Ezcurdia - Introducing Sense (Frege)

Logic, Truth and Arithmetic. Essays on Gottlob Frege.
Marco Ruffino (ed.).
CDD: 149.9
INTRODUCING SENSE
MAITE EZCURDIA
Instituto de Investigaciones Filosóficas

Universidad Nacional Autónoma de México
Ciudad de la Investigación en Humanidades
Coyoacán 04510
MÉXICO
[email protected]
Abstract: In this paper I present what I take to be the best argument for
the introduction of the semantic category of sense. This argument, or a
version of it, can be extracted from Frege’s renowned ‘On Sense and
Reference’, but has not been properly understood or appreciated. I begin
by discussing the Russellian objections to other versions of the argument
attributed to Frege, in order to expound the argument which fends off
such objections.
Key-words: Frege. Millikan. Sense. Reference. Content. Cognitive value.
In his renowned “On Sense and Reference” (‘OSR’ from now on)
Frege famously introduced what he intended to be a semantic level or con-
tent level different from that of reference, viz. that of sense. The argument
put forth by Frege has been called (amongst other things) ‘the paradox of
identity’ by Burge (1977) or ‘Frege’s puzzle’ by Salmon (1986) and has
been widely discussed (Burge, 1977; Evans, 1982; Millikan, 1991; Sainsbury,
1983; Salmon, 1986, amongst others). The category of sense itself has also
been widely discussed (Evans, 1981 and 1982; Perry, 1977; Salmon, 1986
Manuscrito – Rev. Int. Fil., Campinas, v. 26, n. 2, pp. 279-312, jul.-dez. 2003.
280 MAITE EZCURDIA
amongst many others). But what sense is is determined at least in part by

what job the argument for its introduction sets it out to do, so a discus-
sion of what sense is must take place in the light of the argument for its
introduction. In the present paper I intend to present what I think is the
best argument for the introduction of sense in an attempt at rescuing the
semantic notion of sense. I say the best argument because I think it is the
only argument for the introduction of sense which will be valid and pre-
empt the criticisms of Russellians and neo-Russellians like Millikan and
Salmon.
The argument I think is itself given by Frege, yet I shall not be
concerned here with either showing that it is the argument offered by
Frege, nor with giving exhaustive reasons for why it is the only argument
that will preempt such objections. The first task I have done elsewhere 1
and so I shall speak of the argument for the introduction of sense as
Frege’s argument. Nevertheless, I will not shy away from some interpre-
tive work where needed. The second task I shall leave unfinished, and
will only argue how it is that the argument that I present here avoids the
Russellian and in particular Millikan’s objections.
I begin by presenting two versions of the argument (one offered
by Burge and the other by Millikan), and say why such versions will not
do to introduce sense. In the second section I present what I take to be
the best argument and examine how it avoids those objections. And in the
third section I examine the state of affairs between the Russellian and the
Fregean.
1
Burge (1977) calls Frege’s argument for the introduction of sense,
‘the “paradox” of identity’. He describes it in the following way:
1 See Ezcurdia, 1994.
INTRODUCING SENSE 281
The ‘paradox’ of identity says that, whereas a statement of the form

⎡α=α⎤ is uninformative, a statement of the form ⎡α=β⎤ may be of consid-
erable empirical significance; but ‘α’ and ‘β’ are singular terms that refer to
the same object; so the difference in the statements must go beyond what
is referred to in them. The difference is in the mode with which the de-
noted object is presented to a thinker by the singular terms ‘α’ and ‘β’.
And Frege counted this difference as a difference in sense. (1977, p. 354.)
This interpretation of Frege’s argument has a simple form. It has two

parts. The first concludes that the difference between ⎡α=α⎤ and ⎡α=β⎤
cannot be explained purely in terms of the reference of the signs. It is the
following:
1. ⎡α=α⎤ differs in cognitive or informative value from ⎡α=β⎤.

2. ‘α’ and ‘β’ are co-referring singular terms.
∴3. The difference in cognitive value between ⎡α=α⎤ and ⎡α=β⎤
must be explained in terms of something other than the referents
of ‘α’ and ‘β’
The second part seeks to conclude that the difference between sentences
of those two forms is (according to Burge) accounted for by Frege in
terms of the modes of presentation associated with those singular terms, and
that this difference amounts to a difference in sense.
Many philosophers have supposed that a difference in mode of
presentation is equivalent to a difference in sense, and many have equated
modes of presentation with senses. 2 It is clear in Frege’s texts that he
intends his notion of sense to be essentially semantic. By ‘semantic’ I will
mean here truth-value, truth-conditions, reference or objects of refer-
2 Perry (1977), Evans (1981) and Salmon (1986) are examples. For a discus-
sion of the relation between modes of presentation and senses see Ezcurdia,
1995, 1997, and 2001.
282 MAITE EZCURDIA
ence. 3 By ‘content’ I will mean here any of the latter except truth-value.
However, modes of presentation, the ways in which a subject is presented
with an object, are not clearly semantic. They are primarily psychological.
They include senses but also things like sensations, signs (i.e. syntactical
objects), prototypes, causal chains, and thus many things which Frege
would not have wanted to include as senses. At best the category of
mode of presentation is too broad and includes things which are not
senses, and at worst it is the wrong category in being primarily a psycho-
logical category. So if in the argument for the introduction of sense, Frege
– as Burge thinks – is taking the difference in cognitive value between
⎡α=α⎤ and ⎡α=β⎤ to be one solely of modes of presentation, it is not clear
that he has succeeded in introducing the semantic level of sense. Until we
are given a reason to believe that the particular modes of presentation
involved are also semantically relevant we have no introduction of sense
as a semantic category.
3 This way of being semantic may be seen as too narrow for it seems to ex-
clude semantic rules, viz. those things which are necessary for understanding and
which determine truth-conditions and objects of reference. So one might wish to
expand the category of what is semantic to those things that determine truth-
conditions and objects of reference. Yet this brings the problem of how to ex-
clude things which are involved in the fixing of reference in Kripke’s sense
(Kripke, 1980). Those things may be presemantically important but not semanti-
cally relevant. On the other hand, one may want semantic rules to be captured
truth-conditionally (viz. through the axioms of an interpretive truth-theory), and
so find no need in including as semantic something that determines truth-
conditions and objects of reference. But sometimes, and at least for some inter-
pretive truth-theories, that will not be possible, especially where context-
dependent expressions are concerned. Nonetheless, notice that reference was
included amongst those things which are semantic. Taking reference as the rela-
tion of reference, and taking semantic rules to be rules of reference or relations
of reference, we can see how they are incorporated into the category of the se-
mantic.
Millikan (1991) has challenged a Fregean introduction of sense on

these grounds. She holds that in introducing sense Frege presents us with
either an invalid or a question-begging argument. Either Frege produces
an argument which succeeds only in introducing psychological modes of presen-
tation and not semantic senses, or in order to produce a valid argument he
supplements it with a question-begging premise, viz. a premise that says
that the only way in which the difference between ⎡α=α⎤ and ⎡α=β⎤ is to
be accounted for is in terms of sense. But this, she argues, is not so. She
claims that the syntactical or notational difference between the sentences
of those forms suffice to establish why we take it that ⎡α=α⎤ and
⎡α=β⎤ differ in cognitive value.
Before we begin to look at her version of Frege’s argument, we
need to look at what sort of expressions the argument concerns. If ‘α’ and
‘β’ are complex referential expressions 4 which are co-referential, then it is
not clear that the Fregean would have succeeded in introducing sense.
Take two definite descriptions like ‘The Evening Star’ and ‘The Morning
Star’. 5 These expressions are co-referential and so have in one sense the
4 Some might object to speaking of complex referential expressions, claiming
that the only genuine referential expressions are simple ones, and the rest are
quantified expressions. Although I do believe that there are complex referential
expressions, for the purposes of the present argument one could think of what I
call ‘complex referential expressions’ just as complex noun phrases. The present
argument seems to be present in Russell and more recently in Millikan, 1991.
5 There is an unstable position here concerning the way in which ‘The Eve-
ning Star’ and ‘the Morning Star’ are being taken. Given that they are in capital
letters it would appear that they figure as names, not as descriptions. As such,
even if the reference of ‘the Evening Star’ failed to satisfy the description ‘the
evening star’ due to it being perceived in the afternoon but not in the evenings,
or as is the case, due to it not being a star, that name would still refer to that star,
just as ‘Dartmouth’ refers to the town even if it is no longer at the mouth of the
river Dart. But for the argument to go through these expressions do not func-
284 MAITE EZCURDIA
same semantic content, but not in another sense. They share the same seman-
tic content insofar as they refer to the same object, but differ in semantic
content insofar as they arrive at that content (as Millikan puts it) via differ-
ent routes and from different starting points. Given that ‘evening’ and
‘morning’ refer to different parts or times of day, and hence have different
references, we may say that the complex expressions ‘The Evening Star’
and ‘The Morning Star’ have in some respect different contents or refer-
ences. Yet because they refer to the same object, namely, Venus, they have
the same reference. So these expressions differ in content in some way,
but not in another. They both refer to Venus but they get there via differ-
ent routes and from different starting points: via the different times or
parts of the day to which ‘evening’ and ‘morning’ refer.
This account does not, however, cover all referential expressions,
but only complex ones. Simple referential expressions, that is, expressions
with no identifiable semantic complexity or semantic proper parts, like
‘Hesperus’ and ‘Phosphorus’ or like ‘Tully’ and ‘Cicero’, are not covered
by such an account. The difference in content between these co-referring
expressions cannot be explained in terms of the difference in the refer-
ences of the parts as we did with complex referential expressions. ‘Hespe-
rus’ and ‘Phosphorus’ have the same reference, but because they are prima
facie genuinely simple referential expressions, they do not have semantically
relevant proper parts whose reference we could pick on to explain how
they differ in the way of arriving at their reference. Russellians, and in par-
ticular Millikan, hold that only if the Fregean denies such genuine semantic
simplicity, could she hold a view of simple referential expressions as ellip-
tical for complex referential expressions, and only then could she follow
tion solely as names, but also as descriptions. The descriptive elements in them
are taken to have a semantic role to play in determining or arriving at the refer-
ence of the names.
the strategy above in distinguishing different contents. According to the

Russellian then referring expressions like ‘Hesperus’ and ‘Phosphorus’
could only have the same content in one way and differ in another if we
assume a certain hidden complexity in them; for example, that they are
(elliptical for) definite descriptions (Russell) or complex phrases composed
of a demonstrative and a predicate (Burge).6
But accepting that there is a difference in the semantics of two ref-
erential expressions due to the ‘routes’ via which they arrive at their refer-
ence is not yet to admit a level of content other than the referential level. 7
For the Russellian the difference in routes is still a difference in reference:
in the parts or properties being referred to in order to arrive at the object.
6 Frege need not assume a hidden complexity in proper names. It is true that
in ‘The Thought’ he suggests that the senses of proper names are descriptive,
and were that to be his view then he would have no problem about simple co-
referential expressions. However, there is an alternative interpretation of Frege’s
views on proper names which pays more attention to his OSR in McDowell,
1980.
7 Taschek (1991) has argued that Frege’s argument for the introduction of
sense depends on the reference of sentences being their truth-values, the argu-
ment being the introduction of a difference in the senses of the sentences, viz. in
the thoughts expressed by sentences with the same truth-value. But then there
would be no dispute between the Fregean and the Russellian for it will not be
clear that a different semantic or content level will have been introduced. In fact,
a Russellian can accept that there are two types of references for a sentence, one
being its semantic value and the other its semantic content, the semantic value
for a sentence just being its truth-value and the semantic content being captured
by truth-conditions made up of the references of the meaningful parts of a sen-
tence arranged in a certain way. Such truth-conditions are what I have elsewhere
called ‘purely referential truth-conditions’. (See Ezcurdia, 1994 and 1995.) Thus,
for there to be a genuine dispute between the Fregean and the Russellian, I shall
assume that the reference of sentences are just what the Russellians would take
their semantic content to be (or their purely referential truth-conditions).
286 MAITE EZCURDIA
So even if one were to get rid of genuinely simple referential expressions,

this would not entail that one had introduced a different semantic level by
an appeal to routes. Hence, although Frege does not make the distinction
between complex and genuinely simple referential expressions, if the Fre-
gean is to introduce a semantic level different from that of reference, the
distinction between sense and reference must apply to genuinely simple
referential expressions. Thus, Frege’s argument must be put in terms of
those expressions.
Bearing this in mind, we can now look more closely at Millikan’s
particular interpretation of such argument. Frege’s argument concerns the
informativeness or cognitive value 8 of sentences with a certain logical
form. It was Frege’s realization that sentences of the form ⎡α=β⎤ (where
‘α’ and ‘β’ are to stand for simple co-referential expressions) differed in
cognitive value from sentences of the form ⎡α=α⎤ that led him to believe
that these two syntactically differing sentences differed also at the seman-
tic level or level of content. Assuming for the moment that a difference in
cognitive value between two sentences with the same referential content
entails that a subject who is competent in the language in question and
who is rational can take opposing attitudes to the content of those two
sentences, we can put forth the argument which Millikan thinks is Frege’s
argument for the introduction of sense in the following fashion:
1. ‘α’ and ‘β’ are simple co-referring expressions.
2. It is possible for someone, who is rational and understands the
language in question, to take opposing attitudes towards the
content(s) of sentences of the form ⎡α=α⎤ and ⎡α=β⎤.
∴3. ⎡α=α⎤ and ⎡α=β⎤ differ in cognitive value.
8 I have argued elsewhere that a difference in information content may not
amount to a difference in cognitive value. See Ezcurdia, 1994.
∴4. ⎡α=α⎤ and ⎡α=β⎤ differ in some respect other than their refer-
ence.
∴5. Because ‘α’ and ‘β’ are simple co-referring expressions, and the
sign of identity is common to both sentence-forms, the differ-
ence between sentences of the form ⎡α=α⎤ and ⎡α=β⎤ must be a
difference in the modes of presentation of the reference of ‘α’ and
‘β’, in their senses.
So if ⎡α=α⎤ did not differ from ⎡α=β⎤ then coming to know that Cicero is
Tully would not consist in new knowledge and we would be able to infer
directly ‘Cicero is Tully’ from ‘Cicero is Cicero’. The fact that we can take
opposing attitudes towards the contents of the sentences ‘Cicero is
Cicero’ and ‘Cicero is Tully’ is evidence that these sentences differ in
cognitive value. Since the simple referential expressions in those sen-
tences have the same reference they must differ at another semantic or
content level, viz. at the level of sense or of modes of presentation. Such a
level, Frege will claim, is a semantic level.
But if this argument purports to be an argument for sense as
something semantic, then as it stands it is invalid. Since it overlooks the
fact that the notational or syntactical difference between ‘α’ and ‘β’ in
⎡α=α⎤ and ⎡α=β⎤ may give us the desired difference. For why could it not
be the case that the difference in cognitive or information value of ⎡α=α⎤
and ⎡α=β⎤ be owed solely to a difference in notation? Why couldn’t the
difference in notation or syntax suffice to ‘move the mind differently’? Why
could it not be that our ability to take opposing attitudes to sentences of
these two different forms lie in our failure to know that ‘α’ refers to the
same thing as ‘β’? A notational account of the difference between true sen-
tences of the form ⎡α=α⎤ and ⎡α=β⎤ would say that they differ in their
cognitive value because they differ in their notation. ⎡α=α⎤ contains two
tokens of the same sign whereas ⎡α=β⎤ contains two tokens of different
288 MAITE EZCURDIA
signs. There are two such notational accounts. The first, which I shall call
‘the Begriffsschrift notational account’, is suggested by Frege himself in his Be-
griffsschrift; and the second, which I shall call ‘the psychological notational ac-
count’, is suggested more recently by Millikan, 1991.
Under the first account, the content of sentences of the form
⎡α=α⎤ and ⎡α=β⎤ is not given merely by the references of ‘α’, ‘β’ and ‘=’.
Rather under its content we have it that ⎡α=α⎤ simply states that ‘α’ has
the same reference as ‘α’ and ⎡α=β⎤ states that ‘α’ has the same reference
as ‘β’ (or that the referent of ‘α’ is the referent of ‘β’). Thus, in order to
account for the difference in cognitive value between ⎡α=α⎤ and ⎡α=β⎤
taking account of the reference of ‘α’ and ‘β’ and syntactical notations
would seem to suffice to argue that ‘α’ and ‘β’ have different contents at
one level but not at another. ‘α’ and ‘β’ have the same reference but the
sentences which contain such simple referring expressions and which
involve the identity-sign do not. 9 There ‘α’ and ‘β’ refer to themselves as
well, so to speak. So when a subject takes opposing attitudes to the con-
tents of ‘Cicero is Cicero’ and ‘Cicero is Tully’, she does so because those
sentences differ in their content in virtue of their difference in syntax.
9 Frege takes the notational account he proposes in his Begriffsschrift to ac-
count only for sentences containing the identity-sign. Yet, as Salmon (1986) has
been quick to remark, the argument for the introduction of sense has little to do
with sentences containing the identity-sign, and more with simple referring ex-
pressions. For example, ‘Hesperus is a big star’ and ‘Phosphorus is a big star’ are
sentences which do not concern identity, which have the same reference, but
which still differ in cognitive value. So we could take the Begriffsschrift notational
account to cover all sentences in which simple referring expressions occur and
not just identity-sentences. Wherever a sentence occurs with a simple referring
expression the semantic contribution of that simple referring expression will not
be its referent only, but also itself. Thus, ‘Hesperus is a big star’ will state that the
reference of ‘Hesperus’ is a big star.
Millikan would find even this Begriffsschrift notational account sus-

picious. For why should one assume that, because a subject may take op-
posing attitudes to the contents of two sentences which differ syntacti-
cally, those sentences actually differ in content? According to her, there is
no reason to suppose that sentences of the form ⎡α=α⎤ and ⎡α=β⎤ differ
in content just because a subject who is rational and competent in the
language in question can take opposing attitudes towards their contents.
Rather one may suppose that sentences of those two kinds have the same
content, the same semantics, and that what explains why a subject takes
opposing attitudes towards them is the way in which that content is pre-
sented, viz. the syntactic difference between the sentences that have the
same content. The mind is then supposed to be sensitive to those syntac-
tic differences in taking opposing attitudes to the content of sentences.
Those syntactic modes of presentation could suffice for an account of why a
subject may assent to the content of ‘Hesperus is Hesperus’ or ‘Cicero is
Cicero’ and yet deny or dissent to the content of ‘Hesperus is Phospho-
rus’ or ‘Cicero is Tully’. So when a subject takes opposing attitudes to the
contents of ‘Cicero is Cicero’ and ‘Cicero is Tully’, she does so not in
virtue of those sentences differing in content, but rather in virtue of their
differing syntactically. Under the psychological notational account one has
no reason to claim that ‘Cicero is Cicero’ and ‘Cicero is Tully’ differ se-
mantically in order to account for why it is that a subject may take oppos-
ing attitudes towards them without being irrational. If Frege’s argument is
to be valid, the Fregean needs to rule out both notational accounts of the
difference in cognitive value between two sentences of the form ⎡α=α⎤
and ⎡α=β⎤.
In order to rule out Millikan’s psychological notational account,
Millikan claims that at the very least the Fregean needs the following ad-
ditional premise:
290 MAITE EZCURDIA
6. A rational subject, who understands the language in question,

can only take opposing attitudes to the content(s) of sentences
of the form ⎡α=α⎤ and ⎡α=β⎤ if they differ in semantic content.
But 6 is only justified, according to Millikan, if the Fregean does some

mingling between vehicles of representation and the semantics or con-
tents of representations. The move Millikan envisages is one where the
Fregean projects certain properties of the vehicle or sign onto the content:
she assumes (according to Millikan) that if two vehicles of representation
are different then their contents or semantics are different. Millikan calls
this move ‘the externalization of differents’. However, to externalize differents
is just to beg the question at issue: that the difference between ⎡α=α⎤ and
⎡α=β⎤ involves not merely a syntactical difference, a difference in vehi-
cles, but also a difference in content or semantics. That we can take one
attitude to ‘Cicero is Cicero’ and an opposing one to ‘Cicero is Tully’ is
evidence that there is a difference in content between these two sentences
only if we already assume that we can only take opposing attitudes to sen-
tences which differ in content, and not to sentences which differ merely
in notation. Millikan sums up this point thus:
Cicero is Tully is an informative thought whereas Cicero is Cicero is not, so

these thoughts must have different contents. But, quite transparently,
that begs exactly the question at issue. Of course the thoughts Cicero and
Tully are different (or at least for some people they might be) or they
couldn’t move the mind differently […] The question is whether their
contents must be different in order for this to be so. Might they not differ
as it were merely in notation? […] for this Fregean argument to go
through […] [o]ne has to already believe that content and only content
can move the mind directly. Alternatively, one simply means by ‘content’,
mechanical vehicle or, more abstractly, production or occurrence of a
disposition for the mind to move in a certain way. But then one needs an
argument that this sort of ‘content’ is content – that it catches something
with a semantic dimension, something the sameness of which will imply,
say, same way of helping to determine truth value. That this is the case is
not given. (1991, p. 455.)
In order to produce a valid argument, the Fregean would need to show

why the difference between ‘Cicero is Cicero’ and ‘Cicero is Tully’ is not
merely a difference in notation. So far there seems to be no reason why
this difference in notation could not by itself produce different reactions
to them, different mental attitudes towards them. So a further premise
needed to support 6 is the following:
7. A rational subject, who understands the language, cannot take

opposing attitudes to the content of sentences of the
forms ⎡α=α⎤ and ⎡α=β⎤ if they differ merely in notation.
This, however, cannot be had by the Fregean for in the face of (at least)
the psychological notational account it begs the question at issue. There-
fore, according to Millikan either Frege’s argument for the introduction
of sense as a semantic category begs the question or, as it stands, it is in-
valid. 10 However, even if the Fregean were able to rule out Millikan’s psy-
chological notational account, she would still require an argument against
the Begriffsschrift notational account, where the signs themselves are part of
the content of identity-sentences.
Were Frege’s argument to be the one that either Burge or Millikan
envisage, Frege would not have succeeded in introducing sense. How-
ever, I think that both Millikan’s and Burge’s versions of Frege’s argu-
ment are not the only ones. There is a version of the argument which
avoids the problems here presented.
10 For Millikan something counts as content if it is semantic, and something
is semantic if it is either purely referential truth-conditions, referents or extensions.
292 MAITE EZCURDIA
2
Burge is right in concluding that there are two parts to Frege’s ar-
gument. However, he is wrong in believing that the conclusion to the first
part is just that the difference between ⎡α=α⎤ and ⎡α=β⎤ cannot be a dif-
ference in reference alone. Rather the conclusion is that such a difference
cannot be just a difference in the reference of the terms nor, crucially, a
difference in signs between ⎡α=α⎤ and ⎡α=β⎤. The first part of Frege’s
argument (which I shall call ‘The Negative Phase’ from now on) has three
stages. Its overall aim is to establish that differences in the cognitive value
of sentences with the forms ⎡α=α⎤ and ⎡α=β⎤ cannot be accounted for
either in terms of their syntactic differences or in terms of the references
of ‘α’ and ‘β’. The First Stage of this phase establishes that ⎡α=α⎤ and
⎡α=β⎤ differ in cognitive value. The Second Stage concerns the Be-
griffsschrift notational account of the relation of identity, in terms of which
a notational account of the difference in cognitive value between ⎡α=α⎤
and ⎡α=β⎤ is given. And The Third Stage consists of an argument against
such a notational account of the difference in cognitive value. Finally, the
second part of his argument, which I shall call ‘The Positive Phase’, argues
that the difference in cognitive value between sentences of those two
forms will be owed to a semantic difference, in particular to a difference in
senses, or more specifically, ways of determining reference or ways of referring. Let
us then look at the argument.
The Negative Phase

The First Stage
In the last section we said that a difference in cognitive value has
(at least) some psychological consequences. Prima facie, a difference in
cognitive value and the psychological abilities of a rational subject con-
nect rather nicely in the following way:
(I) If ⎡α=α⎤ and ⎡α=β⎤ differ in cognitive value, then it is possible

for there to be a subject who is competent in the language in
question and rational and who takes opposing attitudes to-
wards the content(s) of ⎡α=α⎤ and ⎡α=β⎤.
(II) If there is (as a matter of fact) a subject who is rational and

competent in the language in question, who understands
⎡α=α⎤ and ⎡α=β⎤, and who takes opposing attitudes towards
the contents of ⎡α=α⎤ and ⎡α=β⎤, then ⎡α=α⎤ and ⎡α=β⎤ differ
in cognitive value.
Given (I) and (II), even if all the English-speaking community were to
know that Hesperus is Phosphorus it will still be possible for there to be
a subject for whom sentences of the form ‘Hesperus is Phosphorus’ will
be cognitively valuable (even if that subject is only a possible subject, not
an actual one). So ‘Hesperus is Hesperus’ will always differ in cognitive
value from ‘Hesperus is Phosphorus’.
However, a difference in cognitive value is not exhausted by (I)
and (II). There is more to it than this. ⎡α=α⎤ is knowable a priori whilst
⎡α=β⎤ is not necessarily knowable a priori, 11 where the latter form of iden-
tity-sentences, but not the former, contain valuable extensions to our
knowledge. This is what makes a sentence cognitively valuable, viz. that it
11 Frege says further that ⎡α=α⎤ is analytic, and whilst he does not explicitly
extend the difference by claiming in OSR that ⎡α=β⎤ is synthetic, in his Be-
griffsschrift he had already recognized its synthetic nature, and since in “Function
and Concept” he had made it clear that the truth of α=β is not immediately rec-
ognizable, we can say that it is not knowable a priori. I take it that such analytic-
synthetic difference is a good indication of a difference in cognitive value, but I
do not take it to be essential to such a difference. For this reason I have left it
out of the main characterization of the difference between sentences of the form
⎡α=α⎤ and those of the form ⎡α=β⎤.
294 MAITE EZCURDIA
may extend our body of knowledge. In the light of this, the question we
must ask ourselves when considering Frege’s argument is whether a mere
difference in notation between ⎡α=α⎤ and ⎡α=β⎤ would suffice for the
following:
(a) to account for the difference in cognitive value between ⎡α=α⎤

and ⎡α=β⎤ by doing justice to the claim that sentences of the
form ⎡α=β⎤ may extend our knowledge; and
(b) to allow for the possibility of a rational and understanding sub-

ject to take opposing propositional attitudes to α=α and α=β.
Millikan’s psychological notational account suffices to account for (b),

but what remains to be seen is whether her account or the Begriffsschrift
notational account suffice for (a) too. This is the first stage of The Nega-
tive Phase of Frege’s argument.
The Second Stage

In his Begriffsschrift Frege took it that the difference between ⎡α=α⎤
and ⎡α=β⎤ is a mere difference in notation. This account of identity was
intended to provide the logical form of sentences with the identity-sign.
Frege noticed in the Begriffsschrift that if there were two co-referring signs
which did not differ trivially and were joined by an identity-sign, then
their logical form would be ⎡α=β⎤, and not ⎡α=α⎤. 12 So sentences of the
form ⎡α=α⎤ contain signs which either differ solely as a matter of formu-
lation or are two tokens of the same type. In this way all sentences with
12 There are other cases where one might think that two signs differ trivially,
viz. when one is an abbreviation of another as in the case of ‘NY’ for ‘New
York’.
this logical form assert identity or equivalence in the reference of two

tokens of the same sign, whereas sentences of the form ⎡α=β⎤ do not.
Sentences of the latter form say rather that two different referential ex-
pressions have the same reference, that ‘α’ and ‘β’ are equivalent in what
they refer to, viz. that ‘α’ has the same reference as ‘β’.
As a logical form proposal this notational account is not intended
as an explanation of the psychological phenomenon which involves the abil-
ity of a rational and competent subject of a language to take opposing
attitudes to the content of sentences of the form ⎡α=α⎤ and ⎡α=β⎤, and
hence of the difference in cognitive value between those of the form
⎡α=α⎤ and those of the form ⎡α=β⎤. Millikan’s psychological notational
account, in contrast, does intend to explain such phenomenon: a subject
can take opposing attitudes to the proposition(s) expressed by sentences
like ‘Hesperus is Hesperus’ and ‘Hesperus is Phosphorus’ because such
sentences have different syntactical forms and the mind is sensitive to
such differences. In OSR Frege does not think that his Begriffsschrift nota-
tional account is able to account for differences in cognitive value. His
reasons will also pose a challenge to Millikan’s own psychological nota-
tional account.
The Third Stage

Under the Begriffsschrift notational account what is said by ⎡α=β⎤ is
something about the expressions contained in ⎡α=β⎤, viz. the relation of
sameness of reference of ‘α’ and ‘β’. Frege notes that this relation be-
tween the expressions would hold only insofar as they referred to the
same thing. The telling paragraph of what he thinks regarding such a no-
tational account is the following (which I have split into three parts):
A But this is arbitrary. Nobody can be forbidden to use an arbi-

trarily producible event or object as a sign for something. In
296 MAITE EZCURDIA
that case the sentence a=b would no longer refer to the subject
matter, but only to its mode of designation; we would express
no proper knowledge by its means. But in many cases this is just
what we want to do.
B If the sign ‘a’ is distinguished from the sign ‘b’ only as an object
(here, by means of its shape) not as a sign (i.e. not by the man-
ner in which it designates something), the cognitive value of
a=a becomes essentially equal to that of a=b, provided a=b is
true.
C A difference can arise only if the difference between the signs
corresponds to a difference in the mode of presentation of the
thing designated. (OSR, p. 57)
The Third Stage of Frege’s argument is contained in this paragraph. C

clearly belongs to the positive part of the argument, so for the present we
shall ignore it. A is where the argument against the Begriffsschrift notational
account will be found. B contains a mixture of the negative and the posi-
tive phases of the argument as we shall presently see.
The interpretation I shall be giving of this paragraph does present
Frege as having reasons for why notational accounts will not do to ex-
plain the difference in cognitive value between sentences of the form
⎡α=α⎤ and ⎡α=β⎤, and will leave the way open for why senses are to un-
derpin such differences in cognitive value.
A concludes, roughly, that if the Begriffsschrift notational account of
identity-sentences is true we would express no proper knowledge when using
⎡α=β⎤. But why would it not express ‘proper knowledge’?
Suppose, for the sake of argument, that identity-sentences claim
merely an identity in the reference of two signs as the Begriffsschrift nota-
tional account says. Then the knowledge we get of ⎡α=β⎤ is purely knowl-
edge of a language; it is purely metalinguistic knowledge (viz. expressed
metalinguistically). Coming to know that α=β is just coming to know that

the reference of ‘α’ is the same as the reference of ‘β’. 13 If this is so then
scientific discoveries would just be discoveries about language. For ex-
ample, coming to know that molecules are different from atoms would
just amount to coming to know that the reference of ‘molecules’ differs
from the reference of ‘atoms’. Coming to know that Hesperus is Phos-
phorus would just be coming to know that ‘Hesperus’ and ‘Phosphorus’
have the same referent. But our intuitions tell us that this is not always
what we take proper knowledge to be. Although there are cases in which
knowledge is linguistic (viz. about a language), there are other cases in
which it is not (in certain genuine scientific discoveries). In other cases,
knowledge is geological, economical, chemical, biological, physical (as
pertaining to physics), or astronomical (amongst others). Thus, when we
take sentences like ‘Hesperus is Phosphorus’ we must explain that they
are cognitively valuable insofar as they yield knowledge in astronomy, and
not knowledge which is metalinguistic, viz. about our language.
But the more crucial (and related) argument against the Begriffsschrift
notational account is that it cannot properly account for the difference in
cognitive value between ⎡α=α⎤ and ⎡α=β⎤. It cannot distinguish properly
between the cognitive value of sentences like ‘Hesperus is Phosphorus’
13 I am interpreting Frege’s expression ‘its mode of designation’ in

In that case the sentence a=b would no longer refer to the subject matter
but only to its mode of designation…
to mean its sign, the way in which the reference is referred to by the sign. ‘Mode
of designation’ as used here differs from Frege’s ‘the manner in which it desig-
nates something’ in B where ‘it’ refers to the sign. The latter, as we shall see,
refers to ways of determining references or to senses. This is not so with the
former since in ‘its mode of designation’ ‘its’ indicates that it concerns the refer-
ence’s mode of designation, the sign, and not the way in which the sign refers to
something.
298 MAITE EZCURDIA
and that of sentences like ‘Hesperus is ’, where ‘’ is a sign introduced
arbitrarily as having the same reference as ‘Hesperus’. In not being able to
distinguish between the cognitive value of these, the Begriffsschrift nota-
tional view cannot account properly for the cognitive value of sentences
of the form ⎡α=β⎤, and so cannot properly account for the way the cogni-
tive value of such sentences differs from that of sentences of the form
⎡α=α⎤. This in a nutshell is what the argument in A is. Let us spell it out.
Anyone can introduce any arbitrary sign to refer to an object, but
coming to know that the new sign referred to the same object that an old
sign did would not constitute proper knowledge.
Compare
(i) Hesperus is Phosphorus
with
(ii) Hesperus is .
(i) and (ii) are both true. Prima facie these two sentences differ in their
cognitive value. The former, but not the latter, gives us proper knowl-
edge; it reports a new and genuine scientific discovery. This does not oc-
cur with (ii). (ii) at most purports to tell us how someone intends us to
use ‘’. But (i) purports to do something different from (ii). What it pur-
ports to do must be described by the account of identity-sentences.
However, if the Begriffsschrift notational account cannot bring out the dif-
ference between (i) and (ii), then it will not be able to account properly
for the cognitive value of (i).
For the Begriffsschrift notational account, (i) is equivalent to
(i´) ‘Hesperus’ refers to the same thing as ‘Phosphorus’,

and (ii) is equivalent to
(ii´) ‘Hesperus’ refers to the same thing as ‘’.
Thus, such an account does not have the tools to distinguish properly (i)
from (ii). It does not say why (i) constitutes an extension in our proper
knowledge but (ii) does not, nor why we think that, although (ii) is really
expressing some metalinguistic knowledge, (i) is not. For the Begriffsschrift
notational account knowing that (i) and knowing that (ii) are just two
forms of the same kind of metalinguistic knowledge and nothing else. So
our intuitions about the cognitive value of (i) being different from that of
(ii) are not vindicated.
Furthermore, if an account cannot distinguish the cognitive value
of sentences like (i) from that of sentences like (ii), then it cannot give a
proper account of the cognitive value of (ii) and hence of the difference
in cognitive value between (i) and
(iii) Hesperus is Hesperus

In a notational account (iii) is equivalent to
(iii´) ‘Hesperus’ refers to the same thing as ‘Hesperus’.
Although (iii´) and (i´) may be said to be non-equivalent in cognitive

value, their non-equivalence will be accounted for in a way that does not
vindicate our intuitions, for it will not vindicate our intuitions concerning
the difference in cognitive value between sentences like (i) and those like
(ii). Furthermore, it will not vindicate our intuitions that when the Baby-
lonians discovered that Hesperus was Phosphorus their discovery was
not one purely about language. Their discovery was a genuine one about
a scientific fact in astronomy, and not one about their language. If by
something’s being equivalent in cognitive value to something else Frege
means that they are equivalent in extending proper knowledge, then we may
300 MAITE EZCURDIA
say that Frege is right in asserting in B that sentences of the form ⎡α=α⎤
and ⎡α=β⎤ are equivalent in cognitive value for the Begriffsschrift notational
account. Both extend our metalinguistic knowledge in the same way. That
they are not thus equivalent is an argument against such an account.
We are now in a position to spell out fully the argument of The
Negative Phase of Frege’s argument for the introduction of sense:
1. ‘α’ and ‘β’ are simple co-referential expressions.
The First Stage

2. ⎡α=α⎤ may differ in cognitive value from ⎡α=β⎤ insofar as the
latter but not the former provides valuable extensions to our
knowledge.
The Second Stage
3. ⎡α=α⎤ may differ from ⎡α=β⎤ insofar as ⎡α=α⎤ is semantically
equivalent to ⎡the sign ‘α’ has the same reference as the sign
‘α’⎤ and ⎡α=β⎤ is semantically equivalent to ⎡the sign ‘α’ has the
same reference as the sign ‘β’⎤. (The Begriffsschrift Notational Ac-
count)
The question is then whether a notational account can properly distin-

guish the cognitive value of ‘Hesperus is Phosphorus’ from that of ‘Hes-
perus is ’, where ‘’ is an arbitrary sign introduced as having the same
reference as ‘Hesperus’. This is The Third Stage:
∴4. ‘Hesperus is Phosphorus’ is semantically equivalent to ‘“Hes-

perus” has the same reference as “Phosphorus”’, and ‘Hes-
perus is ’ is semantically equivalent to ‘“Hesperus” has the
same reference as “”’. (Given 3)
∴5. Coming to know that ‘Hesperus’ has the same reference as

‘Phosphorus’ is as cognitively valuable as coming to know that
‘Hesperus’ has the same reference as ‘’.
6. Coming to know that Hesperus is Phosphorus constitutes a
valuable extension to our knowledge, whereas coming to
know that Hesperus is does not constitute such a valuable
extension to our knowledge.
∴7. The Begriffsschrift notational account cannot account properly
for the cognitive value of ‘Hesperus is Phosphorus’.
It is for this reason that Frege both rejects the Begriffsschrift notational ac-
count, and holds that the difference in cognitive value between ‘Hesperus
is Hesperus’ and ‘Hesperus is Phosphorus’ has to be explained in terms
other than their signs or the reference of ‘Hesperus’ and ‘Phosphorus’.
We can generalize the point to all referring expressions because for
any referring expression e we can introduce an arbitrary sign e´ and stipu-
late that e´ has the same reference as e. In all cases the notational account
will be unable to account for the difference in cognitive value between
those sentences which are of the form ⎡α=β⎤ but which have in the ‘β’-
place of that form an arbitrary sign from those that don’t. For this reason,
we can say further that the difference in cognitive value between ⎡α=α⎤
and ⎡α=β⎤ will have to be explained in terms other than their signs or the
reference of ‘α’ and ‘β’. This concludes The Negative Phase of Frege’s
argument against the Begriffsschrift notational account. But notice that it
can be extended to the psychological notational account.
How could the psychological notational account be capable of dis-
tinguishing between the cognitive value of ‘Hesperus is Phosphorus’ and
that of ‘Hesperus is ’? Millikan claims that a notational difference be-
tween ⎡α=α⎤ and ⎡α=β⎤ is just a difference in modes of presentation
302 MAITE EZCURDIA
which have no semantic relevance, and that such a difference will suffice
for a subject to take opposing attitudes towards sentences of these forms.
She believes that these differences in vehicles or in purely syntactical
modes of presentation will suffice to account for differences in cognitive
value, because they will suffice to ‘move the mind differently’. It is not
clear, however, that such an account (or even supplementing the Be-
griffsschrift notational account with the claim about the different syntactical
modes of presentation) will suffice to bring out the difference in cogni-
tive value between ‘Hesperus is Phosphorus’ and ‘Hesperus is ’. It will
certainly be possible to account for why a subject took opposing attitudes
to these sentences or their contents since her mind would have been sen-
sitive to their differences in modes of presentation, in vehicles. But it will
not be possible for the psychological notational account (nor the supple-
mented Begriffsschrift account) to explain or predict the difference in cogni-
tive value between ‘Hesperus is Phosphorus’ and ‘Hesperus is ’. The dif-
ference in vehicles or modes of presentation does not suffice to explain
why the former, but not the latter, brings new and proper knowledge in the
way explained above. In fact, the psychological notational account (as
well as the supplemented one) will predict that ‘Hesperus is ’ is cogni-
tively valuable because the difference in signs involve a difference in
purely psychological modes of presentation and so a subject may take a
different attitude to it from the one it takes to ‘Hesperus is Phosphorus’
or ‘Hesperus is Hesperus’. But ‘Hesperus is ’ is not cognitively valu-
able. So a supplemented notational account, as it stands, would make the
wrong predictions about such a sentence. Something more is needed.
This is the challenge which Frege’s account will pose to any notational
account (or for that matter, to any account), viz. to bring out the differ-
ence in cognitive value between ‘Hesperus is ’ and ‘Hesperus is Phos-
phorus’.
The Positive Phase

Frege thinks that sense can do the job of distinguishing between the
case in which an identity-sentence involves a sign introduced arbitrarily
from the case in which it does not. His notion of sense is already present
in his Begriffsschrift. There he distinguished between signs which differ
merely as a trivial matter of formulation and those which do not, which
are associated with or have different ways of determining reference:
[…] different names for the same [reference] are not always just a trivial
matter of formulation; if they go along with different ways of determining the
[reference], they are relevant to the essential nature of the case. (p. 12; my
emphasis)
In OSR Frege notices that cases where there is a syntactic difference

merely as a trivial matter of formulation are not the only cases in which
there is no one-one correspondence between signs and ways of determin-
ing reference. The sort of case he has in mind is the one we considered in
The Negative Phase: when a sign is introduced arbitrarily. Since it is pos-
sible to introduce a sign arbitrarily, then it is possible for a difference in
signs not to entail a difference in the ways of determining references or
ways of referring. A sign may be introduced arbitrarily either via osten-
sion or via another sign, viz. as having the same reference as another sign.
So the new sign which is introduced thus will not give us a new way of
referring different from the one of the sign used to introduce it or from
the way it was introduced. Suppose that someone introduces ‘’ as hav-
ing the same reference as ‘Hesperus’ (as I have done here). The way in
which ‘’ is introduced will ensure that the way of determining the refer-
ence of ‘’ is just that of ‘Hesperus’. ‘’ will not have a new way of de-
termining its reference, and so ‘Hesperus is ’ will not provide new proper
or scientific (in this case, astronomical) knowledge to the person who has
304 MAITE EZCURDIA
already been introduced to the sign. Furthermore, were someone to un-

derstand and to take opposing attitudes to the contents of ‘Hesperus is
Phosphorus’ and ‘Hesperus is ’, she would be irrational. Unlike ‘Hespe-
rus is ’, ‘Hesperus is Phosphorus’ does provide new proper knowledge
to the person who has already been introduced both to ‘Hesperus’ and to
‘Phosphorus’. These simple referring expressions have associated ways of
determining their references which are independent of each other, so the
subject who has already been introduced to those signs can learn some-
thing new with ‘Hesperus is Phosphorus’.
Frege thinks that only if ways of determining reference are distin-
guished from the signs, can we distinguish the cases in which identity-
sentences do contain a furtherance in knowledge from those that do not.
These ways of determining reference are what correspond to the semantic
level of sense. It is true that in C Frege speaks of modes of presentation
and not of ways of determining reference. But it is also true that in his Be-
griffsschrift Frege speaks of ways of determining reference as that which is
associated with signs. So although Frege doesn’t distinguish between ways
of determining reference and modes of presentation, the distinction is cru-
cial for, as we said at the outset, modes of presentation are essentially psy-
chological whereas senses are essentially semantic. Having made the dis-
tinction between ways of determining references and modes of presenta-
tion, we must then ask ourselves whether ways of determining references
are essentially semantic or not.14
Frege wants differences in senses to account for the difference in cognitive

14
value between the propositions α=β and α=α, and so to allow for the possibility
of a subject taking opposing propositional attitudes to sentences of those two
forms. If a difference in senses is to allow for such a possibility there must be
some way in which senses are psychologically relevant. So there must be a connec-
tion between psychological modes of presentation and semantic senses or ways
When speaking of ways of determination in his Begriffsschrift Frege

gives an example which would lead us to believe that ways of determining
references are essentially epistemic. Of two ways of determining the same
point in the circumference of a circle, where the point is referred to with
‘A’ when determined in the first way and as ‘B’ when determined in the
second way, Frege writes:
The same point is determined in a double way:
(1) It is directly given in experience.
(2) It is given as point B corresponding to the straight line being perpen-
dicular to the diameter. (Begriffsschrift, p. 11)
(1) and (2) state two epistemically different ways of determining the refer-
ence, each corresponding respectively to ‘A’ and ‘B’. Yet one need not be
committed to the idea that all ways of determining reference are always
epistemic ways of determining the reference as in this example. The impor-
tant point about ways of determination is that they determine the refer-
ence of a name, sentence or predicate, that is, of a meaningful expression
in the language, in a certain way. Such ways of determining reference may
well be essentially semantic. Given what we said was essentially semantic,
ways of determining reference will have to be shown to be either truth-
values, truth-conditions, reference or objects of reference. There are at
least two manners (which do not exclude each other) in which ways of
determining reference may be said to be semantic categories. One of
them is to say that ways of determining reference are just ways of refer-
ring and that the relation of reference is what is included under ‘reference’ in
what is semantic, so that ways of determining reference are just different
ways in which the relation of reference may be implemented. In this case,
we come close to identifying senses as ways of referring with semantic
of determination. For more on the distinction and connection between ways of

determining reference and modes of presentation see Ezcurdia, 1995.
306 MAITE EZCURDIA
rules. Such assimilation may well be threatened by context-sensitive ex-

pressions where a way of referring may not be identifiable with the se-
mantic rule of the expression-type but is rather what is expressed by one
of the tokens of such expression-type. In these cases, such ways of refer-
ring would just be instantiations of the semantic rule in a given context.
Another way of viewing ways of referring as essentially semantic (com-
patible with the latter) is one offered by McDowell (1977 and 1980) and
Evans (1981 and 1982). Such a view takes ways determining reference of
sentences to be capturable as interpretive truth-conditions and of semanti-
cally meaningful expressions as interpretive truth-axioms, that is, truth-
conditions or axioms of an interpretive truth-theory à la Tarski. I do not
have space to go into this in the present paper, so I shall leave it aside for
another time. 15 For our purposes it suffices to point out these two ways
in which ways of referring may be essentially semantic.
Under this account of Frege’s argument, it is easy to see why Frege
says in B above, that if the signs are not distinguished by ‘the manner in
which they designate something’, that is, by their ways of determining
something, but only as syntactical objects, then the difference in cognitive
value between ⎡α=α⎤ and ⎡α=β⎤ cannot be properly accounted for. We
already have reasons from The Negative Phase for believing that neither
reference nor mere syntactical differences will suffice to allow for an ac-
count of the difference in cognitive value between ⎡α=α⎤ and ⎡α=β⎤. And
we have already sketched how differences in ways of determination will
suffice for such a task. Only a difference in ways of determination will
account properly for differences in cognitive value as he envisages in C
because, once reference and signs have been ruled out, that is all that is
left.
We can then spell out the argument for The Positive Phase thus:
15 For a discussion of this see Ezcurdia, 1997.
1. ‘Hesperus is Phosphorus’ differs in cognitive value from ‘Hes-

perus is ’.
2. Syntactic differences cannot account properly for these differ-
ences. (From The Negative Phase)
3. ‘Hesperus’ and ‘Phosphorus’ correspond to or have two different
associated ways of determining the same reference.
4. ‘’ is a sign introduced as having the same reference as ‘Hespe-
rus’, so it will not express a new way of determining reference
different from that associated with ‘Hesperus’.
∴5. ‘Hesperus is ’ does not express new proper knowledge.
∴6. Differences in ways of determination can give us the desired
differences in cognitive value between ‘Hesperus is Phospho-
rus’ and ‘Hesperus is ’.
Until we have a reason then for supposing that something different from
ways of determining reference or senses can allow for the difference in
cognitive value between ‘Hesperus is Hesperus’ and ‘Hesperus is ’, we
may assert with Frege that only differences in ways of determination will
be able to account for such differences in cognitive value. 16
16 Salmon (1986) takes ‘Hesperus is Hesperus’ and ‘Hesperus is Phosphorus’
to be asserting the same proposition or to be encoding the same semantic infor-

mation, viz. the self-identity of Venus. Nevertheless, he takes those sentences to
differ in their pragmatically encoded information. Such a difference is meant to
account for why it is that a subject can, without being irrational, assent to ‘Hespe-
rus is Hesperus’ but not to ‘Hesperus is Phosphorus’. But nothing of what
Salmon says helps us distinguish ‘Hesperus is Phosphorus’ from ‘Hesperus is
’. All it does is say how it is that a subject is able to take opposing attitudes to
one and the same content.
308 MAITE EZCURDIA
3
What is then the state of play between the Fregean and the Russel-
lian?
Nothing of what I have said so far entails that the situation which
Millikan envisages, or which any other Russellian might envisage, is false
and it is not available as an alternative account. Yet it is important to note
that such an account would be a revisionary account not only of our lan-
guage but more importantly of what we take knowledge and genuine sci-
entific discoveries to be. 17 What I have said so far only entails that the
argument for Fregean sense is both non-question begging and valid. If
someone, however, intends to pursue a notational account then she must
face what I shall call ‘the Fregean Challenge’. The Fregean Challenge just
consists in either offering an account which allows for an explanation of
the differences in the knowledge we derive from ‘Hesperus is Phos-
17 Some might try (see Blackburn, 1979, pp. 27-8) to revive the notational ac-
count of the difference between ‘Hesperus is Hesperus’ and ‘Hesperus is Phos-

phorus’, in the following manner:
(i) the way of determination associated with ‘Hesperus’ determines the same
reference as the way of determination associated with ‘Hesperus’
differs in cognitive value from

(ii) the way of determination associated with ‘Hesperus’ determines the same
reference as the way of determination associated with ‘Phosphorus’.
The problem with this is that the cognitive value of ‘Hesperus is Phosphorus’
and of (ii) may not be the same. A subject may know that Hesperus is Phospho-
rus, but not that the way of determination associated with ‘Hesperus’ determines
the same reference as the way of determination associated with ‘Phosphorus’. In
knowing that Hesperus is Phosphorus the subject has knowledge which is sensi-
tive to differences in ways of determination or senses, but this is different from
having knowledge of those ways of determining reference.
phorus’ and ‘Hesperus is ’, or saying why coming to know that Hespe-
rus is Phosphorus does not differ in cognitive value from our coming to
know that Hesperus is (contrary to our intuitions).
Against the Begriffsschrift notational account of sentences there is at
least one prima facie undesirable consequence which needs to be pointed
out. If identity-sentences containing simple referring expressions assert
just equivalence in reference between two signs then identity will turn out
to be contingent, when we in fact think it as necessary. ‘“Hesperus” has
the same reference as “Phosphorus”’ would be only contingent because
those signs or expressions could have had different references from that
of Venus. Furthermore, the metalinguistic fact that ‘Hesperus’ has the
same reference as ‘Phosphorus’ cannot alone explain properly why know-
ing that Hesperus is Phosphorus is having some knowledge in astronomy.
The Begriffsschrift notational account will have to say more than just assert
the metalinguistic fact in order to explain why the knowledge we get is
astronomical. But it will have to do more. For the Fregean our understand-
ing of simple referring expressions will involve knowledge of their associ-
ated ways of determination. But the Russellian will have to explain how it
is that our understanding of simple referring expressions can be framed in a
notational form.
Yet a Russellian may not want to go the notational way in account-
ing for the difference in cognitive value between sentences of the form
⎡α=α⎤ and ⎡α=β⎤. She might just want to claim that there is no difference
in cognitive value between sentences of these two forms – as Salmon (in
some way) and Millikan have done –, thus denying that Frege’s argument
is sound by claiming that premise 2 of The Negative Phase is false. Or she
might just claim that ‘Hesperus is Phosphorus’ is just as cognitively valu-
able as ‘Hesperus is ’. To take either of these two routes are viable op-
tions, yet they are revisionist options. And to take a revisionist option one
needs to be furnished with a good argument against pursuing an account
310 MAITE EZCURDIA
that intends to vindicate our intuitions and accord with the commonsense
and day to day evidence as a Fregean account would intend to do.
REFERENCES
BLACK and GEACH (eds.). Translations from the Philosophical Writings of

Gottlob Frege. Oxford: Blackwell, 1980.
BLACKBURN, S. “Thoughts and Things”. Aristotelian Society, Supple-
mentary Volume, LIII, pp. 23-41, 1979.
BROOK, A. and R. DEVIDI (eds.). Self-reference and Self-awareness. Am-
sterdam: John Benjamins Publisher, 2001.
BURGE, T. “Belief De Re”. Journal of Philosophy, LXXIV, pp. 338-62,
1977.
EVANS, G. “Understanding Demonstratives”. In: H. Parret and J. Bou-
veresse (eds.). Meaning and Understanding. Berlin: De Gruyter, 1981.
Reprinted in P. Yourgrau (ed.). Demonstratives. Oxford University
Press, pp. 71-96, 1990.
———. Varieties of Reference. In: McDowell (ed.). Oxford University
Press, 1982.
EZCURDIA, M. Sense, Indexicals and Action. University of London, 1994
(PhD Thesis).
———. “Modos de Presentación y Modos de Determinación”. Crítica,
XXVII, pp. 57-96, 1995.
———. “Dynamic and Coherent Thoughts”. European Review of Philoso-
phy. Vol. 2 Cognitive Dynamics, San Francisco: CSLI, pp. 105-39,
1997.
———. “Thinking About Myself”. In: A. Brook and R. DeVidi (eds.)

Self-reference and Self-awareness. Amsterdam: John Benjamins Pub-
lisher, pp. 197-203, 2001.
FREGE, G. Begriffsschrift. In: Black et al., 1980.
———. “Function and Concept”. In: Black et al., 1980.
———. “On Sense and Reference”. In: Black et al., 1980.
———. “The Thought”, 1918. In: Salmon et al., 1988.
KRIPKE, S. Naming and Necessity. Cambridge, MA: Harvard University
Press, 1980.
MILLIKAN, R. “Perceptual Content and the Fregean Myth”. Mind, C,
pp. 439-59, 1991.
MCDOWELL, J. “De Re Senses”. Philosophical Quarterly, 34, pp. 283-94,
1977. Reprinted in C. Wright (ed.). Frege: Tradition and Influence.
Oxford: Blackwell, pp. 98-109, 1984.
———. “On the Sense and Reference of a Proper Name”. In: M. Platts
(ed.). Reference, Truth and Reality. Routledge & Kegan Paul, pp. 141-
66, 1980.
PERRY, J. “Frege on Demonstratives”, 1977. In: P. Yourgrau (ed.). De-
monstratives. Oxford: Oxford University Press, pp. 50-70, 1990.
PLATTS, M. (ed.). Reference, Truth and Reality. London: Routledge & Ke-
gan Paul, 1980.
SAINSBURY, M. “On a Fregean Argument for the Distinctness of Sense
and Reference”. Analysis, XLIII, pp. 12-14, 1983.
SALMON, N. Frege’s Puzzle. Cambridge, MA: MIT Press, 1986.
312 MAITE EZCURDIA
SALMON, N. and S. SOAMES. (eds.). Propositions and Attitudes. Oxford:

Oxford University Press, 1988.
TASCHEK, W. “Frege’s Puzzle, Sense and Information Content”. Mind,
CI, pp. 767-91, 1991.
WRIGHT, C. (ed.). Frege: Tradition and Influence. Oxford: Blackwell, 1984.
YOURGRAU, P. (ed.). Demonstratives. Oxford: Oxford University Press,
1990.

Ezcurdia - Introducing Sense (Frege)

Uploaded by

Copyright:

Available Formats

Ezcurdia - Introducing Sense (Frege)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ezcurdia - Introducing Sense (Frege)

Uploaded by

Copyright:

Available Formats

Logic, Truth and Arithmetic. Essays on Gottlob Frege.

Marco Ruffino (ed.).

Instituto de Investigaciones Filosóficas

Key-words: Frege. Millikan. Sense. Reference. Content. Cognitive value.

amongst many others). But what sense is is determined at least in part by

1 See Ezcurdia, 1994.

The ‘paradox’ of identity says that, whereas a statement of the form

This interpretation of Frege’s argument has a simple form. It has two

1. ⎡α=α⎤ differs in cognitive or informative value from ⎡α=β⎤.

Millikan (1991) has challenged a Fregean introduction of sense on

4 Some might object to speaking of complex referential expressions, claiming

the strategy above in distinguishing different contents. According to the

So even if one were to get rid of genuinely simple referential expressions,

8 I have argued elsewhere that a difference in information content may not

amount to a difference in cognitive value. See Ezcurdia, 1994.

9 Frege takes the notational account he proposes in his Begriffsschrift to ac-

Millikan would find even this Begriffsschrift notational account sus-

6. A rational subject, who understands the language in question,

But 6 is only justified, according to Millikan, if the Fregean does some

Cicero is Tully is an informative thought whereas Cicero is Cicero is not, so

In order to produce a valid argument, the Fregean would need to show

7. A rational subject, who understands the language, cannot take

10 For Millikan something counts as content if it is semantic, and something

is semantic if it is either purely referential truth-conditions, referents or extensions.

The Negative Phase

(I) If ⎡α=α⎤ and ⎡α=β⎤ differ in cognitive value, then it is possible

(II) If there is (as a matter of fact) a subject who is rational and

(a) to account for the difference in cognitive value between ⎡α=α⎤

(b) to allow for the possibility of a rational and understanding sub-

Millikan’s psychological notational account suffices to account for (b),

The Second Stage

this logical form assert identity or equivalence in the reference of two

The Third Stage

A But this is arbitrary. Nobody can be forbidden to use an arbi-

The Third Stage of Frege’s argument is contained in this paragraph. C

metalinguistically). Coming to know that α=β is just coming to know that

13 I am interpreting Frege’s expression ‘its mode of designation’ in

(i´) ‘Hesperus’ refers to the same thing as ‘Phosphorus’,

(ii´) ‘Hesperus’ refers to the same thing as ‘’.

(iii) Hesperus is Hesperus

Although (iii´) and (i´) may be said to be non-equivalent in cognitive

1. ‘α’ and ‘β’ are simple co-referential expressions.

The First Stage

The question is then whether a notational account can properly distin-

∴4. ‘Hesperus is Phosphorus’ is semantically equivalent to ‘“Hes-

∴5. Coming to know that ‘Hesperus’ has the same reference as

The Positive Phase

In OSR Frege notices that cases where there is a syntactic difference

already been introduced to the sign. Furthermore, were someone to un-

Frege wants differences in senses to account for the difference in cognitive

When speaking of ways of determination in his Begriffsschrift Frege

of determination. For more on the distinction and connection between ways of

rules. Such assimilation may well be threatened by context-sensitive ex-

15 For a discussion of this see Ezcurdia, 1997.

1. ‘Hesperus is Phosphorus’ differs in cognitive value from ‘Hes-

16 Salmon (1986) takes ‘Hesperus is Hesperus’ and ‘Hesperus is Phosphorus’

to be asserting the same proposition or to be encoding the same semantic infor-

count of the difference between ‘Hesperus is Hesperus’ and ‘Hesperus is Phos-

differs in cognitive value from

BLACK and GEACH (eds.). Translations from the Philosophical Writings of

(ii´) ‘Hesperus’ refers to the same thing as ‘’.