36

Notre Dame Journal of Formal Logic

Volume 28, Number 1, January 1987

Thoughts

DAVID BELL

In August 1919 Bertrand Russell wrote to the author of the Tractatus to

ask, among other things, about the nature of thoughts and their constituent ele-
ments. Wittgenstein replied: "I don't know what the constituents of a thought
are, but I know that it must have such constituents which correspond to the
words of Language. Again the kind of relation of the constituents of thought
and of the pictured fact is irrelevant. It would be a matter of psychology to find
out" ([14], p. 72). Returning to the topic some thirty years later, Wittgenstein
wrote, with more than a little irony:
"Can one think without speaking?"—And what is thinking! — Well, don't
you ever think? Can't you observe yourself and see what is going on? It
should be quite simple. You do not have to wait for it as for an astronomi-
cal event and then perhaps make your observation in a hurry. ([15], p. 327)
Now these remarks, and many others like them scattered throughout Witt-
genstein's writings, raise a number of very general and, I think, profoundly dif-
ficult problems; among them: the problem of how we are to account for the
relation of thought to the language which expresses it; of how to account for
the relation of thought to the reality it sometimes concerns; and, even more gen-
erally, of how to provide a genuinely philosophical theory of what can easily
seem to be a phenomenon of primarily, if not exclusively, psychological con-
cern. What can philosophy legitimately tell us about thoughts, that it would not
be "a matter of psychology to find out"?
In what follows I shall try to set out some of the most general, but there-
with most pressing, demands that an acceptable philosophical theory of thoughts
must meet. I shall then try to establish the extent to which Frege's own theory
successfully meets them. In a sense, however, there are not two tasks here but
only one, for in specifying the general constraints within which a philosophical
theory of thought should work, one is already merely recapitulating what is per-
haps Frege's most lasting and revolutionary contribution to our understanding
of the matter. In an important sense the very subject is Frege's —and our first
task will be to get a bird's-eye view of it.

Received May 16, 1986

 THOUGHTS 37

/ As a rough initial approximation we can, I think, distinguish four problem-

atic areas associated with the notion of a thought. There are problems concern-
ing interiority, objectivity, reβexivity, and rationality. As we shall see, thoughts
become most refractory, least amenable to philosophical treatment, at precisely
those points where the requirements dictated by considerations from within these
different areas meet and overlap.
First, then, interiority. Whatever else thoughts might be, and regardless of
the subsequent demands we might wish to make of them, one thing is surely
clear: thoughts can be had. Thoughts can be grasped, apprehended, entertained,
or understood; which at this point is to say no more than that, quite simply, we
can think. Were this not so, then of course the entire topic would lose its interest
and importance. Dummett has written:

Thought differs from other things also said to be objects of the mind, for
instance pains or mental images, in not being essentially private. . . . It is of
the essence of thought that it is transferable, that I can convey to you exactly
what I am thinking. . . . I do more than tell you what my thought is like —I
communicate to you 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. ([2], pp. 116-117)

There is doubtless a sense in which this is right, and I have chosen the ugly
word "interiority" rather than, say, "subjectivity" to avoid any suggestion that
a thought is some sort of private, mental entity. Nevertheless one can go too far
in this direction. After all, thinking is a mental act; grasping a thought is,
paradigmatically, an event that occurs in the arena of the mind —and a philo-
sophical theory of thoughts must surely recognize this. And yet as a topic its
absence from twentieth-century analytic philosophy has been conspicuous. We
might attribute this absence to the influence, say, of Frege's anti-psychologism;
of the behaviorism of Watson and Quine; of American pragmatism; of the
apparent antipathy of the later Wittgenstein to what is "private"—but at all
events contemporary philosophers have on the whole felt themselves unable or
unentitled to inquire into the phenomena associated with interiority, and have
approached such notions as thought, understanding, meaning, and judgment as
though objectivity, reflexivity, and rationality were the only areas of legitimate
philosophical concern. I shall return later to what I take to be a genuinely phil-
osophical problem related to interiority.
In addition to such noncommital acts as entertaining or grasping a thought,
acts which are without prejudice as to the truth or falsity of what is grasped or
entertained, there are also those which, like asserting, denying, doubting, accept-
ing, and judging, do involve some commitment to truth or falsity. Although it
might seem plausible at this point to admit three basic kinds of propositional
attitude—noncommittal (having a thought), positive (taking a thought to be
true), and negative (taking a thought to be false)—Frege argues persuasively that
in fact only two are necessary: "To make a judgment is to make a choice
between opposite thoughts. Accepting one of them and rejecting the other is one
act. So there is no need [in a perspicuous notation] for a special sign for the
rejection of a thought. We only need a special sign for negation as such ([10],
p. 185). Denial, then, does not stand contrasted with judgment; for to deny that
38 DAVID BELL

p is merely to judge that not-p. With Frege we can also demand that one and
the same thought can now be grasped noncommitally, now judged to be true:
"Making a judgment does not alter the thought that is recognized to be true"
([10], p. 251).
The second category of demands we might make on a theory of thoughts
concerns their objectivity, and here three separate but interrelated families of
issues need to be taken into account. First, as Dummett notes in the above quo-
tation, thoughts are essentially intersubjective: two or more people can have one
and the same thought. In other words, the identity conditions of thoughts, unlike
those of, say, sensations or mental images, make no essential reference to the
identity of the person who has them. The second requirement is that thoughts
be expressible: they can be put into words. Now when I express a thought I do
not, as it were, describe my thought from the outside; rather I embody my
thought in language: the relation between a thought and its linguistic expression
is not an external but an internal relation—a truth that both Frege and Wittgen-
stein attempted to capture by identifying a thought with the sense of the declara-
tive sentence expressing it.
Under the general rubric of the "objectivity" of thoughts there is a further
and vitally important family of issues: what we think about might exist, and
what we think might be true. Frege himself typically expresses these requirements
in terms of the conditions of the existence of thoughts: "In thinking", he writes,
"we do not produce thoughts, we grasp them. For what I am calling thoughts
stand in the closest connection with truth. What I acknowledge as true, I judge
to be true quite apart from my acknowledging its truth or even thinking
it. . . . Therefore that truth cannot have come to be only upon its discovery"
([9], p. 25). There is much that is highly contentious here (some of which I shall
defend below), but at least part of Frege's point can be expressed in more plau-
sible terms: thoughts possess truth conditions which are sometimes fulfilled,
sometimes not, but those truth conditions typically make no essential reference
to any person who has, or who might have, the thoughts in question. (The pre-
cise relation between this point which concerns truth conditions, and the anal-
ogous point about identity conditions made in the preceding paragraph has yet
to be examined.)
The constraints which concern reflexivity can also be dealt with briefly.
In the class of things about which we can think and judge there are of course
thoughts and judgments. This apparently incontrovertible observation creates
a host of intractable problems concerning, e.g., intensional entities, referential
opacity, and the provision of an adequate semantics for expressions in oratio
obliqua. Fortunately such problems are not of present concern. There is one
point to note in passing, however, as it will be significant in what follows.
Although reflexive thought is certainly possible, we must distinguish sharply
between the function of thoughts as the contents of acts of thinking, and
thoughts as the objects of such acts; between having a thought and thinking
about a thought.1 Clearly not all thinking can be reflexive.
Finally I come to those constraints on a theory of thought which have to
do with rationality. Thoughts are gregarious things: they come in trains; they
stand to one another in relations of compatibility, incompatibility, relevance and
entailment; they get together to form arguments, and so on. In general, of
 THOUGHTS 39

course, it is to their structure that we look for an explanation of such phenom-

ena, and an important ingredient in any philosophical theory of thoughts will
be the provision and justification of a procedure of analysis that will enable us
to recognize such structures and to isolate the repeatable elements which go to
make up our thoughts. We shall require, in other words, a procedure for carving
up compound thoughts into their component thoughts, and atomic thoughts into
their component concepts.
So far, then, I have suggested that it is at least prima facie reasonable to
require of a philosophical theory of thought and judgment that it meet the fol-
lowing condition of adequacy:

(A) It should allow that thoughts can sometimes be

(1) had (in thinking) λ ¥
. .
/ΛW i * u + r - Λ \\ Inferiority
(2) taken to be true (in judging) j
(3) expressed in language Ί
(4) inter subjective \ Objectivity
(5) objectively true/false J
(6) about thoughts 1 Reflexivity
(7) about judgments J
(8) complex, i.e., containing component thoughts^
(9) complex, i.e., containing component concepts r Rationality
(10) logically related one to another J

and preferably, of course, it should explain such possibilities.

There is here a second, overriding requirement, namely

(B) It should employ a univocal notion of thought throughout (l)-(10)

above.

The thoughts we entertain are the very thoughts that we assert, that we express
in language, that are true or false, that stand in logical relations with each other,
and so on.
The third requirement I shall call the Principle of Spontaneity. The con-
straints it places on possible theories of thought are entirely formal: they amount
to no more than an insistence that any such theory shall not generate a vicious
regressive infinity of conditions on the performance of an act. The principle is
this:

(C) If the performance of an act of type Γis learned or rule governed, then
it cannot be a general requirement of my performing an arbitrary
act of type T that I have already performed an act of that type or,
indeed, of any type that, in its turn, requires the performance of an
act of type 7\

So this principle outlaws any theory according to which, for example, criteria
are always applied on the basis of criteria, or concepts are in general understood
in terms of other concepts, or judgments can only be made once we have satis-
fied ourselves that they are warranted, and so on. Crass infringements of the
principle are rare, needless to say, yet Wittgenstein for one was adept at uncover-
40 DAVID BELL

ing the incoherence of views whose contravention of the principle was very far
from obvious. Thus the Investigations opens with a quotation from Augustine's
Confessions which offers a naive but attractive account of how we learn to talk.
Perhaps the major failing of this account is that "Augustine describes the learn-
ing of human language as if the child . . . already had a language, only not this
one. Or again: as though the child could already think, only not yet speak. And
'think' here would mean something like 'talk to itself " ([15], p. 32). Augustine's
account thus clearly contravenes the Principle of Spontaneity by implying that
mastery of a language is a prior condition of aquiring mastery of a language.
The principle may seem uncontentious, even trivial. Indeed, I think it is—though
it is also, I believe, a principle whose consistent application yields a number of
striking results; for it underlies and justifies many of Wittgenstein's conclusions
about ostensive definition, and about what it is to be able to follow a rule. It
is, for example, an instance of the Principle of Spontaneity which forces the con-
clusion that "When I obey a rule I do not choose, I obey the rule blindly"?

2 Frege's theory of thoughts is ambitious. Of the ten requirements listed under

(A) above, it is intended to meet eight maximally, that is, to comprise an expla-
nation of the phenomena involved. And it is intended to meet the remaining two
demands minimally, in that the theory is claimed to be at least compatible with
their possibility, even though in the last analysis it offers no account of them.
Given Frege's vehement antipathy to all attempts to introduce psychological
considerations into logic, it is hardly surprising that the two phenomena for
which he provides no theoretical account are those constitutive of what I've
called "interiority". One of the specific aims of the late essays, and in particu-
lar of "Thoughts", however, was precisely to argue that insuperable difficulties
for his theory are not to be expected from this quarter. Even if thoughts are,
as he believed, abstract (and hence nonmental) entities, still they can be grasped:
"This is a process in the inner world of a thinker which may have further con-
sequences in this inner world, and which may also encroach on the sphere of the
will and make itself noticeable in the outer world as well" ([9], pp. 28-29). Into
the precise nature of such events, processes, and states, however, Frege like Witt-
genstein thinks it no part of his brief to enquire: "Both grasping a thought and
making a judgment are acts of a knowing subject, and are to be assigned to psy-
chology" ([10], p. 253).
Frege's theory of thought is not only ambitious in scope, it is quite remark-
ably elegant in structure. The keystone of the edifice is the identification of a
thought with the sense of a declarative sentence. Indeed Frege goes further, and
identifies the structure of the thought with the structure of the sentence which
expresses it, on the grounds that "even a thought grasped ( . . . ) for the very
first time can be put into a form of words which will be understood by some-
one to whom the thought is entirely new. This would be impossible, were we
unable to distinguish parts in the thought corresponding to the parts of the sen-
tence, so that the sentence serves as a model of the structure of the thought" ([7],
p. 55). This principle is then supported, on the one side, by a number of theses
concerning sentence structure, and on the other side, by a number of theses con-
cerning the nature of senses. Thus the essential elements of Frege's theory of
thought can, I believe, be captured in terms of the following six theses:
 THOUGHTS 41

Thesis 1 A thought is isomorphic with the sentence whose sense it is.

Thesis 2 Only in the context of a sentence does a word mean something.
Thesis 3 Every unambiguous sentence has a unique function-argument
analysis.
Thesis 4 The sense of an expression is the condition which anything must
meet if it is the reference of that expression.
Thesis 5 The reference of an expression is that in virtue of whose identity
expressions can be intersubstituted for that expression, salva veritate, throughout
any context of the appropriate kind.
Thesis 6 Λssertoric force is to be sharply distinguished from predicative
power.

Taken together (and duly expanded) these six doctrines yield a theory which in
essence meets all the desiderata that we have so far laid down—or which at least
appears to do so.
Although, as I mentioned above, Frege was generally speaking unconcerned
with problems arising in connection with "interiority", he nevertheless saw clearly
that Theses 1 and 6 together provide a powerful account of the difference
between merely grasping or entertaining a thought, and judging or asserting its
truth. The model for the difference between these mental acts is the distinction
between an assertoric and an unassertoric occurrence of a sentence; and crucial
to this distinction is the difference between the role performed by a predicate
within a sentence, and any force possessed by the sentence as a whole. For if
assertoric force belongs to some subsentential element (whether predicate, cop-
ula, or indeed any other), then a sentence containing that element will ipso facto
possess that force. And yet there is simply no element present within an asser-
toric sentence "P", which is absent from that sentence when it occurs unasserted,
say, in " P D Q". Both occurrences of " P " must possess the same sense, and the
same truth-value, otherwise of course modus ponens fails. Frege concluded that
a thought linguistically presented contains no assertoric element or component,
and possesses a truth-value independently of its making an assertion. Moreover
one and the same thought can be expressed by indefinitely many different sen-
tence tokens. The need for and the justification of these doctrines and distinc-
tions as they concern thoughts (senses) linguistically presented are then extended
by Frege to thoughts as they are mentally presented in acts of thinking and judg-
ing: one and the same thought can be both grasped, nonassertively, and taken
to be true in an act of judgment. Moreover, one and the same thought can be
the content of indefinitely many different mental act tokens. As these mental
acts can be performed by indefinitely many different thinkers, we have here one
of the ingredients of Frege's account of the objectivity of thoughts, i.e., of their
intersubjectivity.
In fact we earlier distinguished three aspects to the problem of the objec-
tivity of thoughts, under the respective heads of "expressibility", "intersubjec-
tivity", and "truth". Now Thesis 1 solves at a stroke problems to do with
expressibility, i.e., with the relation in which a thought stands to a sentence such
42 DAVID BELL

that the latter can embody and communicate the former; for a thought just is
the sense of a declarative sentence. As yet, however, this is little more than a
slogan. It needs to be enriched by a fuller account of sentential sense.
Thesis 2 determines that our theory of sense will indeed be, ab initio, a the-
ory of sentential sense; and Thesis 4 tells us that this theory will be articulated
by appeal to the conditions on being the reference of a sentence. Thesis 5, in
turn, tells us exactly what those conditions are. In normal contexts of direct use
(I ignore for the moment contexts involving modality, direct or indirect quota-
tion), a sentence is intersubstitutable salva veritate with any other which is
materially equivalent to it. According to Thesis 5, therefore, its truth-value is
the reference of such a sentence; and Thesis 4 now yields the result that the sense
of a sentence is the condition on its possessing the truth-value it does possess—its
"truth condition" for short. Frege famously expressed it thus: "Every such name
of a truth-value expresses a sense, a thought. Namely, by our stipulations it is
determined under what conditions the name denotes the True. The sense of this
name—the thought—is the thought that these conditions are fulfilled" ([5], pp.
89-90). But as its sense is what we understand when we know what a sentence
means, we can say more elegantly with Wittgenstein: "To understand a sentence
means to know what is the case if it is true" ([13], 4.024).
We are now in a position to offer a fuller explanation both of the inter-
subjective accessibility of thoughts, and of the possession by them of an objec-
tive truth-value. Indeed, surprisingly, within a Fregean framework it turns out
that these two issues are in fact one and the same.
The possibility of the inter subjectivity of thoughts is guaranteed, Frege
believed, by the nature of their identity conditions: "A thought does not have
to be owned by anyone. The same thought can be grasped by several people"
([10], p. 251). For a thought to be intersubjective, in other words, it is neces-
sary that its identity condition contain no specification of the identity of the per-
son who has it. As we noted earlier, this is perhaps the most crucial difference
between thoughts and, say, sensations or mental images. On the other hand, a
thought is objectively true (or false) insofar as it possesses a truth condition
which can obtain (or not) independently of the existence of, or any subjective
state of, the thinker. Now intuitively these two matters seem quite distinct, and
yet time and again Frege appears to run them together: "In order to be true,
thoughts . . . not only do not need to be recognized by us as true: they do not
have to have been thought by us at all. A law of nature is not invented by us,
but discovered, just as a desolate island in the Arctic Ocean was there long
before anyone set eyes on it" ([10], p. 133). ". . . the thought expressed by the
Pythagorean theorem is timelessly true, independently of whether anyone takes
it to be true. It needs no owner" ([7], p. 17). "In thinking we do not produce
thoughts, we grasp them. For what I am calling thoughts stand in the closest
connection with truth. What I acknowledge as true, I judge to be true quite apart
from my acknowledging its truth . . . therefore that truth cannot have come to
be only upon its discovery" ([7], p. 25). What, one might ask, has the indepen-
dent existence of thoughts got to do with their possessing a content that can be
true or false? Surely my thinking that snow is white is a mental act which, as
such, depends for its existence upon me, the person who performs it; and yet
this seems to be a distinct issue from questions concerning the content of that
 THOUGHTS 43

act, and in particular as to whether the content of that act is capable of being
objectively true or false. It seems to me, however, that Frege is quite consistent
on this point. Because a thought is an abstract object, because it has no mate-
rial properties, the condition on its existence is the same as its condition of iden-
tity, which in turn just is a truth-condition. Quite simply, there is nothing more
to a Fregean thought than its content. But as this content is exhaustively spec-
ified in terms of a truth condition (which makes no essential reference to an
owner or bearer), this in turn comprises its identity condition. For Frege, the
possibility of inter subjective accessibility and possession of objective truth con-
ditions are one and the same.
I turn now to the way in which Frege's theory of thought satisfies the
requirements listed above under the heading of "reflexivity". The outline of his
theory is familiar, so I shall be brief. Theses 4 and 5 tell us that the sense of an
expression is to be explained by appeal to the condition which its reference must
satisfy, and that an expression's reference is to be explained in terms of its inter-
substitutability salva veritate. Now typically we think about a thought (as dis-
tinct from straightforwardly having it) either by predicating something of it ("the
thought that/? is an interesting one"), or by ascribing it to someone ("so-and-
so thinks that/?"). Thesis 5 determines that the reference of an expression in ora-
tio obliqua is its normal sense, for it is identity of sense that guarantees inter-
substitutability salva veritate in such contexts. But the sense of a sentence is a
thought, so the reference of a sentence in an oblique context is the thought which
that sentence would normally express. The sense of such a sentence is, via The-
sis 4, to be elucidated in terms of the identity condition of its normal sense. But
because of the fact that, for thoughts, identity conditions and truth conditions
collapse into one another, it follows that a sentence in oratio obliqua has the
same sense as that sentence in a context of direct use. Frege's theory therefore
captures and explains two important intuitions: first, that the reference of an
expression in an oblique context is different from its reference in a normal con-
text (the rules for substitutivity change); but, secondly, that if one already has
a grasp of the normal sense of some expression, E, then one requires nothing
further to understand occurrences of E within oblique contexts.3 The word
"cats" has exactly the same sense in the sentence "cats purr" as it does in the sen-
tence "John thinks cats purr"; otherwise of course the first sentence would not
express what John thinks.
Finally I come to Frege's contributions to our understanding of the prob-
lems mentioned above concerning "rationality". Thesis 3 is crucial in this respect,
for Frege analyzes both molecular thoughts into their component thoughts, and
atomic thoughts into their component concepts4 in terms of the relation
between a function and its arguments. The function-argument model distin-
guishes between subsentential functions ((£)2; the capital of (£)), sentential
functions ((£) is red; (£) = 2), and truth-functions (If (£) then (f)), between one-
and more-placed functions, and between first- and higher-level functions. The
one model thus provides a unified account of the logical form and semantic role
of singular terms, predicates, relations, logical connectives, and quantifiers. In
a sense this is the engine which drives virtually all of Frege's philosophy; but as
it defies brief summary I shall here assume that the details of the theory are
known.
44 DAVID BELL

In summary, then, Frege's six theses are intended to introduce a univocal

notion of thought that will satisfy requirements (3)-(10) maximally, and require-
ments (1) and (2) minimally. The theory claims, in other words, to explain how
thoughts are related to reality, to language, and to one another. The marriage
of such explanatory power to so simple and elegant a theory would indeed have
been a magnificent achievement. I want now to examine, however, a number
of points at which the theory breaks down.

3 Theses 1 and 3 together imply that every (unambiguous) thought has a deter-
minate function-argument structure, and that this corresponds to the function-
argument structure of the sentence whose sense that thought is. This doctrine
is consistent, however, only as long as sentences with different structures can-
not express one and the same thought. Unfortunately, as Frege himself repeat-
edly acknowledged, there are good reasons for believing that the latter is in fact
very far from impossible. "Let us never forget", he wrote, "that two different
sentences can express the same thought ([10], p. 143).
There are three interpretations of this last remark under which it is true:
two of them are unproblematic for Frege's theory of thoughts; the third, how-
ever, is an acute embarrassment. First, then, the remark is true in the weakest
possible sense, i.e., insofar as different sentence tokens of the same type can
express the same thought. As we noted earlier, this claim is so weak as to be
unobjectionable, and we have already incorporated it into our list of plausible
demands. Secondly, two sentences of different types can express the same
thought if, say, both have the same deep structure or fundamental logical form,
so that their differences result merely from some superficial transformation. The
example which Frege typically cites in this context is the difference between a
sentence in the active voice on the one hand, and its corresponding passive form
on the other. The two sentences

(1) Brutus killed Caesar

and

(2) Caesar was killed by Brutus

can be allowed to express the same thought for they have identical truth con-
ditions, and would indeed receive the same translation into a logically perspic-
uous notation. Frege contends, surely rightly, that the differences between
sentences such as (1) and (2) are insufficient to threaten his theory: ". . .in
speech the same thought can be expressed in different ways, by making now this
proper name, now that one, the grammatical subject. No doubt we shall say that
these differing phrasings are not equivalent. This is true. But we must not for-
get that language does not simply express thoughts; it also imparts a certain tone
or colouring to them. And this can be different even where the thought expressed
is the same" ([10], p. 193).
Before dealing with the third kind of case in which different sentences can
express one and the same thought, we can usefully remove a potential source
of confusion. Frege writes:
 THOUGHTS 45

If several proper names occur in a sentence, the corresponding thought can

be analyzed into a complete and an unsaturated part in different ways. The
sense of each of these proper names can be set up as the complete part over
against the rest of the thought as the unsaturated part. ([10], p. 192)
The possibility Frege seems to be invoking here is that one and the same sen-
tence may possess more than one function-argument structure; and if this is so,
then it clearly contradicts Thesis 3. The sort of case Frege envisages is one in
which a sentence such as (1) can be analyzed into a complete component "Bru-
tus", and an unsaturated function-name "(£) killed Caesar"; but which can
equally be analyzed into a complete component "Caesar", along with the
function-name "Brutus killed (£)"• In fact, however, there is no inconsistency
here, for both these analyses are merely partial. Thesis 3 should be taken to
claim no more than that, in Wittgenstein's words, "A proposition has one and
only one complete analysis" ([13], 3.25; my emphasis). So although a given sen-
tence may be decomposed into function and argument in a number of ways, in
the last analysis it nevertheless possesses a determinate function-argument struc-
ture. In the example above the sentence consists determinately of two complete
signs or proper names, "Brutus" and "Caesar", and a two-place, first-level
function-name: "(£) killed (Γ)". 5
I come now, thirdly, to the sense in which the consistency of Frege's the-
ory of thought is threatened by the possibility of different sentences expressing
one and the same thought—namely, the case in which the sentences in question
have radically different function-argument structures. This is embarrassing not
only because, in general, there are apparently sound intuitive reasons for allow-
ing this possibility, but also, more specifically, because a number of Frege's own
procedures and doctrines depend crucially upon it. I shall give just three exam-
ples. In the Grundlagen Frege writes:

The judgment "line a is parallel to line b, or using symbols: a//b, can be

taken as an identity. If we do this, we obtain the concept of direction, and
say: "the direction of line a is identical with the direction of line b". Thus
we replace the symbol // by the more generic symbol =, through removing
what is specific in the content of the former and dividing it between a and
b. We carve up the content in a way different from the original way, and this
yields us a new concept. ([6], pp. 74-75)

Evidently, then, Frege at this time saw the necessity of allowing that
(3) Line a is parallel to line b
and
(4) The direction of a = the direction of b
are sentences with the same conceptual content. And this doctrine was not given
up with the introduction of the distinction between sense and reference; for in
Grundgesetze ([5], p. 36) the notion of a Wertverlauj'is likewise introduced via
the move from
(5) The functions φ(ξ) and ψ(ζ) have always the same value for the same
argument
46 DAVID BELL

or
(5') -^φ(a)^φ(a)
to
(6) The function φ(ξ) has the same Wertverlauf as the function ψ(ξ)
or
(60 eφ(e) = άψ(a).
And, one last example, according to Frege "affirmation of existence is in fact
nothing but denial of the number nought" ([6], p. 65). In other words,
(5) There exist unicorns
and
(6) The number of unicorns is not zero
express the same thought.
It would appear, then, that there is a fundamental inconsistency in Frege's
theory of thought: it entails that certain thoughts both do and do not possess
a determinate structure, a unique function-argument analysis. Frege might per-
haps console himself (as he did when confronted by another contradiction) with
the thought solatium miseris, socios habuisse malorum: for indeed the problem
is a quite general one, and not one which results merely from idiosyncratic fea-
tures of Frege's own theory. It results, that is, from our wanting to hold two
equally plausible but apparently incompatible principles, namely: (i) that we can
distinguish parts in the thought corresponding to the parts of the sentence
expressing it, so that the sentence serves as a model of the thought, and (ii), that
one and the same thought can be expressed by quite different sentences. Some-
thing, it seems, must be given up here. As I believe that both (i) and (ii) have
sufficient intuitive plausibility to warrant their being saved, my own preference
is to discard requirement B, the requirement that there be a single, univocal
notion of thought capable of simultaneously satisfying each of the conditions
6
we have introduced.
I turn now to the second incoherence which is generated, I believe, by a
Fregean theory of thought. It involves the incompatiblity of, on the one hand,
a condition on the objectivity of thoughts with, on the other hand, a condition
on the possibility of interiority, specifically with condition (C) above, which I
have called the Principle of Spontaneity. In its fully Fregean form the incoher-
ence arises as follows. Thinking is grasping or apprehending a thought, and
a thought is an object. Although Frege himself nowhere explicitly asserts that
thoughts are objects, this follows immediately from his identification of a
thought as the reference of a singular term of the form: "The sense of the sen-
tence 'S' " ([4], p. 59). And yet, of course, a thought is also the sense of a name
of an object, i.e., the sense of a sentence whose reference (if it has one) is a
truth-value. It is about these two roles that Frege advances incompatible theses;
for it is impossible for an object to fulfill the role of a sense ([4], p. 64). Accord-
ing to Frege, to have an object in mind is to have grasped the sense of some
expression which has that object as its reference. But a vicious infinity of such
 THOUGHTS 47

acts of grasping is generated immediately if we maintain, with Frege, that this

sense is, in its turn, merely an object we have in mind; for in this case the sense
would likewise have to be grasped via the sense of some some expression, which
in its turn, as an object, would have to be grasped via the sense of some expres-
sion . . . and so on. Clearly this theory contravenes the formal constraints
expressed in the Principle of Spontaneity: such thoughts would be unthinkable.
In fact the incoherence here is one we guarded against earlier (p. 3): it is the
incoherence of construing a thought (albeit tacitly) as what we think of, that
is, as the object rather than as the content of an act of thinking.
Now it might be thought that there is little of contemporary interest here,
insofar as this difficulty seems to arise from a number of peculiarly Fregean (and
perhaps independently objectionable) theses, such as, e.g., that sentences are
names of objects called truth-values, or that senses are themselves abstract
objects. It might seem, then, that the problem can be circumvented by our sim-
ply abandoning such theses as these. This appearance is, I believe, misleading,
for the difficulty is by no means peculiarly Fregean but is, rather, one that
threatens any attempt to reconcile: (i) what is required if thoughts are to be
objective, with (ii) what is required if interiority is to be possible.
If objectivity of thought is to be possible, then thought must conform to
what Kant called "the universal condition of rules ([11], p. 179 (^4135 =
5174)), that is, its content must be determined solely by its conformity to those
rules and conditions which are, precisely, constitutive of objectivity. On the
other hand, however, if interiority is to be possible, then thinking must be spon-
taneous, in the sense that it must conform to the purely formal Principle of
Spontaneity. In and of themselves these conditions are, I believe, entirely uncon-
tentious; but when we try to amalgamate them, to apply them simultaneously,
we are presented with a situation that appears paradoxical. We seem to have to
take seriously, in other words, the claim that thinking must manifest a rule-
determined spontaneity—and that looks like an oxymoron.
The strangeness of this situation, the tension which exists between the
demand for spontaneity on the one hand, and the requirement that thought be
rule-governed on the other, was discovered, and precisely characterized, by Kant:

If understanding in general is to be viewed as the faculty of rules, judgment

will be the faculty of subsuming under rules; that is, of distinguishing
whether something does or does not stand under a given rule. . . . General
logic contains, and can contain, no rules for judgment. . . . If it sought to
give general instructions how we are to subsume under these rules, that is,
to distinguish whether something does or does not come under them, that
could only be by means of another rule. This in its turn, for the very reason
that it is a rule, again demands guidance from judgment. And thus it appears
that, though understanding is capable of being instructed, and of being
equipped with rules, judgment is a peculiar talent which can be practiced
only, and cannot be taught. ([11], p. 177 (.4132 = £171))

Now the impossibility of there being, in general, rules for the following of
rules has, I believe, profound and far-reaching consequences for our understand-
ing of such notions as thought, judgment, understanding, meaning, objectivity,
and the like. But in the present, Fregean context I shall examine only one such
48 DAVID BELL

consequence: it concerns the form which the analysis of acts of thinking and
judging should take and, in particular, how we should construe the notion of
an object of judgment. This is what Frege says:
We are probably best in accord with ordinary usage if we take a judgment
to be an act of judging, as a leap is an act of leaping. . . . Judging, we may
say, is acknowledging the truth of something; what is acknowledged to be
true can only be a thought. . . . If a judgment is an act, it happens at a cer-
tain time and thereafter belongs to the past. With an act there also belongs
an agent, and we do not know the act completely if we do not know the
agent. ([8], p. 42n)
The passage is interesting for it implicitly contains two incompatible models of
thinking. If we take the analogy with leaping seriously, then the model will
clearly be intransitive: just as there are no "objects of leaping", so there will be
no objects of thought. When I leap I do not enter into a relation with an inde-
pendently existing object called "a leap", and analogously, when I entertain a
thought or pass judgment, I do not enter into a relation with something that
exists independently of my acts of thinking and judging, something called "a
thought". In this case the appropriate schema for the analysis of judgment will
be:
agent + act.
On the other hand, however, Frege here seems tacitly to assume the applicability
of a quite different model according to which we should take the notion of an
object of thought quite literally, and adopt the schema:
agent + act + object.
Although in the Begriffsschrift Frege adopted the former, intransitive model,
he eventually came to believe that objectivity could only be protected by adop-
tion of the latter model, which accordingly came to dominate his account of
thinking. He concluded that "when [a person] grasps or thinks a thought he does
not create it but only comes to stand in a certain relation to what already existed"
([9], p. 18n). But it is precisely this model which i$ untenable, for it is this model
which is incompatible with the possibility of interiority, with the possibility, that
is, of thinking. If a thought were an object, then thinking would be, as Frege
insisted, an act that refers to or picks out such an object: "Although the thought
does not belong with the contents of the thinker's consciousness, there must be
something in his consciousness that is aimed at the thought" ([9], p. 26). But this
imposes structurally the same vicious regressive infinity of conditions on suc-
cessful thinking as the directly analogous account of sentential sense imposes on
the successful expression of thought; for if we claim that a sentence always refers
to its sense then, as Wittgenstein argued in the Tractatus, this would imply that
"whether a sentence had a sense would depend on whether another sentence were
true", and "in that case we could not sketch out any picture of the world (true
or false) ([13], 2.0211-2.0212; my emphasis).
The moral here would seem to be this: adoption of a transitive model of
thinking, one, that is, according to which an ineliminable role is to be ascribed
to the object of thought, will fail to meet the Principle of Spontaneity, and hence
 THOUGHTS 49

will fail to allow for the possibility of what I have called the "interiority" of
thought. It will be incompatible with the possibility that thoughts can be had.
If the objections which I have suggested are valid, then of the original
requirements outlined above (p. 4), Frege's theory of thought and judgment
fails to meet (A) (1) and (2), (B), and (C). And of these, the failure to fulfill
condition (C) is the most serious, for it is this which explains its failure to meet
the others.

NOTES

1. In case this point seems too obvious to deserve mention, it may be worth noting that
the list of philosophers who have, albeit tacitly, contravened this requirement is long
and distinguished: it includes Husserl, Moore, Russell, and indeed Frege himself (see
below).
2. See [15], p. 219. For justification of these remarks, see [1], pp. 133-139.
3. For a similar conclusion, though on different grounds, see [3], pp. 96-97.
4. Clearly I am here using the term "concept" in its normal, i.e., non-Fregean sense,
to mean roughly: nonpropositional component of a proposition.
5. This seems to dispose of Ramsey's "paradox", see [12], pp. 118-119.
6. I have explored elsewhere some of the further reasons for, and some of the conse-
quences of, distinguishing two quite different notions of thought (or sense). See [1],
pp. 112-125.

REFERENCES

[1] Bell, David, Frege's Theory of Judgement, Clarendon Press, Oxford, 1979.
[2] Dummett, M. A. E., Truth and Other Enigmas, Duckworth, London, 1978.
[3] Dummett, M. A. E., The Interpretation of Frege's Philosophy, Duckworth, Lon-
don, 1981.
[4] Frege, G., "On Sense and Reference," in Translations from the Philosophical Writ-
ings ofGottlob Frege, edited by P. T. Geach and M. Black, Blackwell, Oxford,
1960.
[5] Frege, G., "Grundgesetze der Arithmetik," in The Basic Laws of Arithmetic, trans-
lated by M. Furth, University of California Press, 1964.
[6] Frege, G., "Die Grundlagen der Arithmetik," in The Foundations of Arithmetic,
translated by J.L. Austin, Blackwell, Oxford, 1974.
[7] Frege, G., "Compound Thoughts," in Logical Investigations, edited by P. T.
Geach, Blackwell, Oxford, 1977.
[8] Frege, G., "Negation," in Logical Investigations, edited by P. T. Geach, Blackwell,
Oxford, 1977.
[9] Frege, G., "Thoughts," in Logical Investigations, edited by P. T. Geach, Black-
well, Oxford, 1977.
50 DAVID BELL

[10] Frege, G., Posthumous Writings, edited by H. Hermes et aL, translated by P.

Long and R. White, Blackwell, Oxford, 1979.
[11] Kant, Immanuel, Critique of Pure Reason, translated by N. K. Smith, Macmillan,
London, 1933.
[12] Ramsey, F.P., The Foundations of Mathematics, edited by R. B. Braitwaite, Rout-
ledge & Kegan Paul, London, 1931.
[13] Wittgenstein, L., Tractatus Logico-Philosophicus, translated by D. F. Pears and
B. F. McGuinness, Routledge & Kegan Paul, London, 1961.
[14] Wittgenstein, L., Letters to Russell, Keynes and Moore, edited by G. H. von
Wright, Blackwell, Oxford, 1974.
[15] Wittgenstein, L., Philosophical Investigations, translated by G. E. M. Anscombe,
Blackwell, Oxford, 1974.

Department of Philosophy
Sheffield University
Sheffield S10 2TN
Great Britain