My topic of
concern is the logical form called predication not the relation--though the two
notions surely are closely related. In particular the concern here is with the
role that logical form has played in ontology in this century. So it is
appropriate to begin with some historical remarks designed to show the
philosophical importance of the connection between predication and what there
is.
One mark of unusual philosophical capacity is the
ability to see the
significance
of a premise in philosophical reasoning that others take for
granted.
Russell had this capacity, and had he never done anything else, it would have marked
him as an outstanding thinker. In the twentieth century it was Russell who
best dramatized the significance of premises about logical form in arguments
about what there is.[1] Concerning those who seemed to take little
notice of this kind of premise in ontological reasoning, Russell could be
scathing.
Consider, for instance, his denunciation of Hegel's
notions of identity in difference and the concrete universal. He wrote:
Hegel's argument in this portion his “Logic” depends throughout upon confusing
the “is” of predication, as in “Socrates is mortal” , with the “is” of identity, as in “Socrates is the philosopher who drank the hemlock”
. Owing to this confusion, he thinks
that “Socrates” and “mortal” must be identical. Seeing that they are different,
he does not infer, as others would, that there is a mistake somewhere, but that
they exhibit “identity in difference”. Again, Socrates is particular, “mortal”
is universal. Therefore, he says, since
Socrates is mortal, it follows that the particular is the universal-- taking
the “is” to be throughout
expressive of identity. But to say “the
particular is the universal” is self-contradictory. Again Hegel does not suspect a mistake but proceeds to synthesize
particular and universal in the individual, or concrete universal. This is an example of how, for want of care
at the start, vast and imposing systems of philosophy are built upon stupid and
trivial confusions, which, but for the almost incredible fact that they are
unintentional, one would be tempted to characterize
as puns.2
Among those who acknowledged the importance of premises
about
logical form in ontological reasoning but who took
certain cases for
granted was Meinong--or so Russell thought anyway.
Although admiring
Meinong's audacity--Russell once said, vis a vis Meinong
and the round
square,
It is not
customary for philosophers to face the round square
with so much courage; and indeed few logicians
can withstand its onset3
--although admitting Meinong's audacity, Russell
believed Meinong to
have been badly misled because of a mistake about
logical form. This is made clear in the
chapter on descriptions in An introduction to Mathe-
matical Philosophy, but his attitude
was already evident in his famous essay of a decade and a half earlier, ‘On
Denoting’ . Here is what he had to say
about the consequences of Meinong's uncritical view that not only statements such
as
Vulcan is a
planet,
and
The round nonround thing does not exist
are predications,
but also statements such as
The round
nonround thing is round.
He writes, first, in
‘On Denoting’
The difficulties concerning denoting are, I
believe, all the result of the wrong analysis of propositions whose verbal
expressions contain denoting phrases.4
Now definite descriptions are denoting phrases, and
in ‘On Denoting’
Meinong is Russell's first example of how one can go
wrong by mis-analy-
zing statements containing denoting phrases. Meinong, Russell reports,
regards definite descriptions as genuine subject terms
in statements such
as
The round nonround thing is round.
That is, in Russellian language, Meinong is reported
(correctly) as regarding this statement as a statement of the form
x is round
and the expression
The round
nonround thing
as its subject term. But, says Russell, such a view is
disastrous because then the denotation of the definite description ‘the round nonround thing’ must be taken to
be a full-fledged object; “yet such objects, admittedly, are apt to infringe
the law of contradiction”.5
In the case
of Meinong it is important to realize that it is not the
actual truth-value Meinong attached to these statements
that is crucial.
Meinong
counted all three statements true, but even if he had thought they were all false, one could still have deduced the round
nonround thing and Vulcan. For the
essential mark of a predication in the case of both for Russell and Meinong is
that the subject terms in the verbal
expression of the proposition-like entity constituting the meaning of
that verbal expression (a statement)
stand for constituents of that proposition-like entity and not whether
the statement is true or false.
It is not
difficult to find current philosophers who have stressed the importance of
predication in other arguments other than ontological arguments. For instance, in the philosophy of science,
Jules Vuillemin, if I have understood him correctly, believes that the
Einstein-Podolsky-Rosen paradox and the Bell inequalities demand abandonment of
certain aspects of the traditional notion of predication. And, of course, in the philosophy of
language there is Quine who believes that whether to treat statements containing psychological verbs as
referentially transparent is essentially the question whether to treat the
statement containing that verb as a predication.6
To return
to ontological reasoning, there is one influential contemporary philosopher who
can only be interpreted as rejecting the Russellian and Meinongian belief that
the predicational character of statements like the
The round
nonround thing is round
provides a route by which objects such as the thing
which is round but
nonround can sneak into the universe. That philosopher
is W. V. Quine.
I don't mean that Quine would deny that Russell or
Meinong drew the
correct conclusions about the objecthood of the thing
which is round but
nonround from statements such as ‘The round nonround thing is round’
given their understanding of predication. What I mean is that Quine can
only be interpreted as believing both Russell and
Meinong to have misun-
derstood predication itself, and that when properly
understood, nothing
about the object-hood of the round nonround thing
follows from the fact
that the statement ‘The round nonround thing is
round’ is a predication or at least from that fact alone.
Indeed, Quine has never been persuaded that definite descriptions are not
singular terms, the essence of Russell's diagnosis about how Meinong went wrong
in classifying statements like ‘The
round nonround thing is round’ as predications. So before regimentation of the language it is Quine's position
that though Meinong was right about regarding the statement in question as a
predication, ontological commitment to the round nonround thing is demanded
rather by other aspects of Meinong's theory, for example, by the truth of the
non-atomic statement
There is an
object the same as the round nonround thing.
What is the
evidence for this view of Quine's position?
Let me cite just three pieces.
First, there is simply no question that Quine, in contrast to Russell,
regards definite descriptions, fulfilled or unfulfilled, as genuine singular
terms.7 Second, even in early papers such
as ‘Designation and existence’ and
earlier books such as Methods of Logic, Quine holds that
the inference pattern called Particularization is valid only when a
singular term constituting the subject term of a predication refers. Finally,
his definition of predication in Word and Object
specifically allows statements containing irreferential singular terms to be
predications.8 In general, in contrast
to Meinong and Russell, Quine does not take predication--at least simple (or
atomic) predications--to be ontically committing. His view might be put this
way (though he would find the ensuing language probably unsatisfactory): the statement ‘The round nonround thing is round’ qualifies as a predication because its truth-value, true or
false, would depend on whether the general term round
is true (or false) of the referent ofthe singular
term ‘The round nonround thing’ were
that singular term to refer. And the
same holds for, say, ‘Vulcan has no
oxygen in its atmosphere’. By contrast, in the Meinong-Russell conception there
is no would or were about it because genuine singular terms always
refer.
What
accounts for these different conceptions of predication ? It has to do with propositions, and in
particular with what David Kaplan and Kit Fine, among others, call Russellian singular propositions--or that
kind of thing. For Meinong and Russell
all meaningful sentences mean proposition-like entities. Meinong called them ‘objectives’ and Russell
called them ‘propositions’ (and
sometimes ‘facts’) . (By the way, Russell's notion of proposition derives from
Meinong's notion of objective via G. E. Moore.9) The point is that the notion of a statement
having the logical form of a predication is a derivative notion based on
the fundamental notion of a proposition (or objective) having the logical form
of predication. In short, a sentence
containing singular terms and a general term, simple or complex, will be a
predication just in case the proposition (or objective) expressed by that
statement is a predication, that is, just in case it is a singular proposition
(or singular objective). Neither
Meinong nor Russell believed that propositions (or objectives) could have holes
in them. So Meinong, working from the language down to the world, and believing
the statement ‘The round nonround thing
is round’ to be a predication, concluded that there must be singular
propositions (objectives). Russell,
working from the world up to language, thought there could be no singular
propositions with the round nonround thing as constituent and thus concluded
that the statement ‘The round nonround
thing is round’ is not a predication.
Quine, on the other hand, does not
believe in propositions, be they
construed
as meanings or merely as truth-value makers.
For him, predi-
cation
as a logical form of statements--or as he would prefer to call it, a ‘construction’-- is not a derivative notion
at all; it is the primary one. There is
no compulsion, philosophical or otherwise, to find referents for prima facie
irreferential expressions like ‘Vulcan’ or
‘The round nonround thing’ , or to look for a deeper (or at least a
different)
logical
structure in statements such as ‘Vulcan
is a planet’ or ‘The
round
nonround thing is round’ than the
predicational form suggested by
the
logical grammar in the traditional logic of terms.
This is
perfectly consistent with the ultimate adoption of what he calls ‘the canonical
language’ because it is his pre-regimentation view that is at issue now. That the only predications in the canonical
language should consist of sentences joining general terms to free variables does
not conflict with the view that in the pre-regimented language statements such
as ‘Vulcan spins’ should be regarded as predications. Indeed, it is partly
because of the problems associated with such predications, for example,
truth-value gaps, and, as shown elsewhere, the threat of nonextensionality10,
that preference for the narrower canonical language is, in part, motivated.
I don't mean to suggest that the issue of whether
predication is ontically committing need be tied in any essential way with
propositions or that sort of thing. It
is not. For instance, the issue arises
in free logic
where
talk of propositions seems not to have arisen in any important way.
Free logics are logics that do for the logic of singular
terms what
symbolic
logic since Boole and Schröder did for the traditional logic of
general
terms; they make the existence assumptions in classical
reasoning
involving singular terms explicit. In
this kind of logic there are
two
ways of thinking about expressions such as
‘Vulcan’ and ‘The
object
which is both round and nonround’ . One
way is to treat them in
a
Meinong-like fashion, that is, as referential but not referring to existent
objects. The other way is to treat them as not
referring to anything,
existent
or otherwise. The question now is: what
motivates these diffe-
rent
policies about how to treat singular terms such as the grammatically
proper
name ‘Vulcan’ --the name of the
putative planet, not the name of the god--and the definite description ‘the object which is both round
and
not round’ ? This very issue came to the surface a few years ago in a
dispute
between Richard Grandy, who favored the Meinong-like approach, and Tyler Burge,
who favored the pre-regimented Quine-like intuition. Grandy, as a matter of
fact, located the difference between himself and Burge precisely in different
concepts of predication11.
Thus consider the statement
Vulcan is Vulcan .
Both
consider this statement a (relational) predication, but only Grandy felt
compelled to find a referent for Vulcan (though a nonexistent one). I
think
Grandy is right in his assessment, though not necessarily in his semantic
conclusion. It seems abundantly clear
from an inspection of Burge's essay[2]
that he (Burge), in contrast to Grandy, thinks it bad policy to assimilate all
relational predications to the
kick,scratch, bite and hammer model in which the relata actively modify each
other (to paraphrase words by van
Fraassen and me written in another place at an earlier time[3]). And the same sentiment holds for
non-relational predications such as
‘Vulcan spins’ and ‘The round but nonround thing is round’. So the issue
about how best to conceive predication arises in free logic where all hands are
in agreement that it is statements (as opposed to propositions) that are the
primary vehicles of predication.
Ignoring Healy's first law of holes—the law that says
that when you are in one, stop digging—let me plunge in deeper. First, one
should not jump to the conclusion that Meinong's opinion that statements such
as ‘Vulcan spins’ and
‘The round nonround thing
is round’ are predications is absurd. Russell's argument that the second of these
statements violates the law of non-contradiction is questionable as is now
widely acknowledged and Meinong should not have been so fast to agree with
Russell's critique[4].
Moreover, there are now various adequate Meinong-like theories around that are
provably consistent contra Gilbert Ryle.
(By adequate, I mean a theory that does not identify all nonexistents as
does, say, Scott's version (vis a vis definite descriptions) of Frege's chosen
object theory). Terry Parsons
has got one, though that theory does not go quite far enough.[5] The round square is in the ontology of
Parson's theory but not the round nonround thing. Still, since his theory breaths on the shadowy neck of the round
nonround thing, it would be foolhardy now to say that a natural theory
countenancing a distinctive round but nonround object is bound to fail.
Secondly, I think it is quite clear that the two
concepts of pre-
dication
I have discussed, the ontically loaded one or its neutral competitor, deserve
much closer study to see which option, philosophically speaking, is the
better. I know that Quine has made
up
his mind in favor of the former notion; so much is reflected in his decision
about the character of the canonical language.
But that decision relies heavily on his elimination program for singular
terms. It is far from clear that that program is feasible. I think Quine's
elimination program
is at least very puzzling, and at most involves a significant loss of explanatory power.
More importantly, I think Quine's decision about the character of the
canonical language is precipitous. We
don't know enough about the consequences of these two policies about predication
to make a sound judgment. And this
brings me to my last point, a plea for further research.
As far as I know there is no treatment of predication
rigorous
enough
to enable us to see clearly what the implications of the two
policies
are. Quine has made interesting but
undeveloped proposals; so
has
Arthur Prior. Perhaps there lurks somewhere in the formal work of Michael Dunn
and his associates on relevant predication decisive factors in favor of one of
these theories over the other. I do not know.
So, I conclude, we need to search for more--and moreover more formal
arguments--to help us see what the relation between predication and what there
is really is.
2
Bertrand Russell,
Our Knowledge of
the External World, George
Allen and
Unwin Ltd., 1914; Fifth
Impression, 1969, p. 42.
3 Review of A. Meinong's
Untersuchungen zur Gegenstandstheorie und Psychologie,
Mind,
NS 14,1905, p. 537.
4 ‘On Denoting’
in Logic and Knowledge (Ed. R. C.
Marsh), 4th Impression, Allen
and Unwin, 1968, p. 43.
5
Ibid., p. 45.
7 Ibid., Word and Object. See the chapter on
Regimentation.
8 Ibid., Word and Object, p.
96.
9 Bertrand Russell,
‘Meinong's theory of
assumptions and complexes’, Mind,1904, p. 206.
10 Karel Lambert,
‘Predication and extensionality’, Journal of Philosophical Logic, 3,
1974, pp. 255-64
11 Richard Grandy, ‘Predication
and singular terms’, Noûs, 11 ,1977, pp. 163-67
[2] Tyler Burge, ‘Singular terms and truth’, Noûs,
8, 1974, pp. 309-25
[3]
Karel Lambert
and Bas van
Fraassen, Derivation and
Counterexample,
Dickensen, Encino, CA, 1972, p. 213.
[4]
See, for example, Terence Parsons, Nonexistent Objects, Yale
University Press, New Haven, 1980.
[5]
Ibid., Nonexistent Objects, 1980.