From The Begriffsschrift To "Über Sinn Und Bedeutung": Frege As Epistemologist And Ontologist

ABSTRACT

Frege presents the eponymous distinction of "Über Sinn und Bedeutung" (USB) as a correction to the theory of identity he held in the Begriffsschrift.¹ His retrospective assessment has prompted different reactions from commentators. Many have simply taken him at his word.² Others, who have lingered over the relevant texts, have encountered a difficulty: in the Begriffsschrift Frege appears to raise the same problem with which he begins USB and to respond to it in words that are strikingly similar to those found in the later work.³ Thus, a number of important scholars have been led to argue that Frege misrepresents his earlier view: contrary to what he says in USB, the Begriffsschrift already contains the distinction between Sinn and Bedeutung.⁴ Despite their obvious differences, the logic of both interpretations requires that the two works be responsive to the same problem. In this paper I challenge that fundamental premise.⁵ If I might be allowed to express my thesis tersely but suggestively: the problem of the Begriffsschrift is epistemological; that of USB, ontological.⁶ The two problems and Frege's solutions to them shall be explained in what follows.

Keywords:
Frege; Sinn; Ontology

I.

Frege's argument for the Sinn-Bedeutung distinction is found in the opening paragraph of USB. In this section I offer a partial reading of that text. It is partial in that my sole aim is to bring out Frege's criticism of the Begriffsschrift. That criticism begins from an obvious fact: identity statements express what Frege calls "proper knowledge." He contends that the theory of identity put forward in the Begriffsschrift cannot account for their doing so. It will be helpful to have the full paragraph before us:

Identity gives rise to challenging questions which are not altogether easy to answer.⁷ Is it a relation? A relation between objects, or between names or signs of objects? In my Begriffsschrift I assumed the latter. The reasons which seem to favour this are the following: a=a and a=b are obviously statements of different cognitive value; a=a holds a priori and, according to Kant, is to be labeled analytic, while statements of the form a=b often contain very valuable extensions of our knowledge and cannot always be established a priori. The discovery that the rising sun is not new every morning, but always the same was one of the most fertile astronomical discoveries. Even today the reidentification of a small planet or a comet is not always a matter of course. Now if we were to regard identity as a relation between that which the names 'a' and 'b' designate, it would seem that a=b could not differ from a=a (i.e. provided a=b is true). A relation would thereby be expressed of a thing to itself, and indeed one in which each thing stands to itself but to no other thing. What we apparently want to state by a=b is that the signs or names 'a' and 'b' designate the same thing, so that those signs themselves would be under discussion; a relation between them would be asserted. But this relation would hold between the names or signs only insofar as they named or designated something. It would be mediated by the connection of each of the two signs with the same designated thing. But this is arbitrary. Nobody can be forbidden to use any arbitrarily producible event or object as a sign for something. In 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 that is just what we want to do. 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 manner in which it designates something), the cognitive value of a=a becomes essentially equal to that of a=b, provided a=b is true. A difference can arise only if the difference between the signs corresponds to a difference in the mode of presentation of the thing designated. Let a, b, c be the lines connecting the vertices of a triangle with the midpoints of the opposite sides. The point of intersection of a and b is then the same as the point of intersection of b and c. So we have different designations for the same point, and these names ('point of intersection of a and b', 'point of intersection of b and c') likewise indicate the mode of presentation; and hence the statement contains actual knowledge.

The first three sentences of the paragraph are a false start. They express a concern with the nature of identity; in particular, with whether it is a relation and, assuming that it is, with the specification of its terms. Despite appearances, the nature of identity cannot be at issue here. Support for that view comes from a variety of quarters, and a full defense of it will emerge from my detailed readings of the opening paragraph of USB and the correlative section of the Begriffsschrift. For the moment one straightforward and powerful argument will suffice: Identity, Frege states, is a relation; and he explicitly excludes relations from consideration in USB.⁸

This reading is confirmed by the development of the opening paragraph. That development is fraught with interpretive difficulties, but it is clear that Frege's attention shifts from identity to identity-statements. The epistemic vocabulary that emerges into prominence shows that the specific problem at issue has to do with knowledge of those statements. One of the crucial phrases in that transition is 'proper knowledge'. Unpacking it requires distinguishing knowledge that is "proper" from knowledge that is not-from what I shall call "improper knowledge." Frege first employs the distinction in the following passage:

What we apparently want to state by a=b is that the signs or names 'a' and 'b' designate the same thing, so that those signs themselves would be under discussion; a relation between them would be asserted. But this relation would hold between the names or signs only insofar as they named or designated something. It would be mediated by the connection of each of the two signs with the same designated thing. But this is arbitrary (willkurlich). Nobody can be forbidden to use any arbitrarily producible event or object as a sign for something. In 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.

Frege is here considering a view - one that he claims he held in the Begriffsschrift - according to which one who knows a statement of the form 'a=b', knows only that the signs 'a' and 'b' designate the same object. Such knowledge is improper because what a sign designates is "arbitrary." Frege's point is simple, but his use of 'arbitrary' tends to obscure it. Only in anomalous circumstances would it be appropriate to describe the sign-object relation as arbitrary. Yet Frege plainly thinks he has identified a pervasive feature of language. A better way to make his point is to say that the sign-object relation is a matter of convention. For example, it is not arbitrary that the name 'Montague' applies to Romeo and the name 'Capulet' to Juliet. It is, however, purely a matter of convention.⁹ Improper knowledge is knowledge of such conventions. As Frege puts it, a sentence that expresses improper knowledge does not "refer to the subject matter, but only to its mode of designation." To put the same point in a way that brings out the contrast more effectively, proper knowledge is knowledge of things; improper knowledge is knowledge of what those things are called.

Frege's criticism of the Begriffsschrift can now be stated more fully. In USB he points out that there are identity-statements - 'the sun that rose in the sky this morning is the same as the sun that rose in the sky yesterday morning' and 'the point of intersection of a and b is the same as the point of intersection of b and c' - that express proper knowledge. His earlier theory, Frege contends, entails that identity-statements describe only linguistic conventions; that is, they express only improper knowledge.¹⁰

II.

The earlier Frege presents his theory of identity in §8 of the Begriffsschrift. There his specific concern is whether the formalism he is developing requires the identity sign. He begins by presenting an argument that it does not. (For ease of reference I have split the text into two parts: the first contains the argument proper; the second, the conclusion that follows from it.)¹¹

[A] Identity of content differs from conditionality and negation in that it applies to names and not to contents. Whereas in other contexts signs are merely representatives of their content, so that every combination into which they enter expresses only a relation between their respective contents, they suddenly display their own selves when they are combined by means of the sign for identity of content; for it expresses the circumstance that two names have the same content. Hence the introduction of a sign for identity of content necessarily produces a bifurcation in the meaning (Bedeutung) of all signs: they stand at times for their content, at times for themselves.

[B] first we have the impression that what we are dealing with pertains merely to the expression and not to the thought, that we do not need different signs at all for the same content and hence no sign whatsoever for identity of content.

[A] is the primary textual basis for interpretations that find the Frege of USB to be correct in attributing to the Begriffsschrift the thesis that identity is a relation between names. Adopting the terminology of Thau and Caplan, I shall call that thesis the "name view." The identification of the Begriffsschrift theory of identity with the name view is flawed in two respects. First, it uncritically accepts the misleading suggestion of the first three sentences of USB that the name view is concerned with the nature of identity.¹² As I shall try to show, the name view is better glossed as the thesis that identity-statements express only improper knowledge.¹³ Second, Frege does not endorse the name view.¹⁴ Defending those two interpretive claims will substantiate a third: the Begriffsschrift theory of identity is immune to the criticism of USB. It does not follow, however, that the distinction between Sinn and Bedeutung is already found in the Begriffsschrift. To the contrary, Frege's Begriffschift theory of identity does not require that distinction in order to deflect the criticism ventured in USB.

It is understandable that [A] is taken to express Frege's view. The thesis found there is the one that USB attributes to the Begriffsschrift. Further, the phrasing of the first sentence ("Identity of content differs from conditionality..."), taken in isolation, does not allow for [A] to be read in any other way than as a statement of the author's position. However, placing [A] in its larger dialectical context supports a different reading. The argument developed in [A] leads to the conclusion drawn in [B] that the vocabulary of the Begriffsschrift need not contain the identity sign. Yet the burden of §8 is to justify the inclusion of that sign. It thus follows that in [A] Frege is not presenting an argument with which he agrees but considering one to which he will respond. A more detailed consideration of [A] and [B] will show that the argument has nothing to do with the nature of identity.

Frege notes that if "identity of content... applies to names and not to content" then introducing the identity sign into the vocabulary of the Begriffsschrift "produces a bifurcation in the meaning of all signs: they stand at times for their content, at times for themselves." For example, in 'Fa' the sign 'a' stands for an object; in 'a=b' it stands for itself. To say that the identity sign produces a "bifurcation of meaning" is thus simply another way of saying that it renders the signs of the Begriffsschrift ambiguous. But ambiguity is one of the main flaws of natural language that the Begriffsschrift is supposed to correct. As Frege puts it elsewhere,

Language proves to be deficient, however, when it comes to protecting thought from error. It does not even meet the first requirement which we must place upon it in this respect; namely being unambiguous.¹⁵

The ambiguity introduced by the identity sign is easily corrected, however. As Frege goes on to point out, on the view he is considering, an identity-statement "expresses the circumstance that two names have the same content." In other words, 'a=b' turns out to be elliptical for the following:

'a' stands for O and 'b' stands for O

The proposed analysis shows how to eliminate the identity sign and with it the ambiguity it creates. A more direct way to achieve the same result is to ban co-referring names from the vocabulary of the Begriffsschrift. That is the remedy Frege proposes: "we have absolutely no need for different symbols of the same content, and thus no [need for a] symbol for identity of content either."¹⁶

The conclusion drawn in [B] poses an obvious problem for those who take the name view to be a thesis concerning the nature of identity. To bring out the difficulty, let me express the view in expanded form. It consists of three claims. Identity exists. It is a relation. Its terms are names. [B] states the implications of the name view for how identity is to be represented in the formalism of the Begriffsschrift. Yet [B] concludes that the vocabulary of that formalism contains neither the identity sign nor co-referring names. How that conclusion fits with the thesis that identity exists; that it is a relation, and that its terms are names is a mystery, to put it mildly.

Not only does the name view clash with [B], further reflection shows that, taken independently of worries over whether the vocabulary of the Begriffsschrift contains the identity sign, it is incoherent. Read literally and without much probing, the natural way to construe the claim that identity is a relation between names is as asserting that when an identity-statement is true, the descriptive signs that occur in it are identical to one another. For example, consider the true identity statement 'a=b'. According to the name view, that statement is more perspicuously written as the following true identity statement:

'a' is identical to 'b'

To be sure, most commentators do not impute this absurd view to the Begriffsschrift.¹⁷ They avoid doing so, however, not by supplying the text with an improved account of the nature of identity but by tacitly dropping that issue from consideration and proceeding as if the name view were concerned with the knowledge expressed by identity-statements. Michael Dummett provides a particularly clear illustration of this procedure:

In Begriffsschrift Frege held that identity was a relation between names and not between things. His motive for this view was to give an explanation of the informativeness of a true identity-statement.¹⁸

As Dummett would have it, in the Begriffsschrift Frege holds that the information content expressed by an identity-statement is that two signs designate the same object. Or, to use the terminology introduced earlier, that identity statements express only improper knowledge.

Two points need to be stressed here. First, the words that are commonly used to give content to the name view - 'identity is a relation between names' - ought to be cast aside, as they suggest a theory of the nature of identity that is neither intelligible nor taken seriously by those who use them. Second, the argument for the name view (understood now as the thesis that identity statements express only improper knowledge) does not make use of any premise concerning the nature of identity. As my account in Part I shows, the basis for the distinction between proper and improper knowledge is the claim that the sign-object relation is conventional. If that claim is correct, then it follows that there is nothing for identity-statements to describe (and hence nothing to be known) other than the linguistic convention that the descriptive signs occurring in them designate the same object.

This reading of the name view is confirmed by Frege's response to it, which is exclusively concerned to analyze the circumstances under which identity statements express proper knowledge. As one would expect, his counter-argument challenges the view that the sign-object connection is always conventional. Having granted in [B] that we have the "impression that... we do not need different signs at all for the same content and hence no sign whatsoever for identity of content," Frege continues,

[C] To show that this is an empty illusion I take the following example from geometry. Assume that on the circumference of a circle there is a fixed point A about which a ray revolves. When this ray passes through the center of the circle, we call the other point at which it intersects the circle the point B associated with the position of the ray at any time; this point is such that continuous variations in its position must always correspond to continuous variations in the position of the ray. Hence the name B denotes something indeterminate so long as the corresponding position of the ray has not been specified. We can now ask: what point is associated with the position of the ray when it is perpendicular to the diameter? The answer will be: the point A. In this case, therefore, the name B has the same content as has the name A; and yet we could not have used only one name from the beginning, since the justification for that is given only by the answer. One point is determined in two ways: (1) immediately through intuition and (2) as a point B associated with the ray perpendicular to the diameter.

The "illusion" Frege aims to dispel is that identity statements record nothing other than linguistic conventions. However, that point and his argument for it are clouded by three difficulties with the symbolism he introduces in [C]. Resolving them is necessary to bring out the general argument embedded in the example.

(i) Frege uses 'B' ambiguously. It is introduced first as a variable. As Frege puts it, "the name B denotes something undetermined as long as the corresponding position of the straight line is not yet specified." However, in the rest of the passage he speaks of it as a descriptive sign (as a replacement instance for the variable). To capture the distinction Frege has in mind, in what follows I shall use 'BX' for the variable and 'B1', 'B2'... for the descriptive signs that replace it (for convenience I shall refer to them as "B-signs").

(ii) The B-signs are descriptive signs as 'A' is. However, unlike 'A', they are abbreviations-abbreviations for definite descriptions of the form 'the point associated with the position of the ray that is at a angle to the diameter'. Though Frege does not explicitly state that the B-signs are abbreviations, his practice makes evident that they are. This is most easily seen if one attempts to spell out what is expressed by an 'A=BX' statement. The only way to do so is by substituting a definite description for the B-sign.

(iii) The distinction drawn in (2) points to an additional difficulty with Frege's symbolism and the terminology he uses to describe it. Frege refers to 'A' and the B-signs as "names." In one sense that is unproblematic. As noted, both are descriptive signs. In addition, both are syntactically simple and both obey the same formation rules. However, (2) shows that identity statements of the form 'A=BX' are always elliptical for those of the form 'A=the point corresponding to the ray that is at a angle to the diameter'. Thus, in another sense, Frege's use of the B-signs is misleading in that it pushes into the background an essential feature of the identity statements he uses in his argument. Those statements are always of the form 'A=BX', and hence the descriptive signs occurring in them always consist of a name ('A') and a definite description.

The preceding clarifications suggest that the distinction between names and definite descriptions is at the heart of Frege's argument. The significance of that distinction, I shall argue, is that the connection between names and what they designate is conventional; the connection between definite descriptions and what they designate is not.¹⁹ Russell (apart from his unfortunate use of the word 'arbitrary') makes the point forcefully and effectively:

You sometimes find people speaking as if descriptive phrases were names, and you will find it suggested, e.g., that such a proposition as 'Scott is the author of Waverley' really asserts that 'Scott' and 'the author of Waverley' are two names for the same person. That is an entire delusion... if that were what were meant, the proposition would be like 'Scott is Sir Walter', and would not depend upon any fact except that the person in question was so called, because a name is what a man is called... the fact that he [Scott] was the author [of Waverley] was a physical fact, the fact that he sat down and wrote it with his own hand, which does not have anything to do with what he was called. It is in no way arbitrary. You cannot settle by any choice of nomenclature whether he is or is not to be the author of Waverley, because in actual fact he chose to write it and you cannot help yourself. That illustrates how 'the author of Waverley' is quite a different thing from a name.²⁰

With the distinction between names and definite descriptions now clearly in view, let me return to Frege's example. Frege there considers a series of identity statements:

A=B₁

A=B₂

A=B₃

A=...

An infinite number of the identity statements in the series are false. One, however, is true-the one in which the B-sign ('B ₆', say) abbreviates the definite description 'the point corresponding to the ray that is at a 90° angle to the diameter'. Frege explains the significance of this at the end of [C]:

What point corresponds to the position of the straight line when it is perpendicular to the diameter? The answer will be: The point A. Thus, in this case, the name B has the same content as the name A; and yet we could not have used only one name from the beginning since the justification for doing so is first given by our answer.

The contrast with justification is stipulation. This is shown by unpacking what it would mean to "have used only one name from the beginning." Here being able to "use only one name" means knowing that one can substitute 'B' for 'A' while preserving truth-value. "From the beginning" means "independently of the geometric construction described in [C]." The only way in which one could use a name - say, 'P' - "from the beginning" would be to stipulate that it designates A. Since 'A' is also interpreted by means of a stipulation, there is obviously no point in carrying out the procedure, but it does serve to bring out the contrast between 'A' and 'B6'. 'B6' designates the point that it does because of the definite description it abbreviates. That definite description, in turn, designates what it does, not as a matter of stipulation, but as a matter of geometrical fact. What is true of the definite description is then true of the sign that abbreviates it: 'B6' also designates the point that it does not as a matter of stipulation but of geometrical fact. For that reason, justification (in this case the justification provided by the construction Frege describes) is required if one is to use "only one name" for the point; that is, if one is to substitute 'B6' for 'A'. Frege draws just that conclusion at the end of §8 when he says that on the assumption that 'A=B' is true "we can everywhere put 'B' for 'A' and conversely."²¹

The example in [C] contains the substance of Frege's argument against the name view. It remains for him to generalize the results of that argument and draw out its implications for the vocabulary of the Begriffsschrift. And, in fact, that is what he does in the next passage, [D]. Unfortunately, [D] is written in such a way as to suggest that Frege does something more. As a first step in clarifying matters, I shall restate the lesson of [C] in a concise way that makes clear its connection to the name view. Having done so, I shall present [D] as I believe Frege should have written it.

[C*] The identity statement 'A=B6' expresses proper knowledge because 'B6' abbreviates a definite description and because the connection between that definite description and the point it designates is not a matter of stipulation but of geometrical fact.

[D*] [C] shows that identity statements express proper knowledge if and only if at least one of the signs flanking the identity sign is (or abbreviates) a definite description. The reason for this is that the connection between a definite description and the object it designates is non-conventional. Thus, a language requires the identity sign if and only if its vocabulary contains co-referring signs and at least one of those signs is a definite description.²²

There are two points about [C*] and [D*] I wish to accentuate. The first is that they provide a compelling argument for including the identity sign in the vocabulary of the Begriffsschrift.²³ The second is that the resources of that argument are quite modest. Making them explicit will show just how modest. The gist of [D*] is that the connection between definite descriptions and objects is non-conventional. Since that non-conventional connection is the defining feature of definite descriptions, [D*] does nothing more than point out that the vocabulary of the Begriffsschrift contains definite descriptions and spell out the implications of that as regards the identity sign.²⁴

The interpretive question that now arises is whether in [D] Frege says anything more than what I have attributed to him in [D*]. He writes,

[D] One point is determined in two ways: (1) Immediately through intuition and (2) As a point B associated with the ray perpendicular to the diameter.

To each of these modes of determining (Bestimmungweisen) the point there corresponds a particular name.²⁵ Hence the need for a sign for identity of content rests upon the following consideration: the same content can be completely determined in different ways; but that in a particular case two modes of determining it really yield the same result is the content of a judgment. Before this judgment can be made, two distinct names corresponding to the two modes of determining the content, must be assigned to what these modes determine. The judgment, however, requires for its expression a sign for identity of content, a sign that connects these two names. From this it follows that the existence of different names for the same content is not always merely an irrelevant matter of form; rather, that there are such names is the very heart of the matter if each is associated with a different mode of determination.

The expression 'mode of determination' is the source of the belief that [D] is an important, new step in the argument of §8.²⁶ It is also the basis for the belief that in §8 Frege makes the Sinn-Bedeutung distinction. Critiquing the first belief will thus constitute a partial evaluation of the second.

Let me focus the issue by returning to the conclusion of [D]:

the existence of different names for the same content is not always merely an irrelevant matter of form; rather, that there are such names is the very heart of the matter if each is associated with a different mode of determination.

According to the interpretation ventured in [D*] that passage says the following:

The existence of co-referring signs is not always merely an irrelevant matter of form; rather, that there are such signs is the very heart of the matter if at least one is a definite description.

That gloss is deflationary since it removes the expression 'mode of determination' and with it the suggestion that the argument in [D] depends on something more than a rather humble fact about language. I believe the deflationary reading is to be preferred in light of the difficulties that arise in trying to say what that something more might be.²⁷ At the beginning of [D] Frege identifies the two ways in which the point in [C] is determined. He then states, "To each of these modes of determining the point there corresponds a particular name." One of those names is 'B6'. Frege appears to be saying that in addition to (i) 'B6', there is (ii) the point it designates (in this case A) and (iii) a mode of determination. The mode of determination provides the connection between 'B6' and A. But, in the example, 'B6' is linked to A by means of the definite description 'the point corresponding to the ray that is at a 90° angle to the diameter'. In other words, the mode of determination is identical to a definite description. This is not to say, of course, that a mode of determination is always a definite description. In the case of 'A', Frege clearly holds that it is not. Yet in that case as well, the attempt to specify the mode of determination yields not the exotic but the familiar-a simple case of a sign serving as a mere label for its object.

One might object to the preceding argument on the grounds that that it depends on the status of 'B6' as an abbreviation. Abbreviations, the objection continues, are eliminable from the language within which they occur. In a language without abbreviations it is no longer so easy to identify modes of determination with definite descriptions. The objection amounts to requesting that the discussion be limited to a language consisting of 'A' and definite descriptions. There is no harm in honoring that request by simply letting 'B6' be a definite description rather than abbreviating one. (That it is orthographically unlike a natural language definite description is obviously unimportant.) The objection holds that in addition to (i) 'B6' there is (ii) the point it designates and (iii) a mode of determination. What the objection comes to depends on how own understands (i) and (iii). If 'B6' is detached from its mode of determination then it is not a definite description at all but a mere ink-mark. The role of the mode of determination is to connect that ink-mark with A and in so doing turn it into a definite description. There is no way to achieve this, while preserving Frege's argument, other than to associate 'B6' with the definite description 'the point corresponding to the ray that is at a 90° angle to the diameter'. Once again, the mode of determination has collapsed into the definite description. And, indeed, in recent discussions of the Begriffsschrift this collapse is evident, albeit not acknowledged as such:

In the geometrical example from the Begriffsschrift, the mode of determination associated with 'A' determines the point "immediately through intuition," whereas the mode of determination associated with 'B' is "the point when the line is perpendicular to the diameter."²⁸

Frege sets out an example in which two names for a single geometrical point correspond to two ways of determining or describing it. (emphasis mine)²⁹

We can grasp the distinction between the man Socrates and the way of determining it as the content of an expression, which is included in the name 'the great teacher of Plato'.³⁰

To sum up this interpretation of §8: The purpose of that section is to show that the vocabulary of the Begriffsschrift requires the identity sign. Frege begins by considering an argument that it does not. That argument holds that identity statements do not express "proper knowledge". He responds by pointing out that they do, provided at least one of the descriptive signs occurring in them is a definite description. Definite descriptions are essential because they, unlike names, designate their objects non-conventionally.

III.

One who comes to USB from the Begriffsschrift encounters a peculiar state of affairs. Frege attributes to his earlier work a view it rejects and then attacks that view on the very same grounds as before. That attack is then offered up as the philosophical motivation for the distinction between Sinn and Bedeutung. Since Frege's self-critique is identical to the argument he uses in support of his view in the Begriffsschrift, it follows that the distinction is found there as well.

In this section and the next I shall try to show that the preceding account of the relationship between the Begriffsschrift and USB is mistaken. On my reading, Frege does misrepresent that relationship. However, his error lies not in failing to recognize that in the earlier work he had already effected the Sinn-Bedeutung distinction but in suggesting that the two works are responsive to the same problem. The Begriffsschrift, I have argued, addresses an epistemological problem. USB, I shall try to show, addresses an ontological problem. The distinction between Sinn and Bedeutung is responsive only to that problem.

The first burden of my interpretation is to show that the view Frege puts forward in USB differs from that found in the Begriffsschrift. To do so I shall consider what might seem to be a rather different issue: is the distinction he makes between Sinn and Bedeutung at the beginning of USB the same as that between a "proper name's" meaning and its referent?³¹ It is widely held that the answer to that question is "yes." Michael Dummett is the most well-known proponent of that position:

Frege... argues that the sense of a proper name cannot merely consist in its having the reference that it has. His argument is set out in terms of the notion of 'cognitive value', that is information content. Frege asks how, if the sense of a proper name consisted just in its having the reference that it does have, any true statement of identity could be informative. The notion of 'information' being appealed to here does not require any elaborate explication: I acquire information when I learn something which I did not previously know, and Frege is asking how it is possible that I may be in a position to know the sense of an identity-statement, i.e. to understand it, and yet learn something that I did not know before by being told that the statement is true. On the theory that the sense of a proper name consists just in its having the reference that it has, this cannot be explained: for then my understanding of the two names connected by the sign of identity would consist just in my associating with each the object that was its referent... I should thus be unable to understand an identity statement without recognizing it as true or as false...

...In invoking the notion of information to support his contention that the sense of a name cannot consist merely in its having the reference which it does have, Frege is tacitly connecting the notion of sense with that of knowledge; and this is the justification for our representing Frege's views by saying that sense is an ingredient in meaning, where meaning is that which a man knows when he understands a word.³²

It is plain that the "information" Dummett has in mind is not meta-linguistic. Thus, the question he imputes to Frege (how can true identity statements be informative?) turns out to be the same as that raised in the Begriffsschrift (how can identity statements express proper knowledge?). According to Dummett, Frege answers that question by distinguishing between a proper name's meaning and its referent. As I now propose to show, that answer is, in essentials, the same as the one Frege defends in §8 of the Begriffsschrift.

Let me begin by focusing on a verbal connection between the Begriffsschrift and USB. In the Begriffsschrift Frege says that signs are associated with "modes of determination" (Bestimmungweise). In USB he says that they are associated with "modes of presentation" (Art des Gegebenseins). So far as the mere terminology goes, the difference between the expressions makes no difference. They can therefore be treated as synonyms. Concerning modes of presentation, Dummett writes:

The sense of an expression is the mode of presentation of the referent: in saying what the referent is, we have to choose a particular way of saying this, a particular means of determining something as the referent.³³

Dummett identifies a Sinn with a mode of presentation, which in turn he identifies with the criterion for the application of a proper name. The identification of Sinn/mode of presentation with a criterion is made more explicit in the following passage:

Frege's theory is that a proper name, if it is to be considered as having a determinate sense, must have associated with it a specific criterion for recognizing a given object as the referent of the name; the referent of the name, if any, is whatever object satisfies that criterion.³⁴

With this chain of identities (meaning/Sinn/mode of presentation/criterion) in place, we are in a position to discern the overlap between Dummett's reading of the opening of USB and §8 of the Begriffsschrift.

For Dummett, the meaning of a proper name is what serves to connect it to the object it designates.³⁵ He also holds that an identity-statement expresses proper knowledge if and only if its descriptive signs refer to the same object but differ in meaning. Replacing 'meaning' with 'connection' yields the following: An identity-statement expresses proper knowledge if and only if its descriptive signs refer to the same object but have different connections to that object. This last formulation is close to the view Frege puts forward in §8. The similarity is most obvious where the identity statement consists of a name and a definite description. As I pointed out in Section II, the connection between name and object is conventional; that between definite description and object, non-conventional. To put the point using Dummett's terminology, the name and the definite description have the same referent but differ in meaning. The similarity is less obvious, but no less present, where the identity statement consists of two definite descriptions. In such cases the descriptive signs have different non-conventional connections to the object they designate. For example, the connection between 'the author of Cymbeline' and Shakespeare is provided by his authoring of Cymbeline. The connection between 'the author of Hamlet' and Shakespeare is provided by his authoring of Hamlet.³⁶ Last, identity statements consisting of two names cannot express proper knowledge because the names, as labels, have the same connection to their referent. For example, 'Mary Ann Evans' and 'George Eliot' are connected to the same object and in the same way. The signs differ then, not in their meaning, but merely in terms of their shape.³⁷ The view Dummett finds in the opening of USB turns out to be the view Frege embraces in the Begriffsschrift, and his interpretation - despite his own position on the matter - thus vindicates those who think the Sinn-Bedeutung distinction predates its explicit formulation in USB.

There is, however, a serious textual obstacle to identifying Sinn with meaning. Above I pointed to a difference between the Begriffsschrift and USB - the presence in the former of the expression 'mode of determination' and in the latter of the expression 'mode of presentation' - and noted that it was merely terminological. There is, however, an additional verbal difference in Frege's use of those phrases that is of more moment. In the Begriffsschrift Frege simply speaks of signs being associated with different "modes of determination." In USB, however, he states,

It is natural, now, to think of there being connected with a sign (name, combination of words, written mark), besides that which the sign designates, which may be called the Bedeutung of the sign, also what I should like to call the Sinn of the sign, wherein the mode of presentation is contained. (¶2)

I have italicized the crucial word. Frege's use of it is metaphorical. The following passage from later in USB provides a clue as to what the metaphor means.

In indirect speech one talks about the sense, e.g., of another persons's remarks. It is quite clear that in this way of speaking words do not have their customary Bedeutung but designate what is usually their Sinn. (¶7)

The customary Bedeutung of the signs in question are objects; for example, Socrates, the moon, the pen on my desk and so on. In holding that Sinne are Bedeutungen, Frege is gesturing at a similarity between them and objects. The only one that is in evidence at this level of abstraction is that both are full-fledged existents.

Frege's hypostatization of modes of presentation is the most striking difference between the Begriffsschrift and USB.³⁸ It is also the feature of his thought that is most difficult to accommodate on a epistemological reading like Dummett's; for it is not necessary to treat meanings as entities in order to explain how identity-statements can be informative. Nor does Dummett claim that it is. For him, Frege's reification of meanings is a feature of his thought that can and should be expunged from it.³⁹ This he proceeds to do by interpreting Sinne as the "means," "criteria" or "ways" in which we refer to objects. "Means," "criteria" and "ways" are certainly not entities, but they provide all that is necessary for understanding Frege's doctrine of Sinn.⁴⁰ Though I shall argue that Dummett's approach to Frege is inadequate, his reading is helpful in clarifying the primary challenge facing any interpretation that takes the status of Sinne as entities seriously: supplying the opening of USB with a problem to which such entities are a relevant solution. It is to that task that I now turn.

IV.

In this section I argue that the distinction between Sinn and Bedeutung is responsive to an ontological problem. Ontology, as I understand it here, begins with a class of true statements. The task of the ontologist is to offer an analysis of the truth-makers for those statements. Ontological analysis is the search for the constituent entities of those truth-makers. ⁴¹

The foregoing account surely requires elaboration, and I shall attempt to provide what is needed in the context of an engagement with the specific problem Frege confronts in the opening paragraph of USB. For the moment let me stress what should be obvious: if Frege is pursuing the project of ontological analysis, then the very nature of his enterprise dictates the positing of entities. An ontological reading thus handles with ease the feature of the text that Dummett and others are forced to dismiss. Yet such a reading faces a formidable textual difficulty of its own: the pervasiveness of epistemological vocabulary ("cognitive value," "a priori," "discovery," "reidentification," "proper knowledge") in the opening paragraph of USB. That difficulty can be overcome, I shall argue, by heeding Bergmann's remark that "Epistemology is but the ontology of the knowing situation."⁴² In this section I shall attempt to show that in USB Frege offers an ontology of the knowing situation. In doing so he defends three theses:

(α) Sinne exist.

(β) They represent objects

(γ) They are not mental entities.

My interpretation is structured around an examination of those theses.

(α)

Frege argues for (α) in the first paragraph of USB. As I have already pointed out, the beginning of that paragraph is extremely misleading as a statement of the problem that is moving him. Contrary to what Frege suggests, his concern is not with the nature of identity but with the knowledge of identity-statements. Let us then ignore his description of his problem and linger over what he says in support of the view he claims (falsely) to have held in the Begriffsschrift-that identity is a relation between names:

a=a and a=b are obviously statements of different cognitive value; a=a holds a priori and, according to Kant, is to be labeled analytic, while statements of the form a=b often contain very valuable extensions of our knowledge and cannot always be established a priori.

Frege holds that the distinction between Sinn and Bedeutung is required to account for the fact that statements of the form 'a=a' and 'a=b' differ in their "cognitive value." That expression is thus badly in need of explication.⁴³ Frege appears to suggest that differences in cognitive value derive from the different types of justification appropriate to the different types of statement. Statements of the form 'a=a' are established a priori; those of the form 'a=b' are established a posteriori. Because they are justified in those different ways, they differ in cognitive value. That, however, cannot be Frege's view since arithmetical truths - for example, '7+5=12' - are of the form 'a=b' and are established a priori. Thus, if "cognitive value" were glossed as "mode of justification," then '7+5=12' and 'Socrates is Socrates' would have the same "cognitive value," a view which is at odds with the whole thrust of the passage.

In light of these difficulties, it is tempting to place the emphasis on "valuable extensions of our knowledge." Statements of the form 'a=b', however they are justified, express extensions of our knowledge, while statements of the form 'a=a' do not. That difference is the basis for ascribing a difference in cognitive value to statements of the two forms. Yet that reading also fails to make sense of 'cogntive value' since the contrast between statements that extend our knowledge and those that fail to dos so is merely another way of stating that some identity statements express proper knowledge and others do not-the very issue Frege raises and resolves in §8 of the Begriffsschrift.

A step toward clarifying Frege's notion of cognitive value is made possible by attending to a subtle difference between his presentation of his problem in the Begriffsschrift and his presentation of it in USB. In the former he considers only identity statements of the form 'a=b'. In the latter he thinks it important to contrast such statements with those of the form 'a=a'. Why does Frege find such statements helpful when before he did not?⁴⁴ An answer is suggested by the passage that immediately follows:

The discovery that the rising sun is not new every morning, but always the same was one of the most fertile astronomical discoveries. Even today the reidentification of a small planet or a comet is not always a matter of course.

What is of interest here is the rhetoric of discovery that is so prominently on display. The difficulty is to determine how that rhetoric figures in the argument for (α). One possibility is that Frege intends to call attention to what is discovered. That will not do since what the astronomer discovers is that the morning star is the evening star, not that Sinne exist. Nor will it suffice to hold that (α) is a further discovery, one made on the basis of the discovery that the morning star is the evening star. That line of argument is not wrong but empty. It says merely that there is some connection between the example Frege considers and the thesis which he defends; it says nothing to specify what that connection is. Another possibility is that Frege's focus is on the process of discovery. That construal must also be rejected on the grounds that (α) is irrelevant to the means whereby an astronomical discovery is made and to the cognitive capacities of the discoverers. With those possibilities eliminated, all that remains of the example is the bare contrast between the state of the astronomer's knowledge before and after the discovery. Before the discovery she does not know that the morning star is the evening star; after it she does.

The contrast between knowing and not knowing explains why Frege introduces statements of the form 'a=a' into the argument of the opening paragraph. Consider, for example, the following: 'Shakespeare is Shakespeare' and 'Shakespeare is the author of Hamlet'. As I have pointed out, the important difference between them is not that the former is known a priori while the latter is known a posteriori; nor is it that one extends our knowledge and the other does not. The relevant difference is that it is impossible not to know the first but possible not to know the second. Frege begins with the contrast between 'a=a' and 'a=b' because it allows him to bring out the contrast between knowing and not knowing in a stark fashion.⁴⁵ When he states that 'a=a' and 'a=b' differ in cognitive value, he is not characterizing those sentences but rather the state of mind of the knower: she knows the first but does not know the second. There are thus two cognitive values for Frege: knowing and not-knowing.

This understanding of 'cognitive value' fits with its other explicit occurrence in the opening paragraph:

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 manner in which it designates something), the cognitive value of a=a becomes essentially equal to that of a=b, provided a=b is true. A difference [in cognitive value] can arise only if the difference between the signs corresponds to a difference in the mode of presentation of the thing designated.

The assumption that 'a' and 'b' differ only in shape means that they are names rather than definite descriptions. As such, they are labels for the objects they represent. One who knows the meaning of those signs therefore knows their referent. It follows that one who understands the language of which 'a=a' and 'a=b' are a part can no more fail to know the one than the other. The sentences thus have the same cognitive value.⁴⁶ Frege concludes that one can understand an identity-statement and yet not know whether it is true only if one of its descriptive signs is a definite description.

Significantly, the distinction between knowing and not-knowing is present in two other passages in which Frege argues for the existence of Sinne. One is found in USB. In it Frege turns from a consideration of definite descriptions to "entire assertoric sentences." His argument is that such a sentence

contains a thought. Is this thought now to be regarded as its Sinn or its Bedeutung? If we now replace one word of the sentence by another having the same Bedeutung, but a different Sinn, this can have no effect upon the Bedeutung of the sentence. Yet we can see that in such a case the thought changes; since, e.g. the thought in the sentence 'The morning star is a body illuminated by the Sun' differs from that in the sentence 'The evening star is a body illuminated by the sun'. Anybody who did not know that the morning star is the evening star might hold the one thought to be true, the other false. The thought, accordingly, cannot be the Bedeutung of the sentence, but must be considered as its Sinn. (¶15)⁴⁷

This is close to a word for word restatement of Frege's first published defense of the Sinn-Bedeutung distinction in "Function and Concept" [FC]:

If we say 'The Evening Star is a planet with a shorter period of revolution than the Earth', the thought we express is other than in the sentence 'The Morning Star is a planet with a shorter period of revolution than the earth'; for somebody who does not know that the morning star is the evening star might regard one as true and the other as false. And yet the Bedeutung of both sentences must be the same; for it is just a matter of interchange of the words 'Evening Star' and 'Morning Star', which have the same Bedeutung, i.e are proper names of the same heavenly body. We must distinguish between Sinn and Bedeutung.⁴⁸

The question these two passages and the opening paragraph raise is how the distinction between knowing and not-knowing supports (α).

My question presupposes that the argument of the three texts is the same. Since in USB Frege suggests that the one concerning "entire assertoric sentences" differs from the one presented in the opening paragraph, that presupposition stands in need of justification. Four differences between the opening paragraph and the companion passages from USB and FC are evident. (i) As Frege points out, the opening paragraph makes the Sinn-Bedeutung distinction for definite descriptions; the later passage (and thus its companion) makes the distinction for sentences. (ii) The argument of the companion passages makes explicit use of the substitution of co-referring definite descriptions; the opening paragraph does not. (iii) Those substitutions are made in sentences that occur in an epistemic context (that is, a context created by expressions of the form '... knows that ...'). The argument of the opening passage seems to consider sentences taken "on their own." (iv) Identity statements are the object of reflection in the opening paragraph whereas statements of subject-predicate form are considered in the companion passages.

Let me take the third difference first. The epistemic context is missing from the opening paragraph only in the most literal sense; in fact, there is no way to make sense of its argument - however one construes it - without implicitly introducing precisely such a context. This is best seen by considering the directly opposed view of Dummett. He asserts that in the absence of an argument for the Sinn-Bedeutung distinction in non-intentional contexts, an argument for it in intentional contexts is futile:

Why then do we need a notion of sense as well as a notion of reference?

A very bad answer, sometimes given to this question, would be that the notion of sense is needed by Frege to explain operators which... create opaque contexts - expressions like 'necessarily' and '...believes that...'. It is true enough that Frege does deploy his notion of sense in treating such expressions: he says that in opaque contexts, a term stands for what, in ordinary contexts, constitutes its sense. But obviously, it would be useless to offer any such explanation unless it had been established that there is something which, in ordinary contexts, constitutes the sense of a term.⁴⁹

It is therefore significant that when Dummett gives his own reading of Frege's argument for the distinction, he is forced to reintroduce the epistemic context that he purports to have banished from consideration. As he says in the passage quoted at the beginning of Part III, if there were no distinction between Sinn and Bedeutung then

my understanding of the two names connected by the sign of identity would consist just in my associating with each the object that was its referent... I should thus be unable to understand an identity-statement without immediately recognizing it as true or as false.⁵⁰

In other words, it would be impossible to understand 'a=b' and fail to know whether it is true or false.⁵¹

Disposing of the third difference disposes of the second as well. In the opening paragraph Frege considers a pair of sentences like the following:

Cordelia knows that a=a.

Cordelia knows that a=b

Frege's point - though one that he does not express clearly - is that in substituting 'b' for 'a'

one passes from a true sentence to a false one. That, of course, is exactly the phenomenon to which he calls attention in the companion passages.

The remaining differences can now easily be dispatched. The fourth is simply irrelevant. What is crucial in all three passages is the transition from a true sentence to a false one by means of the substitution of one co-referring expression for another. The difference in the form of the embedded sentence in which the substitution is made is an inessential feature of the argument.

The first difference is illusory. Frege's suggestion that there are two arguments - or one argument applied to different types of expression (definite descriptions and sentences) - is without support. In all three passages he arrives at a distinction between Sinn and Bedeutung on the basis of the substitution of co-referring definite descriptions in sentences that are embedded in epistemic contexts.

This way of reading Frege's argument raises a problem with respect to the text of USB. For it now turns out that he presents the argument for (α) twice. What purpose, if any, is served by the repetition; and, assuming it is repetition, why does he fail to recognize it as such? Taking a bird's-eye view of the structure of USB provides an answer. Frege's argument for (α), (β) and (γ) runs for thirteen paragraphs. There then follows an interlude - to my mind an irrelevant one - in which he takes up a skeptical objection to his view ("Idealists or sceptics will perhaps long since have objected..."). Having dismissed that objection, Frege repeats his argument for (α). However, what follows that argument makes clear that his primary concern is not with the existence of Sinne but with a different ontological claim: the existence of the True and the False. In context, then, the repetition is not designed to advance the overall argument of USB. Rather, it serves only to build a bridge back to the line of thought that had been broken off by the skeptical interlude. Indeed, nothing would be lost if the repetition were to be excised and replaced with something like the following: "Having shown that sentences express a Sinn we must now consider whether they have a Bedeutung and, if so, what it is like."

To this point my examination of the three passages has been neutral between rival interpretations of Frege's argument for the Sinn-Bedeutung distinction. In particular, it will fit with either an epistemological or ontological construal of that argument. On both readings, the problem Frege raises in the opening paragraph of USB is posed by reflection on pairs of sentences such as the following:

Cordelia knows that Shakespeare is the author of Hamlet.

Cordelia knows that Shakespeare is the author of Cymbeline.

The first sentence is true; the second, false. Frege's question is how that is possible. There are, however, two ways of unpacking the content of that possibility. Epistemological readings place the emphasis on knowing. They ask how it is possible for Cordelia to know that Shakespeare is the author of Hamlet and yet fail to know that he is the author of Cymbeline. As I have pointed out, distinguishing between the meaning and referent of 'the author of Hamlet' and 'the author of Cymbeline' is adequate to answer that question. Epistemological readings thus fail to supply the opening paragraph with an argument for (α).

An ontological reading places the emphasis on truth and falsity. It asks how it is possible for the first sentence to be true and the second to be false. To facilitate the development of such a reading it will be helpful to consider an example that differs terminologically and grammatically from Frege's but does not alter his argument in any fundamental respect. First, in the opening paragraph Frege considers sentences containing 'know'. In the context of that paragraph knowledge is identical to true belief. Thus, sentences containing 'believes' will do just as well as those Frege uses. Second, though Frege's argument makes use of substitutions in embedded sentences, there is no reason it has to do so. His argument can, in fact, be set out much more clearly if we place a definite description rather than a full sentence in the intentional context. In the interest of readability we may thus replace 'believes' with 'thinks'. I alter the previous sentences accordingly:

(1) Cordelia is thinking about the author of Hamlet.

(2) Cordelia is thinking about the author of Cymbeline

Given that the sentences differ in truth-value, their truth-makers must differ. Frege's question is how that difference is possible.

The force of the question and Frege's justification for his answer to it are best explicated by introducing some simple devices for representing the results of ontological analysis. (Henceforth I shall refer to this as the "ontological notation.") Frege, we know, takes "believings," "assertings," etc. to be relations. Let 'T' stand for the relation of "thinking." One term of that relation is a mind. Let 'm' signs stand for the category, with subscripts indicating the specific members. At first it will be tempting to think of the other term as an object: a table, tree or person as the case may be. Let objects be represented by 'o' signs, with subscripts functioning as before. Given those ontological resources, the truth-maker for (1) is the following:

(1^ON) T(m₇, o₃)

However, since the author of Hamlet is identical to the author of Cymbeline, it follows that the truth-maker for (2) is the following:

(2^ON) T(m₇, o₃)

The proposed analysis fails because it does not differentiate the truth-maker for (1) from the truth-maker for (2). Put more pointedly, it does not explain how (2) can be false while (1) is true. From this dilemma Frege concludes that in addition to objects there must be "something else." Those "something-else's" he labels 'Sinne':

It is natural, now, to think of there being connected with a sign... besides that which the sign designates, which may be called the Bedeutung of the sign, also what I should like to call the Sinn of the sign. (¶2)

The ontological notation must now be enriched to accommodate this new ontological category. To do so, let s-signs stand for Sinne. The truth-makers for (1) and (2) are:

(1^ON*) T(m₇, s₆)

(2^ON*) T(m₇, s₈)

Sinne thus allow Frege to differentiate the truth-makers for (1) and (2). Despite the complexity of the interpretation that is needed to unveil it, Frege's argument for (α) is as simple as that.⁵²

Before turning to his argument for (β), there are three issues that need to be addressed.

(i) Though the Sinn-Bedeutung distinction is not the same as the distinction between an expression's meaning and its referent, there is a connection between the two. Frege's argument for (α) depends on the substitution of co-referring proper names in intentional contexts. Those substitutions raise ontological problems only when the substitution changes the truth-value of the sentence in which the substitution is made. And that will be the case only when the proper names differ in meaning; that is, when at least one is a definite description. These considerations provide the real connection between the Begriffsschrift and USB: it is only because identity-statements express proper knowledge that the problem Frege solves by means of Sinne arises.

(ii) My talk of truth-makers and their constituents is likely to prompt an objection. Truth-makers so understood are plainly facts, which Frege, in a well-known passage, denies exist:

"What is a fact? A Fact is a thought that is true."⁵³

Thus, whatever Frege may be doing in USB, it cannot be ontological analysis as I have described it.

A full response to the objection is outside the scope of this paper. For present purposes the following remarks will serve. Frege states that a fact is a thought that is true. Put slightly differently, a fact is a thought that "names" the True. Yet clearly Frege does not think that a thought's naming the True is unanalyzable.For example, the thought expressed by the statement 'Socrates is wise' names the True because an object (Socrates) falls under a concept ("wisdom"). Whether that means that Frege is tacitly committed to facts and, if he is, what one is to make of his denial that they exist are matters that can be safely left to a future occasion. So far as matters here are concerned, all one need accept is that Frege must offer a correlative analysis for intentional statements. That analysis, I have claimed, provides the argument for Sinne.

(iii) 'Sinn' is a label for an ontological category. As such, it is a technical term and has only the meaning that the theorist gives to it. There is no doubt that Frege's choice is dictated by the apparent affinities between the ordinary meaning of 'Sinn' and the meaning it has in USB. Nevertheless, there is nothing in his argument that justifies using that ordinary notion to unpack his technical notion.⁵⁴ The only conclusion to which Frege is entitled is that something other than objects exists. What Frege chooses to call those entities is purely a matter of nomenclature. This point is particularly important to keep in mind if (β) is to be interpreted properly.

β

The argument of the opening paragraph of USB is followed by the conclusion drawn in the second:

It is natural, now, to think of there being connected with a sign (name, combination of words, written mark), besides that which the sign designates, which may be called the Bedeutung of the sign, also what I should like to call the Sinn of the sign, wherein the mode of presentation is contained.

The conclusion asserts not only (α), but (β): Sinne represent objects. That, I take it, is what is expressed in the claim that Sinne "contain" modes of presentation. But Frege does not merely state that there is a relation between entities; he also holds that those entities are related to signs. He expands on the relations between sign, Sinn and object three paragraphs later:

The regular connection between a sign, its Sinn and its Bedeutung is of such a kind that to the sign there corresponds a definite Sinn and to that in turn a definite Bedeutung. (¶5)

And at the conclusion of the first part of USB Frege summarizes what he takes his argument to have established:

A proper name (word, sign, sign combination, expression) expresses its Sinn, means or designates its Bedeutung. By employing a sign we express its Sinn and designate its Bedeutung. (¶13)

That passage makes explicit what is implicit in the first two: the connection between Sinn and object mediates the connection between sign and object; or, to use the more familiar terminology, "sense determines reference."⁵⁵ In this section I shall argue that those two claims - that Sinne represent objects and that "sense determines reference" - must be detached from one another. The first is an integral part of Frege's theory and follows directly from the considerations of the opening paragraph. Matters are quite different with respect to the second. As I shall try to show, it receives no support from the argument of USB and has only the most tenuous connection to its text.

The difficulties with Frege's second claim emerge in considering ¶6 and ¶7, texts that have not been treated with the care they warrant. Let me begin with a passage discussed earlier.

In indirect speech one talks about the Sinn, e.g., of another person's remarks. It is quite clear that in the way of speaking words do not have their customary Bedeutung but designate what is usually their Sinn. (¶7)

The passage prompts a philosophical objection and raises an interpretive difficulty. I take up the objection first.

Frege's point is not limited to indirect speech but applies to intentional contexts generally. Consider, for example, the following sentences:

(1) Cordelia is thinking about the author of Hamlet

(3) The author of Hamlet was born in 1564

According to Frege, in (3) 'the author of Hamlet' stands for an object; in (1), for a Sinn. Thus, he holds that definite descriptions are essentially ambiguous: outside of intentional contexts they stand for objects; inside of them they stand for Sinne.

Frege's claim is puzzling. If it is taken to apply to the meaning of definite descriptions as they are used in everyday discourse, then it is plainly false. As reflection on ordinary speech situations shows, the meaning of a definite description does not change depending on whether it occurs inside or outside of an intentional context. For example, no sensible translator - whose aim, after all, is to preserve meaning - would be tempted to translate 'the author of Hamlet' in one way when it occurs in (3) and then cast around for different words to render its occurrence in (1).⁵⁶

Yet it is doubtful that Frege's claim should be taken as calling attention to ambiguity of the ordinary sort. That is shown by turning to the interpretive difficulty the passage raises. At first glance Frege seems to be continuing the project of ontological analysis begun in the opening paragraph, extending it to include "indirect speech." Yet that "extension" is no extension at all. Indirect speech is merely another example of an intentional context, and as such it introduces no problems not already raised in the opening paragraph. Thus, Frege's interest in "speaking about" Sinne must serve some other purpose. That purpose is brought out by placing the passage in its immediate context:

In order to speak of the Sinn of an expression 'A' one may simply use the phrase 'the Sinn of the expression "A" '. In indirect speech one talks about the Sinn, e.g., of another person's remarks. It is quite clear that in the way of speaking words do not have their customary Bedeutung but designate what is usually their Sinn. (¶7)

The proposal of the first sentence is a development of the discussion initiated in ¶6. That paragraph begins,

If words are used in the ordinary way, what one intends to speak of is their Bedeutung. It can also happen, however, that one wishes to talk about the words themselves or their Sinne.⁵⁷

The rest of ¶6 is given over to the introduction of a device (quotation marks) for mentioning expressions. That device is then employed in the first sentence of ¶7 in order to fashion signs to stand for Sinne. However, the next sentence masks the reason why such signs are needed. It suggests that the notational innovation facilitates reporting the speech of another. The suggestion is obvious nonsense-ordinary language needs no supplementing to carry out that task. For example, someone in attendance at Antony's funeral oration who wishes to relay to his neighbor what was said will simply state, "Antony said that Brutus is an honorable man." He has no need to "speak of a Sinn," much less any notation for doing so.

The one who wants to "speak of the Sinn of an expression" is not the person reporting the content of Antony's speech but the ontologist who is analyzing the situation in which that report is made (a situation described by the sentence 'Antony said that Brutus is an honorable man').⁵⁸ The ambiguity Frege notes exists only in that context. Moreover, it is only the language that is analyzed that is shown to be ambiguous; the language expressing the results of that analysis need not and should not be. That is a theme that Frege has struck earlier in USB. In its second footnote he points out that ambiguity is a failing of ordinary language, one that a properly constructed language (a "perfect language," as he terms it) will avoid. Frege's purpose in ¶6 and ¶7 is to construct a "perfect language" and to use that language to express the results of ontological analysis.

Frege's innovation yields a language that is unambiguous. This is best brought out by replacing the s-signs of the ontological notation with Frege's. (In what follows I shall call this the USB notation.) The truth-makers for (1) and (3) are represented in that notation as follows:

(1^USB) T(m₇, the Sinn of 'the author of Hamlet')

(3^USB) F₄o₃⁵⁹

Though hardly elegant, the USB notation avoids the flaw from which (1) and (3) suffer: using the same sign to stand for different entities.

Oddly, having taken a small step toward the construction of a "perfect language," Frege promptly abandons it. Rather than employ the USB notation, he reverts to using the definite descriptions of ordinary language, thus embracing rather than eliminating the ambiguity they create:

We distinguish accordingly the customary from the indirect Bedeutung of a word; and its customary Sinn from its indirect Sinn. The indirect Bedeutung of a word is accordingly its customary Sinn. Such exceptions must always be borne in mind if the mode of connexion between sign, Sinn and Bedeutung in particular cases is to be correctly understood. (¶7; emphasis in the original)

Frege's procedure is plainly at odds with the considerations that lead to the construction of the USB notation. Moreover, his injunction to "bear in mind" the "exceptions" shows that it is inadvisable. In a properly constructed notation there would be no exceptions to keep in mind; and, since the exceptions are eliminable, they surely should be eliminated. Why Frege chooses not to do so is of peripheral concern here.⁶⁰ There are, however, three points about his choice that should be stressed: First, for the reasons given, it is a mistake. From the first a second follows: USB ought to be interpreted according to the spirit of ¶6 and ¶7; and that spirit demands a "perfect language" of the sort Frege has in mind. Third, the ontological notation satisfies the requirement for being a "perfect language" just as surely as does the USB notation. It is thus the proper device to use in interpreting Frege's argument.

Though "perfect," the ontological notation is not adequate. As it stands, it lacks the resources to express (β).⁶¹ As will now emerge, that defect is easily corrected. Return to (1) above. It describes a situation in which Cordelia is thinking about an object. An adequate analysis of (1) must account for that aspect of it. Merely asserting (α) does not do so. Taking Sinne to be representatives of objects does. The ontological notation - expressing as it does only (α) - lacks the resources to depict that representing relation. Interestingly enough, a few years after the publication of USB, Frege pointed out a neat way to enrich it:

In indirect speech (oratio obliqua) every word has not its ordinary (direct) Bedeutung but, as I put it, its indirect Bedeutung, which coincides with what is otherwise its Sinn. Let, e.g. 'M' and 'N' (in direct speech) be names of the same class, so that M = N is the true. But I assume that these names have a different Sinn in that they determine the class in different ways. Then 'M' cannot be interchanged with 'N' in indirect speech because their Bedeutungen there are different. To avoid ambiguity we really ought to have different signs in indirect speech, though their connection with the corresponding signs in direct speech should be easy to recognize.⁶²

In the last sentence Frege lays down two requirements for a language adequate to express his doctrine of Sinn. The first is that the shape of the signs for Sinne must differ from those for objects. The second is that "their [the Sinn-signs] connection with the corresponding signs in direct speech [the object-signs] should be easy to recognize." The reason for the second requirement is obvious: it is necessary to depict the connection between Sinn and object.

The s-signs of the ontological notation satisfy the first requirement and are easily modified to satisfy the second. All that is required is to complicate their form by adding left subscripts to them. The left-subscript will be identical to the subscript of the o-sign that stands for the object represented by the Sinn. Thus, no s-signs will share right-subscripts but, at least in the typical case, many s-signs will share a left-subscript. For example, let 'o3' continue to stand for Shakespeare. Now return to the two sentences with which we began:

(1) Cordelia is thinking about the author of Hamlet.

(2) Cordelia is thinking about the author of Cymbeline

The earlier representation of their truth-makers in the ontological notation is modified as follows:

(1^ON*) T(m₇, ₃s₆)

(2^ON*) T(m₇, ₃s₈)

The ontological notation makes one point clear: the signs that express the results of ontological analysis do not have the two sides of Sinn and Bedeutung. Consider, for example, 'o₃' and '₃s₆'. Each is a label for an entity. There is little temptation to say of 'o₃' that it "expresses" a Sinn and "designates" a Bedeutung, and even less temptation to say that of '₃s₆'. Thus, the signs of the ontological notation do not stand in two relations to different entities. Rather, they stand in only one relation to one entity. Since there is only one relation holding between signs and entities, that relation cannot be used to differentiate entities from one another. It follows that 'Sinn' and 'Bedeutung' are poor choices to mark the distinction for which Frege argues in USB. One can, if one wishes, reinterpret his terminology accordingly, making clear that they label ontological categories that are different in kind. An easier way to the same result is to coin a different terminology; for example by replacing 'Bedeutung' with 'object', as I have done throughout this paper.

The argument of the previous paragraph is not merely over terminology. Developing it further shows that the slogan "sense determines reference" makes no contact with Frege's theoretical achievement. The slogan, of course, applies to the signs of ordinary language. To this point I have shown only that it does not apply to the signs of the ontological notation. As I shall now argue, what holds for the ontological notation holds for ordinary language as well. In short, just as it is a mistake to attribute the two sides of "expressing" and "designating" to the s- and o-signs, it is also a mistake to attribute those sides to definite descriptions. To bring this out it is necessary to clarify the relation between ordinary language and the language of ontological analysis.

Consider again (1) and (3) and their correlates in the ontological notation:

(1) Cordelia is thinking about the author of Hamlet (1^ON*) T(m₇, ₃s₆)

(3) The author of Hamlet was born in 1564 (3^ON*) F₄o₃

Ontologically speaking, the sentences of ordinary language and those of the ontological notation do the same thing: they represent their truth-makers. The difference lies only in the way they perform that function. The sentences of the ontological notation represent their truth-makers perspicuously; those of ordinary language do not. In this context that difference amounts to the fact that the signs of ordinary language are ambiguous; those of the ontological notation are not. Yet, to repeat, the job those signs do is the same. Thus, if the signs of the ontological notation do not have the two sides of Sinn and Bedeutung, then neither do definite descriptions.

The passages in which Frege appears to attribute a Sinn and Bedeutung to expressions of ordinary language (¶2, ¶5, and ¶13) must now be revisited. Despite initial appearances, the text where that terminology first appears fits rather well with the reading that I have developed thus far:

It is natural now, to think of there being connected (verbunden) with a sign (name, combination of words, written mark), besides that which the sign designates (Bezeichneten), which may be called the Bedeutung of the sign, also what I should like to call the Sinn of the sign. (¶2)

The passage is helpful in that Frege uses the neutral term 'connected'. Though he also employs 'designates', the emphasis of the passage is on the entities to which the definite description is connected, not the nature of those connections. More precisely, the emphasis is on the claim that in intentional contexts definite descriptions are connected to two entities, not one. That claim now needs exploration, especially as it appears to clash with the interpretation I have presented thus far, which has consistently held that definite descriptions in intentional contexts stand for one entity, a Sinn. That interpretation must be revised. The revision, it will be seen, is merely a matter of making explicit what is implicit in holding that Sinne represent objects.

As a representing entity a Sinn never appears "alone" but always connected to an object.⁶³ Thus, if a Sinn is a constituent of a truth-maker, so too is the object it represents. This is as it should be given the requirement that the analysis of the truth-makers for intentional sentences must connect the mind to the object of thought. The ontological notation expresses this aspect of Frege's view. The left-subscript of the s-signs is an essential feature of them; and that subscript stands for an object. Thus, in point of fact, s-signs stand for two entities: a Sinn and an object.

The second passage can be handled along the same lines:

The regular connection between a sign, its Sinn and its Bedeutung is of such a kind that to the sign there corresponds a definite Sinn and to that in turn a definite Bedeutung. (¶5)

Here again Frege says that a sign is connected to a Sinn and, because the Sinn represents an object, to that object as well. The sign thus stands for a Sinn and an object.

In the third passage the doctrine that "sense determines reference" receives its canonical formulation.

A proper name (word, sign, sign combination, expression) expresses its Sinn, means or designates its Bedeutung. By employing a sign we express its Sinn and designate its Bedeutung. (¶13)

The passage is a mirror of the first. The wording of the two is quite similar and each is intended to state the gist of Frege's view. Yet two important differences are evident. The most noticeable is the replacement of talk of "connections" holding between signs and entities by the apparently more precise terminology of "expressing" and "designating." The terminology is more precise only if it does philosophical work. The second difference suggests the work it is intended to do: Frege states that in using definite descriptions speakers "express" Sinne and in so doing "designate" Bedeutungen. That is, Sinne serve as standards for the application of words, and a competent speaker of the language is in possession of such standards.

The passage, unlike the first two, cannot be reinterpreted so as to comport with an ontological reading of USB. The question is whether it should be taken, as it has been by many, as a touchstone for the argument of the work as a whole. The interpretation presented so far supports a negative answer to that question. The interpretation of (γ) to follow will provide additional support. However, in the remainder of this section I want to present a more focused argument against the importance of ¶13. To do so I shall consider a stretch of text that, at first glance, appears to show that USB is centrally concerned with linguistic competence.

In ¶4 Frege writes:

The Sinn of a proper name is grasped by everybody who is sufficiently familiar with the language or totality of designations to which it belongs.

He then appends the following well-known footnote to that sentence:

In the case of an actual proper name such as 'Aristotle' opinions as to the Sinn may differ. It might, for instance, be taken to be the following: the pupil of Plato and the teacher of Alexander the Great. Anybody who does this will attach another Sinn to the sentence 'Aristotle was born in Stagira' than will a man who takes as the Sinn of the name: the teacher of Alexander the Great who was born in Stagira.

Though Frege uses the term 'Sinn' in the footnote and the paragraph it accompanies, the content of his view can be presented without it. This is seen by noting a minor error in his discussion: in specifying the Sinne that might be attached to the name 'Aristotle', he uses the definite descriptions 'the pupil of Plato and the teacher of Alexander' and 'the teacher of Alexander the Great who was born in Stagira'. The problem is that the pupil of Plato and the teacher of Alexander is a person, not a Sinn. Frege's intentions are easy enough to grasp, however. What he means to talk about is not the pupil of Plato and teacher of Alexander but the Sinn expressed by the definite description 'the pupil of Plato and the teacher of Alexander'. Indeed, as we have already seen, Frege commends just this way of speaking about Sinn a few lines later.⁶⁴

If one rewrites the footnote in conformity with those notational conventions one obtains the following:

In the case of an actual proper name such as 'Aristotle' opinions as to the Sinn may differ. It might, for instance, be taken to be the following: the Sinn of the expression 'the pupil of Plato and teacher of Alexander the Great'. Anybody who does this will attach another Sinn to the sentence 'Aristotle was born in Stagira' than will a man who takes as the Sinn of the name: the Sinn of the expression 'the teacher of Alexander the Great who was born in Stagira'.

The rewritten version of the footnote makes two points clear. First, there is no way to make sense of what Frege says without bringing definite descriptions as well as names ("actual proper names," as Frege calls them) into the discussion. Second, once definite descriptions are introduced, all talk of Sinne can be eliminated, at least insofar as we take the footnote to be concerned with linguistic use. This is best shown by rewriting the footnote once again:

In the case of an actual proper name such as 'Aristotle' different speakers will associate it with different definite descriptions. For instance, one will associate it with the definite description 'the pupil of Plato and teacher of Alexander the Great'; another will associate it with the definite description 'the teacher of Alexander the great who was born in Stagira'. Those people will attach different meanings to the sentence 'Aristotle was born in Stagira'.

Frege's point is straightforward: a name applies to an object only by being associated with a definite description and different speakers will associate the same name with different definite descriptions.

The preceding interpretation shows that Sinne have no role in explaining how speakers apply names to objects. Moreover, there is nothing in the text to suggest that Frege intends to mount an argument that they do. But, if according Sinne such a role is not the point of the footnote and associated text, what is? Let me begin to answer that question by expanding the passage quoted above:

The Sinn of a proper name is grasped by everybody who is sufficiently familiar with the language or totality of designations to which it belongs; but this serves to illuminate only a single aspect of the Bedeutung, supposing it to have one. Comprehensive knowledge of the Bedeutung would require us to be able to say immediately whether any given Sinn attaches to it. To such knowledge we never attain.

Frege's emphasis is not on linguistic competence but on epistemic limitations. For example, a person's knowledge that the definite description 'the author of Hamlet' applies to Shakespeare constitutes partial knowledge of him ("serves to illuminate only a single aspect of the thing meant"). Comprehensive knowledge of Shakespeare would mean being able to say for any definite description whether it applies to him or not. Frege, as he does in the footnote, speaks not of definite descriptions but Sinne. Here too his use of that term is idle. For there is no saying that a Sinn attaches to an object apart from saying that a definite description does; and, once that has been said, nothing more is needed to describe the content of what is known.

The footnote is a continuation of that theme. Given that names must be associated with definite descriptions and given the epistemic limitations of the speakers, it is likely that different speakers will associate the same name with different definite descriptions. This reading of the footnote explains a feature of it that would otherwise be idiosyncratic: the complexity of the definite descriptions Frege (implicitly) uses. Were his goal simply to note that names require descriptive backing and then to point out that the backing will vary from speaker to speaker, definite descriptions such as 'the student of Plato' and 'the teacher of Alexander' would suffice. Yet Frege opts for 'the pupil of Plato and teacher of Alexander the Great' and 'the teacher of Alexander who was born in Stagira'. Those definite descriptions neatly illustrate the epistemic point of Frege's discussion. Each, through its incorporation of additional descriptive content beyond that conveyed by 'the teacher of Alexander' suggests comprehensive knowledge; yet each, by being in the possession of one speaker but not the other, shows that it falls short of attaining that status.

As I have unpacked it, Frege's interest in human epistemic limitations has no obvious bearing on his doctrine of Sinn. His reasons for thinking it does can be explained by once again taking account of the structure of USB. The opening paragraph presents his argument in support of (α) and (β). Those theses are stated explicitly in ¶2. ¶3 indicates that the argument is limited to definite descriptions. ¶6 and ¶7 are given over to the development of the USB notation. Frege's argument for (γ) runs from¶8 - ¶12. ¶13 is a summary. That leaves ¶4 and¶5 in something of a no man's land. Both are best read as an appendix to the line of argument of the first three paragraphs. Having established that Sinne exist and said something about their nature (they represent objects), Frege turns to a consideration of what he regards as two imperfections of ordinary language: (1) its signs are often ambiguous and (2) it allows for unfulfilled definite descriptions.

Only the first defect is relevant to the matters at hand. Frege begins his discussion of it in ¶4 where he provides an example (found in the footnote) and locates its source in human epistemic limitations. Epistemic limitations are, on my reading, central to his argument in the opening paragraph. ¶4 is thus a continuation of that argument. Having used the epistemic limitations of a single person to show that Sinne exist, Frege now considers the epistemic limitations of several people. As the footnote makes clear, he holds that when names are used to describe what a person is thinking about, they serve as abbreviations for definite descriptions. The specific definite description will depend on what the thinker knows. Thus, the presence of the same name in two intentional sentences need not entail that the same Sinn is a constituent of their truth-makers. The upshot of all this is that there may be "more" Sinne than a simple inspection of sentences like the following

(4) Plato is thinking about Aristotle

(5) Thomas Aquinas is thinking about Aristotle.

would lead one to believe. The complication Frege notes is genuine. It is also of little ontological interest. At this point in the text the significant claims are (α) and (β). How many Sinne there are and whichspecific Sinne one is to assign to the truth-makers for sentences such as (4) and (5) are details that Frege could and should have avoided. With that said, the important point here is not that Frege engages with a trivial problem but that he engages with an ontological one; that is, with one that has nothing to do with linguistic competence. Interpretations that see USB as centrally concerned with establishing that "sense determines reference" are thus stripped of one of their most important textual foundations.

γ

Frege's thesis (γ) - his Platonism, if you will - is perhaps the most distinctive facet of the view he propounds in USB. To many it is also the least philosophically appealing. Dummett makes clear why:

Frege's conception of thoughts and their constituent senses is mythological. These eternal, changeless entities inhabit a 'third realm', distinct from the physical universe and equally distinct from the inner world of any experiencing subject...

As long as this perspective is dominant, all is mysterious...There is no way of explaining how we grasp [thoughts]...Above all, there is no way of explaining how we attach senses to words or expressions.⁶⁵

The reaction is understandable. It is indeed difficult to conceive how a thinker who is interested in "explaining how we attach senses to words or expressions" would place those senses in an "eternal, changeless...'third realm'... distinct from the inner world of any experiencing subject." And matters are more puzzling still. It is not simply that there "is no way of explaining how we grasp senses" but that in USB Frege is grandly indifferent to providing such an explanation. Since for Dummett there is little difference between "explaining how we grasp senses" and "explaining how we attach senses to words," he should find Frege's Platonism not only mythological but perverse.⁶⁶ Be that as it may, given his understanding of Frege's theoretical ambitions, he is right to dismiss (γ).

There is, of course, another possibility: the mystery Dummett discerns is of his own making, one that derives from his attempt to place (γ) in a theoretical context that is foreign to it. In this section I shall try to show that, situated properly, (γ) is not mysterious but a reasonable solution to an ontological problem.⁶⁷ Let me once again begin by presenting Frege's argument in a way that is neutral between rival interpretations. Its structure is rather simple. Frege believes he has established that in a sentence such as

(1) Cordelia is thinking about the author of Hamlet.

the definite description stands not for an object but for a representative of it.⁶⁸ He recognizes that there will be a strong temptation to say that it stands for something in the mind - an idea, let us say. Frege rejects this:

It might perhaps be said: Just as one man connects this idea, and another that idea, with the same word, so also one man can associate this sense and another that sense. But there still remains a difference in the mode of connection. They are not prevented from grasping the same sense, but they cannot have the same idea. (¶9; my emphasis)

Frege's argument may be stated as follows: Sinne are entities distinct from objects. They do not exist in the mind (they are not ideas). Thus, they exist, as he says elsewhere, in a "third realm." ⁶⁹ The crux of the passage is to argue for the second premise. Frege does so on the basis of two considerations: (A) Two people can think of the same thing. (B) Two people cannot have the same idea. (A), I take it, is unproblematic. It merely allows for the possibility that (1) and

(6) Goneril is thinking about the author of Hamlet

can both be true. The crucial claim, then, is (B).

Surprisingly, commentators have tended to think that (B) is as unproblematic as (A). They regard it as a particular instance of the following general principle, which they also take to be unproblematic: the mental - ideas, emotions, sensations, and so on- is inherently private. Dummett states the view particularly clearly:

Thought differs from other things also said to be objects of the mind, for instance pains or mental images, in not being essentially private. I can tell you what my pain is like, or what I am visualizing, but I cannot transfer to you my pain or my mental image. It is of the essence of thought, however, that it is transferable, that I can convey to you exactly what I am thinking...you as it were take the thought into your mind; I do more than tell you what my thought is like-I communicate to your that very thought. Hence any attempt to investigate thoughts which culminates in a study of what is in essence private, that is, of inner mental experience, must have missed its mark.⁷⁰

Thoughts, Dummett argues, are exceptions to the general rule that mental entities are private. They are exceptions because they alone have non-mental counterparts: the meanings of the words that are used to express them. Dummett thus approves of Frege's non-mentalistic conception of Sinn while rejecting the Platonic form in which it is expressed.⁷¹

The question that must now be raised is why Dummett believes that the mental is "essentially private." An answer can be obtained by attending to the second sentence in the passage. Dummett points out that he can describe his pain to another ("I can tell you what my pain is like"). In other words, his audience can have knowledge by description of it. However, he asserts that he cannot "transfer" his pain to that person. In other words, his audience cannot experience it; that is, cannot have knowledge by acquaintance of it. Whereas physical objects can be experienced by many, only the subject can experience her mental states; and this, it must be emphasized, is as true of thoughts as it is of emotions, ideas and sensations.

Dummett thus offers the traditional other minds problem as Frege's argument for (γ): Frege, recognizing that two people can "think the same" holds that thoughts manage to escape that problem. From this he correctly infers that meanings must be taken out of the mind. His downfall lies in his decision to place them in a "third realm" rather than in the publically observable domain of linguistic use.⁷² Once it is recognized that this intersubjective domain is adequate to secure the objectivity of thought, Frege's Platonism may dismissed as nothing more than an artifact of his limited understanding of the range of philosophical positions available to him.⁷³

Dummett's interpretation seems to have a firm basis in Frege's text. Having asserted that two people cannot have the same idea, Frege writes:

It is indeed sometimes possible to establish differences in the ideas, or even in the sensation, of different men; but an exact comparison is not possible, because we cannot have both ideas together in the same consciousness. (¶9)

Frege thus says what Dummett says he does: one person cannot experience another's idea. Nevertheless, there are two reasons why this cannot be the whole of Frege's argument for (γ). First, placing Sinne in a third realm hardly avoids the problem Frege raises: an "exact comparison" is no more possible for Sinne than for ideas. Second, the thrust of the passage has no obvious connection to the problem that leads Frege to draw a sharp distinction between Sinne and ideas in the first place:

A painter, a horseman and a zoologist will probably connect different ideas with the name 'Bucephalus'. This constitutes an essential distinction between the idea and the sign's Sinn, which may be the common property of many people, and so is not a part or a mode of the individual mind. One can hardly deny that mankind has a common store of thought which is transmitted from one generation to another. (¶8)

Frege's preoccupation in the passage is with a situation in which several people are thinking about the same thing. His question is how that is possible, not how (or whether) one can know that it obtains. The problem raised in ¶8 must now be characterized more precisely. It surely has nothing to do with the psychological processes whereby the painter, the horseman and the zoologist come to think about the same thing or the capacities that make such an achievement possible. Platonic entities fit even more poorly with such causal questions than they do with epistemological ones. An ontological context, however, suits them rather well. As I have noted, Frege is reflecting on a situation like the one described by the following sentences:

(1) Cordelia is thinking about the author of Hamlet.

(6) Goneril is thinking about the author of Hamlet.

As I have also noted, he fastens on a specific feature of that situation: Cordelia and Goneril are thinking about "the same." The analysis of the truth-makers for (1) and (6) must reflect that sameness. That requirement need not lead one to embrace (γ). To show that, I shall propose an analysis that solves the problem raised by (1) and (6) but in a manner very different from Frege's.

The problem with which Frege is concerned is not fundamentally different from that posed by 'This is red' and 'That is red' asserted truly of two material objects. In that case too, ontological analysis must account for "sameness." One way to do so is to hold that 'red' stands for a property and that properties are universals. That universal is a constituent of the truth-makers for both sentences. The same pattern can be directly applied to the analysis of (1) and (6): 'the author of Hamlet' stands for a universal just as 'red' does. In this instance the universal is a property of a mind rather than an object. (1) and (6) describe a situation in which two minds exemplify the same universal. Ideas, construed as universals, solve the problem of "sameness" without invoking a third realm.⁷⁴

Frege, of course, proposes a different solution. The most salient feature of ¶8 and ¶9 is his insistence that two minds cannot "have" the same idea. He thus refuses even to consider the view that takes ideas to be universals. But, as the argument for that view shows, his refusal cannot take it as axiomatic that mental entities can belong to only one mind. Nor does Frege do so. The fundamental move in his argument for (γ) is made earlier in USB.

It is clear from the context that by sign and name I have here understood any designation figuring as a proper name, which thus has as its meaning a definite object (this word taken in the widest range), but not a concept or relation, which shall be discussed further in another article. (¶3)

In the parenthetical Frege points out that he is not using 'object' (Gegenstand) in its ordinary sense. Rather, as the contrast with concepts and relations makes clear, he means to indicate that a definite description always stands for an individual. He thus introduces a constraint on the ontological analysis of sentences containing definite descriptions: no matter what the definite description stands for, that entity must be an individual. It is that ontological thesis, not an epistemological argument rooted in the privacy of the mental, that is the foundation of Frege's argument for (γ).

It does not follow that because an entity is an individual it can be connected to only one mind. Obviously not, for it is essential to Frege's argument for (γ) that many minds can be connected to the same Sinn. The question this raises is why Sinne are treated differently from ideas. It is here that epistemological considerations provide support for Frege's Platonism. If pressed as to why two minds cannot have the same idea, I suspect Frege would hold that to be an immutable feature of the world. If pressed further as to his reasons for so thinking, he would appeal to another immutable feature of the world: "an exact comparison [of ideas] is not possible because we cannot have both ideas together in the same consciousness." The argument is this: suppose Cordelia has a pain-sensation-I₆. Suppose that Goneril wonders whether Cordelia's pain-sensation is more unpleasant than hers-I₇. The only way for her to tell is to bring the two ideas "together in the same consciousness." In doing so, Goneril would have I₆ just as Cordelia does. The impossibility of such a comparison shows that it is impossible for two minds to have the same idea. Ideas are inadequate to serve in the analysis of (1) and (6) because, epistemologically speaking, they are private. However, that matters only because, ontologically speaking, they are individuals. Frege overcomes that epistemological barrier by placing Sinne in the third realm. As Platonic entities they are not answerable to the cognitive limitations of human thinkers. Therefore they can be connected to many minds. Frege's Platonism is not mythological. It is a reasoned solution to a genuine problem.

Conclusion

Philosophically speaking, there is a wide gulf between the Begriffsschrift and USB. Just how wide is obscured by the strong verbal similarity between the texts, a similarity that masks the difference in the problems they address and the solutions they propose. Once those obscurities are cleared away, it can be seen that the restrained epistemologist of the earlier work gives way to the exuberant ontologist of the later. That difference is so striking that one might be forgiven for coming to the view that the author of the Begriffsschrift is not the author of USB.

Bibliography

• ANGELELLI, I. Studies on Gottlob Frege and Traditional Philosophy (Dordrecht: D. Reidl), 1967.
• ______. (ed). Gottlob Frege: Begriffsschrift und Andere Aufsätze (New York: Georg Olms Verlag), 1971.
• ______. (ed). Gottlob Frege: Kleine Schriften (New York: Georg Olms Verlag), 1990.
• BAR-ELLI, G. "Identity in Frege's Begriffsschrift: Where Both Thau-Caplan and Heck Are Wrong," Canadian Journal of Philosophy (38): 269 - 288, 2006.
• BEANEY, M. (ed). The Frege Reader (Oxford: Basil Blackwell Publishing), 1997
• BERGMANN, G. The Metaphysics of Logical Positivism (New York: Longman's Green and Company), 1954.
• ______. "Intentionality," in Bergmann (1959a): 3 - 38, 1955.
• ______. "Frege's Hidden Nominalism," in Bergmann (1959a): 205 - 224, 1958.
• ______. Meaning and Existence (Madison, WI: The University of Wisconsin Press), 1959a.
• ______. "Sameness, Meaning and Identity," in Bergmann (1959a): 132 - 138, 1959b.
• ______. Logic and Reality (Madison, WI: The University of Wisconsin Press), 1964a.
• ______. "Ontological Alternatives," in Bergmann (1964a): 124 - 157, 1964b.
• BURGE, T. "Sinning Against Frege," in Burge (2005a): 213 - 239, 1979.
• ______. "Frege on Sense and Linguistic Meaning," in Burge (2005a): 242 - 269, 1990.
• ______. Truth, Thought and Reason: Essays on Frege (Cambridge, MA: Harvard University Press), 2005a.
• ______. "Introduction," in Burge (2005a): 1 - 68, 2005b.
• BYNUM, T. W. (ed). Gottlob Frege: Conceptual Notation and Related Articles (Oxford: Oxford University Press), 1972.
• DICKIE, I. "Informative Identities in the Begriffsschrift and 'On Sense and Reference'," Canadian Journal of Philosophy (38): 269 - 288, 2008.
• DUMMETT, M. "Can Analytical Philosophy Be Made Systematic, and Ought it to Be?," in Dummett (1978): 437 - 458, 1975a.
• ______. "Frege's Distinction Between Sense and Reference," in Dummett (1978): 116 - 144, 1975b.
• ______. Truth and Other Enigmas (Cambridge, MA: 1978), 1978.
• ______. Frege: Philosophy of Language, 2nd edition (Cambridge, MA: Harvard University Press), 1981a.
• ______. The Interpretation of Frege's Philosophy (Cambridge, MA: Harvard University Press), 1981b.
• ______. "Frege and Wittgenstein," in Dummett (1991): 237 - 248, 1981c.
• ______. "Frege's Myth of the Third Realm," in Dummett (1991a): 249 - 262, 1986.
• ______. Frege and Other Philosophers (Oxford: Oxford University Press), 1991a.
• ______. "The Relative Priority of Thought and Language," in Dummett (1991a): 315 - 324, 1991b.
• ______. The Origins of Analytical Philosophy (Cambridge, MA: Harvard University Press), 1993.
• FREGE, G. "On The Scientific Justification of a Conceptual Notation," in Bynum (1972): 83 - 89, 1882.
• ______. "Function and Concept," in Beaney (1997): 130 - 148, 1891.
• ______. "Thought," in Beaney (1997): 325 - 345, 1918.
• ______. Philosophical and Mathematical Correspondence ed. Brian McGuiness and trans. Hans Kaal (Chicago: The University of Chicago Press), 1980.
• GEACH, P. AND BLACK, M. (eds). Translations from the Philosophical Writings of Gottlob Frege 3rd edition (Oxford: Basil Blackwell), 1980.
• GROSSMANN, R. "Frege's Ontology," Philosophical Review (70): 23 - 40, 1961.
• ______. Reflections on Frege's Philosophy (Evanston, IL: Northwestern University Press), 1969.
• HAAPARANTA, L. AND HINTIKKA, J. Frege Synthesized (Dordrecht: D. Reidl), 1986.
• HECK, R. "Frege on Identity and Identity-Statements: A Reply to Thau and Caplan," Canadian Journal of Philosophy (33): 83 - 102, 2003.
• KLUGE, E.H.W. The Metaphysics of Gottlob Frege: An Essay in Ontological Reconstruction (The Hague: Martinus Nijoff Publishers), 1980.
• KREMER, M. "Sense and Reference: The Origins and Development of the Distinction," in Potter and Ricketts (2010): 220 - 292, 2010.
• MAKIN, G. The Metaphysicians of Meaning: Russell and Frege on Sense and Denotation (New York: Routledge), 2000.
• MARSH, R.C. ed. Logic and Knowledge (London: Unwin Hyman Limited), 1956.
• MOORE, G.E. Some Main Problems of Philosophy (London: George, Allen and Unwin, 1953), 1910/11.
• MENDEHLSON, RICHARD. "Frege's Begriffsschrift Theory of Identity," Journal of the History of Philosophy (20): 279 - 299, 1982.
• POTTER, M. AND RICKETTS, T. The Cambridge Companion to Frege (Cambridge: Cambridge University Press), 2010.
• ROSENKRANTZ, M. "The King of France Restored," Metaphysica 8 (2): 149 - 163, 2007.
• RUSSELL, B. "The Philosophy of Logical Atomism," in Marsh (1956): 177 - 281, 1918.
• SLUGA, H. "Semantic Content and Cogntive Sense" in Haaparanta and Hintikka (1986): 47 - 64, 1986.
• TASCHEK, W. "Frege's Puzzle, Sense and Information Content," Mind (101): 767 - 791, 1992.
• ______. "On Sense and Reference: A Critical Reception," in Potter and Ricketts (2010): 293 - 341, 2010.
• THAU, M. AND CAPLAN, B. "What's Puzzling Gottlob Frege?," Canadian Journal of Philosophy (31): 159 - 200, 2001.
• VAN HEIJENOORT, J. From Frege to Gödel: A Sourcebook in Mathematical Logic 1879 - 1931 (Cambridge, MA: Harvard University Press), 1967
• WILSON, F. "Critical Notice of Michael Dummett, Orgins of Analytic Philosophy," Canadian Journal of Philosophy (27): 377 - 406, 1997.

• 1
  The translation of 'Sinn' and 'Bedeutung' raises a number of philosophical and stylistic complications. To avoid them I shall adopt the expedient of incorporating Frege's terms untranslated and unitalicized into the vocabulary of this paper. I extend that practice to quotations from translations of Frege's works written in 1891 (the year in which he explicitly formulates the Sinn-Bedeutung distinction) and after, but not to those from his earlier writings. A similar approach has been taken by other writers (Taschek (1992) and Beaney (1997)), who interestingly limit it to 'Bedeutung'.
• 2
  Dummett ((1981a) 279, 544) is the most significant representative of this group of interpreters.
• 3
  My use of the phrase 'theory of identity' is designed to do justice to this overlap between the texts (in both of which the word 'identity' figures prominently), while allowing that the content of the theory in each work remains to be determined and may not be the same.
• 4
  Angelelli ((1967) 39 - 40) is, I believe, the first to defend that position. He says of Frege's Begriffsschrift view: "This is already sense and reference: only the famous terminology "Sinn" and "Bedeutung" is lacking here. The point is not trivial, since Frege himself in this respect has done injustice to his text of 1879." His interpretation is endorsed by Bynum ((1972) 66) and Mendelsohn (1982) 283 n. 8. (Mendelsohn argues that USB differs from the Begriffsschrift in extending the distinction from names and definite descriptions to sentences.) More recently, Thau and Caplan ((2001) 197)) have held that the distinction, though not labeled as such, is present in the Begriffsschrift (but see 171, n. 22 for a more qualified assertion). Despite seeming to say that Frege does not distinguish between Sinn and Bedeutung in the Begriffsschrift, Bar-Elli's ((2006) 357, 366)) view is that its domain of application is limited there to singular terms occurring in identity-statements whereas in USB it is applied to all expressions in all sentential contexts. Not all those who have set USB side by side with the Begriffsschrift have concluded that Frege is wrong in finding his later view to be different from his earlier one. For example, Makin ((2000) 81 - 105, esp. 91 - 93 and103 - 105) offers a rich, textually informed argument that the germ of the Sinn-Bedeutung distinction is present in the Begriffsschrift but does not emerge fully formed until USB. One finds a similar, though less developed, view in Grossmann (1969) 23, 155; see also Grossmann (1961) 27 - 29.
• 5
  Compare Makin ((2000) 82)): "Despite [the Begriffsschrift and USB] presenting different solutions, the problem they both confront is exactly the same." Grossmann ((1969) 155) also sees the connection between the two works in that way.
• 6
  Dickie (2006) holds that the Begriffsschrift and USB respond to different epistemological problems: the first concerns the acquisition of knowledge; the second, the making of inferences (see 274 for a concise statement to this effect). She thus sees greater continuity than do I.
• 7
  In this paper I use Black's translation in Geach and Black (1980) and the German version of USB in Angelelli (1990). I shall often refer to passages by paragraph number. Those references will be incorporated parenthetically within the text I have chosen to render Frege's 'Gleicheit' as 'identity' rather than 'equality' as Black does. Frege points out in a footnote that by 'Gleicheit' he means 'Identität'. Since the examples he uses are not confined to mathematical ones, 'identity' seems more appropriate. This also has the obvious advantage of conforming to the terminology of the secondary literature.
• 8
  See ¶3.
• 9
  Bar-Elli ((2006) 358, 363, 367) in passing treats 'arbitrary' and 'conventional' as synonyms. That conflation is unfortunate since it bars the way to a precise characterization of Frege's distinction between proper and improper knowledge, without which it is impossible to evaluate his criticism of the Begriffsschrift.
• 10
  Thau and Caplan ((2001) 172-3, 178) are aware of the importance of improper knowledge in Frege's argument but view it from the widely accepted position that Frege's concern in USB is to differentiate "informative" from "trivial" identity statements. They are thus led to identify improper knowledge with knowledge that is trivial and to assert that all knowledge of linguistic conventions is trivial: "That two names co-refer... is...a rather uninteresting accident of human culture." (172, emphasis in the original) Their judgment will certainly draw the dissent of, among others, the detective who has discovered the alias under which the suspect for whom she is searching is hiding.
• 11
  I here use the Mengleberg-Bauer translation in van Heijenoort (1967) and the German version in Angelelli (1971).
• 12
  For example, Dummett ((1981a) 279) states without comment that "Frege ...maintained in Begriffsschrift that the sign of identity has to be taken as standing for a relation between names." Similarly, Heck ((2003) 88) asserts flatly that the Begriffsschrift theory of identity "[ is] a theory about what relation identity is." (emphasis Heck's; see also 85 and 101)
• 13
  There is thus an obvious awkwardness in stating that the name view holds that identity is a relation between signs-a formulation which cannot help but suggest that it is concerned with the nature of identity. Nevertheless, given the formulation's basis in the Fregean texts and its wide currency, its use is unavoidable, at least provisionally.
• 14
  This second issue has been the subject of much recent scholarly controversy. In their seminal paper Thau and Caplan (2001) contest what they call the "standard interpretation" (see 159 - 164) of the opening paragraph of USB, which takes it to reject the name view. Against the standard interpretation they argue that Frege never abandons the name view. Heck (2003) defends the standard interpretation against their criticisms. Bar-Elli (2006) agrees with Thau and Caplan that in USB Frege does not reject the Begriffsschrift theory of identity but denies that Frege ever subscribed to the name view. On his view (see 356), Frege consistently takes identity to be a relation between objects. Dickie ((2008) see esp. 269 n. 1) endorses Heck's defense and seeks to extend it.
• 15
  Frege ((1882) 84).
• 16
  The argument as I have presented it may appear odd in that it depicts Frege as introducing what he knows to be a flaw into the language of the Begriffsschrift and then proceeding to correct it. Keeping in mind the dialectical structure of §8 will, I believe, remove that feeling of oddness. Those who think Frege endorses the name view have also noticed the consequent ambiguity and faulted Frege accordingly; see Kremer ((2010) 239) and the literature cited there. Bar-Elli ((2006) 363) rightly holds that because the name view would render the Begriffsschrift language ambiguous, Frege cannot be taken to endorse it; however, he does not seem to recognize that Frege does not merely withhold his assent but develops a cogent argument against the view.
• 17
  Heck ((2003) 85) is an exception: "Frege simply assumes... [that] identity is indeed a (binary) relation. Now, for Frege, a relation is a binary function...So the...question as Frege would have understood it circa 1892, is what the arguments of this function are: the question is whether they may be any objects one likes... or must always be names; the question is whether the relation of identity holds between Hesperus and Phosphorus or between 'Hesperus' and 'Phosphorus'." Heck does not say how one might ever be tempted to ask the second member of either of those question pairs. (Though Heck's remarks concern Frege's views in USB, he endorses that work's attribution of the name view to the Begriffsschrift; see 100-101.) Thau and Caplan ((2001) 177 n. 28) also drift toward this interpretation. Makin ((2000) 96) has pointed out the absurdity of taking the name view as a thesis about the nature of identity.
• 18
  (1981a) 544. The recent literature on the Begriffsschrift cited in n. 14 conforms to the pattern found in Dummett. This is true even of Heck (2003) who is the most insistent that Frege's concern is with "what identity is" (88; see also 85, 101). Despite that insistence, his article focuses on what identity statements "express." At the conclusion of his paper he makes a brief attempt to link the two issues. His argument is unconvincing since it does not try to make the claim that identity is a relation between signs intelligible.
• 19
  That point, though basic to Frege's argument, is overlooked in most of the literature on the Begriffsschrift I have cited in this paper. Grossmann ((1969) 19-20) is an exception, though he seems to regard the matter as so obvious as to scarcely be worthmentioning . He also holds that Frege, despite acknowledging the two different ways in which a sign can be connected to an object, does not distinguish between names and definite descriptions. For the reasons given above, I think that Frege fails only to coin the requisite terminology. His failure may well be why the issue has not received due attention. Most commentators on §8 the Begriffsschrift follow Frege's lead in speaking of "names" without recognizing that the term applies to both genuine names and definite descriptions. They are thus led to slight the importance of the distinction between conventional and non-conventional designating relations.
• 20
  (1918) 244 - 245.
• 21
  I have taken the liberty of using single quotes to make clear that Frege is mentioning signs.
• 22
  Bergmann ((1959b) 134) makes the point in a refreshingly straightforward way: "while 'same' is rather useless when it stands between names, such sentences as 'The evening star and the morning star are the same' serve a purpose."
• 23
  Others have taken a dimmer view of Frege's achievement. For example, Heck ((2003) 100-101) holds that the Begriffsschrift has no argument against the charge that identity statements express only improper knowledge. Makin ((2000) 92-94) grants that there is an argument but deems it inadequate. The views of other commentators will be discussed shortly; see n. 24, n. 26 and n. 27.
• 24
  Thau and Caplan ((2001) 178 - 180), believing something more lavish is necessary and taking Frege to believe the same, find his argument to be unconvincing. Kremer ((2010) 258-9) expresses similar reservations. I comment in more detail on this issue below; see n. 27.
• 25
  The Bauer-Mengleberg translation uses 'way of determining'; Bynum (1972) opts for 'mode of determination'. Since the latter has become standard in the literature, I have altered the translation accordingly.
• 26
  See again Thau and Caplan ((2001) 178 - 180) and Kremer ((2010) 258-259).
• 27
  This issue is underexplored in the literature. For example, none of the recent discussions of §8 cited above (see n. 14 ) provide a precise account of it. The gap is particularly glaring in Thau and Caplan ((2001) 178 - 180) who, as I have indicated (see n. 24), find the argument I have presented to be incomplete. What needs to be added, they argue, is Frege's concept of a "mode of determination." Yet they do not say what a mode of determination is, and thus don't make clear what it adds. Moreover, their criticism makes it doubtful that more could or need be added. They deny that the example in [ C] is an instance of proper knowledge on the grounds that "you might know that the mode of determination associated with 'A' and the mode of determination associated with 'B' determine the same object without having any idea what either mode of determination is...[thus] knowing that the two modes of determination determine the same object will not give you any geometrical knowledge at all." (179) The only sense one can make of the possibility thus imagined is that the person knows that 'A' and 'B' apply to the same object but does not know what 'A' and 'B' mean. But Thau and Caplan do not identify mode of presentation and meaning. That is made clear when they argue that Frege does not consider the possibility they raise because he "tacitly assumes that understanding a name [i.e. knowing its meaning] requires knowing which mode of determination is associated with it." Kremer's criticisms are less developed. He notes ((2010) 259) that "Frege never goes beyond the metaphor of 'mode of presentation' to provide a more detailed theory of sense." As I hope to have shown, properly interpreted, §8 says all that needs to be said.
• 28
  Thau and Caplan ((2001) 179; see also 162 n. 7 and 163 n. 8) The original makes use of single quotes throughout (presumably as a result of the journal's editorial conventions). I have used double quotes where Thau and Caplan quote Frege, and retained single quotes where they mention expressions.
• 29
  Dickie ((2008) 280); the point and the phrasing are repeated on 281.
• 30
  Bar Elli ((2006) 361).
• 31
  In USB Frege uses 'proper name' for singular terms generally. I shall follow him in that use, reserving 'name' for signs like 'Socrates', 'Shakespeare', and so on.
• 32
  (1981a) 94-95. In this section and the next I shall use Dummett's reading as a foil for developing my own. Though his approach to Frege is no longer as dominant as it once was, it is nonetheless determinative for the scholarship that has followed in its wake and thus remains fundamental. Moreover, with respect to the specific issue of the Sinn-Bedeutung distinction, later writers have not developed a genuine alternative to Dummet. Though critical of his particular way of understanding the distinction, they agree with him in taking it to be the same as that between an expression's meaning and its referent. For example, Tyler Burge ((1979) 222) rejects Dummett's identification of Sinn with "linguistic meaning" (the meaning of expressions in their ordinary uses) but allows that Frege "would perhaps have granted that meaning and sense are identical in a perfect, context-free language." (See his ((2005b) 36) for a less qualified statement.) Such a language would be found in a community of "ideal speakers" ((1990) 261ff), who would use it to express a "completely fundamental and completely satisfactory (true) theory." (266) Sluga ((1986) 63 n. 4) vehemently rejects Dummett's claim that Sinn is a semantic concept; but his own thesis - that Sinne are the ways objects are presented to knowers (53 - 54) - is simply Dummett's with the linguistic item dropped out (cf. Dummett ((1981a) 110), a passage I discuss below). Sluga's subsequent development of his argument (54ff) further collapses the difference between his position and Dummett's, as he concedes that Frege's notion of Sinn is incorporated within a semantic theory. Taschek ((1992) 768 - 769) argues that Frege's distinction is designed to account for the ways in which beliefs and assertions are logically appraised. However, as his gloss on the opening paragraph of USB (785 - 787) shows, his account does not differ in essentials from Dummett's. For example, Taschek notes that even if 'a=b' is true, it is possible for someone who is ignorant of that truth to believe 'Fa' and '~Fb' without being guilty of logical inconsistency. His explanation is that 'Fa' and 'Fb' differ in their "logically relevant properties." But the only "logically relevant property" that is evident is the meaning of 'a' and 'b'. Whatever the differences between these writers may be on points of detail, they all take Frege's concerns to be epistemological. Burge (see, for example, (1979) 213, 235 n22, 239 and (1990) 245 n5) and Sluga do so explicitly. Tascheck, in a more recent piece ((2010) (293-294, 321-322, 340 - 341), presents his interpretation as a dissent from what he takes to be the prevailing epistemic orthodoxy whose primary representative is Dummett. He continues to identify Sinne with the "logically relevant properties" of expressions (295 - 303). In explaining that view he writes (299), "Frege views the sense of a sub-sentential expression as a way of thinking about its reference - or if one prefers, a way in which the expression presents the reference to the speaker." Thus, his understanding of Sinn, like Sluga's, is essentially the same as Dummett's and is essentially epistemic in character. Taschek's primary disagreement with Dummett is not over his understanding of Sinn per se but with his view that it is part of a theory of linguistic competence (321 - 330). The justification for classifying Dummett's Frege as an epistemologist requires more extended comment; see n. 34 and n. 72. I hope that the obviously epistemic flavor of the passages I quote will suffice in the interim.
• 33
  1981a 227.
• 34
  (1981a) 110. Sinne so understood are obviously heirs to the "concepts" and "ideas" of traditional epistemology. There is some irony in this since Dummett argues ((1981a) 665 - 684, esp. 665 - 669 and (1975a) 441) that Frege achieved a revolution in philosophy by displacing epistemology from the central place it had occupied since Descartes and replacing it with the philosophy of language. In later works ((1991b) 322 and (1993) 5) he acknowledges that Fregean philosophy of language is not as far from epistemology as he had once held.
• 35
  More accurately, the meaning is what connects an ink-mark to an object and in so connecting it turns it into a proper-name.
• 36
  As Dummett puts it ((1981b) 45), "The same object may be picked out in different ways and the different...ways of picking it out constitute different senses that can be borne by different proper names all of which stand for that object." In the example under consideration Shakespeare is "picked out" as the author of Cymbeline and as the author of Hamlet.
• 37
  Or rather, they do provided they are functioning as mere labels. Dummett ((1981a) 97 - 98, 135) takes Frege to deny that the names of ordinary language do so function. On his reading, all such names are associated with a descriptive backing - though this descriptive backing need not be expressible by means of a single definite description - that serves as their sense. This descriptive view of names is a central concern of his (1981a) (see especially 54 - 109); however, he makes the point most straightforwardly elsewhere ((1975b) 126): "there can be no such thing as bare knowledge of the reference of a word." In Section IV (see p. 40ff) I argue that while Frege does accept this descriptivist view of names, it is a relatively unimportant element of his thought.
• 38
  Grossmann ((1969) 155) and Makin ((2000) 103) have also noticed this difference. By contrast, Thau and Caplan ((2001 162 n. 7) find Frege's use of 'contained' to be nothing more than a verbal quirk
• 39
  Dummett expresses that judgment most forcefully in a later paper ((1986) 256): "senses... are not self-subsistent objects at all."
• 40
  Dummett's critics follow him here as well. In none of the works referred to above (see n. 32) does the status of Sinne as entities do any theoretical work.
• 41
  My conception of ontology is indebted to the work of Gustav Bergmann; see Bergmann (1954), (1959a) and (1964a).
• 42
  (1964b) 126. As Bergmann goes on to note, "knowing is used generically for any situation in which an act intends something whatever its species may be." Thus, "assertings," "doubtings", "wonderings" and so on are knowing situations. Frege considers a variety of such "knowing situations" in USB.
• 43
  Unfortunately, commentators have all too rarely felt that need. Instead, they have leaned on the ordinary connotations of the phrase, treating it as equivalent to "what a sentence says" as well as expressing an implied judgment as to the importance of that content. Dummett is a good example. As noted, he identifies the cognitive value of a sentence with its "information content." He also finds ((1981a) 288 - 290) the argument of the opening paragraph of USB to depend on the distinction between sentences that are informative and those that are trivial (lacking in cognitive value).
• 44
  I use the word 'helpful' advisedly; as I shall bring out, the logic of Frege's argument does not require the contrast between the two forms.
• 45
  The starkness is purely rhetorical since nothing in Frege's argument turns on there being statements that it is impossible not to know. All he needs to consider - and all he does consider in texts to be discussed shortly - is a situation in which a person happens to know P but not know Q (further details of that knowing situation will be important but can be set aside for the moment). Insofar as the rhetoric clouds the logic, Frege would have been better off without it.
• 46
  Frege's two other uses of 'cognitive value' require brief mention. One is merely implicit and occurs earlier in the opening paragraph: "Now if we were to regard identity as a relation between that which the names 'a' and 'b' designate, it would seem that a=b could not differ [in cognitive value] from a=a (i.e. provided a=b is true)." Stripped of the misleading talk about the nature of identity, this passage says exactly what the later one does. The other use is found in the final paragraph of USB. There the term occurs twice. The first can be read as the others have been. The second is more troublesome: "the Sinn of 'b' may differ from that of 'a', and thereby the thought expressed in 'a=b' differs from that of 'a=a'. In that case the two sentences do not have the same cognitive value." 'Cognitive value' here is not readily glossed in terms of knowing/not-knowing. Neither, however, is it readily amenable to another reading. For example, if one takes 'cognitive value' to mean "what a sentence says," then Frege asserts the following: if 'a' and 'b' express different Sinne, then 'a=a' and 'a=b' say different things. That fits with the letter of such interpretations but not the spirit; for according to them the difference in Sinn is supposed to explain the difference in what is said. Thus, what Frege ought to have written is that if 'a=a' and 'a=b' differ in cognitive value then 'a' and 'b' express different Sinne and thus so too do the sentences in which they occur. If that correction is made - as I believe it should be - then the passage fits the letter and spirit of any interpretation since it does nothing more than repeat the wording of the first two paragraphs of USB, albeit in condensed form-which is, I suspect, Frege's intention.
• 47
  I have modified the translation slightly.
• 48
  ((1891) 138).
• 49
  ((1981a) 90).
• 50
  ((1981a) 95).
• 51
  Makin ((2000) 109 - 110) also quotes this passage from Dummett and criticizes it on the same grounds as I do.
• 52
  Bergmann ((1958) 214), ((1959b) 135) and ((1964b) 135 n. 11) takes Frege's argument this way but does not attempt to ground his view textually. My reading is also in basic agreement with Makin's ((2000) 86 - 88). Kluge ((1980) 183 - 191) argues that the Sinn of a proper name reflects "the ontological structure of the referent itself." (190) He thinks that for Frege objects are bundles of properties. Therefore, they must always "be referred to...under one aspect or another." (190) Those aspects are Sinne. Kluge's interpretation conflates ontological and epistemological considerations. For example, it is perfectly consistent to hold that objects are particulars exemplifying properties - a position Kluge finds incompatible with the existence of Sinne - and yet to acknowledge that we are not presented with those particulars and therefore must always refer to objects "under one aspect or another." Kluge's understanding of the theoretical role of Sinne turns out to be the same as Dummett's; it differs only in finding that role adequate to secure them ontological status.
• 53
  ((1918) 342.) See Dummett ((1981a) 442) for the standard reading of this passage.
• 54
  For this reason I find Beaney's and Taschek's decision to leave 'Bedeutung' untranslated while rendering 'Sinn' by 'sense' (see n. 1) less than felicitous. Doing so strongly implies that 'Sinn' is not a technical term for Frege. My discussion of (β) will provide further grounds for dissenting from their practice.
• 55
  Since the slogan is so familiar and since its wording so nicely captures its content, I have refrained from altering it in accord with the conventions introduced at the beginning of this paper.
• 56
  Dummett faults Frege on just those grounds: ((1981a) 268): "the sense [meaning] of a word cannot vary from context to context, but is a property of the word apart from any context." He proposes modifying Frege's system accordingly. In doing so he tacitly acknowledges that in intentional contexts Frege's notion of Sinn cannot be identified with an expression's meaning. Later ((1990) 318) he makes that point explicitly.
• 57
  I have modified the translation slightly.
• 58
  The blurring of these two contexts - an ordinary situation and the ontological analysis of it - explains why Frege makes his notational proposal in connection with indirect discourse. His real interest is in designing signs to stand for Sinne. He suggests doing so by mentioning the expressions that are used in the situation to be analyzed. In that sense what the ontologist does has an affinity with what one who reports the speech of another does.
• 59
  I have taken the liberty of introducing F-signs to stand for concepts. Since concepts are not under discussion here, those signs may be treated as mere placeholders.
• 60
  The explanation lies, I believe, in two related failings of his notation: The first is its obvious clumsiness. The second is its incorporation of mentioned definite descriptions within the signs for Sinne. That feature, though philosophically motivated (see n. 61), makes Frege's notation very similar orthographically to ordinary language. Since the USB notation is hard to use and not all that different "visually" from ordinary language, it is all too easy for Frege to lapse in the way I have described. That lapse, I stress, results from Frege's failure as a craftsman, not a philosopher.
• 61
  The USB notation expresses (β) by incorporating mentioned definite descriptions in the signs for Sinne. The mentioned definite description indicates how the sign in which it occurs is to be interpreted: it stands for the Sinn that represents the object to which the definite description applies. It bears mentioning that the mentioned descriptions also serve the same function as the subscripts of the s-signs of the ontological notation.
• 62
  Frege ((1980) 153). I have corrected an obvious use/mention error in the penultimate sentence of the passage.
• 63
  For this reason the analysis of unfulfilled definite descriptions poses a special problem for Frege. An unfulfilled definite description is connected to a Sinn, and a Sinn must represent an object. But in the case of an unfulfilled definite description there is no object for the Sinn to represent. Thus, the Sinn exemplifies the representing relation even though one of the terms of that relation is missing. Elsewhere I have argued (Rosenkrantz (2007)) that Frege cannot surmount this problem. Dummett ((1981b) 45 - 46) notes the difficulty in its philosophy of language variant but is content to rely on an unexplicated metaphor to relieve it: the sense of an expression, he holds, is the route to its referent but sometimes a route "lead[s] nowhere."
• 64
  Frege's misstep is perhaps explained by the fact that he has not yet introduced the requisite devices for avoiding it.
• 65
  ((1986) 251 - 252).
• 66
  It should be clear that, mutatis mutandis, anyone proposing an epistemological reading of USB should find (γ) to be mythological. More frequently, the issue is ignored. For example, it is not discussed by Taschek or Sluga in the works cited above(n. 32). Burge ((1990) 255 n. 9) notes that on his reading (γ) is not an essential component of Frege's theory of Sinn. He does allow, however, that the thesis is "deeper and more interesting than commonly thought." Unfortunately, he does not expand on that assessment.
• 67
  My criticism of Dummett's evaluation of (γ) is indebted to Wilson (1997).
• 68
  Since 'Sinn' inevitably suggests (γ) I avoid using it here.
• 69
  ((1918) 337).
• 70
  ((1975) 116 - 117). It is important to note that Dummett is here speaking in his own voice as well as offering an interpretation of Frege
• 71
  As he ((1993) 25) argues elsewhere, "The importance of the denial of the mental character of thoughts...did not lie in the philosophical mythology to which it gave rise - Frege's myth of the 'third realm'...It lay, rather, in the non-psychological direction given to the analysis of concepts and propositions."
• 72
  The passage from Dummett in n. 71 continues, "One in this position [of taking the mental to be private] has therefore to look about him to find something non-mythological but objective and external to the individual mind to embody the thoughts which the individual subject grasps and may assent to or reject. Where better to find it than in the institution of the common language." These remarks allow for a more precise characterization of how Dummett understands the relationship between Frege and traditional epistemology. For Dummett, epistemology is irredeemably mentalistic and the mental is inherently private. Language is public. Thus, the philosophy of language can give an adequate account of the objectivity of knowledge while epistemology cannot. But this amounts to saying that Frege is able to deliver a superior epistemology. Dummett's Frege is thus not a revolutionary but a reformer; compare n. 34.
• 73
  Dummett writes ((1986) 249), "Given [Frege's] strict dichotomy between the radically subjective and the wholly objective - his rejection of any intermediate category of the intersubjective - it follows from the fact that the senses of words and of sentences can be grasped by different individuals that they exist eternally and immutably in complete independence of us."
• 74
  Bergmann (1955) proposes a view akin to the one of this paragraph. Wilson ((1997) 403 - 406) also puts forward Bergmann's view as an alternative to Frege's. He finds the original source of it in Moore (1910/11).

Publication Dates

• Publication in this collection
  Jul-Sep 2016

History

• Received
  23 Nov 2015
• Reviewed
  12 June 2016
• Accepted
  27 June 2016