one combines all of the piemisses, that is, all
sentences in , using , and then one puts the iesulting conjunction in
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic :
fiont of an aiiow followed by the conclusion. Te iesulting sentence is
logically tiue if and only if the aigument is valid. Tis can be expiessed
moie succinctly as follows:
1uiovim:.:.
I
, . . . ,
n
if and only if
I
. . .
n
is a tautology
(that is, ip
I
. . .
n
is logically true).
I do not give a pioof of this theoiem heie.
I will apply Teoiem :.: to the example above: Fiist, the two pie-
misses aie combined into (P Q} Q. It is necessaiy to ieintioduce
the biackets aiound P Q because otheiwise the iesult would be an
abbieviation foi P (Q Q} as binds moie stiongly than . Next,
the aiiow is put between this conjunction and the conclusion. Tis yields
((P Q} Q} P. Te biackets aiound the conjunction of the two
piemisses aie not ieally necessaiy since binds moie stiongly than ,
but they might make the sentence easiei to iead. Te tiuth table foi this
long sentence looks like this:
P Q ((P Q} Q} P
T T T F F T F T T F T
T F T T T F F F T F T
F T F T F T T T T T F
F F F T T F F F T T F
(:.:)
Tus, the sentence ((P Q}Q} P is valid, that is, it is a tautology.
By Teoiem :.:, it follows that P Q, Q P. Of couise, we know
this alieady fiom tiuth table (:.:).
Diawing tiuth tables foi aiguments oi sentences with many sentence
letteis is cumbeisome: eveiy new sentence lettei doubles the numbei of
lines of the tiuth table, because foi any alieady given line two possibilities
must be consideied: the new sentence lettei can have tiuth-value T oi F.
Tus, a sentence oi aigument with I sentence lettei iequiies only 2 lines,
one with 2 difeient sentence letteis iequiies 4, one with 3 iequiies 8 lines,
and so on. Geneially, the tiuth table foi a sentence oi aigument with n dif-
feient sentence letteis will have 2
n
lines of tiuth-values. In Exeicise ,.o
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,
theie will be an aigument in L
I
with 7 sentence letteis. Tus, wiiting
down the coiiesponding tiuth table would iequiie 2
7
= I28 lines.
To show that an L
I
-sentence is a tautology, one does not need to diaw
a complete tiuth table. One only needs to show that theie cannot be a
line in the tiuth table that yields an F in the main column. In oidei to
iefute the existence of such a line foi Example :.:,, the best stiategy is to
stait with the assumption that the value in the main column is F:
P Q ((P Q} Q} P
F
A sentence of the foim has the tiuth-value F only if has tiuth-
value T and has tiuth-value F. Tus, I must have:
P Q ((P Q} Q} P
T F F
I can continue to calculate tiuth-values backwaids:
P Q ((P Q} Q} P
T T T F F T
I wiite the calculated tiuth-values also undei the fist two occuiiences of
P and Q:
P Q ((P Q} Q} P
T T : T T T F F T
But now theie is no way to continue. Te slot maiked with a question
maik cannot be flled with a tiuth-value: theie should be an F undei the
negation symbol , as Q has tiuth-value T, but theie should also be a T,
because (P Q} and P have Ts. It follows that theie cannot be a line
with an F in the main column. Teiefoie, in the full tiuth table with all
the lines, all tiuth-values in the main column aie Ts. Tis pioves again
that ((P Q} Q} P is a tautology.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic
Since it is not easy (foi an examination maikei, foi instance) to iecon-
stiuct how the tiuth-values have been calculated, it is useful to iecoid the
oidei in which the values weie obtained:
P Q ((P Q} Q} P
T
6
T
3
: T
7
T
I
T
4
F F
2
T
3
Of couise I could have wiitten down the tiuth-values in a difeient oidei.
Foi instance, afei aiiiving at F
2
, I could have calculated the value in the
last column and only then have tuined to the pait pieceding .
Now I will use the same method to show that (P Q} P is not
a tautology. As befoie, an F is wiitten in the main column:
P Q (P Q} P
F
Te following table iesults fiom the fist backwaids-calculation:
P Q (P Q} P
T F F
Tus, P must ieceive the tiuth-value T:
P Q (P Q} P
T F F T
Tus one can also wiite a T undei the fist occuiience of P:
P Q (P Q} P
T T F F T
Since P Q has the tiuth-value T and P also has the tiuth-value T, the
sentence Q ieceives a T, and Q, accoidingly, an F. Hence, the line can
be completed as follows (I will also inseit the obligatoiy indices):
P Q (P Q} P
T F T
4
T
I
T
3
F
6
F F
2
T
3
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,
At any rate, when one has arrived at a possible line one should cal-
culate the truth-values from bottom to top (starting from the truth-values
that have been obtained for the sentence letters) to ensure that one has not
missed a column that cannot given a unique truth-value. Only once this
fnal check has been caiiied out, one knows that the line obtained is a
possible line in a tiuth table.
Te above backwaids-calculations shows that the sentence (P
Q} P has tiuth-value F if P has the tiuth-value T and Q has the
tiuth-value F.
Technically speaking, if A(P} = T and A(Q} = F foi a stiuctuie A,
then (P Q} P is false in A. So, by Defnition :.,(i), (P Q}
P is not logically valid, that is, it is not a tautology.
Sometimes this method of calculating tiuth-values backwaids ie-
quiies moie than one line. Tis is the case in the following example:
P Q (P Q} (Q P}
F
If a sentence is false, theie aie two possibilities: could have
tiuth-value T and tiuth-value F, oi, could have F and could have
tiuth-value T. As such, one has to take these two possibilities into account:
P Q (P Q} (Q P}
I T F F
2 F F T
I have undeilined the tiuth-values that cannot be uniquely deteimined,
and so moie than one possibility (line) needs to be checked.
Te iest is ioutine. Te indices in the table below indicate the oidei
in which I aiiived at the tiuth-values. Te oidei in which the values
aie calculated does not ieally mattei, but the indexing makes it easiei to
follow the ieasoning.
P Q (P Q} (Q P}
I F
3
T
I
: T
6
F T
3
F
2
F
4
2 F
3
F
I
F
4
T
3
F T
6
T
2
:
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic o
Neithei of the two lines can be completed. Tis shows that (P Q}
(Q P} is a tautology.
Of couise it can happen that moie lines aie iequiied and that difeient
cases undei consideiation have to be split up into fuithei subcases.
Te method of backwaids-calculation can also be applied in oidei to
check whethei an aigument is valid oi not. To show that an aigument
is not valid, one has to fnd a line (that is, a stiuctuie) wheie all of the
piemisses aie tiue and the conclusion is false. If theie is no such line,
the aigument is valid. Heie is an example of how to use the method to
deteimine whethei an aigument is valid. I have picked an example that
foices me to considei seveial difeient cases. So, I want to deteimine
whethei
P Q, (P
I
Q} (P P
I
} (P Q} P
I
.
I will stait by wiiting the two piemisses and the conclusion in one table.
I have to check whethei theie can be a line in the table wheie the two
piemisses come out as tiue while the conclusion is false. As such, I should
stait by wiiting Ts in the main columns of the piemisses and an F in the
main column of the conclusion:
P Q P
I
P Q (P
I
Q} (P P
I
} (P Q} P
I
T T F
Now I have to make a case distinction: the fist sentence could be tiue
because P is false oi because Q is tiue. Similaily, in the case of the othei
sentences, theie is no unique way to continue. Given that I can make a
case distinction with iespect to any of the thiee sentences, it is not cleai
how to pioceed. But some ways of pioceeding can make the calculations
quickei and less complicated. It is useful to avoid as much as possible
picking a sentence that will iequiie a new case distinction in the next
step.Ultimately though, so long as all possible cases aie systematically
checked, the oidei in which one pioceeds will not afect the end iesult.
At this stage in the calculation, a case distinction cannot be avoided,
and so I will pick the last sentence: (P Q} P
I
can be false eithei
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic ,
because P Q is false oi because P
I
is false. I will tiy to complete the
line foi the lattei case and leave the foimei foi latei:
P Q P
I
P Q (P
I
Q} (P P
I
} (P Q} P
I
I T T F
I
F
2 T F
3
F
2
T
4
T : F
3
F F
I
Te second line of the table cannot be completed: the second piemiss
must be tiue, but it follows fiom my assumption that both (P
I
Q}
and P P
I
must be false. Tis means that the iules tell me to wiite an F
in the slot maiked by :, because P
I
is false, but they also tell me to wiite
a T foi :, because the entiie sentence (P
I
Q} (P P
I
} is tiue and
(P
I
Q} is alieady false.
Line : is moie complicated, because P Q can be false foi two
ieasons: fist, P can be tiue while Q is false oi, second, P can be false while
Q is tiue. Tus, I need to distinguish the subcases maiked I.I and I.2:
P Q P
I
P Q (P
I
Q} (P P
I
} (P Q} P
I
I.I T
3
T : T T
4
T
2
F
I
F
3
F
2 T F
3
F
2
T
4
T : F
3
F F
I
I.2 T T F
2
F
I
T
3
F
Since P is tiue and P Q is tiue, Q has to be tiue as well. But accoiding
to the assumption, Q is false. Teiefoie, line I.I cannot be completed.
Only case I.2 iemains:
P Q P
I
P Q (P
I
Q} (P P
I
} (P Q} P
I
I.I T
3
T : T T
4
T
2
F
I
F
3
F
2 T F
3
F
2
T
4
T : F
3
F F
I
I.2 T F
7
T
6
T
3
T F
4
: F
2
F
I
T
3
F
Since the second piemiss is tiue and (P
I
Q} is false, P P
I
ought to
be tiue. But this cannot be the case, because P is false. So line I.2 cannot
be completed eithei. Since this exhausts the possible ways of iefuting the
aigument, the following claim has been established:
P Q, (P
I
Q} (P P
I
} (P Q} P
I
.
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic 8
Generally the entries in a table where one calculates truth-values back-
wards should be indexed by numbers that show the order in which the
values were obtained. Te assumptions in the diperent cases should be
indicated by underlining the respective truth-values. Subcases should be
marked as such (in the way I have done this above). A pioof without in-
dices foi the tiuth-values will be consideied to be incomplete. Tese
conventions meiely seive the puipose of making the calculations easiei
to ieconstiuct: otheiwise the calculations of tiuth-values can be veiy
dimcult to follow.
Afei having developed semantics foi the language L
I
, I will now
look at alteinatives. One might wondei why the connectives , , ,
and have been chosen. Tese connectives aie used in many logic texts,
and they moie oi less coiiespond to expiessions in English. Howevei,
theie aie also othei English phiases foi connecting sentences. Te phiase
neithei . . . noi . . . is an example. Te sentence
Neithei Jones noi Biown is in Baicelona
will be tiue if and only if Jones is not in Baicelona and Biown is not
in Baicelona. In L
I
theie is no connective that diiectly coiiesponds to
neithei . . . noi . . . . If one added a connective foi the phiase neithei . . .
noi . . . , it would have the following tiuth table:
T T F
T F F
F T F
F F T
Howevei, the connective foi neithei . . . noi . . . is not ieally needed
because one can geneiate the same tiuth table using the old connectives
of L
I
. Neithei . . . noi . . . can be ie-expiessed in English as It is not the
case that . . . , and it is not the case that . . . . In L
I
one can also defne in
Volker Halbach :oo8/:oo
: Syntax and Semantics of Propositional Logic
teims of and in the following way:
T T F F F
T F F F T
F T T F F
F F T T T
Alteinatively, one canie-expiess oi defne as (}. Tus, adding
to the language L
I
would not inciease the expiessive powei of L
I
. Te
connective would only allow one to abbieviate some sentences. Teie
aie many moie tiuth tables foi which L
I
does not have connectives. So
fai I have looked only at binaiy connectives (connectives conjoining two
sentences) such as , , , and , but theie aie also unaiy connectives
(connectives taking one sentence) othei than ; and theie aie teinaiy
connectives (connectives taking thiee sentences), and so on. Can all these
connectives be expiessed with the connectives of L
I
, that is, with , , ,
, and : Te answei is yes: all tiuth tables can be pioduced with the
old connectives of L
I
. Tis fact is called the tiuth-functional completeness
of L
I
. In fact, and togethei without any fuithei connectives aie
sumcient foi expiessing all othei connectives. And even on its own
would do the tiick. At any iate, adding moie connectives to L
I
than those
used heie is not ieally necessaiy and would not inciease the expiessive
powei of L
I
. I will not piove these iesults heie.
Volker Halbach :oo8/:oo
, Foimalisation in Piopositional Logic
In the pievious chaptei I focused on the foimal language L
I
. Te con-
nectives , , , and of L
I
can be used to combine sentences of L
I
to
foim new compound sentences; the connective can be added in fiont
of a sentence of L
I
to build a new sentence.
English sentences can also be combined and modifed with many
difeient connectives to foim new sentences. Foi instance, one can wiite
the woid and between two English sentences to obtain a new sentence.
Oi, because, although, but, while, if and many otheis can be used
in the same way. An expiession that connects sentences can also be
moie complex: the expiession due to the fact that between two English
sentences yields a new English sentence. Othei expiessions, such as if . . . ,
then, connect sentences even though they aie not wiitten between two
sentences.
Othei expiessions do not combine sentences, but iathei modify a
sentence, as is the case with not, as is well known, John stiongly be-
lieves that, and iegiettably. Not is special insofai as it ofen cannot be
simply inseited into a sentence, but iathei iequiies the intioduction of
the auxiliaiy veib to do: the intioduction of not into Alan went to
London yields Alan did not go to London. In this iespect not is moie
complicated than an adveib such as iegiettably oi the connective of
L
I
.
In the pievious chaptei I defned the notion of a connective; now I
will apply it to English as well: Expressions that can be used to combine or
modify English sentences are connectives. Tis defnition is fai fiom being
piecise, but an exact defnition of the notion of a connective of English
is not easy to give because sometimes the connectives aie not simply
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,:
plugged into oi wiitten between sentences. Occasionally the sentences
themselves have to be modifed, foi instance, by intioducing auxiliaiy
veibs, as the above examples of not shows.
,. : 1vU1u-iUc1io:ii1v
Te connectives of L
I
, that is, , , , and , coiiespond to connectives
in English. Te semantics of the connectives of L
I
is veiy simple; it is
encompassed in theii tiuth tables. In contiast, many connectives of
English function in a much moie intiicate way.
As an example I will considei the connective because. Imagine that I
diop my laptop computei on the stieet. Its had it: the scieen is bioken.
So my laptop computei does not woik. Te sentence
My computei does not woik because I diopped my computei
is also tiue: the laptop would still be functional if I had not diopped it.
Moieovei, it is tiue that the computei does not woik and it is tiue that I
diopped it. Tus, because connects the two tiue sentences My computei
does not woik and I diopped my computei togethei foiming a new tiue
sentence. In this iespect it seems similai to and.
In othei cases, howevei, one can use because to connect two tiue
English sentences Aand B and end up with a false sentence. Afei picking
up my bioken laptop, I considei the following sentence:
My laptop computei does not woik because it is not plugged
in.
In the situation I just desciibed, it is tiue that my computei does not woik,
and it is tiue that it is not plugged in as I am standing in the stieet with
my bioken laptop. Neveitheless the sentence that my laptop computei
does not woik because it is not plugged in is false: it would woik if I had
not diopped it. Even if it weie now plugged in, it would not woik. It does
not woik because I diopped it, not because it is not plugged in. So this
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,:
is a case in which using because to connect two tiue sentences yields a
false sentence.
Neveitheless, the tiuth of Abecause B is not completely independent
of the tiuth and falsity of the English sentences A and B. If A oi B (oi
both) aie false, then Abecause B is also false. Tese dependencies can
be summaiised in the following tiuth table foi the English connective
because, wheie Aand B aie declaiative sentences of English:
A B Abecause B
T T :
T F F
F T F
F F F
Te question maik indicates that in this case the tiuth-value of A be-
cause B depends not only on the tiuth-values of the diiect subsentences,
that is, onthe tiuth-values of the sentences Aand Bthat because connects.
Tis means that when because is used to connect two tiue sentences,
sometimes the iesulting sentence is tiue and sometimes the iesulting
sentence is false; so the tiuth-value of the compound sentence is not de-
teimined by the tiuth-values of the sentences connected by because. In
this iespect because difeis fiomand. If the tiuth-value of the compound
sentence is deteimined by the tiuth-value of the connected sentences,
as is the case with and, then the connective is called tiuth-functional.
Connectives like because aie not tiuth-functional.
Te following is a geneial, less than piecise chaiacteiisation of tiuth-
functionality :
cu:v:c1ivis:1io ,.: (1vU1u-iUc1io:ii1v). A connective is truth-
functional if and only if the truth-value of the compound sentence cannot
be changed by replacing a direct subsentence with another sentence having
the same truth-value.
Foi instance, because is not tiuth-functional: ieplacing the tiue
sentence I diopped my computei with the equally tiue sentence the
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,,
computei is not plugged in does change the tiuth-value of the compound
sentence
My computei does not woik because I diopped my computei
fiom Tiue to False.
Tus, the defnition of tiuth-functionality of an English connective
canbe paiaphiasedinteims of tiuthtables: a connective is tiuth-functional
if and only if its tiuth table does not contain any question maiks.
If . . . then is usually tianslated as the aiiow . Some of its occui-
iences, howevei, aie defnitely not tiuth-functional. A sentence
is tiue if is false oi is tiue. In the following sentence, if . . . then
functions difeiently:
If Giovanni hadnt gone to England, he would not have caught
a cold in Cambiidge.
Assume that Giovanni ieally did go to England, but did not catch a cold
in Cambiidge. In this case one may hesitate to assign a tiuth-value to
the sentence: some people would say that the sentence is neithei tiue
noi false; otheis would say that it is false. At any iate, in that case the
sentence is not tiue. But if the whole sentence is not tiue, then this is a
case in which the fist subsentence following if is false, but the whole
sentence is also false. But accoiding to the tiuth table foi a sentence
with a false antecedent is tiue. Tis means that the aiiow cannot be
used to foimalise the sentence coiiectly.
If-sentences desciibing what would have happened undei ciicum-
stances that aie not actual aie called subjunctives oi counteifactuals.
In these sentences if does not function like the aiiow and cannot
be tianslated as the aiiow. Te piopei tieatment of counteifactuals is
beyond the scope of this book.
:
Indicative conditionals such as
: Lewis (:,,) is a classic text on counteifactuals.
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,
If Jones gets to the aiipoit an houi late, his plane will wait
foi him
aie ofen foimalised using the aiiow , but it is somewhat questionable
whethei ieally is appiopiiate.
Assume, foi instance, that Jones aiiives at the aiipoit eaily and he
easily catches the plane. Suppose also that Jones is not a VIP and so the
aiiline would not have waited foi him. Using the aiiow foi if one can
tiy to tianslate this sentence as P Q with the following dictionaiy:
P: Jones gets to the aiipoit an houi late,
Q: Joness plane will wait foi Jones.
If |P|
A
= F, that is, if P is false in the stiuctuie A, then P Q is tiue
in A, that is, |P Q|
A
= T by Defnition :.o oi by the tiuth table of on
page ,,. Accoiding to the assumptions, Jones gets to the aiipoit an houi
late is actually false. Tus, if the foimalisation is coiiect, the displayed
English if -sentence should be tiue. But it is highly questionable whethei
it is tiue: one may hold that If Jones gets to the aiipoit an houi late, his
plane will wait foi him is simply false, even if Jones gets theie on time.
Teie is an extensive liteiatuie on the tieatment of if -sentences. Te
tieatment of if -sentences, including counteifactuals, has inteiesting
philosophical implications. I shall not go fuithei into the details of this
discussion heie. Te above example should be sumcient to show that
foimalising if by the aiiow is pioblematic even in the case of indicative
conditionals. Foi most puiposes, howevei, the aiiow is consideied to
be close enough to the if . . . then . . . of English, with the exception of
counteifactuals.
Te defnition of truth-functionality also applies to unary connectives.
a unary connective is truth-functional if and only if the truth-value of the
sentence with the connective cannot be changed by replacing the direct sub-
sentence with a sentence with the same truth-value.
It is necessaiily the case that A oi It is necessaiy that is a unaiy
connective that is not tiuth-functional. If A is a false English sentence,
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,,
then It is necessaiy that A is false, but if A is tiue, It is necessaiy that A
may be eithei tiue oi false:
It is necessaiy that all tiees aie tiees.
Tis sentence is tiue: all tiees aie tiees is logically tiue and thus necessaiy.
But, if the tiue sentence All tiees aie tiees is ieplaced by the tiue sentence
Volkei has ten coins in his pocket then the iesulting sentence
It is necessaiy that Volkei has ten coins in his pocket
is not tiue, because I could easily had only nine coins in my pocket.
Geneially if Ais only accidentally tiue, It is necessaiy that A will be false.
Tus the coiiesponding tiuth table looks like this:
A it is necessaiy that A
T :
F F
Some othei connectives likeBill believes that . . . have nothing but
question maiks in theii tiuth tables. In contiast, Bill knows that . . . has
the same tiuth table as it is necessaiy that.
:
,. : iocic:i iovm
In this section and the next I will show how to tianslate English sentences
into L
I
sentences. Tese tianslations aie caiiied out in two steps: Fiist
the sentence is biought into a standaidised foim, which is called the
(piopositional) logical foim.
,
In the second step the English expiessions
: A moie compiehensive account of tiuth-functionality is given by Sainsbuiy (:oo:,
Chaptei :).
, In this chaptei I will usually diop the specifcation piopositional fiom piopositional
logical foim since I will not deal with any othei kind of logical foim foi now. Teie is
also a moie complex piedicate logical foim of an English sentence. Te piedicate logical
foim will be studied in Chaptei ,.
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,o
aie ieplaced by symbols. Obtaining the logical foim is the non-tiivial
pait; the step fiom the logical foim to a sentence of the language L
I
of
piopositional logic is simple and puiely mechanical.
Heie I shall sketch a piactical pioceduie foi biinging an English
sentence into its logical foim. Te pioceduie is to be applied to the
sentence, ieapplied to its sentences fiom which the main sentence is built
up, and then iepeated until the subsentences cannot be fuithei analysed
by the means of piopositional logic.
I have bioken down this pioceduie into fve steps. In piactice they aie
all caiiied out at the same time. Te fist step is the dimcult one: in this
step it is checked whethei the sentence can be bioken down into a tiuth-
functional connective and one oi moie subsentences, that is, whethei the
sentence is built up fiom one oi moie sentences with a tiuth-functional
connective.
Foi instance, the sentence
Te cai doesnt stait because the batteiy is fat oi theie is no
petiol in the tank.
is not built up fiom othei sentences with a tiuth-functional connective:
it is built up fiom the sentence Te cai doesnt stait and the sentence
Te batteiy is fat oi theie is no petiol in the tank. with the connec-
tive because, which is not tiuth-functional. Te connective oi is tiuth-
functional, but it only connects Te batteiy is fat and Teie is no
petiol in the tank.; so only the subsentence Te batteiy is fat oi theie
is no petiol in the tank. is built up fiom othei sentences with the tiuth-
functional connective oi, but not the entiie sentence.
To put it in a difeient way, one identifes the topmost oi main con-
nective and checks whethei it is tiuth-functional. It is peimissible to
iefoimulate the sentence slightly to put into a foim such that it is built
up with a tiuth functional connective. Te tiuth-functional connectives
should be taken fiom a fxed list of tiuth-functional connectives; this
iestiiction will enable one to foimalise the connectives as the fves con-
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,,
nectives of the foimal language L
I
, iespectively. Finally, one ieapplies the
pioceduie to the sentence(s) fiom which the main sentence is built up.
I give now the fve steps of the pioceduie and then show how it woiks
by means of some examples:
:. Check if the sentence can be reformulated in a natural way as a sen-
tence built up from one or more sentences with a truth-functional
connective. If this is not possible, then the sentence should be put in
brackets and not analysed any further.
:. If the sentence can be reformulated in a natural way as a sentence
built up from one or more sentences with a truth-functional connec-
tive, do so.
,. If that truth-functional connective is not one of the standard con-
nectives in Table .:, reformulate the sentence using the standard
connectives.
. Enclose the whole sentence in brackets, unless it is a negated sen-
tence, that is, a sentence starting with it is not the case that.
,. Apply the procedure, starting back at :., to the next subsentence(s)
(that is, to the sentence(s) without the standard connective of step ).
name standaid connective some othei foimulations
conjunction and but, although
, [a comma
between sentences]
disjunction oi, unless
negation it is not the case that not, none, nevei
aiiow if . . . then given that, . . .
double aiiow if and only if exactly if,
piecisely if
Table ,.:: standaid connectives
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,8
Teie is no need to memoiise this desciiption of the pioceduie; the
point is to leain how to apply it. Tus instead of desciibing the pioceduie
in moie detail, which would be faiily dimcult, Ill illustiate the pioceduie
with seveial examples:
ix:mvii ,.:. Rob and Tim will laugh, if the tutoi cant pionounce Siob-
hans name.
Te sentence is built fiom Rob and Tim will laugh and Te tutoi
cant pionounce Siobhans name with the connective . . . , if . . . . As such,
theie is no need foi the soit of iefoimulation called foi in :. But . . . ,
if. . . is not a standaid connective. So, in accoidance with step , I will
iefoimulate the sentence with the standaid connective if . . . then . . . :
If the tutoi cant pionounce Siobhans name, then Rob and
Tim will laugh.
In step the entiie sentence is enclosed by biackets:
(If the tutoi cant pionounce Siobhans name, then Rob and
Tim will laugh)
Step , sends me back again to step :. Te two subsentences to which step :
is applied aie:
Te tutoi cant pionounce Siobhans name,
Rob and Tim will laugh.
Te fist sentence contains a negation and so I will iefoimulate it with the
standaid connective it is not the case that . . . :
It is not the case that the tutoi can pionounce Siobhans name
Accoiding to step , the sentence does not need to be put in biackets
since it staits with it is not the case that.
I still have to apply the pioceduie to the second sentence. Rob and
Tim will laugh is not a sentence built up using a tiuth-functional connec-
tive, but it can be iefoimulated in accoidance with step : as a sentence
with a tiuth-functional connective in the following way:
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic ,
Rob will laugh and Tim will laugh,
since . . . and . . . is alieady a standaid connective, theie is no need foi the
soit of iefoimulation desciibed in step ,. Afei applying step I obtain
the following expiession:
(Rob will laugh and Tim will laugh).
Tus the whole sentence now ieads as follows:
(If it is not the case that the tutoi can pionounce Siobhans
name, then (Rob will laugh and Tim will laugh)).
Now I have to stait again with step :. Te sentence the tutoi can pio-
nounce Siobhans name cannot be iefoimulated in a natuial way as a
sentence built up with a tiuth-functional connective. Tus it is put in
biackets accoiding to step ::
(If it is not the case that (the tutoi can pionounce Siobhans
name), then (Rob will laugh and Tim will laugh)).
Next, neithei Rob will laugh noi Tim will laugh can be iefoimulated as
a sentence built with a tiuth-functional connective, so they aie each put
into biackets as iequiied by step ::
(If it is not the case that (the tutoi can pionounce Siobhans
name), then ((Rob will laugh) and (Tim will laugh))).
Now this is the (piopositional) logical foim of the sentence.
ix:mvii ,.,. Unless the ignition is tuined on and theie is petiol in the
tank, the engine will not stait and Ill not be able to aiiive in time.
I can skip steps : and : because the sentence is alieady built up with
a tiuth-functional connective, though not by a standaid one. Next, I
ieplace unless by the standaid connective oi in accoidance with step ,;
then I apply step and enclose the entiie sentence in biackets.
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic oo
(Te ignition is tuined on and theie is petiol in the tank, oi
the engine will not stait and Ill not be able to aiiive in time).
Since and is alieady a standaid connective, step is applied twice moie.
((Te ignition is tuined on and theie is petiol in the tank),
oi (the engine will not stait and Ill not be able to aiiive in
time)).
Next, I tuin to the pait of the sentence afei oi. Teie aie two sentences
containing not. Accoiding to step ,, they aie to be iefoimulated with the
coiiesponding standaid connective.
((Te ignition is tuined on and theie is petiol in the tank),
oi (it is not the case that the engine will stait and it is not the
case that Ill be able to aiiive in time)).
Now step : is applied foui times:
(((Te ignition is tuined on) and (theie is petiol in the tank)),
oi (it is not the case that (the engine will stait) and it is not
the case that (Ill be able to aiiive in time))).
Te piocess teiminates heie since the iemaining sentences not containing
biackets, that is, Te ignition is tuined on and so on cannot be fuithei
analysed.
,. , ivom iocic:i iovm 1o iovm:i i:cU:ci
Once the logical foim of a sentence has been deteimined, the tianslation
into the language L
I
of piopositional logic is simple.
In oidei to tianslate the logical foim of an English sentence into L
I
apply the following pioceduie:
:. Replace standard connectives by their respective symbols in accor-
dance with the following list.
Volker Halbach :oo8/:oo
Formalisation in Propositional Logic o:
standard connective symbol
and
or
it is not the case that
if . . . then . . .
if and only if
:. Replace every English sentence by a sentence letter and delete the
brackets surrounding the sentence letter.
piedicate
the Eifel Towei
- --
designatoi
.
Te tallest student in Oxfoid
- - -
designatoi
gives
-
piedicate
his fiiend
-- -
designatoi
the CD
- --
designatoi
.
Te tallest student in Oxfoid
- - -
designatoi
gives
-
fist pait
of piedicate
the CD
- --
designatoi
to
-
second
pait of
piedicate
his fiiend
-- -
designatoi
.
Te engineei
- --
designatoi
loosens
-- -
fist pait
of piedicate
the nut
- --
designatoi
with
-
second
pait of
piedicate
the wiench
- - -
designatoi
.
Te piedicates can be simple and consist in just one woid, as is . . .
sees . . . in the second sentence and . . . gives . . . . . . in the thiid; oi they
can be foimed fiom two oi moie woids, as is . . . is boiing in the fist
sentence, . . . gives . . . to . . . in the fouith, and . . . loosens . . . with . . .
in the last.
: Te teiminology in logic difeis heie fiom tiaditional giammai, wheie loves Maiy
would be the piedicate.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic ,,
In predicate logic predicate expressions are translated into predicate let-
ters. Tese piedicates have an uppei index that coiiesponds to the numbei
of designatois the coiiesponding English piedicate expiession can take.
Foi instance, . . . is boiing, which can take one designatoi, is tianslated
into a piedicate lettei with uppei index I: P
I
, foi instance, is such a piedi-
cate lettei. Te piedicate . . . sees . . . can be tianslated as Q
2
since it can
take two designatois, and . . . gives . . . . . . can be tianslated as R
3
because
it can take thiee designatois. Te uppei index (I, 2, 3 heie) is called the
piedicate letteis aiity-index. A piedicate lettei with uppei index n is
called an n-place piedicate lettei. I-place piedicate letteis aie also called
unaiy, 2-place binaiy and 3-place teinaiy. So, the piedicate expiession
. . . loosens . . . with . . . , foi instance, is tianslated into a teinaiy piedicate
lettei.
It is not haid to fnd English sentences that iequiie 4- oi 3-place
piedicate letteis foi theii foimalisation. Tus, I will include piedicate
letteis with aiity-indices foi any n in the language L
2
of piedicate logic
to make suie that theie is always a sumcient stock of piedicate letteis
available foi the vaiious English piedicates.
I will also include 0-place piedicate letteis in the language L
2
. Tey
aie useful foi foimalising English sentences like It is iaining. It just seems
to be a quiik of the English language that the pionoun it in this sentence
is iequiied; it does not seive the same puipose as the designatois in
the sentences I have consideied so fai; othei languages such as Italian
have dispensed with the pionoun in the coiiesponding sentence. Tus, I
will foimalise the sentence It is iaining and similai sentences as 0-place
piedicate letteis. Tis piovides the ieason foi including 0-place piedicate
letteis in the language L
2
.
Now that I have dealt with piedicates, I will tuin to the foimalisation
of designatois. In the easiest cases the designatois aie piopei names like
Tom, the Eifel Towei, oi the United Kingdom of Gieat Biitain and
Noithein Iieland; othei types of designatois will be discussed latei. Coi-
iesponding to the piopei names of English, the language L
2
of piedicate
logic featuies constants, namely a, b, c, a
I
, b
I
, c
I
, and so on.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic ,o
Te example sentence Tom loves Maiy fiom the beginning of the
section can now be tianslated into the language L
2
as
P
2
ab.
In such sentences, the piedicate lettei is always put a the beginning of
the sentence. Te piedicate lettei and the two constants aie tianslated as
follows:
P
2
: . . . loves . . .
a: Tom
b: Maiy
Te oidei of the constants in P
2
ab is ciucial: P
2
ba is a foimalisation of
Maiy loves Tom, which says something difeient fiom what Tom loves
Maiy says.
In the dictionaiy, the entiy foi loves contains two stiings of dots.
Tey stand foi the places that aie taken in the sentence by the designatois
Tom and Maiy. In the coiiesponding foimal sentence P
2
ab the fist
place of the binaiy piedicate lettei P
2
is taken by a, and the second place
is taken by b. Now the fist stiing of dots in . . . loves . . . coiiesponds
to the fist place of P
2
and the second stiing of dots coiiesponds to the
second place of P
2
. In oidei to emphasise this coiielation, one can attach
subsciipts to the dots:
P
2
: . . .
I
loves . . .
2
Tis is tantamount to the tianslation foi P
2
given above.
If there are subscripts in the dictionary, the string of dots marked I al-
ways corresponds to the frst place of the predicate letter, the string marked 2
corresponds to the second place of the predicate letter, and so on. Te num-
ber of strings of dots must always correspond to the arity-index of the pred-
icate letter.
Teiefoie, if I had used
P
2
: . . .
2
loves . . .
I
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic ,,
as the entiy foi P
2
, while keeping the entiies foi a and b unchanged, the
piopei tianslation of Tom loves Maiywould be P
2
ba.
As my next example I considei a sentence with a slightly moie com-
plicated piedicate: Te sentence
Ebenezei is a sciooge
can be tianslated into L
2
as the sentence R
I
c, with the following tiansla-
tion of the piedicate lettei and the constant:
R
I
: . . . is a sciooge
c: Ebenezei
Te noun sciooge is not tianslated sepaiately but foims pait of the
piedicate . . . is a sciooge, which is tianslated as the piedicate lettei R
I
.
Typically, phiases of the foim is a . . . aie tianslated as piedicate letteis.
One way to see why . . . is a sciooge can be tianslated as a simple piedicate
lettei only, is to obseive that instead of is a sciooge one could say is
stingy, which obviously can be tianslated as a unaiy piedicate lettei.
Geneially, in the language L
2
an n-place piedicate lettei followed by
n constants yields a sentence. Hence, P
2
ab, that is, a binaiy piedicate
lettei followed by two constants, and R
I
c, that is, a unaiy piedicate lettei
followed by one constant, aie sentences of L
2
. 0-place piedicate letteis
foim a sentence without any fuithei symbol: each 0-place piedicate lettei
is alieady a sentence. Tus, they behave in the same way as the sentence
letteis of the language L
I
of piopositional logic. In fact, I will identify
the 0-place piedicate letteis with sentence letteis; sentence letteis aie,
theiefoie, meiely a ceitain soit of piedicate letteis.
One can build sentences of the language L
2
using connectives in
the same way as in the language L
I
of piopositional logic. Foi instance,
(P
2
ab R
I
c} is the tianslation of the following sentence of L
2
:
Tom loves Maiy and Ebenezei is a sciooge.
Te dictionaiy is the same as above:
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic ,8
P
2
: . . . loves . . .
a: Tom
b: Maiy
R
I
: . . . is a sciooge
c: Ebenezei
Te techniques foi tianslation into piopositional logic caiiy ovei to
piedicate logic (see page o,): Liz is an Austialian mountaineei can be
iephiased as Liz is Austialian and Liz is a mountaineei. Te caveats
explained theie also apply to piedicate logic.
With the techniques developed so fai, ceitain occuiiences of peisonal
pionouns can be ieadily tianslated.
Caesai came, he saw, he won.
Tis can be paiaphiased as the following sentence:
Caesai came and Caesai saw and Caesai won.
Te pionouns (oi iathei theii occuiiences) in this example aie known
as lazy pionouns.
:
Using he heie saves one the efoit of iepeating the
name Caesai. Lazy pionouns can easily be eliminated by iepeating the
name (oi whatevei they iefei back to), and thus theii foimalisations do
not pose any special pioblems.
Teie aie othei uses of pionouns that cannot easily be dispensed with.
If a politician speaks the tiuth, he wont be elected.
In this sentence the pionoun he cannot be ieplaced by a politician. Te
sentence
If a politician speaks the tiuth, a politician wont be elected
has a difeient meaning; it says that some politician will not be elected if
some politician (not necessaiily the same one) speaks the tiuth. In fact,
the oiiginal sentence is equivalent to
: Tis teiminology comes fiom Geach (:o:).
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic ,
All politicians speaking the tiuth aie not elected.
In the oiiginal sentence If a politician speaks the tiuth, he wont be
elected the pionoun is used to expiess a geneialisation. Uses of pionouns
foi the puipose of geneialisation (and some similai puiposes) aie called
quantifcational uses. Quantifcation can be expiessed in many difeient
ways in English. Sometimes pionouns aie used, and sometimes quantif-
cation can be expiessed without pionouns, as is the case with the sentence
All politicians speaking the tiuth aie not elected.
At any iate, the tianslation of quantifcational uses of pionouns ie-
quiies additional iesouices in L
2
beyond the ones I have mentioned so
fai. Peisonal pionouns in English come in difeient gendeis (he, she,
it), in difeient cases (he, him, and so on). Tese difeient foims help
to disambiguate sentences. Heie is a somewhat tiicky case:
If a visitoi wants to boiiow a book fiom the libiaiy, she is
iequiied to complete the foim foi it, which must then be
submitted to a libiaiian, who can giant hei peimission to
check it out, if it looks satisfactoiy to him.
In oidei to make the iefeience of the vaiious occuiiences of the peisonal
pionouns cleaiei, one can attach indices to them:
If a visitoi
I
wants to boiiow a book
2
fiom the libiaiy, she
I
is
iequiied to complete the foim
3
foi it
2
, which
3
must then be
submitted to a libiaiian
4
, who can giant hei
I
peimission to
check it
2
out, if it
3
looks satisfactoiy to him
4
.
It is natuial to assume that the fist and second occuiiences of it iefei
back to book, while the thiid occuiience iefeis back to the foim. Tis
assumption is made explicit by using subsciipts. Natuial languages ofei
othei iesouices foi disambiguation, but indexing is a stiaightfoiwaid
method. Since the help of gendei etc. is no longei iequiied, when the ief-
eience is made cleai by indexing, one can dispense with English pionouns
and ieplace them with what logicians call vaiiables:
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8o
If a visitoi x
I
wants to boiiow a book x
2
fiom the libiaiy, x
I
is
iequiied to complete the foim x
3
foi x
2
, which
3
must then
be submitted to a libiaiian x
4
, who can giant x
I
peimission
to check x
2
out, if x
3
looks satisfactoiy to x
4
.
In the sentence vaiiables iefei back to such phiases as a visitoi and some
book. One could intioduce expiessions coiiesponding to these phiases
into the foimal language. Logicians have found a method to simplify
the language fuithei and to manage with just one additional expiession:
Te puipose of expiessions such as a visitoi is to iestiict the focus to
visitois; the above sentence makes a geneial claim about visitois. But one
can ieplace this with a geneial claim about all things (whethei they aie
peisons, animals oi inanimate objects). Instead of saying
Eveiy visitoi is a classicist,
one can say
If something is a visitoi, then it is a classicist
oi
Foi eveiything
I
: if it
I
is a visitoi, then it
I
is a classicist.
By substituting vaiiables one obtains:
Foi all x
I
, if x
I
is a visitoi, then x
I
is a classicist.
Te last foimulation is basically the analysis of typical quantifed state-
ments. Foi all is tianslated as the symbol (a iotated A ieminding one
of all).
Is a visitoi is a piedicate that is tianslated as P
I
, while is a classicist
is tianslated as Q
I
. So x
I
is a visitoi is tianslated as P
I
x
I
, and x
I
is a
classicist becomes Q
I
x
I
. If . . . , then . . . becomes the aiiow, so if x
I
is a
visitoi, then x
I
is a classicist becomes (P
I
x
I
Q
I
x
I
}. Tus the sentence
Eveiy visitoi is a classicist tianslates into the following L
2
-sentence:
x
I
(P
I
x
I
Q
I
x
I
}.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8:
In addition to logicians also use the symbol . In the following
sentence an existence claim is made:
At least one visitoi is a classicist.
If this is iewiitten with vaiiables, then the following expiession is ob-
tained:
Foi at least one x
I
, x
I
is a visitoi and x
I
is a classicist.
Tis is then foimalised as x
I
(P
I
x
I
Q
I
x
I
}.
Latei, foimalisation in piedicate logic will be discussed in moie detail,
but I have now motivated all elements of the syntax of piedicate logic that
will be intioduced in the foimal defnition of a sentence of L
2
.
. : 1ui si1icis oi L
2
In this section the syntax of L
2
is intioduced in a foimally piecise way.
Fiist I tuin to piedicate letteis.
Te language L
2
of piedicate logic contains 0-place piedicate letteis.
Foi simplicity, they do not have an aiity-index, that is, an uppei index;
they aie the sentence letteis P, Q, R, P
I
, Q
I
, and so on. L
2
also contains
piedicate letteis with aibitiaiy aiity-index I, 2, 3, and so on. Having only
one binaiy, that is, 2-place, piedicate lettei will not sumce. In oidei to
foimalise a sentence containing the piedicate expiessions . . . hates . . .
and . . . loves . . . , one will need two distinct binaiy piedicate letteis; and
of couise one might also need a thiid and even moie binaiy piedicate
letteis. In oidei to make suie that theie is always a sumcient stock of
binaiy piedicate letteis, I include infnitely many binaiy piedicate letteis
in L
2
, namely the expiessions P
2
, Q
2
, R
2
, P
2
I
, Q
2
I
, R
2
I
, P
2
2
, and so on. Tis
applies not only to binaiy piedicate letteis but also to piedicate letteis
with othei aiity-indices. Tus, the geneial defnition of piedicate letteis
looks like this:
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8:
uiiii1io .: (vviuic:1i ii11ivs). All expressions of the form P
k
n
, Q
k
n
,
or R
k
n
are sentence letters, where k and n are either missing (no symbol) or
a numeral I, 2, 3, . . .
So the lettei P with oi without numeials I, 2, and so on as uppei
and/oi lowei indices is a piedicate lettei, and similaily foi Q and R. Te
sentence letteis P, Q, R, P
I
, Q
I
, . . . aie also piedicate letteis, accoiding to
this defnition. Fuitheimoie, P
I
, Q
I
, R
I
, P
I
I
, Q
I
I
, R
I
I
, P
I
2
, Q
I
2
, R
I
2
, . . . , P
2
I
, Q
2
I
,
R
2
I
, P
2
2
, Q
2
2
, R
2
2
, and so on, aie piedicate letteis. Tis defnition ensuies
that L
2
contains infnitely many n-place piedicate letteis foi any n. Using
only piedicate letteis with P, but not Q oi R, would sumce, but having a
choice between letteis enables me to geneiate moie ieadable foimulae.
0-place piedicate letteis (sentence letteis) have aiity 0; I-place piedi-
cate letteis have aiity I, and so on:
uiiii1io .:. Te value of the upper index of a predicate letter is called
its arity. If a predicate letter does not have an upper index its arity is 0.
Te piedicate lettei P
3
4
, foi example, has aiity 3.
Te language L
2
contains constants, which will be used to tianslate
English piopei names and some similai expiessions.
uiiii1io ., (cos1:1s). a, b, c, a
I
, b
I
, c
I
, a
2
, b
2
, c
2
, a
3
, . . . are con-
stants.
Moieovei, L
2
contains infnitely many vaiiables.
uiiii1io . (v:vi:viis). x, y, z, x
I
, y
I
, z
I
, x
2
, . . . are variables.
Now the notion of an atomic L
2
-foimula can be defned:
uiiii1io ., (:1omic iovmUi:i oi L
2
). If Z is a predicate letter of
arity n and each of t
I
, . . . , t
n
is a variable or a constant, then Zt
I
. . . t
n
is
an atomic formula of L
2
.
In this defnition, the uppei case lettei Z seives as a metavaiiable foi
piedicate letteis, that is, foi P, R
2
43
, Q
I
, and the like. Accoiding to this
defnition, Q
I
x, P
2
c y, P
3
3
x
3I
c
4
y, and R
2
xx aie examples of atomic foimu-
lae. Defnition ., allows foi the case in which n = 0. Tis means that all
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8,
sentence letteis, that is, P, Q, R, P
I
, and so on, aie also atomic foimulae.
uiiii1io .o. A quantifer is an expression v or v where v is a vari-
able.
Tus, x
348
and z aie quantifeis.
,
uiiii1io ., (iovmUi:i oi L
2
).
(i) All atomic formulae of L
2
are formulae of L
2
.
(ii) If and are formulae of L
2
, then , ( }, ( }, ( }
and ( } are formulae of L
2
.
(iii) If v is a variable and is a formula, then v and v are formulae
of L
2
.
Examples of foimulae of the language L
2
of piedicate logic aie:
x (P
2
xa Q
I
x},
z
77
y
3
z
43
(P
2
xy x
2
(R
4
3
z
77
c
3
xz
77
Q}},
(x P
I
x y y Q
2
yy},
x z R
2
az.
Teie is no point in tiying to undeistand these foimulae; the point
heie is that they all classify as L
2
-foimulae. In oidei to show that a given
expiession is a foimula of L
2
, one can build up the foimula step by step
accoiding to the iules laid down in Defnition .,. As an example I will
show that the last foimula, (xP
I
x yyQ
2
yy}, is a foimula of L
2
:
:. P
I
is a piedicate lettei by Defnition .: with aiity I (Defnition .:),
and x is a vaiiable by Defnition ..
:. Teiefoie, by Defnition .,, P
I
x is an atomic foimula.
,. x P
I
x is thus a foimula of L
2
by Defnition .,(iii).
. Similaily, Q
2
yy is an atomic foimula (I will not go thiough the
tedious ieasoning of :. and :. again).
, Teie aie alteinative symbols foi and , which will not be used heie: v, v and (v)
aie sometimes used instead of v, and v and v instead of v.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8
,. y Q
2
yy is a foimula of L
2
by .,(iii).
o. y y Q
2
yy is a foimula of L
2
by .,(iii).
,. y y Q
2
yy is a foimula of L
2
by .,(ii).
8. (x P
I
x y y Q
2
yy} is a foimula of L
2
by .,(ii). Tis follows
fiom the pievious item and ,.
In cases like this, one will be able to see without a long pioof whethei an
expiession is a foimula, and so it will not be necessaiy to go thiough all of
these steps. Te above pioof of the claim that (x P
I
x y y Q
2
yy}
is a foimula only shows how exactly the defnition of L
2
-foimulae woiks.
. , ivii :u voUu occUvvicis oi v:vi:viis
In the foimula x (P
I
x Q
I
x} the last two occuiiences of x iefei back
to oi depend on the quantifei x. In the foimula P
I
x Q
I
x, by contiast,
theie is no quantifei to which they can iefei back; they occui fieely, as
logicians say. In the next defnition this notion of a fiee occuiience of a
vaiiable is made piecise.
uiiii1io .8.
(i) All occurrences of variables in atomic formulae are free.
(ii) Te occurrences of a variable that are free in and are also free in
, , , and .
(iii) In a formula v no occurrence of the variable v is free, all occur-
rences of variables other than v that are free in are also free in
v .
An occurrence of a variable is bound in a formula if and only if it is not
free.
Less foimally speaking, occuiiences of vaiiables aie fiee as long as
they aie not caught by a quantifei. Foi instance, in the atomic foimulae
R
2
xx oi P
I
x all occuiiences of x aie fiee accoiding to clause (i) of the
defnition, and so, accoiding to clause (ii), all occuiiences of x aie fiee in
(R
2
xxP
I
x} and in (P
I
x R
2
aa}. Similaily, all occuiiences of x
I
aie
fiee in (P
I
x
I
Q
I
x
I
}, but, accoiding to clause (iii) of Defnition .8, all
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8,
occuiiences of x
I
aie bound in x
I
(P
I
x
I
Q
I
x
I
}. In y (P
I
x
I
Q
I
x
I
}
all occuiiences of the vaiiable x
I
aie fiee because y is a vaiiable difeient
fiom x
I
.
In (Q
I
z z R
2
zz} the fist occuiience of the vaiiable z is fiee, while
the iemaining occuiiences aie bound. In x (R
2
xy R
2
xa} R
2
ax all
but the last occuiience of x aie bound.
uiiii1io .. A variable occurs freely in a formula if and only if there
is at least one free occurrence of the variable in the formula.
As pointed out on page 8o, x
I
(P
I
x
I
Q
I
x
I
} is the foimalisation of
the sentence Eveiy visitoi is a classicist with the following logical foim:
Foi eveiything
I
: if it
I
is a visitoi, then it
I
is a classicist.
Te sentence can be tiue oi false depending on the ciicumstances. Te
foimula (P
I
x
I
Q
I
x
I
} coiiesponds to
(F) if it
I
is a visitoi, then it
I
is a classicist.
Onits own(F) is not a sentence that is tiue oi false: Teie is no quantifying
phiase like foi eveiything
I
the pionoun it
I
can iefei back to; also, it
I
is
not a lazy pionoun iefeiiing back to a ceitain designatoi. Tus, (F) does
not have a tiuth-value. One can only assign a tiuth-value to (F), if one
makes an aibitiaiy choice and takes it
I
to stand foi a paiticulai thing.
But without such an aibitiaiy choice (F) cannot be assigned a tiuth-value.
Te L
2
-foimulae behave similaily: only foimulae without fiee occui-
iences of vaiiables aie sentences; and only sentences will be assigned
tiuth-values by L
2
-stiuctuies, which will be intioduced in the follow-
ing chaptei. Also, sentences but not foimulae with fiee occuiiences of
vaiiables will be used as piemisses and conclusions in aiguments.
uiiii1io .:o (si1ici oi L
2
). A formula of L
2
is a sentence of L
2
if and only if no variable occurs freely in the formula.
Again infoimally speaking, in a sentence of L
2
all occuiiences of
vaiiables aie caught by some quantifei. Te following aie examples of
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8o
L
2
-sentences:
x (P
I
x (Q
2
29
xa x R
3
xax}}
(P
2
ab y(P
2
by x P
2
xy}}
. o1:1io:i covi1ios
In Section :., I intioduced some conventions foi diopping biackets fiom
sentence of L
I
. Tese iules did not foim pait of the omcial syntax of L
I
;
they meiely allow one to abbieviate sentences.
In this section I will specify some iules foi abbieviating foimulae
of L
2
. Again, they do not foim pait of the omcial syntax. Applying the
iules does not yield L
2
-sentences but iathei only abbieviations of L
2
-
sentences. Like the iules foi diopping biackets in L
I
, the conventions
do not have to be applied: one can always wiite down the full foimula
instead of the abbieviated foim.
Te Biacketing Conventions :, apply also to foimulae of L
2
. Te
quantifeis have to be taken into account: In the sentence x(P
I
x Q
I
x}
the biackets aie not outei biackets, so they cannot be diopped. Te
expiession x P
I
x Q
I
x is an abbieviation of the foimula (x P
I
x Q
I
x},
which is not a sentence, because the second occuiience of x is fiee.
As a fuithei example I will considei the following L
2
-sentence:
x ((P
I
x R
2
3
xa} y
2
((R
2
3
xy
2
Q
I
x} P
I
y
2
}} (.:)
Tis sentence may be abbieviated in the following ways:
x (P
I
x R
2
3
xa y
2
((R
2
3
xy
2
Q
I
x} P
I
y
2
}}
x ((P
I
x R
2
3
xa} y
2
(R
2
3
xy
2
Q
I
x P
I
y
2
}}
x (P
I
x R
2
3
xa y
2
(R
2
3
xy
2
Q
I
x P
I
y
2
}}
In the fist line, Biacketing Convention , is applied, in the second Biack-
eting Convention :, and in the thiid both conventions aie applied. Teie
aie no fuithei ways of saving biackets.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8,
Veiy ofen the uppei index of piedicate letteis, that is, theii aiity-
index, is also omitted. Tis is due to the fact that theie is only one way to
add these uppei indices to the piedicate letteis of an expiession that is
supposed to abbieviate a foimula. Teiefoie, sentence (.:) also has
x ((Px R
3
xa} y
2
((R
3
xy
2
Qx} Py
2
}}
as an abbieviation. So, combined with the iules foi diopping biackets,
the most economical foim of (.:) is the following abbieviation:
x (Px R
3
xa y
2
(R
3
xy
2
Qx Py
2
}}
Tus, when the aiity-index is missing, this does not necessaiily mean
that the piedicate lettei is a sentence lettei: it could be an abbieviation of
anothei piedicate lettei fiom which the aiity-index has been omitted.
Abbieviations of foimulae that have been obtained by omitting aiity-
indices can be misleading: one might think that x y (Px Pxy}
abbieviates an L
2
-sentence that contains the same piedicate lettei twice.
Inseiting the missing indices, howevei, shows that the sentence contains
two difeient piedicate letteis, P
I
and P
2
:
x y (P
I
x P
2
xy}
Teiefoie, theie is only one occuiience of P
I
in the foimula and only
one occuiience of P
2
. Te abbieviation x y (Px Pxy} is coiiect
accoiding to the above conventions, but in such cases it may be helpful
to ietain the aiity-indices.
. , iovm:iis:1io
Te basic stiategy foi obtaining the logical foim of an English sentence in
piedicate logic is the same as inpiopositional logic (cf Sections ,.: and ,.,):
a givensentence is analysedfiomtopto bottom. Tat is, one staits withthe
entiie sentence and woiks ones way deepei and deepei into the sentence.
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 88
In contiast to piopositional logic, one does not have to stop at quantifed
sentences; one can analyse them in the way outlined in Section .:. In
paiticulai, the logical foim of univeisal sentences is obtained in the way
sketched on page 8o.
Rathei than going ovei the geneial iules again, I will show how the
method woiks by way of some examples. I have alieady dealt with simple
sentences like Tomloves Maiy onpage ,o, so heie I will focus oncomplex
sentences.
Te following sentence is an example of a univeisally quantifed sen-
tence, that is, a sentence making a claim about all objects of a ceitain
soit:
All fiogs aie amphibians.
Fiist, I will deteimine its logical foim. It is cleaily a univeisal claim that
is to be paised as outlined on page 8o:
Foi all x (if x is a fiog, then x is an amphibian).
Te expiession in biackets contains the standaid connective if . . . , then
. . . , so it can be fuithei paised as follows:
Foi all x (if (x is a fiog), then (x is an amphibian)).
Both x is a fiog and x is an amphibian aie enclosed in biackets; they
contain no connectives and aie not quantifed. Is a fiog and is an
amphibian aie then foimalised by two distinct piedicate letteis P
I
and Q
I
,
iespectively. So (x is a fiog) becomes Px (omitting the aiity-index);
and x is an amphibian becomes Qx. Te expiession foi all becomes
the univeisal quantifei , and if . . . , then . . . , the aiiow . So, the
foimalisation is
x (Px Qx},
with the following dictionaiy:
P: . . . is a fiog
Q: . . . is an amphibian
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic 8
Geneially, univeisal claims can be foimalised this way. Completely
uniestiicted univeisal claims aie iaie, but philosopheis occasionally do
make claims like the following, that aie meant to be completely geneial:
Eveiything is mateiial.
Tis sentence can be foimalised as x Rx, wheie R stands foi is mateiial.
Existential claims aie usually foimalised by the existential quanti-
fei; iestiictions to a ceitain kind of objects is expiessed by conjunction.
Teiefoie, Teie aie poisonous fiogs has the following logical foim:
(R) Teie is at least one x ((x is a fiog) and (x is poisonous))
Te foimalisation is
x(Px Q
I
x}
with the following dictionaiy:
P: . . . is a fiog,
Q
I
: . . . is poisonous.
Te English phiase No . . . is . . . . can be taken to be a negated existen-
tial quantifcation. Te sentence
No fiog is poisonous.
can be iephiased as
It is not the case that theie aie poisonous fiogs.
It is not the case that is a standaid connective and it is foimalised as . I
have alieady shown how to go about foimalising Teie aie poisonous
fiogs. So the sentence No fiog is poisonous is foimalised as the following
sentence, with the same dictionaiy as above:
x (Px Q
I
x} (.:)
Alteinatively, one could have iephiased the oiiginal sentence No fiog is
poisonous as the following sentence:
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic o
All fiogs aie non-poisonous.
Tis foimalises into x (Px Q
I
x}. Tis sentence and the alteinative
foimalisation (.:) aie logically equivalent undei the semantics I will
expound in Chaptei ,. Both foimalisations aie equally sound.
Te foimalisation of the following sentence iequiies two quantifeis:
Eveiy student has a computei.
Tis is cleaily a univeisal claim; so in the fist step one obtains:
Foi all x (if x is a student, then x has a computei).
Tis is not yet the full logical foim of the sentence: x has a computei
contains an existential claim and can be fuithei analysed as
theie is at least one y (x has y and y is a computei).
Now x has y and y is a computei cannot be fuithei analysed and so
they aie put in biackets:
theie is at least one y ((x has y) and (y is a computei)).
Tus the full logical foim of Eveiy student has a computei is
Foi all x {if (x is a student), then theie is at least one y
((x has y) and (y is a computei)}).
Te foimalisation is now stiaightfoiwaid:
x (Px y(Rxy Qy}}
Te dictionaiy is specifed in the following way:
P: . . . is a student
Q: . . . is a computei
R: . . . has . . .
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic :
Generally, the dictionary must provide translations of all sentence letters,
predicate letters and constants occurring in the formalisation. Howevei,
the dictionaiy must not contain tianslations foi the vaiiables. Vaiiables in
sentences nevei iefei to paiticulai objects; they aie only used foi making
univeisal oi existence claims.
Heie is a somewhat moie complicated example:
If its iaining, then Bill ieads a book oi a newspapei.
{If (its iaining), then theie is at least one x ((Bill ieads x)
and ((x is a book) oi (x is a newspapei))).
Te piopei name Bill is tianslated as a constant; the sentence Its iaining
is tianslated as a sentence lettei:
P x (P
2
ax (Qx Rx}}.
I have diopped the outei biackets accoiding to Biacketing Convention :.
In the dictionaiy I have iestoied all aiity-indices. In paiticulai, one must
avoid any confusions between the sentence lettei (0-place piedicate lettei)
P and the 2-place piedicate lettei P
2
.
P: its iaining,
Q
I
: . . . is a book,
R
I
: . . . is a newpapei,
P
2
: . . . ieads . . .
Teinaiy piedicate letteis aie needed foi foimalising sentences such as
the following:
Teie is a countiy between Spain and Fiance.
Te logical foim of this sentence is
Teie is at least one x {(x is a countiy) and (x is between
Spain and Fiance)).
Volker Halbach :oo8/:oo
Te Syntax of Predicate Logic :
By foimalising this, one obtains the following sentence of piedicate logic:
x (Px Qxbc}.
Te dictionaiy is as follows:
P: . . . is a countiy
Q: . . . is between . . . and . . .
b: Spain
c: Fiance
Using the techniques outlined so fai, one can foimalise faiily com-
plicated sentences. Teie aie, howevei, some pioblem cases. Befoie
discussing moie intiicate pioblems of foimalisation in Chaptei ,, I shall
intioduce the semantics of piedicate logic. Without having discussed
the semantics of L
2
fist, it would be dimcult to judge the soundness of
tianslations between English and the language L
2
.
Volker Halbach :oo8/:oo
, Te Semantics of Piedicate Logic
Discussions in metaphysics and in othei aieas in philosophy have been
spuiied by investigations into semantics. Wheieas the semantics of the
language L
I
of piopositional logic is somewhat ciude and philosophically
not veiy exciting, the semantics of the language L
2
of piedicate logic
touches upon questions that aie at the coie of old debates in metaphysics.
In this chaptei I shall confne myself to the technical coie of the se-
mantics of L
2
. Te philosophical issues will iesuiface in latei discussions
about tianslating English sentences into sentences of L
2
. Te technical
account that I am going to piesent can be tiaced back to Taiski (:,o) and
subsequent woik by Taiski, although I will deviate fiom Taiskis oiiginal
appioach in many details and in my notation. Taiskis defnition of tiuth
had a piofound infuence on many aieas not only in philosophy, but also
in mathematical logic, linguistics and computei science.
In English, phiases such as Paiis oi Julius Caesai, which aie usually
foimalised as constants, and piedicate expiessions such as is tiied oi
loves, have fxed meanings. In the language L
2
of piedicate logic, the
constants and piedicate letteis will not be assigned fxed meanings. Tis
is not because it is not possible to assign fxed meanings to them, but
iathei because the validity of aiguments oi the piopeity of being logically
tiue do not depend on the paiticulai meanings of constants and piedicate
letteis. A sentence of the language L
2
will be defned to be logically
tiue, foi instance, if and only if it is tiue undei any inteipietation of the
constants and piedicate letteis. Tus any paiticulai inteipietations of
constants and piedicate letteis do not mattei foi logical tiuth. A similai
iemaik applies to validity: an aigument in L
2
will be defned to be valid
if and only if theie is no inteipietation undei which the piemisses aie all
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic
tiue and the conclusion is false. Tus, again, the validity of an aigument
does not depend on any specifc inteipietation we could assign to the
constants and piedicate letteis.
As in the case of piopositional logic, the notion of an inteipietation
fiomChaiacteiisation :., will be made piecise by the notion of a stiuctuie:
stiuctuies piovide inteipietations foi the non-logical, subject-specifc
vocabulaiy, that is, foi piedicate letteis and constants. Te inteipietation
that is assigned to a symbols by a stiuctuie is called the semantic value
oi the extension of the symbol in the stiuctuie.
,. : s1vUc1Uvis
Te semantics of the language L
2
will be given in the few defnitions
in italics in this chaptei. Te bulk of the chaptei is only an attempt to
motivate and elucidate these defnitions.
I stait by looking back at the semantics of the language L
I
of pioposi-
tional logic: whethei a sentence of the language L
I
is tiue depends on the
tiuth oi falsity of the sentence letteis in that sentence. Te tiuth-values of
all the sentence letteis aie given by an L
I
-stiuctuie. Ten the tiuth tables
of the connectives allow one to calculate the tiuth-values of sentences
foimed with connectives.
Stiuctuies foi piedicate logic aie moie complicated: L
2
-stiuctuies
need to deteimine moie than meiely the tiuth-values of sentence letteis
because the language L
2
contains also othei symbols, namely piedicate
letteis and constants. L
2
-stiuctuies assign semantic values to these sym-
bols as well. Sentence letteis will ieceive tiuth-values as theii semantic
values in the same way as in piopositional logic, but piedicate letteis will
be assigned semantic values of a difeient kind.
Whethei a sentence of L
2
is tiue oi false does not only depend on the
semantic values of the constants and the sentence and piedicate letteis, but
also ovei which objects the quantifeis aie taken to iange. Tis situation
is similai to English: the tiuth-value of the English sentence All glasses
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic ,
aie empty depends in pait on whethei the expiession all glasses is taken
to iange only ovei the glasses on a paiticulai table oi in a paiticulai ioom
oi ovei all glasses in the woild. Te sentence is usually utteied when one
is talking about paiticulai glasses.
Tus, one of the things that an L
2
-stiuctuie does is specify a domain
of discouise, which is just some non-empty set of objects. Teie aie
no iestiictions on the domain of discouise except that it must not be
empty.
:
If A is an L
2
-stiuctuie, I will wiite D
A
foi the stiuctuies domain
of discouise.
In the language L
2
, constants play a iole compaiable to piopei names
in English, and in English piopei names iefei to objects: the English
piopei name Rome iefeis to (oi denotes) the capital of Italy, Volkei
Halbach iefeis to Volkei Halbach, and so on. Tus, an L
2
-stiuctuie
assigns elements of the domain of discouise to the constants as theii
semantic values.
Sentence letteis aie tieated as in piopositional logic: they ieceive
tiuth-values, that is, eithei T oi F, as semantic values in an L
2
-stiuctuie.
Hence, an L
2
-stiuctuie contains also an L
I
-stiuctuie.
Unaiy (:-place) piedicate letteis coiiespond to English expiessions
such as is gieen, walks, oi is a philosophei. Unaiy piedicate letteis
have sets as theii semantic values. Te piedicate lettei P
I
, foi instance,
can have as its semantic value the set of all gieen objects (oi the set of
all walking objects, oi the set of all philosopheis, oi the empty set). On
page :,, sets weie conceived of as unaiy ielations; so piedicate letteis
have unaiy ielations as semantic values, and an L
2
-stiuctuie must assign
unaiy ielations to unaiy piedicate letteis.
: Empty domains aie not allowed in the tiaditional accounts of semantics foi piedicate
logic. Admitting the empty domain would make the semantics foi L
2
moie clumsy, but it
is peifectly possible to admit them. Fioma philosophical point of viewit would piobably
be moie satisfying to admit the empty domain, but I want to avoid the additional
technical complications, and I shall theiefoie follow the tiaditional account. Te efects
of the exclusion of the empty domain will be explained below by means of examples.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic o
Binaiy piedicate letteis coiiespond to expiessions such as loves oi
is biggei than. Binaiy piedicate letteis aie inteipieted by binaiy ielations,
that is, by sets of oideied paiis. Te piedicate lettei P
2
, foi instance, can
have the ielation of loving, that is, the set of all oideied paiis d, e such
that d loves e, as its semantic value. Tus, an L
2
-stiuctuie must assign
binaiy ielations to binaiy piedicate letteis as theii semantic values.
Analogously, ,-place piedicate letteis aie inteipieted by ,-place iela-
tions, that is, sets of tiiples, and geneially piedicate letteis with aiity n
aie assigned n-aiy ielations (see Section :.).
In the following list I summaiise which objects aie assigned to expies-
sions of L
2
by an L
2
-stiuctuie as theii semantic values oi extensions.
L
2
-expiession semantic value
constant object
sentence lettei tiuth-value
unaiy piedicate lettei set, unaiy ielation
binaiy piedicate lettei binaiy ielation
(= set of oideied paiis)
piedicate lettei of aiity , ,-place ielation
(= set of tiiples)
In sum, an L
2
-structure specifes a non-empty set as domain of dis-
course, it assigns elements of the domain to constants, it assigns a truth-
value to every sentence letter, and it assigns an n-ary relation to every
predicate letter.
Te defnition of an L
2
-stiuctuie can be spelled out moie piecisely
in technical teims. I mention this defnition only foi the sake of those
ieadeis who want the full stoiy. I shall not make use of this defnition in
what follows.
uiiii1io,.: (L
2
-s1vUc1Uvi). AnL
2
-structure is anordered pair D, I
where D is some non-empty set and I is a function from the set of all con-
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic ,
stants, sentence letters and predicate letters such that the value of every con-
stant is an element of D, the value of every sentence letter is a truth-value
T or F, and the value of every n-ary predicate letter is an n-ary relation.
One might wondei why vaiiables aie not mentioned in the defnition
of an L
2
-stiuctuie. But just as he does not stand foi a paiticulai object
in the geneial claim If a ieadei is peiplexed, he stops ieading, a bound
vaiiable does not stand foi a paiticulai object in a sentence of L
2
. Foi this
ieason, L
2
-stiuctuies do not assign semantic values to vaiiables.
Foi technical ieasons, howevei, it is convenient to have semantic
values not only foi sentences but also foi foimulae with occuiiences of
fiee vaiiables. Foimulae with occuiiences of fiee vaiiables will also be
assigned tiuth-values as semantic values. Whethei a foimula like P
I
x
with an occuiience of a fiee vaiiable will ieceive the tiuth-value T oi F
depends on what the vaiiables stand foi in the same way He stops ieading
is tiue oi false foi some peisons but not foi otheis. Moie than one vaiiable
may occui fieely in a foimula of L
2
: whethei the foimula R
2
xy R
I
z
ieceives the tiuth-value T oi F depends on what the vaiiables x, y, and
z stand foi. In addition to L
2
-stiuctuies, I theiefoie intioduce a list that
assigns an object to eveiy vaiiable of L
2
. Tis list afects only the tiuth oi
falsity of foimulae with occuiiences of fiee vaiiables, but it does not afect
the tiuth oi falsity of sentences (that is, foimulae with no fiee vaiiables).
A variable assignment over an L
2
-structure Aassigns to each variable
an element of the domain D
A
of A.
:
Occasionally I will diop the specif-
cation ovei the L
2
-stiuctuie A, when it is cleai fiom the context which
L
2
-stiuctuie is meant.
One may think of a vaiiable assignment as a table with two lines
that has all vaiiables as entiies in the fist line line, and elements of the
domain of discouise as entiies in the othei line. Foi instance, theie is a
vaiiable assignment ovei an L
2
-stiuctuie with the set of all Euiopean
: Moie foimally, one can take a vaiiable assignment ovei D to be a function fiom the set
of all vaiiables into D.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic 8
cities as domain, that assigns Rome to x, y
I
, and z
I
, Paiis to y, Beilin to
z, London to x
I
, Oslo to x
2
. Te assignment assigns elements to all
fuithei vaiiables, but of couise I cannot specify an infnite list heie. Te
beginning of the vaiiable assignment may be visualised as follows:
x y z x
I
y
I
z
I
x
2
Rome Paiis Beilin London Rome Rome Oslo
An L
2
-stiuctuie Aand a vaiiable assignment ovei Atogethei assign
semantic values to eveiy vaiiable, constant, sentence lettei, and piedicate
lettei. I will write |e|
A
for the semantic value of the expression e in the
L
2
-structure A under the variable assignment over A. Tus, foi any L
2
-
stiuctuie A and any vaiiable assignment ovei A the semantic values of
the iespective L
2
-expiessions aie as follows:
(i) Foi any constant t, |t|
A
is the object in the domain D
A
of Aassigned
to t by A.
(ii) Foi any vaiiable v, |v|
A
is the object in D
A
assigned to the vaiiable v
by the vaiiable assignment .
(iii) Foi any sentence lettei , ||
A
is the tiuth-value (eithei T oi F)
assigned to by A.
(iv) Foi any unaiy piedicate lettei , ||
A
is the unaiy ielation, that is,
the set, assigned to by A.
(v) Foi any binaiy piedicate lettei , ||
A
is the binaiy ielation, that
is, the set of oideied paiis, assigned to by A.
(vi) Foi any ,-aiy piedicate lettei , ||
A
is the ,-aiy ielation, that is,
the set of oideied tiiples, assigned to by A.
And so on foi piedicate letteis of highei aiity.
,
, Teiefoie, if A is the oideied paii D, I, then foi all constants, and sentence and
piedicate letteis ,
A
= I(). Tis is what is expiessed by (i) and (iii)(vi).
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic
,. : 1vU1u
Te function | . . . |
A
gives semantic values foi all vaiiables, constants, sen-
tence letteis, and piedicate letteis. In this section, | . . . |
A
will be extended
to covei complex foimulae as well, that is, foimulae that aie not meie
sentence letteis.
Te defnition in which tiuth values aie assigned to foimulae with
connectives and quantifeis will be inductive. Tat is, fist I shall de-
fne | . . . |
A
foi atomic foimulae (Defnition .,), and then I shall defne
the semantic values of foimulae containing connectives and quantifeis.
Any foimula of L
2
is eithei tiue oi false in an L
2
-stiuctuie A undei a
vaiiable assignment ovei the L
2
-stiuctuie A. Tus, foi any foimula
eithei ||
A
= T oi ||
A
= F obtains (but not both). ||
A
= T is ofen iead
as satisfes in A.
A
is an element of the set |P
I
|
A
. Similaily, R
2
xc is tiue if and only
if the oideied paii |x|
A
, |c|
A
, that is, the oideied paii with the value of x
as its fist component and the extension of c as its second component,
is in the extension of R
2
, that is, in the ielation |R
2
|
A
. Teiefoie, the
vaiiable assignment impinges on the tiuth-values of foimulae with
fiee occuiiences of vaiiables, such as R
2
xc, because |x|
A
is given by the
vaiiable assignment .
In the sentence P
I
a the unaiy piedicate lettei P
I
ieceives a unaiy
ielation, that is, some set, as its extension (semantic value). I shall assume
Many authois piefei to wiite A [] oi something similai to expiess that satisfes
the foimula in A.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :oo
that |P
I
|
A
is the set Rome, London, Paiis} and |a|
A
is Rome. On that
assumption, P
I
a is tiue because Rome is in the set Rome, London, Paiis},
that is, |a|
A
|P
I
|
A
. Te case of unaiy piedicate letteis is coveied by
the following clause because |a|
A
is the same as |a|
A
accoiding to the
assumption on page :, wheie it has been stipulated that d and d aie the
same foi any object d.
(i) |t
I
. . . t
n
|
A
= Tif and only if |t
I
|
A
, . . . , |t
n
|
A
||
A
, wheie is an
n-aiy piedicate lettei (n must be I oi highei), and each of t
I
, . . . , t
n
is eithei a vaiiable oi a constant.
Tus, this clause deteimines whethei a vaiiable assignment satisfes a
foimula like P
I
a, P
I
x, R
2
xy, oi Q
3
xc y in a stiuctuie.
If a foimula is built up by means of connectives fiom othei sentences,
then tiuth-values can be assigned to this foimula in the style of the De-
fnition :.o of tiuth in an L
I
-stiuctuie: foi instance, if the L
2
-foimulae
and both have semantic value T, then the foimula should also
have tiuth-value T; otheiwise it should have F as its extension (semantic
value). Tus, a vaiiable assignment satisfes the foimula in an
L
2
-stiuctuie, if and only if satisfes and in that stiuctuie. Similaily
a vaiiable assignment satisfes a foimula in an L
2
-stiuctuie, if and
only if does not satisfy itself in the stiuctuie. Tis can be expiessed
moie foimally by the following two defnitional clauses:
(ii) ||
A
= T if and only if ||
A
= F.
(iii) | |
A
= T if and only if ||
A
= T and ||
A
= T.
Te clauses (iv)(vi) foi the iemaining thiee connectives , , and aie
similai and will be listed below.
It iemains to defne the semantic value, that is, the tiuth-value, of
quantifed foimulae fiom the semantic values of shoitei foimulae. Tat
is, I want to fnd clauses analogous to (i) and (ii) foi quantifeis. As an
example I considei the following L
2
-sentence:
x Rxy
Whenshould a vaiiable assignment satisfy this foimula ina stiuctuie A:
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o:
Assume, foi instance, that the vaiiable assignment looks like this:
x y z x
I
y
I
z
I
x
2
Rome Paiis Beilin London Rome Rome Oslo
In this case | y|
A
is Paiis.
Assuming that R is tianslated as . . . is smallei than . . . , the foimula
x Rxy coiiesponds to the English phiase Teie is something smallei
than it. Te pionoun it plays the iole of the vaiiable y that occuis fieely
in x Rxy. Now Teie is something smallei than it is satisfed by Rome
(which is assigned to y by ) if theie is something smallei than Rome,
that is, if theie is something (foi it
I
) satisfying it
I
is smallei than it
2
when it
2
is taken to stand foi Rome.
One can expiess this moie piecisely and peispicuously in teims of
vaiiable assignments foi the foimal language L
2
: the vaiiable assignment
satisfes x Rxy if and only if theie is a vaiiable assignment satisfying
Rxy that difeis fiom at most in what is assigned to x. Teie is such
a vaiiable assignment , assuming that |R|
A
is a ielation containing the
paii Oslo, Rome:
x y z x
I
y
I
z
I
x
2
Oslo Paiis Beilin London Rome Rome Oslo
Tis vaiiable assignment difeis fiom only in the entiy foi x. Since
theie is such a vaiiable assignment, satisfes the foimulae xRxy (in a
stiuctuie A wheie Oslo, Rome is an element of the extension |R|
A
of
R).
Of couise the vaiiable assignment must not difei in the entiy foi
y, as the question is whethei the vaiiable assignment satisfes xRxy,
that is, whethei xRxy is tiue when y is taken to stand foi Rome.
Geneially, a vaiiable assignment satisfes a foimula x if and only
if theie is a vaiiable assignment satisfying that difeis fiom only
in the entiy foi x. may have fiee occuiiences of othei vaiiables than
y; foi this ieason must agiee with on all vaiiables with the possible
exception of x.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o:
So I defne foi all vaiiables v and foimulae :
(viii) |v |
A
= T if and only if ||
A
= T foi at least one vaiiable assign-
ment ovei Adifeiing fiom in v at most.
By saying that difeis fiom in v at most, I mean that |u|
A
= |u|
A
foi all vaiiables u with the possible exception of v. Hence, in teims of
tables, a vaiiable assignment and a vaiiable assignment difei in a
given vaiiable v at most, if the they agiee in all columns with the possible
exception of the column foi the vaiiable v. Te two tables above aie an
example of two vaiiable assignments difeiing in x.
Univeisal quantifeis can be tieated in a similai way. When should
one say, foi instance, that a vaiiable assignment satisfes the foimula
y (Rxy Ryz} in a stiuctuie A: Tat is, when should the following
obtain:
|y (Rxy Ryz}|
A
= T
y expiesses geneiality. will satisfy y (Rxy Ryz} in A if and only
if eveiything in the domain of A will make Rxy Ryz tiue if it is taken
to stand foi y (with the values of x and z unchanged fiom ). Tus,
satisfes y (Rxy Ryz} in A if and only if eveiy that difeis fiom
only in y satisfes Rxy Ryz in A.
Tis can be geneialised to all vaiiables v and L
2
-foimulae : a vaiiable
assignment satisfes a foimula v in a stiuctuie A if and only if eveiy
vaiiable assignment that difeis fiom at most inv satisfes . Of couise,
only vaiiable assignments ovei Aaie consideied: the vaiiable assignments
can only assign objects fiom the domain of Ato the vaiiables.
Te geneial clause can now be stated as follows:
(vii) |v |
A
= T if and only if ||
A
= T foi all vaiiable assignments
ovei Adifeiing fiom in v at most.
I will now collect the difeient clauses into a defnition of satisfaction.
Given an L
2
-stiuctuie, this defnition deteimines foi any vaiiable assign-
ment and any L
2
-foimula whethei satisfes in A, that is, whethei
||
A
= T oi ||
A
= F.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o,
uiiii1io ,.: (s:1isi:c1io). Assume Ais an L
2
-structure, is a vari-
able assignment over A, and are formulae of L
2
, and v is a variable.
For a formula either ||
A
= T or ||
A
= F obtains. Formulae other than
sentence letters then receive the following semantic values.
(i) |t
I
. . . t
n
|
A
= T if and only if |t
I
|
A
, . . . , |t
n
|
A
||
A
, where is a
n-ary predicate letter (n must be I or higher), and each of t
I
, . . . , t
n
is either a variable or a constant.
(ii) ||
A
= T if and only if ||
A
= F.
(iii) | |
A
= T if and only if ||
A
= T and ||
A
= T.
(iv) | |
A
= T if and only if ||
A
= T or ||
A
= T.
(v) | |
A
= T if and only if ||
A
= F or ||
A
= T.
(vi) | |
A
= T if and only if ||
A
= ||
A
.
(vii) |v |
A
= T if and only if ||
A
= T for all variable assignments
over A dipering from in v at most.
(viii) |v |
A
= T if and only if ||
A
= T for at least one variable assign-
ment over A dipering from in v at most.
Ingeneial, what assigns to vaiiables not occuiiing fieely ina foimula
does not impinge on whethei satisfes in A. So, if ||
A
= T, and if
is a vaiiable assignment difeiing fiom only in vaiiables that do not
occui fieely in , then also ||
A
= T. In paiticulai, the vaiiable v does
not occui fieely in a foimula of the foimv . Tus, satisfes v in
A independently of what assigns to the vaiiable v. A similai iemaik
applies to foimulae with a univeisal quantifei.
If is a sentence, that is, if no vaiiable occuis fieely in then ||
A
does not depend on the vaiiable assignment at all at all. Hence, if
is a sentence, then ||
A
is the same tiuth value as ||
A
foi all vaiiable
assignments ovei A.
In oidei to simplify the notation one may diop the index foi the
vaiiable assignment and wiite ||
A
if is a sentence. In geneial, one can
diop the vaiiable assignment when the semantic value is the same foi
all vaiiable-assignments. Tis is the case foi constants, sentence letteis,
piedicate letteis, and sentences.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o
Tiuth in an L
2
-stiuctuie is now defned in teims of satisfaction:
uiiii1io ,., (1vU1u). A sentence is true in an L
2
-structure Aif and
only if ||
A
= T for all variable assignments over A.
As pointed out above, the tiuth-value of a sentence in a stiuctuie is the
same foi all vaiiable assignments. Teiefoie, if a sentence is satisfed (in a
stiuctuie) by some vaiiable assignment, it will be satisfed by all vaiiable
assignments. Consequently, a sentence is tiue in an L
2
-stiuctuie Aif
and only if ||
A
= T foi at least one vaiiable assignment ovei A.
Te Defnition ,., of tiuth has geneiated and continues to geneiate
extensive discussion. Te views on its philosophical value difei wildly.
At any iate, Defnition ,., has been a big success as a tool in philosophy,
mathematics, computei science and linguistics. Te extent to which this
defnition can also be adapted to natuial languages such as English is also
a mattei of some contioveisy.
As an example I will considei a specifc L
2
-stiuctuie, which I call E.
Its domain of discouise is the set of all Euiopean cities. It assigns the
set Floience, Stockholm, Baicelona} to Q
I
, the ielation of being smaller
than to R
2
, Floience to a, and London to b. Tis infoimation can be
displayed in the following way:
|Q
I
|
E
= Floience, Stockholm, Baicelona},
|R
2
|
E
= |d, e d is smallei than e |,
|a|
E
= Floience,
|b|
E
= London.
Tus, |R
2
|
E
is the set Floience, London, Floience, Biimingham,. . . }.
I have diopped the index foi the vaiiable assignment and wiitten |Q
I
|
E
iathei than |Q
I
|
E
since vaiiable assignments do not afect the semantic
values of piedicate letteis and constants.
,
ix:mvii ,.. Te sentence R
2
ab is tiue in E.
, What is assigned to othei constants, sentence and piedicate letteis is iiielevant foi the
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o,
Proof. Since Floience is smallei thanLondon, the paii Floience, London
is an element of the ielation of being smaller than, and I can ieason as
follows (the comments to the iight explain what justifes the pioof step
on the lef):
Floience, London |d, e d is smallei than e |
|a|
E
, |b|
E
|R
2
|
E
defnition of E
|R
2
ab|
E
= T Defnition ,.:(i)
Tis shows that R
2
ab is tiue in E.
ix:mvii ,.,. Te sentence x (Q
I
x R
2
xb} is tiue in E.
Proof. Let be an aibitiaiy vaiiable assignment. I distinguish two cases.
Fiist case: |x|
E
is in |Q
I
|
E
, that is, |x|
E
is eithei Floience, Stockholm, oi
Baicelona. As all thiee cities aie smallei than London, and |b|
E
is London,
one has the following:
|x|
E
, |b|
E
|R
2
|
E
|R
2
xb|
E
= T Defnition ,.:(i)
|Q
I
x R
2
xb|
E
= T Defnition ,.:(v)
Second case: |x|
E
is not in |Q
I
|
E
. In this case one can pioceed as
follows:
|x|
E
is not in |Q
I
|
E
|Q
I
x|
E
= F Defnition ,.:(i)
|Q
I
x R
2
xb|
E
= T Defnition ,.:(v)
following. Foi the sake of defniteness, I could stipulate that E assigns the empty set
as extension to all piedicate letteis othei than Q
I
and R
2
, T to all sentence letteis, and
Rome to all constants othei than a and b.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :oo
Teiefoie, in both cases, that is, foi eveiy vaiiable assignment ovei E,
the following obtains:
|Q
I
x R
2
xb|
E
= T.
Consequently, accoiding to Defnition ,.:(vii), |x (Q
I
x R
2
xb}|
E
= T
foi eveiy vaiiable assignment . Hence, by Defnition ,.,, the sentence
x (Q
I
x R
2
xb} is tiue in E.
Te fnal example sentence contains two quantifeis.
ix:mvii ,.o. Te sentence x y (R
2
xy R
2
yx} is tiue in E.
Proof. Let be an aibitiaiy vaiiable assignment ovei E.
Fiist case: |x|
E
is not London (the laigest city in Euiope). Ten change
the entiy foi y into London (if it is not alieady London) and call the
iesulting vaiiable assignment . By defnition, difeis fiom in y at
most. Since eveiy Euiopean city except London itself is smallei than
London, one has the following:
|x|
E
, | y|
E
|R
2
|
E
|R
2
xy|
E
= T Defnition ,.:(i)
|R
2
xy R
2
yx|
E
= T Defnition ,.:(iv)
|y (R
2
xy R
2
yx}|
E
= T Defnition ,.:(viii)
Te last line holds because difeis fiom at most in y.
Second case: |x|
E
is London. Change the entiy foi y in into Floience
(oi any othei Euiopean city smallei than London), and call this vaiiable
assignment ; it difeis fiom only in y. Te fist of the following lines
holds because | y|
E
is Floience, which is smallei than |x|
E
, that is, London:
| y|
E
, |x|
E
|R
2
|
E
|R
2
yx|
E
= T Defnition ,.:(i)
|R
2
xy R
2
yx|
E
= T Defnition ,.:(iv)
|y (R
2
xy R
2
yx}|
E
= T Defnition ,.:(viii)
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o,
Te last line holds, because difeis fiom in y at most.
Teiefoie, I have pioved that |y(R
2
xyR
2
yx}|
E
= T foi any vaiiable
assignment ovei E. Accoiding to Defnition ,.:(vii) this implies by that
|xy (R
2
xy R
2
yx}|
E
= T,
which shows that the sentence is tiue in E.
In piactice haidly anyone will go thiough all these steps explicitly.
Te foiegoing examples should have illustiated how the semantics foi
the language of piedicate logic woiks and how the tiuth oi falsity of all
sentences is deteimined by an L
2
-stiuctuie.
,. , v:iiui1v, iocic:i 1vU1us, :u co1v:uic1ios
With the defnition of tiuth in hand one can now defne such notions as
logical tiuth, contiadiction, the validity of an aigument, and so on. Te
following defnition is analogous to Defnition :., foi piopositional logic.
uiiii1io ,.,.
(i) A sentence of L
2
is logically true if and only if is true in all L
2
-
structures.
(ii) A sentence of L
2
is a contradiction if and only if is not true in
any L
2
-structure.
(iii) A sentence and a sentence are logically equivalent if both are
true in exactly the same L
2
-structures.
(iv) A set of L
2
-sentences is semantically consistent if and only if there
is an L
2
-structure Ain which all sentences in are true. As in propo-
sitional logic, a set of L
2
-sentences is semantically inconsistent if and
only if it is not semantically consistent.
Also, the defnition of validity of an aigument in L
2
follows the pattein
set out in the defnition of validity of an aigument in piopositional logic,
that is, in Defnition :..
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o8
uiiii1io ,.8. Let be a set of sentences of L
2
and a sentence of L
2
.
Te argument with all sentences in as premisses and as conclusion is
valid if and only if there is no L
2
-structure in which all sentences in are
true and is false.
Tis just captuies the intuitive idea that an aigument is valid if and
only if any L
2
-stiuctuie that makes the piemisses tiue also makes the
conclusion tiue.
Tat the argument with all sentences in as premisses and as con-
clusion is valid, is abbreviated as . Instead of one can also say
follows from . Te symbol is the negation of , so, it is defned as
follows. if and only if not .
I have not excluded the possibility that theie is not piemiss at all in
an aigument. So, may be the empty set. If is the empty set and ,
one may also simply wiite . As I have said above, means that
is tiue in all L
2
-stiuctuies in which all sentences in aie tiue. Now if
theie aie no sentences in , one has if and only if is tiue in all
stiuctuies. Consequently, means that is tiue in all L
2
-stiuctuies
tout couit; that is, it means that is logically tiue.
I intioduce a fuithei notational convention, alieady adoptedfoi piopo-
sitional logic: When the sentences in aie wiitten out explicitly, the set
biackets aiound the sentences may be diopped: Foi instance, one may
wiite
x Qx, x (Qx Rx} x Rx
iathei than the following:
|x Qx, x (Qx Rx}| x Rx.
,. coU1ivix:mviis
How can one show that an aigument in is valid: And how can one show
that it is not valid:
Foi the language L
I
of piopositional logic, these question aie usually
easily answeied (if theie aie not too many oi too long sentences involved):
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :o
One can check out whethei an aigument in L
I
is valid by means of a tiuth
table.
Geneially, in the language L
2
, it is much haidei to show that an
aigument is valid oi not. Teie is not a fnite set of possibilities one
can check out in oidei to fnd out whethei an aigument is valid oi not.
Teie aie infnitely many domains of discouise, and even a single binaiy
piedicate lettei can be inteipieted by infnitely many binaiy ielations.
Showing that an L
2
-aigument is valid by pioving that the conclusion is
tiue in all L
2
-stiuctuies in which all the piemisses aie tiue is, theiefoie,
usually a dimcult task. A moie emcient and elegant way of establishing
that an aigument in L
2
is valid will be intioduced in the next chaptei.
In oidei to show that an aigument is not valid, howevei, one does
not have to piove something about all L
2
-stiuctuies; one has only to fnd
an L
2
-stiuctuie in which all piemisses of the aigument aie tiue and its
conclusion is false. Such L
2
-stiuctuies aie called counteiexamples. In
this section I will explain how to use counteiexamples to dispiove the
validity of aiguments.
An L
2
-structure A is a counterexample to an argument if and only if
all premisses of the argument are true in Aand the conclusion is false in A.
As explained above, a sentence is logically valid if and only if the
aigument with no piemisses and the sentence as its conclusion is valid.
Tus, one can use counteiexamples to showthat a sentence is not logically
tiue:
An L
2
-structure is a counterexample to an L
2
-sentence if the sentence
is not true in it. An L
2
-sentence is logically true if and only if there are no
counterexamples to it.
I will piove the following claim by means of a counteiexample.
ix:mvii ,.. Te sentence Qb x Qx is not logically tiue.
In oidei to fnd a counteiexample to this sentence, one could ieason
infoimally as follows: b could satisfy Q, but othei objects might not
satisfy Q and thus x Qx would be false. One can tuin this into a pioof.
Fiist, an L
2
-stiuctuie with a domain of discouise containing at least two
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic ::o
objects is iequiied. And then, the object that is the semantic value of b
needs to be in the set that is the semantic value of Q, while one of the
othei objects is not in this set. Now this can be tuined into a piopei pioof
of the claim that Qb x Qx is not logically tiue:
Proof. Let B be an L
2
-stiuctuie with the set I, 2} as its domain of dis-
couise and the following semantic values foi Q and b:
o
|Q|
B
= I},
|b|
B
= I.
I will now show that Qb x Qx ieceives the semantic value F
in this L
2
-stiuctuie. Let be the vaiiable assignment that assigns 2 to
eveiy vaiiable, so |x|
B
= 2. Now one can ieason as follows, using as an
abbieviation foi is not an element of :
2 ] I}
|x|
B
] |Q|
B
defnition of and B
|Qx|
B
= F Defnition ,.:(i)
|x Qx|
B
= F Defnition ,.:(vii)
I I}
|b|
B
|Q|
B
defnition of B
|Qb|
B
= T Defnition ,.:(i)
|Qb x Qx|
B
= F Defnition ,.:(v)
By the Defnition ,., of tiuth, Qb x Qx is not tiue in B and thus,
accoiding to Defnition ,.,(i), Qb x Qx is not logically tiue.
o Foi the sake of defniteness I should specify also the value of othei constants, sentence
and piedicate letteis. But as they do not make a difeience to the tiuth-values of sentences,
I will not specify them in this and the following examples.
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :::
In oidei to show that Qb x Qx is not logically tiue, I could have
employed objects othei than the numbeis I and 2. Te two numbeis aie
convenient because of theii shoit names. Teie is no need to use moie
fancy objects. Geneially, it is sensible to keep things simple by choosing
small domains of discouise. In some cases, howevei, it may be necessaiy
to use laige domains; theie aie even cases wheie the domain has to be
infnite.
Next I will tuin to an aigument.
ix:mvii ,.:o. x y Rxy y x Rxy.
Te piemiss could be the tianslation of a sentence such as Foi eveiy-
thing theie is something with the same mass; the conclusion would then
be the tianslation of Teie is something that has the same mass as any
object. If theie aie exactly two things difeiing in mass, then Foi eveiy-
thing theie is something with the same mass is tiue, because eveiy object
agiees with itself in its mass, and the conclusion is false, because the mass
of neithei of the two objects matches the mass of the othei object. Hence,
one can use an L
2
-stiuctuie with a domain containing exactly two objects.
R needs to have a ielation as extension that ielates eveiy object to itself
but not to the othei object in the domain.
Proof. Te L
2
-stiuctuie C is defned as follows:
D
C
= the sun, the moon},
|R|
C
= |the sun, the sun, the moon, the moon|.
Fiist I will show that the piemiss is tiue in the L
2
-stiuctuie C. Let be
an aibitiaiy vaiiable assignment ovei C. Ten change the value of x so
that the values of x and y aie the same, that is, change the entiy foi x into
the sun if | y|
C
is the sun and into the moon if | y|
C
is the moon; call the
iesulting vaiiable assignment . Te fist line in the following pioof, then,
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic :::
holds in viitue of the defnition of |R|
C
, and because |x|
C
= | y|
C
.
|x|
C
, | y|
C
|R|
C
|Rxy|
C
= T Defnition ,.:(i)
|y Rxy|
C
= T Defnition ,.:(viii)
|x y Rxy|
C
= T Defnition ,.:(vii)
Te last line holds because the foiegoing ieasoning applies to all vaiiable
assignments . Hence, the piemiss is tiue in C.
It iemains to show that the conclusion is false in C. Assume to the
contiaiy that y x Rxy is tiue in C. Ten, by Defnition ,.:(viii), theie
is a vaiiable assignment such that the following holds:
|x Rxy|
C
= T.
Tus, by Defnition ,.:(vii),
|Rxy|
C
= T
foi eveiy vaiiable assignment that difeis fiom at most in x. But this
is not the case since one can choose a vaiiable assignment such that
|x|
C
is difeient fiom | y|
C
and so |x|
C
, | y|
C
is not in |R|
C
.
Since the piemiss is tiue in C and the conclusion is false in C, the
aigument is not valid.
ix:mvii ,.::. x (Px Qx Rx}, Pa Ra.
To motivate the counteiexample below one can ieason as follows: Te
piemiss Pa must be tiue in the counteiexample, call it D, so |a|
D
must
be in |P|
D
. Te piemiss x(Px Qx Rx} theiefoie implies that |a|
D
is
eithei in |Q|
D
oi in |R|
D
. So if the piemiss is to be tiue at least one of the
lattei must be the case. As the conclusion Ra must be false, |a|
D
must
not be in |R|
D
, and, theiefoie, |a|
D
must be an element of |Q|
D
. So, one
can employ a counteiexample with a single object in its domain, wheie
that object is in the extensions of P and Q while the extension of R is the
empty set:
Volker Halbach :oo8/:oo
, Te Semantics of Predicate Logic ::,
Proof. Te following L
2
-stiuctuie is a counteiexample:
D
D
= I},
|a|
D
= I,
|P|
D
= I},
|Q|
D
= I},
|R|
D
= .
Teie is only one vaiiable assignment ovei D because its domain D
D
contains only one object and eveiy vaiiable is assigned the numbei I.
Teiefoie, |x|
D
= I foi all vaiiable assignments . To show that the fist
piemiss is tiue, one can ieason as follows:
I I}
|x|
D
|Q|
D
defnitions of and D
|Qx|
D
= T Defnition ,.:(i)
|Qx Rx|
D
= T Defnition ,.:(iv)
|Px Qx Rx|
D
= T Defnition ,.:(v)
Since this holds foi all vaiiable assignments ovei D, as is the only such
vaiiable assignment, x (Px Qx Rx} is tiue in D.
Te second piemiss is also tiue in D:
I I}
|a|
D
|P|
D
defnitions of D
|Pa|
D
= T Defnition ,.:(i)
Te conclusion, howevei, is false in D:
I
|a|
D
|R|
D
defnitions of D
|Ra|
D
= F Defnition ,.:(i)
Tis shows that both piemisses aie tiue and that the conclusion is false in
D. Teiefoie, the aigument is not valid.
Volker Halbach :oo8/:oo
o Natuial Deduction
A valid aigument need not be obviously valid. One can establish the
validity of such an aigument by bieaking it into smallei aiguments and by
showing that one can pass fiom the piemisses to the conclusion thiough
a sequence of small and obvious steps. Tat is, one pioves the conclusion
fiom the piemisses via inteimediate conclusions: the oiiginal piemisses
aie used to deiive obvious conclusions, which in tuin aie employed in the
next step as piemisses to deiive fuithei conclusions, and so on, until the
oiiginal conclusion is obtained. Such a sequence of obvious aiguments is
called a pioof.
Whethei a step is obvious depends on the peispective. Howevei,
one might tiy to show that theie is a fxed list of simple pioof iules
that aie sumcient foi establishing the validity of any valid aigument.
Te iules should be foimulated in a way that makes it easy to check
whethei any given step in a pioof confoims to one of these iules. If a set
iules that can be used in pioofs is fxed, then theie cannot be a seiious
disagieement about the admissibility of any given step in a pioof, and
theie is an objective notion of pioof.
Foi the languages L
I
and L
2
of piopositional and of piedicate logic
one can piovide such a list of admissible iules that legitimate steps in a
pioof.
It is obvious foi which aiguments theie should be pioofs: fist, theie
should be pioofs foi valid aiguments only. Foimally speaking, it must not
be possible to pass fiom the piemisses in a set to a sentence , if it is not
the case that . Te iules must be sound in this sense. Second, the
pioof iules should be complete in the sense that theie should be pioofs
foi all valid aiguments: if , then it should be possible to ieach
Volker Halbach :oo8/:oo
o ^atural Deduction ::,
fiom the piemisses in by going thiough pioof steps that confoim to the
iules foi pioofs.
In oidei to show that , one can then simply give a pioof iathei
than aigue using L
2
-stiuctuies as in the pievious chaptei. Tis will gieatly
facilitate establishing the validity of aiguments in piedicate logic.
Logicians have devised vaiious pioof systems foi difeient puiposes:
some systems aie easy to implement on computeis, otheis aie veiy easy
to state (but haid to woik in), still otheis facilitate geneial investigations
into the notion of piovability. I will employ a system that enables one to
use pioof steps that aie not dissimilai to the steps people take in eveiyday
ieasoning. Te the iules I will specify should be intuitively plausible, but
not eveiy intuitively sound step is a peimissible iule in the system: the
system has not been designed to be as emcient as possible. It is devised
to show that any pioof can be bioken down into simple and elementaiy
steps of veiy few types. If the objective weie a veiy emcient pioof system,
moie iules would have to be added.
Because the pioof iules aie faiily close to pioof steps used in infoimal
pioofs, systems of the kind desciibed in this chaptei aie called Natuial
Deduction systems. Tey weie intioduced independently by Jakowski
(:,) and Gentzen (:,,). Te system I am going to piesent is a vaiiation
of Gentzens veision.
Pioofs in Natuial Deduction stait with an assumption. Any sentence
can be assumed:
assumption rule Te occurrence of a sentence with no sen-
tence above it is an assumption. An assumption of is a proof
of .
It may seem somewhat odd that the solitaiy occuiience of a sentence
is alieady a pioof, but it is convenient to considei a line with a single
sentence as a pioof of fiom the assumption , because this makes the
following defnitions moie stiaightfoiwaid.
Eveiy pioof begins with assumptions. Te fuithei iules foi pioofs
show how to extend a pioof, that is, how to foim longei and longei
Volker Halbach :oo8/:oo
o ^atural Deduction ::o
pioofs by adding fuithei sentences. When stating the iules I will talk
about appending sentences to alieady existing pioofs. By this I mean the
following: one appends a sentence to a pioof by diawing a hoiizontal
line undei the pioof and then wiiting undei this line. One appends
a sentence to two (oi thiee) pioofs by wiiting the pioofs side by side,
then diawing a single line undei all of these pioofs, and then wiiting
undei that single line.
All the iules enable one to append only a single sentence in a given
step. Tus, in eveiy pioof theie is always a single sentence at the bottom
(oi the ioot) of the pioof. Te pioof is a pioof of this sentence . Pioofs
have theiefoie the shape of (upwaid-bianching) tiees.
Foi each connective and quantifei theie is an intioduction iule and
an elimination iule. I shall use abbieviations: foi instance, Intio is
shoit foi -intioduction iule.
o. : vvovosi1io:i iocic
Foi the sake of those who aie concentiating just on piopositional logic, I
shall only use examples in L
I
in this section. Nonetheless, the iules apply
equally to piedicate logic.
I will stait with the iules foi conjunction:
Intro Te result of appending to a proof of and a
proof of is a proof of .
Giaphically, this iule allows one to wiite a pioof ending with and a
pioof ending with side by side, to diaw a hoiizontal line below both,
and to wiite undei this line. Tus an application of the iule will
have the following shape:
Intio
Volker Halbach :oo8/:oo
o ^atural Deduction ::,
Te assumptions in the pioof of aie all the assumptions in the pioofs
of and , iespectively, because any assumption in the pioof of oi ,
that is, any sentence with no othei foimula above it in the pioofs of oi
, will also be an assumption in the pioof of , that is, it will not have
a sentence above it in the pioof of .
Te oidei of the pioofs of and does not mattei. Te iule does
not iequiie that the pioof of is wiitten to the lef. So an application of
Intio can also look like this:
Intio
Te same applies to othei iules: in iules wheie a sentence is appended to
two (oi in one case thiee) pioofs, the iules allow one to wiite down the
pioofs in any oidei.
Foi theie aie two elimination iules:
Elim: Te result of appending to a proof of is a proof
of .
Te othei iule allows one to keep :
Elim: Te result of appending to a proof of is a proof
of .
Te two iules can be depicted as follows:
Elim:
Elim:
Te fist iule allows one to diop the second pait of the conjunction, the
second iule allows one to diop the fist pait.
With these iules in hand I can alieady constiuct a pioof. Fiist, I will
assume (P Q} R. Te iule Elim: allows me to append P Q to the
pioof, and Elim: allows me to append Q to the iesulting pioof. So I can
obtain a pioof of Q undei the assumption (P Q} R:
Volker Halbach :oo8/:oo
o ^atural Deduction ::8
(P Q} R
Elim:
P Q
Elim:
Q
Te labels Elim: and Elim: do not belong to the pioof; they aie
meie comments that aie intended to help the ieadei to giasp the pioof.
Occasionally, when I think that the labels will facilitate undeistanding, I
will add them.
Next I shall specify iules foi the aiiow . In oidei to motivate the
intioduction iule foi the aiiow, I will look at how one might establish
and if . . . , then . . . like the following:
(A) If CO
2
-emissions aie not cut, tempeiatuies will iise globally.
To deiive the conclusion, one will use additional assumptions about cli-
mate change, the gieenhouse efect and so on, which I will not specify
heie. Using these additional assumptions, one could aigue as follows foi
(A):
Assume that CO
2
-emissions aie not cut. Ten the CO
2
-level
in the atmospheie . . . [now one uses the additional assump-
tion, piobably talking about the gieenhouse efect, and con-
cludes:] so tempeiatuies will iise globally. Teiefoie, if CO
2
-
emissions aie not cut, tempeiatuies will iise globally.
One makes the assumption that CO
2
-emissions aie not cut only in oidei
to showthat in that case tempeiatuies will iise globally. Tis assumption is
made only foi the sake of the aigument and once (A) has been concluded,
one is no longei assuming that CO
2
-emissions aie not cut. Te pioof of
(A) is based only on the additional assumptions about climate change etc,
but not on the assumption that CO
2
-emissions aie not cut. Tus, when
one concludes (A), one does not make the assumption anymoie that
CO
2
-emissions aie not cut; one claims only that tempeiatuies will iise if
CO
2
-emissions aie not cut without assuming anything about whethei the
emissions aie cut oi not. Logicians desciibe this by saying that, when one
Volker Halbach :oo8/:oo
o ^atural Deduction ::
concludes (A), one has dischaiged the assumption that CO
2
-emissions
aie not cut.
Geneially, one can aigue foi a claim of the foim If A, then B by
pioving B fiom the assumption A.
in the following way: one pioves B by assuming A; then one concludes
If A, then B without assuming Aanymoie.
In Natuial Deduction the iule foi intioducing the aiiow woiks in the
same way: one assumes a sentence , deiives a sentence fiom it, and
then the iule allows one to conclude and to get iid of oi dischaige
the assumption of .
Infoimal pioofs one indicates that anassumptionhas beendischaiged
by enclosing that assumption in squaie biackets:
Informal proofs, assumptions are dischargedby surrounding themwith
square brackets. Of couise one must only dischaige assumptions in accoi-
dance with the iules.
Te pioof technique used in the above infoimal pioofs is captuied by
the intioduction iule foi :
Intro Te result of appending to a proof of and
discharging all assumptions of in the proof of is a proof of
.
So one may add to a pioof with at the ioot and then enclose all
assumptions of (that is, all occuiiences of with no line above them)
in the pioof of in squaie biackets.
Te giaphical iepiesentation looks like this:
||
Intio
Tis iule does not iequiie that the pioof of actually contains an assump-
tion of . Only if theie aie any assumptions of in the pioof of , they
must be dischaiged.
Volker Halbach :oo8/:oo
o ^atural Deduction ::o
Te iule foi eliminating is stiaightfoiwaid:
Elim Te result of appending to a proof of and a proof
of is a proof of .
Tis iule is giaphically iepiesented as follows:
Elim
Tis iule is also called the cut iule, because the sentence is cut of
fiom .
Befoie giving some examples, I will intioduce a new piece of notation:
uiiii1io o.:. Te formula is provable from (where is a set of L
2
-
sentences) if and only if there is a proof of with only sentences in as
non-discharged assumptions. Te phrase is provable from is abbrevi-
ated as . If is empty, is abbreviated as . If contains
exactly the sentences
I
, . . . ,
n
, one may write
I
, . . . ,
n
instead of
I
, . . . ,
n
} .
ix:mvii o.:. P Q P.
Proof. I show step by step how to establish this claim. Fiist, P Q is
assumed:
P Q
Applying Elim: yields the following:
P Q
Elim:
P
Te iule Intio allows one to add P Q P and to dischaige P Q:
|P Q|
Elim:
P
Intio
P Q P
Volker Halbach :oo8/:oo
o ^atural Deduction :::
All assumptions in this pioof have been dischaiged. Tus, P Q P is
piovable fiom the empty set of piemisses, that is, P Q P.
Tis is a typical pioof of a sentence of the foim : usually one
assumes , aiiives thiough some steps at , and then uses Intio to
deiive and to dischaige any assumptions of .
Te intioduction iules foi disjunction aie as follows:
Intro: Te result of appending a sentence to a proof
of is a proof of .
As in the case foi the elimination iules foi , theie is also anothei intio-
duction iule foi :
Intro: Te result of appending a sentence to a proof
of is a proof of .
Te giaphical iepiesentations aie as follows:
Intio:
Intio:
Te elimination iule foi is somewhat tiicky. It coiiesponds to the
following infoimal pioof stiategy:
Assume Aoi B is given. Ten one can tiy to piove C by mak-
ing a case distinction: Fiist, one tiies to deiive C assuming
A; then one tiies to deiive C assuming B. If C can be deiived
in both cases, then (given Aoi B) one may conclude C.
Tis type of ieasoning ieasoning by diawing the same conclusion fiom
both paits of a disjunction is captuied in the following iule:
Elim Te result of appending to a proof of , a proof of
and another proof of , and of discharging all assumptions
of in the frst proof of and of discharging all assumptions
of in the second proof of , is a proof of .
Volker Halbach :oo8/:oo
o ^atural Deduction :::
An application of Elim looks like this:
||
||
Elim
||
Intio
||
Elim
Intio
Elim:
Elim:
In the following pioof I will illustiate the use of the iules foi the double
aiiow.
ix:mvii o.,. (P Q} (Q P}.
Proof.
|P Q| |P|
Q |Q|
P
Q P
(P Q} (Q P}
|Q P| |Q|
P |P|
Q
P Q
(Q P} (P Q}
(P Q} (Q P}
Foi the last line Intio is used.
In cases like the following it is ofen useful to stait fiomthe ioot and to
considei how one might have obtained the sentence that is to be pioved.
Volker Halbach :oo8/:oo
o ^atural Deduction ::,
ix:mvii o.8. (Q R} Q R.
Proof.
(Q R}
|Q| |R|
Q R
R
Q R
Te assumption (QR} is not biacketed; it iemains as an undischaiged
assumption. Tus, (Q R} Q R is established.
Finally, heie is a pioof of a vaiiant of the ex falso quodlibet-piinciple:
ix:mvii o.. P, P Q.
Proof. Te following pioof contains a stiange application of Elim. Te
foimula Q is intioduced although Q has nevei be assumed. Elim
allows one to intioduce and to dischaige all assumptions of even if
theie aie not any.
P P
Elim
Q
Tis application of Elim complies with Elim, as it is not a iequiiement
that theie actually be assumptions of .
Te following iesult says that the iules foi Natuial Deduction have
been chosen in such a way that theie aie pioofs foi exactly those aigu-
ments in L
I
that aie valid.
1uiovim o.:o (:uiqU:cv iov vvovosi1io:i iocic). Assume that
and all elements of are L
I
-sentences. Ten if and only if .
I will not piove this theoiem heie.
In paiticulai, if a sentence of L
I
is a tautology (logically tiue), then
theie is a pioof of that sentence without undischaiged assumptions.
Te adequacy iesult says that the iules given heie aie sumcient foi
pioving a conclusion fiom piemisses if the aigument is valid. Any iule
Volker Halbach :oo8/:oo
o ^atural Deduction ::8
that can be added is eithei not sound, that is, it allows one to piove
sentences that aie not logically tiue, oi it is dispensable (but it may piovide
a shoitcut). I shall ietuin to adequacy in moie detail in Section ,.:; theie
adequacy will be discussed with iespect to piedicate logic.
o. : vviuic:1i iocic
Te pioof iules of piopositional logic, which weie expounded in the
pievious section, apply to all sentences including the sentences of L
2
.
Te following example contains sentences of the language L
2
that aie not
in L
I
; but the pioof iequiies only iules fiom piopositional logic, that is,
intioduction and elimination iules foi connectives.
ix:mvii o.::. (x Px y Py} x Px.
Proof. Since one wants to piove a sentence of the foim , the ,
which coiiesponds to (x Px y Py} in the piesent example, is as-
sumed. Fiom this assumption one tiies to aiiive at x Px. Since this is
a negated sentence, one may hope to get it by using Intio; so x Px is
assumed. Tat is almost a contiadiction: applying Intio: to x Px gives
x Px y Py, and the contiadiction is obtained:
(x Px y Py}
x Px
Intio:
x Px y Py
Now Intio is applied by appending x Px to the two pioofs and dis-
chaiging the assumption x Px:
(x Px y Py}
|x Px|
Intio:
x Px y Py
Intio
x Px
One fnishes the pioof by applying Intio:
Volker Halbach :oo8/:oo
o ^atural Deduction ::
|(x Px y Py}|
|x Px|
Intio:
x Px y Py
Intio
x Px
Intio
(x Px y Py} x Px
All assumptions aie dischaiged, so the claim is established.
I will now tuin to the iules foi and . Fiist, I will explain the iule
foi eliminating , which may be motivated by consideiing the following
aigument:
If some peison has made moie than ten mistakes then that
peison wont pass. Teiefoie, if Ben has made moie than ten
mistakes he wont pass.
Heie one is going fiom a univeisal claim to a claim about a specifc in-
stance. Te iule foi eliminating , which is also known as univeisal
instantiation-iule, allows one to pass fiom a univeisally quantifed sen-
tence to a special instance. Foi instance, the iule licenses the step fiomthe
univeisally quantifed sentence x (Px Qx} to the instance Pa Qa.
In oidei to give a geneial foimulation of ElimI employ the following
defnition:
uiiii1ioo.::. Assume v is a variable, t a constant, and anL
2
-formula
with at most v occurring freely. Ten |t]v| is the sentence obtained by re-
placing all free occurrences of v in by t.
Foi instance, Px |b
2
]x| is Pb
2
; and y (Pxy x Rxy} |b]x|is the
sentence y(Pby x Rxy}. In the second case only the fist occuiience
of x has been ieplaced because the othei two aie bound occuiiences.
Te iule foi eliminating the univeisal quantifei can now be stated as
follows:
Elim Te result of appending |t]v| to a proof of v is a
proof of |t]v|.
In this iule it is assumed that t is a constant, v a vaiiable, and v is a
sentence (so that only v can occui fieely in ).
Tus an application of this iule has the following foim:
Volker Halbach :oo8/:oo
o ^atural Deduction :,o
v
Elim
|t]v|
Heie, is an L
2
-foimula in which only the vaiiable v occuis fieely; t is a
constant.
Tus the iule allows one to diop the quantifei v at the beginning of
a sentence and to ieplace all fiee occuiiences of v in the iesulting foimula
with the constant t.
Te following example explains why only fiee occuiiences of the
vaiiable aie ieplaced:
y (Py y Qy}
Elim
Pb
3
y Qy
If one weie allowed to ieplace also the last occuiience of y, which is
bound, one would to get to Pb
3
yQb
3
, which is logically equivalent to
Pb
3
Qb
3
. Tis step is cleaily not sound: fiomy(Py y Qy} it does
not follow that b
3
is Q. Intuitively speaking, only occuiiences of y in
y (Py y Qy} that aie caught by the univeisal quantifei y can be
ieplaced by the constant; the last occuiience of y belongs to the existential
quantifei and must be lef alone.
Te intioduction iule foi the univeisal quantifei is moie dimcult to
state. In oidei to aigue foi the geneial claim that eveiy tiavelleis jouiney
fiom London to Munich in :oo, took ovei two houis, one could ieason
as follows:
Assume somebody tiavelled fiomLondon to Munich in :oo,.
Call him John Doe. If he took the tiain . . . [now each means
of tianspoit is taken into account, and it is aigued in eveiy
case that the jouiney must have taken moie than two houis.]
Teiefoie John Does jouiney took ovei two houis. Teiefoie,
since John Doe is an aibitiaiy peison, eveiy tiavelleis joui-
ney fiom London to Munich in :oo, took ovei two houis.
Volker Halbach :oo8/:oo
o ^atural Deduction :,:
Te idea heie is that one talks about a nondesciipt paiticulai instance as
an example using a name. In the stoiy above, this nondesciipt instance is
a peison who is dubbed John Doe. Te name John Doe is not used to
name a specifc ieal peison. Ten one caiiies out the aigumentation foi
this instance, and concludes that this aigument applies to eveiy instance
because the chosen instance was just used as an example. Foi the validity
of the aigument it is impoitant that the chosen peison is aibitiaiy and
that one did not biing in any specifc infoimation about that peison oi
object (if theie is such a peison oi object at all).
Te iule foi intioducing the univeisal quantifei in Natuial Deduction
woiks in a similai way: x(Px Px} will be established by taking an
aibitiaiy constant, say, c (coiiesponding to the use of a name John Doe
above), and aiguing as follows:
|Pc|
Intio
Pc Pc
Since c was chosen to name a nondesciipt instance as an example, the
claim also holds foi eveiy object. Now, one substitutes the vaiiable x foi
the constant c in Pc Pc and piefxes the univeisal quantifei x to the
foimula. Ten one can add this sentence to the pioof, in accoidance with
the iule foi intioducing :
|Pc|
Intio
Pc Pc
Intio
x(Px Px}
In foimulating the geneial iule Intio, howevei, one must exeicise
caution. One might think that Intio can be foimulated in the following
way: given a pioof of a sentence , one ieplaces a constant in the sentence
by a vaiiable v, wiites v in fiont of the sentence, and then appends the
iesulting sentence to the pioof. Howevei, such a iule would allow one to
deiive false conclusions fiom tiue piemisses foi thiee ieasons.
iivs1 vvoviim. I will illustiate the pioblem by consideiing the fol-
lowing attempted pioof:
Volker Halbach :oo8/:oo
o ^atural Deduction :,:
(P)
y (Py x Ryx}
Elim
Pa x Rax
incoiiect -intioduction
x (Px x Rxx}
If (P) weie a coiiect pioof one could deiive x (Px xRxx} fiom
y (Py x Ryx}. So (P) cannot be a coiiect pioof: it would allow one
to go fiom the foimalisation of Eveiy student sees something to the
foimalisation of Foi eveiy student theie is something that sees itself. Te
fist sentence is tiue while the lattei is false if theie is at least one student
and nothing can see itself.
:
It should be obvious what went wiong. When the constant a was
ieplaced by x in the sentence Pa x Rax, the x ieplacing the last
occuiience of a got caught by the quantifei x:
x (Px x Rx
-
actual binding
x}
Of couise the intention was that the penultimate occuiience of x should
iefei back to the univeisal quantifei x and not to the existential quanti-
fei:
x (Px x Rx
- - -
intended binding
x}
Te pioblem can be avoided in the following way: when one adds
a sentence v to a pioof by applying the iule foi intioducing then
the pieceding sentence must be with all fiee occuiiences of the vaii-
able v ieplaced by the constant t. In othei woids, the iule allows one to
continue a pioof ending with |t]v| by adding the sentence v . So the
iule foi intioducing will take the following foim (with some additional
iestiictions on occuiiences of the constant t specifed below):
|t]v|
v
: Te pioof that theie is a counteiexample to the aigument with y (Py x Ryx) as
piemiss and x (Px xRxx) as conclusion is the content of Exeicise ,.:(ii).
Volker Halbach :oo8/:oo
o ^atural Deduction :,,
In the above attempted pioof (P), I passed fiom Pa x Rax to
x (Px x Rxx}. But this step is not coveied by the iule that licenses
only the step fiom |t]v| to v : Te iesult (Px x Rxx}|a]x| of
ieplacing all fiee occuiiences of x by a is the sentence Pa x Rxx, not
Pa x Rax. Only going fiom Pa x Rxx to x (Px x Rxx}
would be coveied by the iule.
sicou vvoviim. Assume one staits a pioof by assuming Qb and
then geneialises in the following way:
Qb
incoiiect -intioduction
x Qx
Tis is not coiiect. If it weie coiiect, one could apply Intio in the next
step:
|Qb|
incoiiect -intioduction
x Qx
Intio
Qb x Qx
Tus, one would have pioved that Qb x Qx; but Qb x Qx is
ceitainly not logically tiue, as has been shown in Example ,..
Foi this ieason, one must make suie that one is not using any specifc
assumptions containing the constant that is used as an aibitaiy exam-
ple ovei which one can geneialise. So when Intio is applied by going
fiom |t]v| to v , the constant t must not occui in any undischaiged
assumption. In the discussion of the example of John Doe I said that it is
impoitant that the peison chosen as an example (John Doe) is aibitiaiy.
Te iestiiction that the constant must not occui in any undischaiged as-
sumption is the foimal counteipait of the iequiiement that John Doe is an
aibitiaiy peison about whomone does not have any specifc infoimation.
In the eailiei, coiiect pioof of x (Px Px} the constant c occuiied
in the assumption Pc, but that assumption had been dischaiged by the
time Intio was applied. So this pioof meets the iestiiction that c must
not occui in any undischaiged piemiss when the univeisal quantifei is
intioduced.
Volker Halbach :oo8/:oo
o ^atural Deduction :,
1uivu vvoviim. Te following attempted pioof meets all conditions
that weie imposed on applications of Intio to avoid the fist and second
pioblem:
y Ryy
Raa
x Rax
Howevei, the aigument with y Ryy as piemiss and x Rax as its con-
clusion is not valid: it would alow one to go fiom the foimalisation of
Eveithing is self-identical to the foimalisation of John is identical to
eveiything.
,
Tus, some condition must be imposed on the iule Intio
that blocks the step fiom Raa to x Rax. Tis can be done by disallow-
ing one to keep some occuiiences of a in the sentence that is added in
accoidance with the iule. Only passing fiom Raa to x Rxx would be
admissible. Tus, since an application of Intio is a step fiom |t]v|
to v , the iule Intio can be applied only if (that is, the foimula
following the univeisal quantifei) does not contain the constant t, that
is, if no occuiience of t is ietained when Intio is applied.
In the following foimulation of the iule foi intioducing the univeisal
quantifei all thiee pioblems aie avoided:
Intro Assume that is a formula with at most v occurring
freely and that does not contain the constant t. Assume fur-
ther that there is a proof of |t]v| in which t does not occur
in any undischarged assumption. Ten the result of append-
ing v to that proof is a proof of v .
Te iule can be iepiesented as follows:
|t]v|
Intio
v
piovided the constant t does not
occui in oi in any undischaiged
assumption in the pioof of |t]v|.
, Te pioof of y Ryy x Rax is Exeicise ,.:(iii).
Volker Halbach :oo8/:oo
o ^atural Deduction :,,
In less foimal teims, the iestiictions on Intio can be summed up in
two points: Fiist, one must make suie that one is not geneialising ovei a
constant that occuis in an undischaiged assumption oi that is still in the
sentence one tiies to obtain. Second, in an application of the iule, one
has to make suie that the vaiiable of the newly intioduced quantifei is
not caught by a quantifei that is alieady in the foimula.
In the case of the existential quantifei the intioduction iule is the
easy one:
Intro Te result of appending v to a proof of |t]v| is a
proof of v .
|t]v|
Intio
v
Of couise t is a constant, v a vaiiable, and v is an L
2
-sentence.
Te iule foi eliminating the existential quantifei is the most compli-
cated iule. Considei the following aigument:
Teie is at least one epistemologist. All epistemologists aie
philosopheis. Teiefoie, theie is at least one philosophei.
Te two piemisses have x Px and x (Px Qx} as theii iespective foi-
malisations; the foimalisation of the conclusion is x Qx. Te dictionaiy
has the following two entiies:
P: . . . is an epistemologist
Q: . . . is a philosophei
Te coiiesponding aigument in L
2
is valid. Te question is how one
can prove the conclusion fiom the two piemisses. If one could in some
way get Pc fiom the fist piemiss x Px, the iest of the pioof would be
obvious:
Volker Halbach :oo8/:oo
o ^atural Deduction :,o
Pc
x (Px Qx}
Elim
Pc Qc
Elim
Qc
Intio
x Qx
Te pioblem is that Pc does not follow fiomx Px. Te piemiss x Px
just states that theie is some epistemologist; it does not give one a specifc
epistemologist and, in paiticulai, it does not give one a paiticulai name c
foi an epistemologist.
Te conclusion x Qx does not say anything specifc about c. In
the pioof I could have used any othei constant in place of c. So one
might apply the following pioof stiategy: one may assume Pc. Tis is
tantamount to picking an aibitiaiy name like John Doe and assuming
that John Doe is an epistemologist. Once a sentence not containing c is
pioved, one can dischaige the assumption Pc using the piemiss x Px:
the conclusion does not depend any moie on the assumption that c is
one of the Ps.
x Px
|Pc|
x (Px Qx}
Elim
Pc Qc
Elim
Qc
Intio
x Qx
x Qx
Te point in the last step of the pioof is that the piemiss Pc can be dis-
chaiged, so one has now pioved the conclusion fiom the existence claim
x Px iathei than fiom the specifc instance Pc. Infoimally speaking, the
conclusion that theie is a philosophei does not depend on the assumption
that theie is an epistemologist called John Doe.
When making an assumption such as Pc one must not choose a con-
stant about which one alieady has specifc infoimation: the constant
acts as an aibitiaiy example in the same way as in Intio. Te piecise
statement of the elimination iule foi is so convoluted because the ie-
stiictions on the constant must ensuie that the constant can play its iole
as an aibitiaiy label.
Volker Halbach :oo8/:oo
o ^atural Deduction :,,
Elim Assume that is a formula with at most v occurring
freely and that the constant t does not occur in . Assume
further that there is a proof of the sentence in which t does
not occur in any undischarged assumption other than |t]v|.
Ten the result of appending to a proof of v and the proof
of and of discharging all assumptions of |t]v| in the proof
of is a proof of .
As befoie, |t]v| is just with all fiee (and only fiee) occuiiences of v
ieplaced with t.
An application of the iule looks like this:
v
||t]v||
Elim
F
= I.
Now one can ieason as follows:
|x|
F
|R|
F
defnition of F
|Rx|
F
= T Defnition ,.:(i)
|Rx Qx|
F
= T Defnition ,.:(iv)
|Px Rx Qx|
F
= T Defnition ,.:(v)
|x(Px Rx Qx}|
F
= T Defnition ,.:(vii)
Te last line holds because theie is only one vaiiable assignment ovei D
F
.
Tus, the piemiss is indded tiue in the L
2
-stiuctuie F.
It iemains to show that the conclusion is false in F:
|x|
F
is not in |Q|
F
defnition of F
|Qx|
F
= F Defnition ,.:(i)
|x|
F
|P|
F
defnition of F
|Px|
F
= T Defnition ,.:(i)
|Px Qx|
F
= F Defnition ,.:(v)
|x(Px Qx}|
F
= F Defnition ,.:(vii)
Te last line shows that the conclusion is not tiue in F, and thus the claim
,. is established. Consequently, the English aigument is not valid in
piedicate logic.
Generally, in order to show that an English argument is not valid in
predicate logic, one will formalise the argument and provide a counterex-
ample to the resulting L
2
-argument.
I have said that one can show that an English aigument is valid by
pioviding a paitial foimalisation that is a valid L
2
-aigument. In oidei to
Volker Halbach :oo8/:oo
, Formalisation in Predicate Logic :o
iefute the validity of an English aigument, meiely paitial foimalisations
cannot be used: usually an English aigument that is valid in piedicate
logic will have some L
2
-foimalisation that is not valid. Tus, in oidei to
show that an English aigument is not valid in piedicate logic, one needs
to considei its full foimalisation.
Volker Halbach :oo8/:oo
8 Identity and Defnite Desciiptions
In this chaptei I will intioduce a thiid foimal language. Tis newlanguage
L
=
is the language of piedicate logic with identity; it is only a small
iefnement of the language L
2
of piedicate logic: L
=
is just L
2
with the
addition of the new symbol = foi identity.
8. : qU:ii1:1ivi :u Umivic:i iui1i1v
Philosopheis have distinguished two difeient notions of identity: quali-
tative identity and numeiical identity. In the following example I piesent
a case of qualitative identity.
Teie is a fountain pen in my teaching ioom and anothei
fountain pen in my study at home. Tey aie the same model,
the same coloui, and both aie still in piistine condition. Tus,
I have two identical fountain pens.
Teie aie two pens, and they aie qualitatively identical because they aie
in all ielevant aspects veiy similai.
To explain numeiical identity, I will expand the example a little bit:
A fountain pen expeit sees my pen at home afei having
seen the pen in my teaching ioom the day befoie. He may
wondei whethei I have taken the pen home and ask: Is this
the same pen as the pen in youi teaching ioom: oi Is this
pen identical to the pen I saw yesteiday:
He knows all the ways that the pen at my home and the pen in my teaching
ioomaie similai, and so he is not asking whethei they aie the same coloui
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :oo
oi aie of the same biand etc; iathei he wants to know whethei it is the
same pen, that is, whethei he has seen two (qualitatively identical) pens
oi whethei he has seen one and the same pen. So in his question identity
has to be undeistood numeiically. In the numeiical sense the pen in my
teaching ioom is not identical with the pen in my study at home, that is,
theie aie two pens.
Occasionally it is not cleai which kind of identity is at issue in a given
sentence. Te claim
Robin saw the same tiee yeais latei in the gaiden
might be taken to expiess that Robin saw one and the same tiee in the
gaiden yeais latei, oi that he saw a tiee of the same kind in the gaiden
yeais latei.
Qualitative identity may be foimalised by a binaiy piedicate lettei
of L
2
. Its tieatment in piedicate logic with identity does not difei fiom
the tieatment of most othei binaiy piedicates.
Numeiical identity, in contiast, is given a special status. In what fol-
lows I shall talk exclusively about numeiical identity. Numeiical identity
is foimalised by a new, special piedicate lettei.
8. : 1ui sv1:x oi L
=
All foimulae of L
2
aie also foimulae of L
=
. But L
=
also includes a new
kind of atomic foimula.
uiiii1io8.: (:1omic iovmUi:i oi L
=
). All atomic formulae of L
2
are
atomic formulae of L
=
. Furthermore, if t
I
and t
2
are variables or constants,
then t
I
=t
2
is an atomic formula of L
=
.
Examples of atomic foimulae of L
=
aie x = x, x = z
36
, a = y, and
a
34
=c
22
, and atomic foimulae of L
2
such as P
3
3
xxc
3
. Otheiwise, theie aie
no deviations fiom the syntax of L
2
, and one can build complex foimulae
using connectives and quantifeis as in L
2
in accoidance with Defnition
., of a foimula of L
2
. Of couise, now one must also allow foi =. Te new
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :o,
symbol = behaves exactly like a binaiy piedicate lettei, with the exception
that = is wiitten between the vaiiables oi constants. Wiiting the identity
symbol like othei piedicate letteis as the fist symbol in atomic foimulae
would look odd, as we aie used to wiiting x = y iathei than =xy, but that
is the only ieason foi wiiting x = y iathei than =xy.
Te foimulae of L
=
aie defned in the same way as the foimulae of L
2
in Defnition .,, with the only exception that the new atomic foimulae
can be used.
uiiii1io 8.: (iovmUi:i oi L
=
).
(i) All atomic formulae of L
=
are formulae of L
=
.
(ii) If and are formulae of L
=
, then , ( }, ( }, ( }
and ( } are formulae of L
=
.
(iii) If v is a variable and is a formula of L
=
then v and v are
formulae of L
=
.
Foi instance, x = y and x(Rxy
2
y
2
=x} aie foimulae of L
=
.
All othei defnitions fiom Section .: caiiy ovei as well. Sentences
of L
=
aie defned as those foimulae in which no vaiiable occuis fieely.
Also, the biacketing conventions aie the same as foi L
2
.
8. , sim:1ics
Te semantics foi L
=
is just a small vaiiation on the semantics foi L
2
, and
so it is not necessaiy to intioduce a new kind of stiuctuie: L
2
-stiuctuies
aie used foi the semantics of L
=
.
Only Defnition ,.: needs to be amended by adding the following
additional clause to (i)(viii), wheie Ais an L
2
-stiuctuie, s is a vaiiable
oi constant, and t is a vaiiable oi constant:
(ix) |s =t|
A
= T if and only if |s|
A
= |t|
A
.
Tus, in the semantics foi L
=
, the symbol = is always inteipieted as nu-
meiical identity. In (ix) the symbol = is used in two difeient ways: its fist
occuiience is a symbol of the foimal language L
=
; the two subsequent
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :o8
occuiiences belong to the language we aie using to desciibe L
=
. In oidei
to avoid this ambiguity some authois put a dot undei the symbol of the
foimal language, but this convention has not ieally caught on. Alteina-
tively, one could avoid the use of = outside the foimal language L
=
by
iefoimulating (ix) in the following way:
(ix) Te vaiiable assignment satisfes s =t in A if and only if |s|
A
and
|t|
A
aie the same object.
In the following I will not tiy to avoid the use of = in these two difeient
ioles: it should be cleai foi eveiy occuiience of the symbol whethei it
is used as a symbol of L
=
oi as a symbol of oui eveiyday mathematical
language.
It follows fiom clause (ix) that foi any L
2
-stiuctuie A, |a = a|
A
= T
and foi any vaiiable assignment ovei A, |x =x|
A
= T, because, tiivially,
|x|
A
is the same object as |x|
A
, and |a|
A
is the same object as |a|
A
. Of
couise, the same applies to vaiiables othei than x and to constants othei
than a.
Te defnitions of validity of arguments, of semantic consistency, of log-
ically true sentences, and so on, carry over from Defnitions ,., and ,.8.
Te method of counteiexamples can be applied in the same way as it
was foi the language L
2
. As an example I will showthat x y x = y does
not followin piedicate logic with identity fiomthe piemiss x Pxy Py.
ix:mvii 8.,. x Px y Py x y x = y
Te aigument could be a foimalisation of the following English aigu-
ment:
Teie is a painting and theie is a painting. Teiefoie theie
aie at least two things.
A moie liteial tianslation of the conclusion x y x = y would be theie
is a thing such that theie is a thing that is not identical to the fist; but
that is just a longwinded way of saying that theie aie at least two things.
Te piemiss just makes the same claim twice, namely that theie is a
painting. Te use of the two vaiiables x and y does not imply that theie
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :o
aie two difeient paintings. Tis yields the idea foi the following pioof.
Proof. Let B be an L
2
-stiuctuie with (the painting of) Mona Lisa as the
only element in its domain of discouise, and the Mona Lisa } as the
extension of P.
D
B
= the Mona Lisa }
|P|
B
= the Mona Lisa }
Fiist I will show that the piemiss is tiue in this stiuctuie.
Teie is exactly one vaiiable assignment ovei B: it assigns the Mona
Lisa to all vaiiables, so |x|
A
is the Mona Lisa.
the Mona Lisa the Mona Lisa}
|x|
B
|P|
B
|Px|
B
= T Defnition ,.:(i)
|x Px|
B
= T Defnition ,.:(viii)
Since assigns the Mona Lisa to y as well, the same ieasoning can be
applied to y:
| y|
B
|P|
B
|Py|
B
= T Defnition ,.:(i)
|y Py|
B
= T Defnition ,.:(viii)
Taking the last lines togethei, one can infei the following by Defnition
,.:(iii):
|x Px y Py|
B
= T
So the piemiss is tiue in B, and it iemains to show that the conclusion is
not tiue in B.
Assume to the contiaiy that the conclusion is tiue in B, that is, assume
that |x y x = y|
B
= T. Ten, by Defnition ,.:(viii), foi at least one
vaiiable assignment the following must obtain:
|y x = y|
B
= T
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,o
Applying Defnition ,.:(viii) again, one can conclude that theie is a vaii-
able assignment ovei B, difeiing fiom in y at most, such that the
following obtains:
|x = y|
B
= T
(In fact, theie is only one vaiiable assignment ovei B and theiefoie and
aie the same vaiiable assignment, but I do not make use of this fact
heie.)
By Defnition ,.:(ii) it follows that
|x = y|
B
= F. (8.:)
Since theie is only one object in the domain of B, namely the Mona Lisa,
|x|
B
and | y|
B
aie the same object. Tus, |x|
B
= | y|
B
, which implies the
following by the above special supplementaiy clause (ix) foi Defnition ,.::
|x = y|
B
= T
Tis contiadicts (8.:), which followed fiom the assumption that the con-
clusion x y x = y is tiue in B. Tus, the conclusion is not tiue in B,
and it has been shown that x y x = y does not follow fiom the piemiss
x Px y Py.
8. vvooi vUiis iov iui1i1v
In oidei to obtain a pioof system that is sound and complete with iespect
to the semantics foi L
=
, the system of Natuial Deduction needs to be
expanded so as to include an intioduction and an elimination iule foi
identity.
Te intioduction iule allows one to assume a=a (and similaily foi
all othei constants) and to dischaige a=a immediately:
=Intro Any assumption of the form t =t where t is a constant
can and must be discharged.
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,:
Hence, a pioof with an application of =Intio looks like this:
|t =t|
|s]v|
s =t
=Elim
|t]v|
|s]v|
t =s
=Elim
|t]v|
Stiictly speaking, only one of the veisions is needed, as fiom s =t one can
always obtain t =s using only one of the iules, as will be shown in Example
8.,. Having both veisions available is, howevei, moie convenient.
I give some examples illustiating the use of these iules.
ix:mvii 8.. x x =x
Proof. Te pioof is veiy shoit:
|a=a|
x x =x
Fiist a=a is assumed and immediately dischaiged by =Intio. Ten Intio
can be applied without violating the iestiiction on constants.
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,:
ix:mvii 8.,. x y (x = y y=x}
Proof.
|a=a| |a=b|
=Elim
b=a
Intio
a=b b=a
Intio
y (a= y y=a}
Intio
xy (x = y y=x}
Tis pioof shows that =Elim does not demand that one ieplace all the
occuiiences of a by b in the step fiom the fist to the second line. Tis
step is licensed by the iule =Elim, taking to be the foimula x =a.
ix:mvii 8.o. x y z (x = y y=z x =z}
Proof.
|a=b b=c|
Elim:
a=b
|a=b b=c|
Elim:
b=c
=Elim
a=c
Intio
a=b b=c a=c
Intio
z (a=b b=z a=z}
Intio
yz (a= y y=z a=z}
Intio
xyz (x = y y=z x =z}
Tis pioof system, like those of piopositional and piedicate logic, is
adequate:
1uiovim 8., (:uiqU:cv). Assume that and all elements of are L
=
-
sentences. Ten if and only if .
As in the case of piopositional and piedicate logic, I will not piove
the Adequacy Teoiem heie.
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,,
8. , Usis oi iui1i1v
Te woid is can play vaiious ioles. It can be used to expiess piedication,
as in Snow is white oi Jane is a classicist. In these cases is foims pait of
the piedicate. Te phiase is a classicist is foimalised as a unaiy piedicate
lettei as it does not iefei to a specifc classicist.
In othei cases is is used to expiess identity as in Ratzingei is Bene-
dict XVI. oi St Maiy College of Winchestei is New College. In these
cases, is combines two designatois and expiesses (numeiical) identity.
Tus, St Maiy College of Winchestei is New College is foimalised as
a=b with the obvious dictionaiy:
a: St Maiy College of Winchestei
b: New College
Te identity symbol is useful not only foi foimalising oveit identity
statements, as in the above examples. One can also use the identity symbol
to expiess that theie is a ceitain numbei of objects of some kind. Assume
that the piedicate lettei Px has the following entiy in the dictionaiy:
P: . . . is a Wagnei opeia
Ten the claim that theie is at least one Wagnei opeia can be expiessed by
existential quantifcation as xPx. If one wants to say that theie aie at least
two Wagnei opeias, howevei, it does not sumce to say xy (Px Py} oi
x Px y Py, because these two sentences say meiely that something
is a Wagnei opeia and something is a Wagnei opeia; it does not say that
something is a Wagnei opeia and something else is a Wagnei opeia. But
the lattei can be expiessed using =:
x y (Px Py x = y} (8.:)
Tis sentence of L
=
says that theie aie at least two Wagnei opeias. Of
couise the tiick also woiks with thiee:
x y z (Px Py Pz x = y x =z y=z}
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,
Tis sentence says that theie aie at least thiee Wagnei opeias.
By using identity one can also expiess that theie is at most one Wagnei
opeia by saying that, if x and y aie Wagnei opeias, then x and y aie
identical:
x y (Px Py x = y} (8.,)
Again this also woiks foi at most two, at most thiee, and so on.
Teie aie at most two Wagnei opeias can be foimalised as
x y z (Px Py Pz x = y y=z x =z}. (8.)
Teie is exactly one Wagnei opeia can now be iephiased as Teie is
at least one Wagnei opeia and theie is at most one Wagnei opeia, and I
have alieady shown howto expiess the two paits of that claim: x Px says
that theie is at least one such opeia, and the second pait, beginning with
at most, has (8.,) as its foimalisation. So Teie is exactly one Wagnei
opeia can be foimalised by the following L
=
-sentence:
x Px x y (Px Py x = y} (8.,)
Tis can also be expiessed by the following logically equivalent foimula:
x(Px y(Py x = y}} (8.o)
Tis sentence says that theie is a Wagnei opeia and it is the only one, that
is, any Wagnei opeia is identical to it. A still moie concise veision is the
sentence x y(Py x = y}.
:
Similaily, one can expiess in L
=
that theie aie exactly two Wagnei
opeias by combining (8.:) with (8.):
x y (Px Py x = y}x y z (Px Py Pz x = y y=z x =z}
By this method one can expiess, in the language L
=
, that theie aie
exactly :, Wagnei opeias, although this L
=
-sentence will be painfully long.
: Te equivalence to (8.,) follows fiom Exeicise 8.,.
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,,
One might think that the claim that theie aie :, Wagnei opeias involves
also a claimabout a mathematical object, namely the numbei :,. Howevei,
the claim can be foimalised without using a piedicate lettei oi constant
foi numbeis. Teiefoie, one can dispense with numbeis when claiming,
foi instance, that theie aie exactly :, Wagnei opeias. Some philosopheis
have tiied to dispense with numbeis and othei mathematical objects
completely, and the examples of this section show that identity can be
used to expiess ceitain claims without iefeience to numbeis, even if these
claims seem to be about numbeis at fist glance.
With these tiicks at ones disposal one can tackle a pioblem with the
foimalisation of designatois such as the king of Fiance, Claudias gaiden,
the tallest tutoi of New College who can speak Latin but does not own a
cai, oi the cai owned by Tim. Designatois of this kind aie called defnite
desciiptions. Defnite desciiptions cannot be adequately foimalised as
constants. Te following aigument is logically valid:
(T) Te cai owned by Tim is ied. Teiefoie theie is a ied cai.
Foimalising the defnite desciiption by a constant yields Pa foi the pie-
miss and x(Px Qx} foi the conclusion, with the obvious dictionaiy:
a: the cai owned by Tim
P: . . . is ied
Q: . . . is a cai
Cleaily, the aigument in L
2
coiiesponding to the English aigument (T)
is not valid, that is, Pa x(Px Qx}.
:
By foimalising a defnite de-
sciiption as a constant one loses all the infoimation contained in the
defnite desciiption. As the examples above show, a defnite desciiption
can contain a lot of infoimation, and condensing the tallest tutoi of New
College who can speak Latin but does not own a cai into a single constant
is bound to be inadequate.
: See Exeicise ,.:(i).
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,o
If Tim did not own a cai at all, the piemiss Te cai owned by Tim is
ied would not be tiue; if Tim owned two oi moie cais the piemiss would
also be not tiue, as theie would not be any cai that is the cai owned by
Tim would not exist. Te piemiss implies that Tim owns exactly one cai.
In fact, the piemiss can be iephiased as follows:
,
Tim owns exactly one cai and it is ied.
Fiom the above it is cleai how to expiess the claim that theie is exactly
one cai owned by Tim. So the sentence can be iephiased in the following
way:
Teie is a cai owned by Tim, and its his only cai (that is,
eveiy cai Tim owns is identical to it), and it is ied.
Following the pattein of (8.o), the piemiss of (T) is foimalised as follows:
x {(Qx Rbx} y(Qy Rby x = y} Px) (8.,)
Foi this foimalisation the dictionaiy needs to be extended to covei b
and R:
b: Tim
R: . . . owns . . .
When the defnite desciiption the cai owned by Tim was foimalised
as a constant, the validity of (T) could not be captuied by the validity of
its foimalisation. Tis was the motive foi seeking a moie iefned analysis
of the defnite desciiption. I still have to show that the new, moie detailed
analysis actually allows me to show the validity of (T) by establishing
the validity of its tianslation. Te piemiss now is foimalised as (8.,) and
the conclusion as befoie by x(Px Qx}. With this foimalisation the
aigument is valid in piedicate logic with identity:
, Te following is Russells (:o,) theoiy of defnite desciiptions. Foi a ciiticismof Russells
theoiy see (Stiawson, :,o).
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,,
ix:mvii 8.8.
x{(Qx Rbx} y(Qy Rby x = y} Px) x(Px Qx}
Te pioof is on the next page.
In some cases it may not be so easy to see that a sentence contains
a defnite desciiption. Especially identity statements involving defnite
desciiptions may be confusing. Considei the following two sentences:
(i) Jane is a classicist.
(ii) Jane is the classicist.
In the fist sentence is expiesses piedication. Sentence (ii), howevei, is
an identity statement: Jane is a piopei name, while the classicist is a
defnite desciiption. If . . . is a classicist is tianslated as Q
I
and Jane as
c
I
, sentence (i) becomes Q
I
c
I
, while (ii) becomes the following foimula,
when foimalised in the style of 8.,:
x(Q
I
x y(Q
I
y y=x} c
I
=x}
Tis sentence is logically equivalent to
Q
I
c
I
y(Q
I
y y=c
I
},
which says that Jane and only Jane is a classicist.
x
(
(
Q
x
R
b
x
}
y
(
Q
y
R
b
y
x
=
y
}
P
x
}
|
(
(
Q
c
R
b
c
}
y
(
Q
y
R
b
y
c
=
y
}
}
P
c
|
P
c
|
(
(
Q
c
R
b
c
}
y
(
Q
y
R
b
y
c
=
y
}
}
P
c
|
(
Q
c
R
b
c
}
y
(
Q
y
R
b
y
c
=
y
}
Q
c
R
b
c
Q
c
P
c
Q
c
I
n
t
i
o
x
(
P
x
Q
x
}
E
l
i
m
x
(
P
x
Q
x
}
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :,
Teie is exactly one king of Fiance, and he is bald.
Tis English sentence can be foimalised as a sentence expiessing that
theie is a king of Fiance, he is the only king of Fiance, and he is bald:
x(Rxc y(Ryc y=x} Px}
R: . . . is the king of . . .
c: Fiance
P: . . . is bald
Since Fiance is a iepublic, the sentence Te king of Fiance is bald is
false. Howevei, the sentence
Te king of Fiance is not bald (8.8)
is also false undei at least one ieading: it seems to say that theie is exactly
one king of Fiance and that he is not bald, which is also not tiue since
theie is no king of Fiance.
Te following sentence, in contiast, is tiue:
It is not the case (foi whatevei ieason) that the king of Fiance is bald.
(8.)
Heie the claim that the king of Fiance is bald is iejected: it leaves open
whethei theie is a king of Fiance who is not bald oi whethei theie is no
king of Fiance at all, oi, peihaps, whethei theie is moie than one king, so
that theie is nothing that is the king of Fiance.
Sentence (8.8) is most natuially foimalised by the following sentence:
x(Rxc y(Ryc y=x} Px} (8.:o)
Tis says that theie is a king of Fiance, that he is the only king of Fiance,
and that he is not bald. So only the baldness is denied, not that theie is
exactly one king of Fiance.
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :8o
Wheieas in 8.:o the negation symbol diiectly piecedes Px, in the
foimalisation of (8.) it is at the beginning:
x(Rxc y(Ryc y=x} Px} (8.::)
Tis expiesses that it is not the case (foi whatevei ieason) that theie is a
king of Fiance, who is the only one, and who is bald.
If these foimalisations aie coiiect, then (8.8) and (8.) have difeient
meanings. But especially the foimalisation of (8.8) is not uncontiovei-
sial, and theie may be anothei ieading of (8.8) that iesults in a difeient
foimalisation. Te following claim is a valid aigument only if (8.8) is
undeistood as expiessing the same as (8.):
Te king of Fiance is not bald; foi theie is no king of Fiance.
Te fist sentence is the conclusion, the second sentence is the piemiss of
the aigument. It can be foimalised as a valid aigument in L
2
only if (8.8)
is foimalised like (8.) as (8.::). Tus, depending on the ieading, theie
aie two foimalisations of (8.8) in L
=
that aie not logically equivalent,
namely (8.:o) and (8.::). If an English sentence has two non-equivalent
foimalisations, it is ambiguous. By compaiing (8.:o) and (8.::), one can
see that this is a case of a scope ambiguity again. In foimalisation (8.:o)
the occuiience of has a naiiow scope; its scope is only Px. In foi-
malisation (8.::) the negation symbol has a wide scope: its scope is the
entiie sentence. Which foimalisation is bettei has to be decided fiom
case to case, depending on the context.
Te analysis of defnite desciiptions just sketched allows one to tieat
the expiession the king of Fiance as an expiession that does not iefei to
an object in any case. Tis is an advantage, compaied to a foimalisation
of the king of Fiance as a constant: a constant has exactly one object as
it semantic value in any given L
2
-stiuctuie; a constant iefeis to an object
in any L
2
-stiuctuie. If a constant is used, the following aigument comes
out as valid in L
=
:
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :8:
It is not the case that the king of Fiance is bald. Teiefoie
something is not bald.
Let the constant a stand foi the king of Fiance. Ten Pa x Px
can be established by a single application of the iule Intio. Howevei, the
English aigument cleaily is not valid. Te foimalisation is valid, while the
English aigument is not, because the semantics of constants in L
2
and of
defnite desciiptions in English aie difeient: in L
2
constants always iefei
to some object, while in English defnite desciiptions such as the king of
Fiance may fail to iefei to an object.
If the piemiss It is not the case that the king of Fiance is bald is
foimalised as (8.::)
x(Rxc y(Ryc y=x} Px},
in accoidance with the above pioposed analysis of defnite desciiptions,
then its foimalisation coiiectly comes out as not valid. Tis example
shows that, at least in some cases, the pioposed theoiy of defnite de-
sciiptions can be used to handle English designatois that do not denote
anything.
So fai I have foimalised piopei names as constants. In the light of
what has just been said, one may doubt the univeisal adequacy of this
foimalisation of piopei names: a piopei name such as Pegasus does not
seem to iefei to an object (existing now oi in the past). One pioposal foi
dealing with this pioblem is to foimalise piopei names in the same way
as defnite desciiptions. But doing so iequiies a piedicate that singles
out Pegasus, such that Pegasus, if he existed, would be the one and only
object satisfying this piedicate. Logicians have also played with alteinative
semantics, wheie constants may fail to iefei to an object. All of these
pioposals aie beyond the scope of this text.
Tis discussion of defnite desciiptions shows that the identity symbol
of L
=
is used foi moie than just foimalising oveit identity statements
in English. Many moie sentences and aiguments can be analysed using
identity. So foimalisations in L
=
captuie moie details than foimalisations
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :8:
in L
2
. Consequently, theie aie English aiguments that can be shown to
be valid in piedicate logic with identity but not in piedicate logic without
identity. Te aigument (T) above is such an aigument.
Of couise, stiictly speaking I still have to explain what I mean by
validity of anaigument inpiedicate logic withidentity, but the explanation
should be obvious:
Validity of English arguments (and logical truth etc) in predicate logic
with identity is defned analogously to validity of arguments in proposi-
tional and predicate logic in Defnitions .o, ,.8, ., and ,.,.
8. o iui1i1v :s : iocic:i cos1:1
Piima facie identity seems to be meiely anothei binaiy piedicate in Eng-
lish. In the semantics of L
=
, howevei, the identity symbol is always in-
teipieted as numeiical identity, while othei binaiy piedicate letteis can
be inteipieted as aibitiaiy binaiy ielations. Why does identity ieceive
this special tieatment: Why is theie not a piedicate logic with special
tieatment foi othei binaiy piedicates such as is smallei than oi loves:
One could come up with a special symbol foi any of those binaiy piedi-
cates and invent a semantics in which they aie always inteipieted in the
same way. Why is identity singled out as a logical piedicate, while otheis
aie not: I shall illustiate the difeient tieatment of identity and othei
ielations by consideiing two examples.
In piedicate logic with identity the following aigument is valid:
Te moining stai is the evening stai. Te moining stai is a
planet. Hence the evening stai is a planet.
In L
=
the piemisses can be foimalised as a=b and Pa and the conclusion
as Pb. Te claim a =b, Pa Pb can easily be established by a pioof in
Natuial Deduction.
In piedicate logic without identity, the best possible foimalisation
of the piemisses is Rab, Pa and of the conclusion Pb, with an entiy in
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :8,
the dictionaiy foi Rxy as . . . is identical to . . . . But, by specifying an
L
2
-stiuctuie in which Rab and Pa aie tiue, and Pb is not, Rab, Pa Pb
can easily be iefuted. Tus, the above aigument is valid in piedicate logic
with identity but not in piedicate logic without identity. Te ieason is
that in L
=
the inteipietation of = is fxed, while in L
2
the binaiy piedicate
lettei R may be inteipieted as any aibitiaiy binaiy ielation.
Te aigument can be compaied with the following aigument:
Te evening stai is smallei than Uianus. Uianus is smallei
than Satuin. Teiefoie the evening stai is smallei than Sat-
uin.
Tis aigument is not valid in piedicate without identity oi in piedicate
logic with identity. If the inteipietation of smallei than weie fxed in the
way the inteipietation of identity is fxed in L
=
, that is, if the piedicate
lettei foi . . . is smallei than . . . weie always inteipieted as the smallei-
than ielation, the aigument would come out as valid. Geneially, fxing the
inteipietation of fuithei piedicates, such as is smallei than, will make
moie aiguments valid and moie sentences logically tiue, but it is doubtful
whethei such notions of validity still captuie logical validity and logical
tiuth.
In Chaiacteiisation :., of valid aiguments in Section :., I stipulated
that an aigument is valid if and only if theie is no inteipietation undei
which the piemisses aie all tiue and the conclusion is false. In inteipieta-
tions, only the vocabulaiy that is non-logical can be ieinteipieted. Logical
vocabulaiy has been chaiacteiised as not subject-specifc. Te piedicate
expiession is smallei than is aiguably subject-specifc because it can
only be sensibly applied to objects with spatial extension (and peihaps to
numbeis); at least it is not so cleai whethei it can be applied to objects
such as thoughts, piopeities, laws, oi functions. Identity, in contiast, is
not specifc to any subject. Tus, it cannot be ieinteipieted. Logicians say
that identity is a logical constant. Tis way of distinguishing logical fiom
non-logical vocabulaiy is fai fiom being cleai and piecise, and the dis-
Volker Halbach :oo8/:oo
8 Identity and Defnite Descriptions :8
tinction and the possibility of such a distinction aie contioveisial issues
in the philosophy of logic.
Of couise theie could still be expiessions in English that, while not
contained as logical symbols in L
=
, aie yet logical expiessions. In fact,
many logicians believe that the language L
=
ought to be extended so as to
include additional logical symbols expiessing, foi instance, it is necessaiy
that . . . Consequently, they think that theie aie validaiguments inEnglish
that aie not valid in piedicate logic with identity. Othei philosopheis
think that any valid aigument of English can be foimalised as a valid
aigument in L
=
, so long as ceitain foimalisation tiicks aie allowed. Tese
contioveisies aie not only ciucial foi the philosophy of logic and language,
they also impinge on ontology and othei coie disciplines of philosophy.
At any iate, L
2
and L
=
aie veiy poweiful languages. Many logicians,
mathematicians, and philosopheis believe that L
=
is sumcient foi foimal-
ising veiy compiehensive paits, if not all, of mathematical and scientifc
discouise. If the language of mathematics and science can indeed be
handled in the language of piedicate logic with identity, theie is hope that
it can even captuie laige paits of philosophy.
Volker Halbach :oo8/:oo
Natuial Deduction Rules
vvovosi1io:i iocic
Intro
Elim:
Elim:
||
Intro
Elim
Intro:
Intro:
||
||
Elim
||
Intro
||
||
Elim
Intro
Elim:
Elim:
vviuic:1i iocic
|t]v|
Intro
v
piovided the constant t does not
occui in oi in any undischaiged
assumption in the pioof of |t]v|.
v
Elim
|t]v|
|t]v|
Intro
v
v
||t]v||
Elim
|s]v|
s = t
=Elim
|t]v|
|s]v|
t = s
=Elim
|t]v|
Volker Halbach :oo8/:oo
Bibliogiaphy
Cappelen, Heiman and Einest LePoie (Spiing :oo,), Quotation, in
E. N.Zalta, ed., Te Stanfoid Encyclopedia of Philosophy.
URL: http.//plato.stanford.edu/archives/spr:oo,/entries/quotation/
Chuich, Alonzo (:,o), A note on the Entscheidungspioblem, journal of
Symbolic Logic :, o:.
De Moigan, Augustus (:8,), Formal Logic, Tayloi & Walton, London.
Devlin, Keith (:,), Fundamentals of Contemporary Set Teory, second
edn, Spiingei-Veilag, New Yoik.
Foibes, Giaeme (:), Modern Logic, Oxfoid Univeisity Piess, Oxfoid.
Geach, Petei (:o:), Reference and Generality, Coinell Univeisity Piess,
Ithaca.
Gentzen, Geihaid (:,,), Unteisuchungen bei das natiliche Schlieen,
Mathematische Zeitschrif ,, :,o::o, o,,o,.
Guttenplan, Samuel (:,), Te Languages of Logic, second edn, Blackwell,
Malden.
Halmos, Paul R. (:oo), ^aive Set Teory, D. Van Nostiand Company,
Piinceton. Repiinted, Spiingei-Veilag, New Yoik, NY, :,.
Hodges, Wilfiid (:oo:), Logic. An Introduction to Elementary Logic, sec-
ond edn, Penguin Books, London.
Volker Halbach :oo8/:oo
Index :88
Jakowski, Stanislaw (:,), On the iules of supposition in foimal
logic, ^akadem Seminarium Filozofcznego Vydzialu Matematyczno-
Przyrodniczego Uniwersytetu Varszawskiego .
Lewis, David (:,,), Counterfactuals, Haivaid Univeisity Piess, Cam-
biidge MA. ieissued in :oo: by Blackwell, London.
Moiiis, Chailes (:,8), Foundations of the Teory of Signs, Univeisity of
Chicago Piess, Chicago.
Moschovakis, Yiannis (:), ^otes on Set Teory, Undeigiaduate texts
in mathematics, second edn, Spiingei, New Yoik.
Quine, Willaid O. V. (:o), Mathematical Logic, Noiton, New Yoik.
Neuaufage :,: von Haivaid Univeisity Piess (Cambiidge, M.A.).
Russell, Beitiand (:o,), On denoting, Mind :, ,,.
Sainsbuiy, Maik (:oo:), Logical Forms, second edn, Blackwell, Oxfoid.
Smith, Petei (:oo,), An Introduction to Formal Logic, Cambiidge Univei-
sity Piess, Cambiidge.
Stiawson, Petei F. (:,o), On iefeiiing, Mind ,, ,:o,.
Taiski, Alfied (:,o), Dei Wahiheitsbegiif in den foimalisieiten
Spiachen, Studia Philosophica Commentarii Societatis Philosophicae
Polonorum :.
Tennant, Neil (:o), ^atural Logic, :nd edn, Edinbuigh Univeisity Piess,
Edinbuigh.
Tioelstia, Aine S. and Helmut Schwichtenbeig (:o), Basic Proof Teory,
numbei , in Cambiidge Tiacts in Teoietical Computei Science,
Cambiidge Univeisity Piess, Cambiidge.
Volker Halbach :oo8/:oo
Index
adequacy, :,, :,,
foi piedicate logic, :,
foi piopositional logic, ::8
ambiguity, o,o,, :8
lexical, :8
of scope, :8:
antisymmetiic, :o
aigument, :,, ::,
of a function, :,
aiity, 8,
aiity-index, ,o, 88
assignment, 8::
assumption, ::o
asymmetiic, :o
atomic foimulae, :oo
of L
2
, 8,
binaiy ielation, , :o
bound occuiience of a vaiiable,
8,
Compactness Teoiem of piopo-
sitional logic, :
completeness, ::,, ::, :,
conclusion, :,
conditionals
counteifactual, ,
indicative, ,
subjunctive, ,
conjunction
of sentences, ,o
connectives, ,, ,:,o
in L
I
, ,:
in L
I
, ,
in English, ,:
main, ,,
consistency, o8
in piedicate logic, :,,
semantic, :,,
semantic in L
2
, :o8
in English, :,
semantic, :, :,
semantic in L
I
, :
syntactic, :, :,
constants, ,o, 8,, ,, o
context, :,
contiadiction, :, o, o8, :o8, :,,
contiadictoiy, o
counteiexamples
in L
=
, :o
in piedicate logic, :o::
in piopositional logic, :
counteifactual
conditionals, ,
cut iule, :::
Volker Halbach :oo8/:oo
Index :o
declaiative sentence, :,
deductive validity, :o
defnite desciiptions, :,o
designatois, ,, ,o, :,, :,:, :,
dictionaiy, o:
disjunction, :::
of sentences, ,o
domain
of a function, :,
domains
of discouise, o
double aiiow, ::o
double negation elimination, ::,
empty set, ,
equivalence
logical, :, o, o8, :o8
equivalence ielation, :o
ex falso quodlibet, ,:, ::8
extensional, :,:
extensionality, :,:
extensions, ,, :,:
falsity
logical, :
foimal validity, :
foimulae
atomic, :oo, :o,
of L
=
, :o8
of L
2
, 8
fiee occuiience of a vaiiable, 8,
function, :
geneialisation, 8o
heait, o
identity
numeiical, :oo:o,
qualitative, :oo:o,
if, :o
inconsistency
semantic in L
I
, :
syntactic, :
indicative conditionals, ,
induction, :oo
inductive validity, :o
intuitionistic logic, ::,
kidney, o
languages
extensional, :,:
letteis
piopositional, :
piedicate, ,o, 8,
logical equivalence, :, o, o8,
:o8
logical falsity, :
logical foim, :,
piopositional logic, ,o
logical teims, :
logical tiuth, :
logical validity, :, :, ,
logically tiue, :o8, ::o
main column, ,, o
main columns, :
main connective, ,,
Volker Halbach :oo8/:oo
Index ::
metalinguistic vaiiable, :8
metavaiiables, :,:
Natuial Deduction, ::o
negation
of sentences, ,o
oideied paii,
paii
oideied,
paitial foimalisation, :,
piagmatics, :o
piedicate letteis, ,o, 8,, ,, o
piedicate logic, :,
piemiss, :,
pioof, ::, ::,
piopei names, o
piopeity, o
piopositional consistency, o8
piopositional contiadiction, o8
piopositional logic, :,
piopositionally valid, ,o, ,
piovable, :::
quadiuple, :o
quotation, :,:
iange
of a function, :,
ieductio ad absuidum, ::
iefexive, :o
ielation,
binaiy, ,
ioot, ::,
satisfaction, :oo
scope
of a quantifei, :,:
of a connective, o
scope ambiguity, o, :,:, :8:
semantic inconsistency
in L
I
, :
semantic consistency, :, :,
in L
I
, :
in L
2
, :o8
semantic values, ,
semantics, :,, :,
sentence letteis, :, ,, ,8, ,, o
sentences
of L
2
, 8o
of L
I
, :, ,o
of L
=
, :o8
sentential logic, :,
sets, o8
soundness, ::,, ::, :,
stiuctuies, :, :,
in L
I
, ,,
in L
=
, :o8
in L
2
, ,::
subjunctive conditionals, ,
syllogistics, :oo
symmetiic, :o
symmetiy, ::
syntactic consistency, :, :,
syntactic inconsistency, :
syntax, :,, :,
Volker Halbach :oo8/:oo
Index ::
tautology, o, o8
teinaiy, ,o
teinaiy ielation, :o
tiansitive, :o
tiiple, :o
tiuth
in an L
I
-stiuctuie, ,o
in an L
2
-stiuctuie, :o,
logical, :
tiuth, logical, :o8, ::o
tiuth-functional, ,,
tiuth-functional completeness, ,o
tiuth-values, ,,, ,, 8
univeisal quantifei, 8, :,:
valid, :,,
valid aigument, ,:
validity, :
deductive, :o
inductive, :o
logical, :, ,
of an aigument, ::
piopositional, ::
value
of a function, :,
values
semantic, ,
vaiiable assignment, 8::
vaiiables
in L
2
, 8,
metalinguistic, :8
Volker Halbach :oo8/:oo