Beruflich Dokumente
Kultur Dokumente
AN ANATOMY OF WITTGENSTEIN'S
PICTURE THEORY
The idea of deconstruction is just fine, but it has been given a bad name
by the soi-disant deconstructionists. There are plenty of Humpty-Dumpty
ideas in philosophy and its history, that is, ideas which not only have
several distinguishable ingredients but which are such that they cannot
be put together any longer once the difference between the different
factors is discovered and recognized. There is an abundance of such
concepts and conceptions ripe to be deconstructed which are incom-
parably more important philosophically than the notions with which
Derrida and his ilk have occupied themselves. What I shall do in this
paper is to present a case study in the kind of deconstructive method I
just indicated.
Indeed, it is the main thesis of this paper that what is usually discussed
under the unitarian heading of "Wittgenstein's picture theory" involves
several different and largely independent ideas. The inconclusiveness
of most of the earlier discussions of Wittgenstein's so-called "theory"
is due to a failure to separate these different ingredients from each
other. This failure is to some extent shared by Wittgenstein himself,
for he apparently began to pay serious attention to differences between
the different picture ideas only when he was forced to give some
of them up during his middle period. In fact, discussions of whether
Wittgenstein "gave up the picture theory" in his later philosophy offer
an instructive example of the confusion one inevitably runs into if
one does not distinguish the different components of the syndrome
that usually goes by the name "Wittgenstein's picture theory". Since
Wittgenstein gave up only some of the different picture ideas, it makes
no sense to ask whether he gave up "the picture theory" or not.
To express my thesis more explicitly, I shall argue that Wittgenstein's
so-called "picture theory" is a combination of at least the following
different tenets:
223
C.C. Gould and R.S. Cohen (eds.), Artifacts, Representations and Social Practice, 223-256.
© 1994 Kluwer Academic Publishers.
224 JAAKKO HINTIKKA
(7) The true logic of human language and human thinking is a higher-
order quantificational language.
A more specific assumption which Wittgenstein shares with the
majority of twentieth-century analysts can likewise be considered a part
of the same Frege-Russell syndrome. It is expressed by Wittgenstein
as follows (4.221):
It is obvious that the analysis of propositions must bring us to elementary proposi-
tions which consists of names in immediate combination.
(8) The whole logic of our conceptual system can be reduced to truth-
functional logic.
This thesis is to all practical purposes identical with Proposition 5 of
the Tractatus:
A proposition is a truth-function of elementary propositions.
226 JAAKKO HINTIKKA
Logical forms are not for him independent self-sustaining entities. They
are not constituents of propositions. This is what Wittgenstein means
by such pronouncements as 4.0312:
My fundamental idea is that the logical constants are not representatives; that there can
be no representatives of the logic of facts.
For what the so-called logical constants were supposed to denote are
logical forms (or ingredients thereof).
But what is it in Wittgenstein's theory that can do the job that Russell's
self-sustaining forms were supposed to do? Here the Frege-Russell
paradigm of Begriffsschrift again comes into play. For according to it
the rock bottom of one's universe - the kind of universe which we can
speak of in a Frege-Russell language - is a set of objects, including
both individuals (particulars) and higher-order objects. Hence it is
tempting to use these basic ingredients of a Russellian universe as the
ultimate building-blocks of all logical forms, too. More specifically, it
is natural to think that all other logical forms can be assembled from
the forms of the basic (simple) objects. This is precisely what
Wittgenstein argues for in the Tractatus. In this way, Wittgenstein can
so to speak have his logical forms and dispense with them, too. In other
words, he can have in his theory a class of logical forms all right, viz.
the logical forms of simple (Le., unanalyzable) objects. These forms
are no longer self-sustaining, however. They are forms of objects; they
do not exist independently of the objects whose forms they are; they
are given to me together with the objects whose forms they are.
This simple account shows a crucial difference between Russell's
and Wittgenstein's simple objects (objects of acquaintance). Both were
simple in the sense of being unanalyzable into other, more basic objects.
However, Wittgenstein's objects were not simple in the sense of being
formless.
If you step back and look for a moment at the view thus sketched, you
are likely to have a deja vu experience. What the rigmarole of objects
in a fact being related in the same way as their names in an elemen-
tary proposition amounts to is in effect the first clause of a Tarski-type
truth-definition, viz. the clause for atomic sentences. This clause pre-
supposes that a valuation of the formal language in question is given.
Applied to a sentence like "aRb" it says, roughly speaking, that it is
true iff the relation which is the value v(R) of "R" holds between the
values v(a) and v(b) of "a" and "b", respectively. At first, this looks
precisely the same as Wittgenstein's picture idea as applied to elemen-
tary propositions. On both accounts, "aRb" is true if and only if the
configuration of the linguistic symbols matches the configuration of
the entities in the world they represent.
This comparison between Tarski-type truth-definitions and Wittgen-
stein's picture view brings out what seems to me the most important
idea included in it. As we have seen, this idea is also a direct response
to the historical situation in which Wittgenstein's theories were formed.
In the light of hindsight, however, there is one respect in which the
Tarski-Wittgenstein analogy is highly misleading. To put the point in a
nut-shell, Tarski-type truth-definitions are formulated for one "world"
230 IAAKKO HINTIKKA
It is obvious that thesis (1) does not do the whole job Wittgenstein needed
to be done. He wanted the entire role of Russellian independent logical
forms to be played by the logical forms of simple objects. These logical
forms spell out how objects can be combined with each other. One
particular elementary proposition reflects one such combination. But this
is not enough. Obviously it must be the case that the totality of all
elementary propositions must reflect the totality of possible facts.
WITTGENSTEIN'S PICTURE THEORY 231
use it, viz. in discussing the idea of logic as consisting of mere tautologies
in 6.13 and in discussing all the different forms of propositions in 5.511
and 5.514. How Wittgenstein's notion of tautology and his thesis of the
tautological character of logical truth is connected with the mirroring
thesis (2) will be indicated below in section 12.
But, what is the basis of the mirroring thesis in the Tractatus? Why
does Wittgenstein think that all and only ways of combining names are
matched by possible configurations of objects? Sight unseen, one can
think of different answers to these questions. For instance, it might
be thought that the mirroring thesis is a guideline as to how a truly
representative langauge, perhaps an "ideal langauge", ought to be
constructed. The rules for combining names might be conventional, but
the conventions governing them must be chosen in accordance with the
mirroring thesis. This thesis would then be true by an enlightened fiat,
not by necessity.
This is not Wittgenstein's view, however. There is a deeper founda-
tion for the mirroring thesis (2) - or perhaps a much closer guideline
for choosing our symbolism.
The thesis says that all the possible combinations of simple objects
must be matched by possible combinations of names and vice versa. Now
what determines the totality of combinations of objects? Wittgenstein's
answer is contained in propositions like 2.0123 and 3.315. They show
that what governs the possibilities of combining an object with others
is its logical form. As was emphasized above, this form is given together
with the object.
From this it follows that the mirroring thesis is valid as soon as each
object is represented by a name which has the same logical form as it.
Only then can it be the case that, e.g., "a picture contains the possi-
bility of the situation that it represents".
This, then, is the basis of the mirroring thesis in the Tractatus:
Wittgenstein is in it requiring that each simple object must be represented
by a symbol (name) which shares its logical form. This is precisely the
import of the categorial matching thesis (3).
The consequences of this observation depend on Wittgenstein's notion
of logical form. He does not say very much about it, but some of the
things he says are quite striking. For instance witness 2.0251:
WITTGENSTEIN'S PICTURE THEORY 233
e.g. Phil. Remarks, VII, sec. 68: "The language itself belongs to the
second [i.e. physicalistic] system".)
9. PICTURES BY SIMILARITY?
At first sight a proposition - one set out on the printed page for instance - does not
seem to be a picture of the reality with which it is concerned. But no more does musical
notation at first sign seem to be a picture of music, nor our phonetic notation (the
alphabet) to be a picture of our speech.
And yet these sign-languages prove to be pictures, even in the ordinary sense
[emphasis added] of what they represent.
Thus the mirroring idea and the categorial matching thesis (thesis
(3» are closely intertwined. The latter is a' motivation for the former,
but Wittgenstein was led to it via the same considerations by which he
was led to the picture idea (thesis (1», i.e., via a rejection of Russell's
view of logical forms as independent objects.
It is nevertheless important to note that the core idea of Wittgenstein's
picture view, i.e., the thesis (1), does not entail either of the two theses
(2)-(3). The primary application of picture ideas in the Tractatus thus
is to elementary propositions. It can be extended to names only by
courtesy of further assumptions that are independent of thesis (1).
10. ATOMICITY
essentially what must be the case. Hence some version of the mirroring
idea is clearly in keeping with Leibniz's thinking.
But the ultimate conclusions of Wittgenstein and Leibniz are never-
theless entirely different. For Wittgenstein, the only a priori truths (if
they can be called truths) are tautologies, whereas for Leibniz we can
know a priori a number of things about the structure reflected by monads,
for instance, that this structure exemplifies the simplest (metaphysi-
cally) possible laws.
Where do the two philosophers part company? The answer lies in
Wittgenstein's assumption of a radical contingency, or perhaps rather
atomicity, of the world. The logical form of each simple object governs
the ways it can be combined with other objects. But for two different
objects these possibilities are independent of one another. If a and b
are combined with each other in a certain way, nothing can be inferred
concerning whether c and d are in fact combined in a certain possible
(but not necessary) way. This is the atomicity thesis (4) mentioned above.
As we have seen, it is independent of theses (1)-(3).
In the jargon of logic, the atomicity thesis asserts that different
elementary propositions are independent of each other. From the truth
or falsity of one elementary (atomic) proposition one can never validly
infer the truth or the falsity of another. Metaphorically speaking,
according to the atomicity thesis, the grand mirroring relation, with all
possible total states of the world being mirrored by all possible maxi-
mally consistent sets of propositions, reduces in the Tractatus to a local
mirroring with the totality of elementary propositions reflecting the
totality of basic facts which can obtain or fail to obtain completely
independently of each other. Thus the atomicity thesis (4) is in effect a
sharpened form of the mirroring thesis (3).
Why did Wittgenstein believe in the atomicity thesis? No explicit
answer can be found in the Tractatus, nor is one likely to be deduced
from the text. It is not even clear whether Wittgenstein himself fully
realized at the time of the writing of the Tractatus that the atomicity thesis
is really independent of his other assumptions. Some informed guesses
concerning its background are nevertheless possible. I have argued (with
Merrill B. Hintikka) that the simple objects of the Tractatus were
phenomenological objects, rather like Russell's "objects of acquaintance".
They were not phenomenalistic objects. that is phenomena as dis-
tinguished from reality, but they were given to me in my immediate
experience.
WITTGENSTEIN'S PICTURE THEORY 237
There is only one way of making sense about this situation. It is to realize
that Proposition 6 is Wittgenstein s way of extending the picture idea from
242 JAAKKO HINTIKKA
This last point deserves to be spelled out more fully. It is not only the
case that Wittgenstein could (to his own satisfaction) extend the picture
WITTGENSTEIN'S PICTURE THEORY 245
Here we can also answer a question which may have bothered my readers
for a while about the Russell-Wittgenstein link. A major part of Russell's
philosophical project in 1910-16 was his epistemological and semantical
reduction to acquaintance, or perhaps its mirror image, the logical
construction of the rest of the world out of the objects of acquaintance.
What is the counterpart to this two-way street between objects of
acquaintance and everything else in Wittgenstein? The apparent lack of
a counterpart has been adduced as a reason why Wittgenstein's phi-
losophy in the Tractatus cannot be considered as an extension and
derivation of Russell's theory of acquaintance.
An answer is very simple. Wittgenstein is offering us a linguistic
counterpart of the reduction to acquaintance. That is what his exten-
246 JAAKKO HINTIKKA
sion of the picture theory to the entire language amounts to. What is
essential to the language Wittgenstein envisages in the Tractatus is not
that it is an ideal language or that it is our actual language. What matters
is that it is a Begriffsschrift in Frege's sense: a codification of our entire
conceptual framework. For this reason, Wittgenstein can try to practice
what he preaches and to carry out the entire reductions to acquaintance
as an exercise in the logical syntax of our language or, strictly speaking,
of Wittgenstein's Begriffsschrift. The extension of the picture theory to
the entire language is Wittgenstein's "reduction to acquaintance".
If this is explanation of the Wittgenstein reduction to acquaintance
appears too simple to be true, it is instructive to recall that at one time
Russell thought that he could carry out his reduction to acquaintance
by the sole means of logical analysis. In Russell's case, this analysis
was an application of his theory of denoting, including his theory of
definite descriptions. Toward the end of his famous paper "On Denoting"
Russell writes:
One interesting result of the above theory of denoting is this: when there is anything
with which we do not have immediate acquaintance, but only definition by denoting
phrases, then the propositions in which this thing is introduced by means of a denoting
phrase do not really contain this thing as a constituent, but contain instead the
constituents expressed by the several words of the denoting phrase. Thus in every
proposition we can apprehend (i.e. not only those whose truth or falsehood we can
judge of, but in all that we can think about), all the constituents are really entities
with which we have immediate acquaintance [emphasis added].
But, although the presence of the syntacticity thesis (6) in the Tractatus
is obvious, its precise role is not.
At first sight, thesis (6) of syntactical form as pictorial form does
not play an especially conspicuous role in the Tractatus, but it is
nevertheless important to appreciate it. It is closely connected with the
question as to whether, and if so in what sense, Wittgenstein was
advocating an "ideal language" in the Tractatus. As Frank Ramsey
emphasized against Russell (who nonetheless has since been followed by
a host of interpreters), Wittgenstein's ideas in the Tractatus were not
by any means restricted so as to apply to an ideal language only. The
framework used here enables us to state precisely what the situation is
and in what sense Wittgenstein's "theory" was supposed to apply also
to ordinary language. For Wittgenstein, most of the picture theses apply
both to natural and ideal languages. In particular, in so far as it makes
sense to apply theses (1)-(6) to natural languages, they are applicable.
How can we see this? One way of doing so is to recall that
Wittgenstein's picture view way calculated to apply also to thoughts
and not only to language. Witness 3-3.001, 3.02.
A logical picture of facts is a thought.
'A state of affairs is thinkable' - this means that we can picture it to ourselves.
A thought contains the possibility of the situation of which it is the thought.
What is thinkable is possible too.
The other theses (1)-(5) can likewise be seen to apply to thoughts and
through them to natural languages.
The only exception is the syntacticity thesis (6). This can be seen, e.g.,
from 3.323:
In everyday language it very frequently happens that the same word has different modes
of signification, and so belongs to different symbols - or two words that have
different modes of signification are employed in propositions in what is superficially
the same way.
(The conceptual notation of Frege and Russell is such a language though, it is true,
it fails to exclude all mistakes.) (Tractatus, 3.325.)
The application of logic decides what elementary propositions there are. (5.557.)
The tacit conventions on which the understanding of everyday language depends are
enormously complicated. (4.002.)
Language disguises thought. So much so, that from the outward form of the clothing
it is impossible to infer the form of the thought beneath it. (4.002.)
In fact, all the propositions of our everyday language, just as they stand, are in perfect
logical order. (5.5563.)
These observations are interesting also because they throw sharp light on
other facets of Wittgenstein's philosophy. It is sometimes said that in
the transition from his early philosophy to a succession of later
positions Wittgenstein changed his philosophical methodology. The
suggestion seems to be, in the crudest possible terms, that he changed
from a logician into an ordinary-language analyst.
I shall not argue here that Wittgenstein underwent such a transition
or that he did not. What is clear is that whatever changes there were in
his mode of philosophical methods they were but consequences of
changes in his substantive philosophical views.
250 JAAKKO HINTIKKA
This can be seen from what was said earlier. In his early work
Wittgenstein was concerned with logical forms because he thought that
the syntactical form of a proposition in a logically correct language
contains everything we want to know about it - certainly everything
we need to understand it. This explains Wittgenstein's preoccupation
in the Tractatus with the logical (syntactical) forms of propositions such
as they would be in a logically correct language.
But Wittgenstein later came to believe that the true logical form of
a proposition can only be shown by the way it is used, no matter how
ideal the language is that we are employing. Then the major emphasis
will obviously have to be on those facts Of use and usage which in-
directly show its logical form. And to study them will be to study
"ordinary use". Hence, Wi ttgenstein' s alleged change of methodology
is nothing but a corollary to certain changes in Wittgenstein's actual
substantial views.
Here we also have an illustration of an extremely important point
to be kept in mind in trying to understand Wittgenstein. In spite of all
the changes, sometimes dramatic ones, in Wittgenstein's philosophical
views, almost invariably his later position grew out of seeds implicit
in his earlier ideas. In the present case, the idea that "what signs fail
to express, their application shows" prefigures his later view of meaning
as use.
Similar anticipations of later views can be found elsewhere in
Wittgenstein. For instance, his eventual explanation of how "we can refer
to sensations and to give them names" is anticipated by his ideas about
"the language of gestures" in the early thirties.
If you think of propositions as instructions for making models, their pictorial nature
becomes even clearer.
We might now express ourselves thus: the method of projection mediates between
the drawing and object. - But if the method of projection is a bridge, it is a bridge
which isn't built until the application is made.... What we may call 'picture' is the
blueprint together with the method of its application .... So I am imagining that
the difference between proposition and reality is ironed out by the lines of projec-
tion belonging to the picture, the thought, and that no further room is left for a
method of application, but only for agreement and disagreement.
Thus Wittgenstein is here rejecting only the syntacticity thesis (6), not
the other ingredients of the picture syndrome.
And the idea that "the method of representation" has to be included
in the proposition is not even new for Wittgenstein. It is already asserted
in the Tractatus; witness, e.g., 3.13: "A proposition includes all that
the projection includes, but not what is projected". The only essential
difference is that this "method of representation" now involved (accord-
ing to Wittgenstein) human activities. In the Tractatus, "The method
of projection so to think of the sense of the proposition". Later, it
involves according to Wittgenstein calculus-like activities and eventually
actual applications of language (Cf. the last displayed quotation from
Wittgenstein.) .
We have thus seen how each of the six different thesis (1)-(6) plays a
different role in the Tractatus. Recognizing the differences between
their several contributions to Wittgenstein's line of thought also helps
to clarify essentially the argumentative structure of the Tractatus.
But does this suffice to show that the several theses are really separate?
Perhaps not. However, eloquent further evidence comes from an inter-
esting source: later Wittgenstein. For as I have noted above, in his later
development Wittgenstein rejected some of the theses (1)-(6) but left
others intact. Furthermore, history has rendered her judgement in other
ways. For instance, anyone who believes in a (suitably modalized)
version of Tarski-type truth-definitions in effect accepts thesis (1), but
few, if any, of such analysts in these days accept any theses like (2)
and (3).
To return to Wittgenstein, I have argued elsewhere that the first main
change in his views was to abandon, probably sometime in 1928, the
atomicity thesis (4). This means, trivially, giving up the categorial
WITTGENSTEIN'S PICTURE THEORY 255
matching thesis (3) and the mirroring thesis (2). Less trivially, it means
that the thesis (5) of complex propositions as pictures becomes, not
invalid, but inoperative. It no longer serves the function of showing
that all propositions in a well-formed language are pictures. Moreover,
we saw above that Wittgenstein gave up the syntacticity thesis (6).
Furthermore, because of Wittgenstein's incipient holism "the concept
of an elementary proposition now loses all of its earlier significance",
as Wittgenstein puts it, in Philosophical Remarks I, sec. 83. This seems
to jeopardize thesis (1). Does anything remain about Wittgenstein's earlier
theory?
It is perhaps understandable that many interpreters have taken these
developments to mean that Wittgenstein "abandoned the picture theory".
A closer look shows something quite different, however. It shows
Wittgenstein's remarkable persistence in clinging to the picture idea.
We saw instances of the persistence earlier in sections 9 and 19.
However, the way in which Wittgenstein was able to salvage some
of his picture view has not yet been diagnosed fully. What happened
was that by giving up the syntacticity thesis (5) Wittgenstein was able
to save some extra ingredients of the picture idea. Even though a propo-
sition as a syntactical entity cannot be construed as a picture, the situation
is different if we take it together with the activities that connect it with
the world. These activities are what Wittgenstein calls a "method of
projection". Then a proposition could still be a picture of a fact, albeit
not a "picture by similarity" (cf. sec. 9 above).
Another opening for maintaining the picture idea and even extending
it was created by Wittgenstein's abandoning the idea of the logical
forms of simple names which is the main presupposition of thesis (3).
By giving up this idea, Wittgenstein made it possible for himself to
acknowledge pictures which are not pictures by similarity, as we saw
in sec. 9 above. In other words, the names could now be related to their
objects conventionally, without presupposing any intrinsic identity of
logical forms. (Of course, the logical form of propositions could still
reflect the structure of the corresponding facts.)
This enables Wittgenstein to extend the picture idea from propositions
whose names represent definite objects to propositions where they stand
for indefinite objects, e.g. objects of a certain kind. This is what is
involved in later Wittgenstein's distinction between "historical pictures"
and "genre pictures". This pair of contrasting concepts is one of the
recurring themes in Wittgenstein's later thought. (See, e.g., Philosophical
256 JAAKKO HINTIKKA
Boston University