Professional Documents
Culture Documents
(Kit Fine (Auth.), Andrea Bottani, Massimiliano CA
(Kit Fine (Auth.), Andrea Bottani, Massimiliano CA
TOPOl LIBRARY
VOLUME 4
Managing Editor:
Ermanno Bencivenga, University of California, Irvine, U.S.A.
Editorial Board:
Daniel Berthold-Bond, Bard College, AnnantkUe-on-Hudson, U.S.A.
William James Earle, Baruch College, City University of New York, New York, U.S.A.
Ann Ferguson, University of Massachusetts, Amherst, U.S.A.
David Lloyd, Scripps College, Claremont, U.S.A.
Topoi Library is sponsored by the Department of Philosophy and the School of Humanities at
the University of California, Irvine
Scope:
Like the journal TOPOl, the TOPOl Library is based on the assumption that philosophy is a lively,
provocative, delightful activity, which constantly challenges our inherited habits, painstakingly
elaborates on how things could be different, in other stories, in counterfactual situations, in
alternative possible worlds. Whatever its ideology, whether with the intent of uncovering a truer
structure of reality or of shooting our anxiety, of exposing myths or of following them through, the
outcome of philosophical activity is always the destabilizing, unsettling generation of doubts, of
objections, of criticisms.
It follows that this activity is intrinsically a dialogue, that philosophy is first and foremost
philosophical discussion, that it requires bringing out conflicting points of view, paying careful,
sympathetic attention to their structure, and using this dialectic to articulate one's approach, to
make it richer, more thoughtful, more open to variation and play. And it follows that the spirit
which one brings to this activity must be one of tolerance, of always suspecting one's own
blindness and consequently looking with unbiased eye in every comer, without fearing to pass a
(fallible) judgment on what is there but also without failing to show interest and respect.
It is no rhetoric then to say that the TOPOl Library has no affiliation to any philosophical school
or jargon, that its only policy is to publish exciting, original, carefully reasoned works, and that its
main ambition is to generate serious and responsible exchanges among different traditions, to have
disparate intellectual tools encounter and cross-fertilize each other, to contribute not so much to
the notarization of yesterday's syntheses but rather to the blossoming of tomorrow's.
INDIVIDUALS,
ESSENCE AND IDENTITY
Themes of Analytic Metaphysics
edited by
ANDREA BaITANI
Universita di Bergamo, Italy
MASSIMILIANO CARRARA
Universitiz di Padova, Italy
and
PIERDANIELE GIARETIA
Universita di Padova, Italy
Introduction vii
Part One: Ontology and Analysis
KIT FINE / The Question of Realism 3
ACHILLE C. VARZI / Words and Objects 49
Part Two: Essence and Existence
ENRICO BERTI / Being and Essence in Contemporary
Interpretations of Aristotle 79
DAVID CHARLES / Some Comments on Prof. Enrico Berti's
Paper "Being and Essence in Contemporary Interpretations of
Aristotle" 109
ALEX ORENSTEIN / Existence, Identity, and an Aristotelian
Tradition 127
MAURO MARIANI/Orenstein on Existence and Identity 151
STEPHEN YABLO / Abstract Objects: A Case Study 163
E. JONATHAN LOWE / Kinds, Essence,
and Natural Necessity 189
KATHE TRETTIN / Kinds of Necessity: a Commentary
on EJ. Lowe's Paper 207
Bibliography 479
Andrea Bottani
Massimiliano Carrara
Pierdaniele Giaretta
one has to look at what kinds of entities are quantified over through
expressions like "there is" and "everything". It is the criterion of on-
tological commitment, also expressed as "to be is to be a value of a
bound variable" (Quine [1953], pp.14-5) The criterion is only a test for
detecting what entities one is committed to. Quine's criterion has been
widely accepted, but has generated also discussion concerning the re-
lation between existence as expressed by "there is" and other possible
senses of existence. As we have already said, not everyone has ac-
cepted that the sense of existence is unique and is expressed by the
existential quantifier.
The other principal innovations brought to the fore by modem
analytic research in ontology concern identity. In analytic metaphysics
it is taken into account from three new points of view. First, identity is
seen as an ontological relation presupposed by our basic practice of
identifying entities. This kind of topic was initiated by Strawson and
its ontological implications more fully pursued by Wiggins. The most
original aspect of this topic is given by the connection of identity with
our way of conceptual ising entities. Second, in this context, Wiggins
took identity as a relation that primitively applies also to entities
picked out at different times. So identity, sortally qualified, grounds
both identifying and reidentifying practices. Third, a methodological
question concerning identity was raised by Quine: the clarity with
which identity can be explicated, in short the definiteness of identity
criteria, confers ontological legitimacy on the entities for which the
identity criteria are stated. As we have already said, not all these theses
have been fully accepted, but each has constituted a new and original
topic for metaphysics.
These are some themes of "analytic metaphysics". But they are not
the only ones. Other themes can be mentioned which are as fully de-
bated even if not completely new. One of these is, for example, the
distinction between intrinsic and extrinsic properties.
However, rather than mention the specific topics dealt with by
analytic philosophers, it could be useful to give a glimpse of the way in
which metaphysics is taken and pursued by them. Let us try to formu-
late some ingredients of a provisional picture.
xii Individuals, Essence, and Identity
All these questions directly relate to the notion of entity, which, since
Quine, has been naturally connected to (indeed, elucidated in terms of)
that of identity. So, the second section turns to the concept of identity.
Can identity be reduced to something else (indiscernibility, community
of parts or some other relation)? Can '=' be eliminated from our lan-
guage with no semantic loss? And what kind of relation is identity?
What connections are there - if any - between identity and sortal
predication? Can identity itself be vague? Can any objects be neither
identical nor different? The third section enlarges the field, bringing to
the fore modal themes and issues. What is essence? What is necessity?
What is a natural law? What do natural laws have to do with natural
kinds? Since the relation between essence and existence is one of the
central themes of ancient metaphysics, the questions of the nature of
existence and of the meaning of the existential quantifier find a place
in this section. In the last section, all the topics and issues discussed in
the previous sections find an application - though surely a very central
one - to the problem of identity, change and persistence across time.
The theme is taken up over a wide range of questions, concerning the
existence of temporal parts of ordinary objects like cats or trees, the
possibility of a genuine identity across time, the compatibility between
ordinary change and Leibniz's Law and some lessdiscussed issues such
as the possibility of intermittent existence.
According to some metaphysicians, the identity of a thing is given
by its origin. The origin of this book was a conference with the same
title, held in Bergamo (Italy) on June 22-24, 2000. Though the majority
of the papers published in the volume derive from the talks given at the
conference - and conversely the majority of the talks given at the con-
ference have become papers in this volume - this is not a volume of
proceedings, because a few of the contributors to the volume did not
take part in the conference.
We are indebted to many people for different reasons. First of all
we thank the authors who contributed to the conference or to the vol-
ume. We think to have been very lucky to gather so many distin-
guished philosophers together. Secondly, thanks to Alberto Castoldi,
rector of the University of Bergamo, for the firm support given to the
conference from which the book originated. Thanks also to Vittorio
Morato, Marzia Soavi, and Elisabetta Bonadeo: they were very helpful
Introduction xv
during the conference. Vittorio Morato did also almost all the work for
the camera-ready copy of the book. Carlo Nizzo checked most of the
editing work. Richard Davies made some corrections and gave some
advice about the English of some papers. Many thanks to them.
Thanks to the anonimous reader for the Kluwer Publisher for a wide
range of useful comments. And last, but not least, thanks to the series
editor of "Topoi Library" Ermanno Bencivenga for having supported
this long-standing enterprise.
The preparation of this book and the related conference have been
made possible by grants from the Department of Education and Com-
munication Sciences of the University of Bergamo and from a national
research project on "Knowledge and Cognition", coordinated by Paolo
Parrini (University of Firenze) - more specifically, from the local proj-
ects, on "Language, Thought, and Normativity", coordinated by Diego
Marconi (University of V ercelli) and on "Reference and Thought" co-
ordinated by Pierdaniele Giaretta (University of Padova). The projects
were sponsored by the Italian Ministry for University and Scientific
Research (MURST).
Notes
1 Chisholm [1996], p.8.
Part One
Kit Fine
3
A. Bottani et al. (eels. J. Individuals. Essence and Identity. 3-48.
© 2002 Kluwer Academic Publishers.
4 Individuals, Essence, and Identity
conclude with some remarks on the essential unity of these two ques-
tions and of the means by which they are to be answered (§1O).
1. REALITY
conflict or not? And if not, then how is the apparent conflict between
them to be dispelled?
If we take the conflict to be genuine, we obtain what has been
called an "eliminative" or "skeptical" conception of antirealism. The
antirealist will be taken to dispute what we ordinarily accept, the real-
ist to endorse it. Thus the antirealist about numbers will be taken to
deny, or to doubt, that there are prime numbers between 2 and 6; and
likewise, the moral antirealist will be taken to deny, or to doubt, that
killing babies for fun is wrong.
Of course, the mere rejection of what we ordinarily accept is per-
verse and so presumably the interest of antirealism, on this conception,
must derive from the assumption that philosophy is able to provide us
with some special reasons for dOUbting what we ordinarily accept.
Thus the antirealist may attempt to convince us that we have no good
reason to believe in a realm of abstract objects with which we can have
no causal contact or that, in moral matters, we can have no justification
for going beyond the mere expression of approval or disapproval. Our
world-view will therefore be the product of dealing with these doubts,
either by laying them to rest or by retreating into skepticism.
Anti-realism, as so understood, has a long and illustrious history;
and certainly its interest is not to be denied. However, in this age of
post-Moorean modesty, many of us are inclined to doubt that philoso-
phy is in possession of arguments that might genuinely serve to un-
dermine what we ordinarily believe. It may perhaps be conceded that
the arguments of the skeptic appear to be utterly compelling; but the
Mooreans among us will hold that the very plausibility of our ordinary
beliefs is reason enough for supposing that there must be something
wrong in the skeptic's arguments, even if we are unable to say what it
is. In so far, then, as the pretensions of philosophy to provide a world-
view rest upon its claim to be in possession of the epistemological high
ground, those pretensions had better be given up.
Is there room for another form of antirealism - and another ac-
count of philosophy's pretensions - that does not put them in conflict
with received opinion? If there is, then it requires that we be able con-
sistently to affirm that something is the case and yet deny that it is
really the case. I It requires, in other words, a metaphysical conception
of reality, one that enables us to distinguish, within the sphere of what
6 Individuals, Essence, and Identity
is the case, between what is really the case and what is only apparently
the case.
But what might this metaphysical conception of reality be? Two
main answers to this question have been proposed. According to the
first, metaphysical reality is to be identified with what is "objective" or
"factual". The antirealist, on this conception, denies that there are any
facts "out there" in virtue of which the propositions of a given domain
might be true. The propositions of the domain are not in the "business"
of stating such facts; they serve merely to indicate our engagement
with the world without stating, in objective fashion, how the world is.
As familiar examples of such a position, we have expressivism in eth-
ics, according to which ethical judgements are mere expressions of at-
titude; formalism in mathematics, according to which mathematical
statements are mere moves within a system of formal rules; and in-
strumentalism in science, according to which scientific theories are
mere devices for the prediction and control of our environment. Ac-
cording to the second conception, metaphysical reality is to be identi-
fied with what is "irreducible" or "fundamental". On this view, reality
is constituted by certain irreducible or fundamental facts; and in de-
nying reality to a given domain, the antirealist is claiming that its facts
are all reducible to facts of some other sort. Thus the ethical naturalist
will claim that every ethical fact is reducible to naturalistic facts, the
logicist that every mathematical fact is reducible to facts of logic, and
the phenomenalist that every fact about the external world is reducible
to facts about our sense-data.
We might see the anti factualist and reductionist as indicating two
different ways in which a proposition may fail to "correspond" to the
facts. For it may fail even to point in the direction of the facts, as it
were; or it may fail to indicate, at the most fundamental level, how the
facts are. In the one case, the propositions of a given domain will not
even represent the facts, while in the other, the propositions will not
perspicuously represent the facts - there will be some divergence be-
tween how the facts are "in themselves" and how they are represented
as being. If either of these metaphysical conceptions of reality is vi-
able, then it would appear to provide a way of upholding a non-
skeptical form of antirealism. For it will be perfectly compatible with
affirming any given proposition to deny that it is genuinely factual or
Kit Fine, The Question of Realism 7
2. FACTUALITY
linguistic contexts. But the same point applies. For in the ordinary
sense of 'believe', 'assert', etc., we do have moral beliefs and make
moral assertions, we do draw moral conclusions, and we do embed
moral propositions in larger linguistic contexts; and similarly for the
propositions of mathematics or of science or of other disputed areas.
Indeed, once given one of these characteristics, the rest seem to follow
- their possession is, for the most part, a "package deal". There there-
fore seems to be no reasonable hope of identifying a non-skeptical
form of factuality in terms of the possession of some of these charac-
teristics as opposed to the others.
Any reasonable, non-skeptical view should therefore grant that
propositions of the kind that figure in realist disputes will possess all
of the obvious trappings of factuality: they will be capable of being
true or false, believed or asserted, embedded in larger linguistic con-
texts, and so on. The antifactualist should therefore be a quasirealist
and attribute to the nonfactual all those features that were traditionally
thought to belong to the factual. But if the nonfactual is not to be dis-
tinguished from the factual in terms of the obvious trappings of factu-
ality, then how is it to be distinguished? What, in a word, is the differ-
ence between quasi realism and genuine realism?
Various more sophisticated criteria that appear to avoid these dif-
ficulties have been proposed. 4 I shall focus on one, Dummett's, ac-
cording to which realism for a given area of discourse is primarily a
matter of its conforming to the Principle of Bivalence, the principle
that every statement of the discourse should be either true or false. s I
shall argue that the criterion, even when supplemented, is unsatisfac-
tory, and then attempt to draw some broader conclusions.
We should note right away that the proposal, even if otherwise ac-
ceptable, does not answer to our needs. For we were after a non-
skeptical form of antirealism, one that was not at odds with received
opinion. But in regard to many areas of discourse, the received opinion
is that the statements are subject to Bivalence; and so any form of
Dummettian antirealism must to that extent be skeptical. Thus in so far
as philosophers have wished to espouse a completely non-skeptical
form of antirealism, the Dummettian criterion must be considered un-
satisfactory.
10 Individuals, Essence, and Identity
ence, but he is not a realist about the external world, which is what is
here in question.
The second strategy is to provide a semantic supplement to Biva-
lence: not only must Bivalence hold, but it must be a semantical matter
that it holds. But our previous counter-example still stands, since the
validity of Bivalence is a semantical matter under the antirealist se-
mantics that I proposed (given that there is a semantical underpinning
for applying the Law of the Excluded Middle to statements of war-
rant). It therefore appears necessary to go deeper into the mechanism
by which the language is to be interpreted. But what might that be?
One suggestion is that we require "acceptance of classical two-valued
semantics [... ] in its entirety" (Dummett [1993], p.468). This would
require not only Bivalence but the exclusion of empty terms, the stan-
dard clauses for the connectives, and so on. But the problem now is
that, given that our antirealist is willing to accept Bivalence, it is not
clear why he should be unwilling to accept the rest of the classical se-
mantics - though, of course, under his own understanding of what this
comes to. Another suggestion is that we require that our understanding
of the language should be truth-conditional, that our grasp of the
meaning of a statement should consist in knowledge of its truth-
conditions. But although philosophers use this phraseology as if they
knew what it meant, it is not at all clear that it can be explained in such
a way as to both imply realism and yet not presuppose that the truth-
conditions are already to be understood in realist fashion. One can, of
course, insist that the relevant notion of truth should conform to Biva-
lence or be evidence-transcendent. But this then leads us back to our
previous difficulties.
What, I believe, has made these various criteria so appealing is
that it is often hard to see how one could plausibly maintain that a
given criterion is satisfied (or not satisfied) and yet still be a realist (or
antirealist). Realism about mathematics, for example, is a reason - and
perhaps the only good reason - for holding that every mathematical
statement is true or false or that there might be mathematical truths
that are beyond our ken. But it should be recognized that, even though
the existence of an external reality may make it plausible that our lin-
guistic and epistemic contact with that reality is of a certain sort, this is
not in what the externality of the reality consists. In thinking about
Kit Fine, The Question of Realism 13
3. REDUCIBILITY
We tum to the second of the two metaphysical conceptions of reality,
the conception of reality as irreducible. This conception can be no
14 Individuals, Essence, and Identity
better off than the concept of reduction with which it is associated; and
so we may ask, "What is it for one proposition (or statement or sen-
tence) to be reducible to others?"
Three main lines of response to this question have been proposed.
According to the first, reduction is a matter of logical analysis. To say
that one sentence reduces to, or is analyzable in terms of, another is to
say that they express the same proposition but that the grammatical
form of the second is closer to the logical form of the proposition than
the grammatical form of the first. 9 Thus reduction reveals a discrep-
ancy between the "apparent" grammatical form of the sentence and the
"genuine" logical form of the proposition and serves to bring the two
in closer alignment. To take a paradigm example, the sentence 'The
average American is 5 feet tall' will reduce to 'The sum of heights of
all Americans divided by the number of Americans is 5 feet', since the
latter brings us closer to the logical form of the proposition that is ex-
pressed.
This approach suffers from at least two problems of detail. First, it
is unable to deal with one-many reductions. The philosopher who does
not believe in conjunctive facts will want to say that the truth of the
conjunction S /\ T reduces to the truth of its conjuncts S and T. But
here there is no question of a single proposition being expressed on left
and right. It is also unable to deal with one-one reductions in which the
reducing sentence is merely sufficient for the sentence to be reduced.
For example, when S is true and T false, we may wish to say that the
truth of the disjunction S v T reduces to the truth of the disjunct S. But
again, no single proposition is expressed.
A second difficulty concerns de re reductions. Just as there are de
re modal claims that are to be distinguished from their de dicto coun-
terparts, so there are de re reduction claims that are to be distinguished
from the corresponding de dicto claims. Thus we may wish to claim
not merely that the sentence 'The couple Jack and Jill is married' is
reducible to the sentence 'Jack is married to Jill', but also that the sat-
isfaction of the open sentence 'z is married' by the couple Jack and Jill
is reducible to the satisfaction of the open sentence 'x is married to y'
by Jack and Jill. But it is hard to see, on the proposed view, how this
could be a case of logical analysis. For the proposition expressed by
the open sentence 'z is married' under the assignment of the couple to
Kit Fine, The Question of Realism 15
that the value of each parameter is reducible to the value of the others.
As a particular example, we might take the parameters to be the mass,
volume and density of a given body.15
Finally, the approach is no better able than the first to capture the
antirealist import of reductive claims. For as long as reduction is re-
garded as getting us closer to what is real, we will wish to deny the re-
ality of any fact that reduces to something else. But then how could the
mere existence of certain modal connections between one class of
propositions and another serve to establish the unreality (or the reality)
of the facts from either class?
Indeed, a broader conclusion may be justified, just as in the case
of factuality. For it is hard to see how there could be any sufficient
condition for one proposition to be reducible to others (trivial and cir-
cular cases aside). For whatever the sufficient condition might be, its
satisfaction would appear to be compatible with the adoption of a
thorough-going realist position, one which took every single fact to be
real, and hence compatible with the rejection of any given reductive
claim. 16
We see from the previous discussion that the prospects for defining the
notions of factuality and reducibility in fundamentally different terms,
or even for providing conceptually unproblematic sufficient conditions
for their application, do not look good. Although we arrived at this
negative conclusion as a result of detailed investigation, the conclusion
could perhaps have been anticipated from the start. For we were after a
form of antirealism that was not necessarily skeptical, at odds with re-
ceived opinion. Now, presumably there is nothing special about the
opinion's being received in this regard. In so far as an antirealist posi-
tion is independent of such opinion it should be independent of all
similar opinion, whether received or not. Indeed, central to our present
understanding of antirealism is a distinction between what one might
call "first-order" propositions, which merely say how things are with-
out regard to their metaphysical status, and the corresponding "second-
Kit Fine, The Question of Realism 19
derive, not from their lack of content, but from their distinctive char-
acter. Indeed, it is hard to avoid the impression, once one surveys the
various attempts at definition, that we have, in the conception of real-
ity as objective or fundamental, a distinctively metaphysical idea.
From this point of view, the attempt to define these notions in other
terms would be akin to the naturalistic fallacy; and just as it would be a
mistake to infer the unintelligibility of normative notions from the dif-
ficulty of defining them in naturalistic terms, so it would be a mistake,
in the present case, to infer the unintelligibility of the notions of factu-
ality and reducibility from the difficulty of defining them in non-
metaphysical terms.
Of course, the quietist may have a general hostility to metaphysi-
cal concepts, but he, of all philosophers, is not in a good position to
justify such hostility in a principled way. For the usual basis for re-
jecting the intelligibility of a whole sphere of concepts is that they
cannot be rendered intelligible within some chosen world-view - one
that sees only the physical or only the psychological, for example, as
real. Now one might attempt to motivate the rejection of metaphysical
concepts by adopting a world-view that sees only the first-order facts-
as given in ethics or mathematics or science, etc. - as real. But the
adoption of such a view already presupposes the intelligibility of a
metaphysical concept of reality. Thus there is a real danger that the
quietist's position rules out as unintelligible the only ground that could
possibly make it plausible.
There is also strong intuitive evidence in favor of intelligibility;
for the fact that a notion appears to make sense is strong prima facie
evidence that it does make sense. Indeed, the indispensability of the
notions in formulating certain metaphysical issues would appear to
make their intelligibility almost impossible to deny. Consider the issue
dividing the "A-theorist" and the "B-theorist" as to whether temporal
reality is intrinsically tensed. This is an issue that cannot be rendered
intelligible without invoking the metaphysical conception of 'fact'. For
the A-theorist will want to affirm, and the B-theorist to deny, that there
are tensed facts in the world; and it is only the metaphysical rather
than the ordinary notion of 'fact' that can properly serve to represent
what is here at issue. I might also note that appeal to the metaphor of
an Archimedean standpoint is almost irresistible in this context. For
Kit Fine, The Question of Realism 21
5. GROUND
Its being the case that Britain and Germany were at war in 1940
consists in nothing more than ... ,
where ' .. .' is a compendious description of the warring activity of
various individuals.
In such cases, we say that the propositions on the right (collec-
tively) ground the proposition on the left and that each of them partly
grounds that proposition. I shall normally assume that the grounded
proposition and its grounds are true, though one may also talk of
"ground" when the grounding propositions would ground the grounded
proposition were they true.
The notion of ground should be distinguished from the strict no-
tion of reduction. A statement of reduction implies the unreality of
what is reduced, but a statement of ground does not. Thus in saying
that the fact that P /\ Q reduces to the fact that P and the fact that Q,
we are implying that the conjunctive fact is unreal; but in saying that
the fact that P /\ Q is grounded in, or consists in, the fact that P and the
fact that Q, we are implying no such thing. We are adopting a meta-
physically neutral stand on whether there really are conjunctive facts
(or truths). Thus our view is that there is sense in which even a realist
about conjunctive facts may be willing to concede that the fact that P
/\ Q consists in the fact that P and the fact that Q; there is a position
here that may be adopted by realist and antirealist alike. 21
The notion of ground, like the notion of reduction, is also to be
distinguished from logical analysis. Indeed, the paradigm of logical
analysis ("the average American") is not for us a case of ground, since
the propositions expressed on both sides of the analysis are presuma-
bly the same and yet no proposition can properly be taken to ground
itself. For us, the potentially misleading surface appearance of gram-
mar is entirely irrelevant to questions of ground, since we are looking
to the propositions expressed by the sentences rather than to the sen-
tences themselves. Thus we distinguish between the essentially lin-
guistic matter of determining which proposition is expressed by a
given sentence (whether, for example, a term is a genuinely referring
expression) and the essentially metaphysical matter of determining
what grounds what.
Kit Fine, The Question of Realism 25
proceed to the second stage. Ifhe says "No", we ask him, "Which ba-
sic propositions collectively ground the proposition P?" Granted:
there will be some basic propositions that ground P; and so let us sup-
pose that they are Q, R, S, ....
If the antifactualist denies that these propositions ground P, then
we already have a disagreement on ground. So suppose he agrees that
they ground P. Then he must take one of them, say Q, to be nonfac-
tual, since
The factualist, on the other hand, will take all of them to be factual,
since
(c) no factual proposition IS partly grounded III a nonfactual
proposition.
Thus the two philosophers will differ on the factual status of the
proposition Q; and we may substitute the present Q for the previous P
and proceed to the next stage.
Stage 2. Our factualist and antifactualist will differ on the factual
status of the proposition P, which the factualist takes to be basic.
Given that the antifactualist takes the proposition P to be nonfactual,
he must acknowledge that at least one of its constituents is nonfac-
tual/ s since
Thus in the case of the proposition that abortion is wrong, the nonfac-
tual constituent would presumably be the attribute wrong.
Let C, D, ... be the constituents of the proposition which our anti-
factualist takes to be nonfactual. Now it is conceivable that our factu-
alist might also take some of the constituents C, D, ... to be nonfactual
despite believing the given proposition to be factual. But any plausible
28 Individuals, Essence, and Identity
factualist and anti factualist position will surely agree on the question
of whether the given proposition would be nonfactual if the constitu-
ents C, D, ... were nonfactual. Thus even though they may disagree on
the factuality of the proposition that abortion is wrong, they will agree
that the proposition would be nonfactual if the attribute wrong were
nonfactual. But given that the factualist holds the given proposition to
be factual, and since he agrees with the antifactualist that the constitu-
ents C, D, ... , if nonfactual, would render the proposition nonfactual,
he must take one of those constituents to be factual.
Stage 3. Our factualist and antifactualist disagree on the factual
status of some constituent, say C, of the given proposition P. Say that a
proposition essentially contains a given constituent if its replacement
by some other constituent induces a shift in truth-value. Thus Socrates
is a essential constituent in the proposition that Socrates is a philoso-
pher though not in the proposition that Socrates is self-identical. Now
any plausible antifactualist view will presumably maintain
(e) any nonfactual constituent C is essentially contained in some
true factual proposition P+.
In the case of the ethical antifactualist, P+ might be the proposition
that so-and-so said that abortion is wrong (or attributed wrongness to
abortion) or the proposition that the word 'wrong' refers to wrongness.
The factualist, moreover, is plausibly taken to agree with the antifactu-
alist on this matter. There need be no disagreement about the proposi-
tion is essentially containing the given constituent or its truth; and the
factualist would appear to have even less reason than the antifactualist
for taking the proposition P+ to be nonfactual. Indeed, let us suppose
that the only other constituents in the proposition, besides C, are those
that both agree are factual. Then the proposition P+, for the factualist,
will only contain factual constituents and must therefore be factual.
Stage 4. Our factualist and antifactualist agree on the factuality of
the proposition P+ and yet disagree on the factual status of its con-
stituent C. Say that a proposition is imperfectly factual if it is factual
but contains a nonfactual constituent and that it is perfectly factual if it
is factual and contains only factual constituents. Thus the proposition
P+, for the antifactualist, is imperfectly factual. But then he will be-
Kit Fine, The Question of Realism 29
lieve that it has a perfectly factual ground, i.e., one consisting entirely
of perfectly factual propositions, since
(t) any true imperfectly factual proposition has a perfectly factual
ground
The factualist, on the other hand, is plausibly taken to believe that the
proposition has no ground none of whose propositions involve the
constituent C, since
with the rules of the game. Thus, even in these cases, there is some.-
thing of a factual sort that counts as getting things right.
Similarly, there would appear to be a factual sense in which some-
one may be said to be responsive to the facts. Again, the most obvious
sense of being responsive to the facts for the antifactualist is one that is
itself nonfactual, since it will rest upon what one takes the facts to be.
But in the shadow of this sense, there is a another sense that is factual.
For our epistemic activity must somehow engage with the real world;
there must be something that we aim for - in aiming to be well-placed
epistemic agents - whose realization is a factual matter. So for the ex-
pressivist, it might consist in being appropriately sensitive to one's
(implicit) commitments in the formation of one's ethical beliefs; while,
for the formalist, it might consist in being appropriately responsive to
the rules of the game in executing a proof.
Granted that they can agree upon a common practice, the factualist
and anti factualist each owes us an account of what it is, of that in
which it consists. Or to put it more precisely, each should provide us
with an account of what might ground the propositions from the ex-
tended domain. But their accounts are subject to very different con-
straints. The antifactualist must provide an account of the practice
without making any reference to the contested constituents. The ex-
pressivist, for example, must be able to say what having a moral belief
might consist in without making any use of moral vocabulary, and the
formalist must be able to say what possessing a mathematical proof
might consist in without making any use of mathematical vocabulary.
It is this constraint that explains why a antifactualist must be able to
provide some alternative to a truth-conditional account of our under-
standing of language; for truth, in its application to the sentences of a
given nonfactual domain, is nonfactual and must therefore be elimi-
nab Ie. It also explains why the standard formulations of antifactualist
positions - expressivism, constructivism, formalism etc. - are com-
monly taken to be antifactualist, even though this is not strictly implied
by the formulations themselves. For they provide the general means by
which the constraint may be seen to be satisfied. It is immediately
clear from the expressivist's position, for example, how he would wish
to account for ethical belief without making use of ethical terms.
Kit Fine, The Question of Realism 37
8. REALITY AS FUNDAMENTAL
Of course, the extreme quietist will not be happy with the concept of
Reality in itself. But it should be recalled that our target is the moder-
ate quietist. We have thrown conceptual caution to the winds and our
only question is: Given that the concept is intelligible, then how is its
application to be settled? How are we to determine, from all the possi-
ble truths, which are descriptive - or possibly descriptive - of such a
reality, and which are not?
In attempting to determine what is real in this way, we cannot
simply appeal to the fact that a given proposition is basic. For a basic
proposition may be nonfactual; and clearly no nonfactual truth is de-
scriptive of fundamental reality. However, any basic factual proposi-
tion will be real. For any true factual proposition is real or grounded in
what is real; and so the proposition, if basic, will be real.
Our previous methodology for determining what is factual can
therefore be of help in determining what is real. For once it is resolved
that a given domain is factual, then any basic propositions from that
domain can be taken to be real. But even without any help from that
methodology, there is perhaps a general presumption in favor of any
given proposition's being factual; and so the mere fact that a proposi-
tion is basic will give us reason to believe that it is real, in the absence
of any reason to the contrary.
Just as we cannot read off what is real from what is basic, so we
cannot read off what is unreal from what is nonbasic. Indeed, it is pos-
sible to imagine metaphysical scenarios in which the nonbasic, or
grounded, is plausibly taken to be real. Suppose, to take one kind of
case, that Aristotle is right about the nature of water and that it is both
indefinitely divisible and water through-and-through. Then it is plausi-
ble that any proposition about the location of a given body of water is
grounded in some propositions about the location of smaller bodies of
water (and in nothing else). The proposition that this body of water is
here, in front of me, for example, will be grounded in the proposition
that the one half is here, to the left, and the other half there, to the
right. But which of all these various propositions describing the loca-
tion of water is real? We cannot say some are real and some not, since
there is no basis upon which such a distinction might be made. Thus
42 Individuals, Essence, and Identity
we must say either that they are all real or that none are. But given that
the location of water is a factual matter, we should take all of them to
be real, notwithstanding the fact that each is grounded in propositions
of the very same sort. 38 Another kind of case involves "horizontal"
rather than "vertical" considerations. Imagine an ontology that takes
certain simple events and the causal relationships between them to be
real. Suppose now that one simple event causes a compound of simple
events. Then this presumably consists in its causing one component of
the compound and in its causing the other component. Now suppose
that a compound of simple events causes a simple basic event, even
though no component of the compound causes the event. Then it is not
clear what the ground might be. But if this causal relationship is taken
to be basic and hence real, then compound events should also be taken
to be real, and so causation of the compound - which is a real relation-
ship between real relata -should also be taken to be real, notwith-
standing its being grounded in other causal relationships.
So given that one proposition is grounded in others, how are we to
ascertain whether or not it is real? What I would like to suggest is that
there is a general presumption in favor of the grounded not being real.
In the absence of any reason to the contrary, such as those illustrated
by the cases above, we should assume that any given grounded propo-
sition is unreal.
The presumption may be justified by reference to the general aims
of realist metaphysics. For the distinction between what is and is not
real represents a general strategy for making metaphysical sense of the
factual world. For, of all of the structure that the world exhibits, some
may be taken to be real, to belong to the world itself, and some to be
only apparent and to be understood by reference to what is real. Let us
call a division of all propositions into those that are real and those that
are unreal a world-view. Thus a world-view will correspond to a par-
ticular attempt to see the world as intelligible in terms of the distinc-
tion between what is and is not real.
Let us now call a factual proposition moot if it is grounded and if
there is no special reason to think it real. There are then three possible
world-views one might adopt: the minimalist, which takes each moot
propOSItIon to be unreal; the maximalist, which takes each moot
proposition to be real; and the middling, which takes some moot
Kit Fine, The Question of Realism 43
Notes
1 I here ignore the possibility that reconciliation is to be achieved by modifying
our view of received opinion, either by not taking it to be matter of what we believe
or by supposing that its content is other than what we naturally take it to be. These
attempts at reconciliation, to my mind, merely shift the conflict with received opin-
ion to another place.
2 Many philosophers do not take reduction to have antirealist import. Their con-
cept of reduction seems to correspond more closely to what I later call 'ground'.
3 I here slide over the difficulty of whether the bearers of nonfactuality can
properly be said to be propositions.
Kit Fine, The Question of Realism 45
4 A useful survey of such criteria is given in Wright [1992] and some general
critiques of them are to be found in Rosen [1994], Dworkin [1996] and Stroud
[2000], Chapters 1-2.
5 A similar proposal has been made by Gaifman [1975]. It should be noted that
Dummett [1993], p.467, is not inclined, as I am, to assimilate the nonfactualism of a
sophisticated form of expressivism with the nonfactualism of a constructivist posi-
tion. However, none of my criticisms will turn upon making this assimilation.
6 See Edgington [1980-81] and Winkler [1985] for further criticisms of Biva-
lence as a sufficient condition. I also believe that there are problems with Bivalence
as a necessary condition for realism even when the obvious sources of truth-value
gaps (such as vagueness or reference-failure) are removed, but this is not something I
shall discuss.
7 Let us use 't F A' for 't warrants A' and 't ~ s' for 't improves s' (I assume
that any situation is an improvement upon itself). Molecular formulas within the pro-
posed semantics may then be subject to the following clauses:
(i) t FB 1\ C ifftFB and t FC
(ii) t FBv Ciff(Vt* ~ t)(3u~ t"') (u FB or u FC)
(iii) t F~B iff (Vt* ~ t) (not-(t* FB)).
Given that atomic formulas p satisfy the non-foreclosure condition:
(*) t FP iff (Vt* ~ t)(3u ~ t*) (u FP)
then so will every formula A.
s Dummett [1978], pp.365-67, has reservations about making the transition from
the Law to the Principle, but 1 do not believe that they apply in the present case.
9 Sometimes the emphasis is on a correspondence with "facts" rather than
propositions.
10 The significance of de re reductive claims has not been properly appreciated.
They enable one to achieve a huge simplification in the formulation of many reduc-
tions and in certain cases - such as the bundle theory of particulars - they are essen-
tial to understanding the very point ofthe reduction.
11 Similarly, it is perfectly in order for the ethical antifactualist to express his
view that there are no genuine moral properties in the form "For any moral property
and any possible bearer of the property the bearer's possession of the property is a
nonfactual matter". We quantify over all moral properties in order to express the
view that none of them are "real".
12 Dummett propounds such a conception in [1993], pp.56-7.
13 Or to put it more linguistically, between the satisfaction of the condition 'z is mar-
ried' by c and the satisfaction of the condition 'x is married to y' by a and h.
14 Advocates of this approach include Armstrong [1997], p.12, Chalmers
[1996], p.48, and Jackson [1998], p.5. Many philosophers, I should note, do not take
supervenience to capture a metaphysically significant notion of reduction.
15 For further discussion of these points, see Kim [1993], pp.I44-46, and the
references contained therein.
46 Individuals, Essence, and Identity
16 We should note an asymmetry in the two cases, however. For even though
there may be no guarantee of nonfactuality, there will be a guarantee of nonreduci-
bility. For if R is reducible to S then S will necessarily imply R and so the possible
truth of Sand -,R (a purely modal fact) will guarantee that R is not reducible to S.
17 Ronald Dworkin has pointed out to me that first-order morality is naturally
taken to include the claim that if there are no objective moral facts then anything
goes (or some other such moral conclusion) and, given that this is so, the second-
order claim that there are no objective moral facts will not be independent of moral-
ity. However, for present purposes, I would not take the above conditional to be
strictly first-order.
18 Their number include Blackburn [1992], pp. 7, 34, 168, and [1998], p.319;
Dworkin [1996]; A. Fine [1984], pp.97-100; Putnam [1987], p.19. Other philoso-
phers, such as Rosen [1994], have flirted with quietism without actually embracing
it.
19 Rorty [1979], p.311, is someone who has espoused a more moderate form of
quietism.
20 It is an oddity in the logical positivist's critique of metaphysics that these two
charges were linked. For if there is no way of settling metaphysical questions, then
who cares whether or not they make sense?
21 Some philosophers have thought of supervenience as a metaphysically neutral
counterpart to the notion of reduction and to this extent, at least, what they have in
mind may correspond to our notion of ground.
22 I should note that I do not take all judgements of ground to be a priori. Thus
the philosophical investigation of reality should only be based upon those judge-
ments that are a priori or that can be given some kind of a priori backing.
23 I hold a similar view concerning the notions offactuality and irreducibility.
24 For simplicity, we assume that P (and its successor p+) are true, though all
that is strictly required for the argument is that they be possibly true.
25 Intuitively, a nonfactual constituent is one which can be a source ofnonfactu-
ality in a proposition to which it belongs. Perhaps the factuality of constituents can
be defined in terms of propositional factuality in the following way. With any con-
stituent c may be associated the class of propositions Pc in which c has its primary
employment. A constituent c is then nonfactual iff any proposition of Pc is nonfac-
tual.
I have presupposed that propositions are structured entities built up from their
constituents. Those who do not like this assumption may conduct a parallel argument
with sentences and their terms in place of propositions and their constituents.
26 It then becomes less clear whether the factualist need take the constituent C to
be "ineliminable" at stage 4. But as we shall see, the only plausible cases in which it
is eliminable are ones in which it gives way to constituents which the factualist and
anti factualist will recognize as equivalent, in their factuality status, to the given con-
stituent.
27 When P is the disjunction PI v P z of a true factual proposition PI and a false
nonfactual proposition P 2' it will be grounded by PI alone and we will then wish to
Kit Fine, The Question of Realism 47
say, in conformity with (b), that P is factual. Thus factuality, on our current under-
standing, is a contingent matter. (On an alternative understanding, a proposition
might be taken to be nonfactual when it could be nonfactual in our current sense). I
might add that we do not think of vagueness as a source, per se, of nonfactuality,
since a vague proposition may still be aimed at the real world. Thus cases in which a
vague truth is grounded in a precise truth are also not counterexamples to (b).
28 The truth of P v -.P when P is a nonfactual truth is also not a counterexample
to (c) since P v "OP, in this case, should be taken to be nonfactual.
29 Consider the proposition that the last thing the Pope said is true and suppose
that the last thing that he said was that abortion is wrong. Then this proposition is
nonfactual (if ethics is nonfactual) and so we are obliged by this assumption to treat
some constituent of the proposition - presumably either 'thing' or 'true' - as non-
factual. This is awkward, since it means that the proposition is imperfectly factual
even when the last thing that the Pope said was factual. There is a related difficulty
for the thesis that no 'ought' can be derived from an 'is'. For from the assumptions
that the last thing the Pope said is true and that the last thing the Pope said is that
abortion is wrong, we can derive the conclusion that abortion is wrong. Thus the first
assumption should be taken to be an 'ought' even though it apparently contains no
moral terms. Perhaps there is some other way of dealing with such cases.
30 A related assumption (which he regards as a fallacy) has been adumbrated by
Horwich [1998], p.21. This is that "whenever a fact has a certain component, then
whatever constitutes this fact must contain either the same component or alterna-
tively something that constitutes it". Assumption (g) says that the constituting fact
must contain the same component when that component is itself fundamental, i.e.,
such as to occur in an unconstituted fact. But if the constituting fact does not contain
something that constitutes the component in his sense, there is no reason to suppose
that it is fundamental in my sense. Thus accepting (g) is perfectly compatible with
rejecting his fallacy.
31 On certain deflationary views, of the sort proposed by Field [2001], Chapters
4 and 5, these propositions would not even be taken to involve a relationship be-
tween a term or concept and an entity.
32 Thus Putnam's argument for scientific realism ([1978], p.lDO) and the argu-
ment that Harman [1977], Chapter 1, considers against moral realism both turn on
whether the best explanation of some phenomenon (the success of science, our moral
responses) does or does not involve reference to the facts that are in dispute. But in
so far as the relevant notion of explanation is not metaphysical, it is unclear why the
factualist or antifactualist should differ on this question. I also doubt, though this is a
separate matter, that these arguments can plausibly be brought to bear on the skepti-
cal form of antirealism.
33 Another aspect of our "practice" concern the metaphysical status of the
propositions from the given domain. Thus even if a antifactualist takes a proposition
to be nonfactual, he may still take it to be a factual matter that it is nonfactual or that
it is contingent or that it is grounded in certain other propositions. The question of
the metaphysical status of these metaphysical claims, though of enormous philo-
48 Individuals, Essence, and Identity
sophical interest, is not normally so relevant to adjudicating disputes over the factu-
ality of a given first-order domain.
34 Other philosophers (principally Dummett, passim, and Blackburn [1984],
p.169, and [1998], p.50) have also emphasized the role of practice in adjudicating
realist disputes. What is distinctive about my view is the precise way in which it ar-
ticulates what a practice is and what is involved in accounting for a practice.
35 This is merely a picture. It need not commit one to the view that there are
facts in the world whose structure might correspond to the structure of the proposi-
tions or sentences by which they are described, as in the standard exposition of logi-
cal atomism (Wisdom [1969]).
36 See Williams [1978], p.66, and [1985], p.241.
37 One might express the notion of reality by means of a connective 'it is con-
stitutive of reality that .. .', just as in the case of the notions of factuality and ground.
Considerable interest would then attach to developing the logic of such notions.
38 In making out this case, it is essential to distinguish between ground and re-
duction. If a ground is taken to bring one closer to what is real, then it is hard to see
how there could be an infinite regress of grounds (with nothing at the bottom). For
how can one get closer to what does not exist? But once grounds are taken to be
metaphysically neutral, there is no more difficulty than in the case of cause in con-
ceiving them to form an infinite regress. (Cf. the discussion of Leibniz's position on
this question in Adams [1994], pp.333-8).
39 It is possible to envisage a semi-quietist position that accepts the concept of
factuality but rejects the concept of fundamentality. The world would divide into an
objective and nonobjective part, on this view, but the objective part would be an un-
differentiated mass and could not be meaningfully be taken to possess any particular
intrinsic structure. The above definition would not then be available and the study of
what is factual would have to proceed independently of any consideration of what is
fundamental.
40 I have been very fortunate with the assistance I have received in writing this
paper. Participants of seminars at UCLA and Princeton and of talks at the Universi-
ties of Columbia, Michigan, Pennsylvania, and Rutgers made many excellent com-
ments; I have had valuable conversations or correspondence on the topics of the pa-
per with Rogers Albritton, Paul Boghossian, Ruth Chang, Hartry Field, Mark
Johnston, David Kaplan, David Lewis, Mark Moyer, Gideon Rosen, Stephen Schif-
fer and Bartosz Wieckowski; and, to top it all, Ronald Dworkin, Andre Gallois, Gil
Harman, Paul Horwich, Joshua Schechter and two referees from the journal Imprint
provided me with an extensive array of written comments which were of great help
in revising the paper. This paper first appeared in the electronic journal Philospher's
Imprint (voU, no. 1).
WORDS AND OBJECTS
Achille C. Varzi
1. INTRODUCTION
know what sorts of things there are or could be? Then let's see what
sorts of things there must be in order for what we truthfully say to be
true. This, in tum, requires that we pay due attention to what it is that
we say when we say something. For obviously we do not want to be
misled by the idiosyncrasies of the language we speak. Obviously we
do not want to say that there must be such things as age differences
(for example) just because the English sentence 'There is a difference
in age between John and Tom' is true. We do not want to say that there
exist such things as holes just because the sentence 'There is a hole in
that piece of cheese' is true. First we have to uncover the logical forms
of these sentences. We have to look at the deep, semantic structures
underlying their superficial, grammatical structures, and then we can
look at what sorts of things must exist in order for those semantic
structures to agree with the facts. For example, we can look at the ref-
erents of the logically proper names and at the values of the bound
variables. Do these include ages? Do they include holes? Does the do-
main of reference and quantification underlying our English assertions
include such things in addition to chairs, people, and chunks of
cheese? Does it include conferences, hurricanes, stabbings? Does it
also include numbers? Jokes? Haircuts? What else?
The second tool developed by philosophers is the one known as
Leibniz's law, broadly understood as a principle of substitutivity salva
veritate. Is this chair something over and above the mereological fu-
sion of the molecules that constitute it? Well, let us see - is there any
statement that is true of the chair but not of the mereological fusion, or
vice versa? If there is no such statement - if the chair and the
mereological fusion are indiscernible -then they are one and the same
thing, or so we may suppose. Otherwise they must be two things. Is
Brutus's stabbing of Caesar the same event as his killing of Caesar?
Well, if whatever is true of the stabbing is also true of the killing, and
vice versa, then the answer is: Yes, they are one and the same event;
otherwise the answer is: No. Of course it may actually be impossible to
check every statement, so in actual circumstances it may be impossible
to be sure that we have one thing or one event rather than two or more.
That's the obvious difficulty with one half of Leibniz's law, the one
that asserts the identity of the indiscernibles. (This is why it would be
good to have some identity criteria that are tailor-made for material
Achille Varzi. Words and Objects 51
Let me first focus on the business of logical form. The guiding idea,
here, is one that goes back to Frege and especially to Russell. We
know that we must make room for those things whose existence is im-
plied or presupposed by any statement that we can truthfully make. But
ordinary-language sentences may have a deceptive grammatical form
and therefore questions about their aboutness - questions concerning
their ontological commitment, as some like to say - only arise upon
52 Individuals, Essence, and Identity
Sentences such as these - we are to understand - are not about the en-
tities they seem to be about. It's not that there is this guy, Mr. Nobody,
whom Alice is seeing ("And at that distance too!", as the King would
remark in a fretful tone.) It is not that there is this fellow, the present
king of France, who does not exist (as Meinong and others supposedly
held). To think so would be to be misled or deceived by superficial
grammatical features. It would be to treat as designators expressions
which are not, in fact, meant to designate anything at all. Deep down
such sentences have an entirely different form and that is the form that
matters when it comes to assessing their ontological import - or so the
story goes. The first sentence is the negation of an existential percep-
tual report:
(1') It is not the case that there is at least one person whom
I can see on the road.
fonn 'There is a P' or 'There are Ps'. The first example I mentioned
earlier was from Morton White:
(3) There is a difference in age between John and Tom.
Let us suppose that this sentence is true. Do we want to say that this
fact requires the existence of a suitable entity satisfying the existential
quantifier-the existence of an age difference? Of course not. That
would be falling in a trap. Here is what we are supposed to do instead:
We might begin by saying that we understand the relational predicate 'is
as old as' and that we test statements of the form 'x is as old asy' without
having to see that x has some queer thing called an age, thaty has one, and
that these ages are identical. In that event, the belief of the ordinary man
that there is a difference in age between John and Tom would be rendered
in language that is not misleading by saying instead, simply, 'It is not the
case that John is as old as Tom' [ ... ] We need not assert the existence of
age differences [ ... ] in communicating what we want to communicate.
(White [1956], pp.68-69)
predicates' ... is perforated' and 'there are holes in ... ' - just like any
other shape-predicate, say' ... is a dodecahedron' - may truly be predi-
cated of pieces of cheese, without any implication that perforation is due
to the presence of occult, immaterial entities. (Lewis and Lewis [1970],
p.4)
So we can truly assert sentence (4). But the underlying logical form is
not an existential statement but a simple subject-predicate statement:
Here are some more examples of logical analyses of this sort, taken
somewhat randomly from the literature:
(8) This tomato and that fire engine have the same color.
(8') This tomato and that fire engine agree colorwise. 4
Nor are these the only sort of cases that philosophers have been wor-
rying about. All of these are examples that illustrate an eliminativist
strategy: they show that we can analyze sentences which seem to in-
volve ontological commitment to certain entities as expressing propo-
sitions that are, in fact, ontologically neutral with respect to those enti-
ties. But there are also cases where the analysis goes in the opposite
direction, i.e., cases where the logical form discloses a hidden quanti-
Achille Varzi, Words and Objects 55
All of this is standard lore. Some of these analyses are so naturally ac-
cepted that they are now found in introductory textbooks in logic. This
is true of Russell's analysis of definite descriptions but also, for exam-
ple, of Davidson's analysis of action sentences. 6 Still, I think that there
are severe problems with this general picture. Some of these problems
56 Individuals, Essence, and Identity
are well known but some are less obvious and, I think, rather worri-
some, especially if we keep in mind the tasks of the working metaphy-
SICIan,
One obvious problem is that we can hardly eliminate or introduce
anything by mere armchair speculation. Philosophers do speak as
though we could banish entities from existence just by helping our-
selves with Occam's razor, or bring entities into existence just by
adding some quantifiers. But this is eerie and we should better take it
as a jat:;on de parler. All we can do is to show how the existence or
non-existence of certain entities would allow us to explain certain
facts, just as scientists sometimes posit the existence or non-existence
of certain entities in order to explain certain natural phenomena. In any
case, it is obvious that logical analysis per se can do very little. Para-
phrasability of sentences about holes does not per se eliminate holes
from the world just as the assertibility of such sentences does not
automatically introduce holes. Paraphrasability is a necessary condition
if we want to avoid commitment to such things, and assertibility is a
sufficient condition if we want to proclaim commitment, but neither is
necessary or sufficient to affect the ontology itself.?
A second obvious problem concerns the very idea that the logical
form of a sentence may allow us to withhold our commitments while
still communicating what we want to communicate. What test can we
apply to see whether a given English sentence can be understood as
having a certain logical form? How do we know, for example, whether
the logical form of a sentence such as (3) is correctly represented by
(3'), where there is no mention of age differences? Pretty clearly, if we
want to use (3') to communicate what we would be communicating
using (3), then (3') must express the same proposition as (3). It must be
the case that in uttering (3') we would be making the same assertion as
we would make if we uttered (3). Otherwise (3') would not represent a
legitimate analysis of (3). But then, in uttering (3') we would be talking
about the very same things we would be talking about if we uttered (3).
And why should one utterance be better than the other? Why should
(3') be ontologically more transparent than (3)? As William Alston
pointed out a while ago, this bears more than a passing resemblance to
the paradox of analysis:
Achille Varzi, Words and Objects 57
We can also put the problem as follows. Whether or not the truth of
our statement implies (or presupposes) the existence of age differences
does not depend on the words that we utter to make that statement - it
doesn't depend on the sentence that we use. So let us assume that sen-
tence (3') can be used to make the same statement as sentence (3),
though in a way that does not mention age differences explicitly. Then
we may as well say that (3) itself is a sentence that can be used to make
the same statement as (3 '), though in a way that does mention such
things explicitly. So from left to right (so to say) the analysis results in
an elimination; but from right to left it results in an introduction. How
do we choose?
Let me elaborate. 8 The idea behind the use of logical analysis is
that in order to assess the ontological commitment of ordinary-
language sentences one must first provide suitable logical paraphrases
that are "intrinsically non-misleading" (as Ryle put ie) and therefore
ontologically transparent. This amounts to a sort of linguistic recon-
structivism: the truth conditions of our sentences are determined by the
truth conditions of the corresponding paraphrases and do not therefore
require an independent ontology. Very well. The question is: Where do
these paraphrases come from? On what grounds should we look for the
logical forms that underlie our ordinary statements? Plainly, the very
issue of which sentences must be logically paraphrased - let alone how
they ought to be paraphrased - can only be addressed against the back-
ground of one's own philosophical inclinations. When Russell, for ex-
ample, says that (2) must be paraphrased as (2') it is because Russell
holds that the former, as it stands, is incompatible with our sense of
reality, with that robust "feeling for reality which ought to be preserved
even in the most abstract studies"lo (and which lies behind that
"aesthetic sense of us who have a taste for desert landscapes"ll, as
Quine later put it). The analysis yields no ontological discovery. It is
Russell's own ontological convictions that lead him through the quest
for an appropriate logical form for (2), not vice versa. So much so that
a philosopher such as Meinong might feel no need to take any action.
58 Individuals, Essence, and Identity
For him the grammatical fonn of (2) may well coincide with its logical
fonn because for him the present king of France does have ontological
dignity. It is just one among many characters that a complete inventory
of the world should include. 12
If things are so, however, then here is the problem: How shall we
go about detennining whether or not a given sentence can be taken at
face value? How do we know whether or not it needs to be analyzed or
rephrased before we can look at its ontological import? In the case of a
sentence such as (3) we may be inclined to look for a paraphrase that
avoids any reference to age differences because these would be "queer
things" (in White's own tenninology) to be included in our inventory.
But if we thought that such things are not queer, then we wouldn't
need any paraphrase. We could still accept the analysis but we might
be more inclined to read it from right to left and observe that whoever
says that John is not as old as Tom is actually asserting the existence of
an age difference between them.
For another example, let's go back to the holes in the cheese. Ifwe
are inclined to rewrite (4) as (4') it is because our strong "feeling for
reality" prevents us from taking this to be a statement about a hole: if
holes are not real then our statement can only be about the cheese. This
is understandable and may even justify the enonnous amount of work
that will be involved in analyzing every natural-language sentence that
seems to refer to or quantify over holes as expressing a proposition
which only involves reference to or quantification over perforated ob-
jects. Consider, for instance:
about what there is rather than attribute our views to the language that
we speak, and hence to the speakers who share our language. What
would entitle us to do that?
guage that is intelligible at all. For the picture would be this: our daily
language - the language that we have learned and made ours since our
very first contacts with the surrounding world - that language would
consist of sentences whose real meaning often eschews us. It would
consist of sentences most of which are only acceptable as loose talk. It
would at best qualify as a sort of metalanguage with regard to the
regimented language of philosophy, the latter being the only genuine
object language and thus the only language that can express our genu-
ine ontological commitments. Is this an acceptable picture?
We thus come to what I regard as the main problem with the
whole idea of ontological transparency. As it turns out, both strategies
involve a duplication of languages. For neither is willing to give up
natural language altogether. Whether you are a revolutionist or a her-
meneuticist, you want to carry on speaking with the vulgar, hence you
are going to emphasize the pragmatic indispensability of ordinary lan-
guage against the philosophical value of the regimented language (on-
tologically impeccable but practically unspeakable). However, this du-
plication of languages only works fine for the revolutionist. For only
the revolutionist is always in a position to tell which language is being
spoken.
Take the Lewisean hole-e1iminativist once again. When speaking
with the vulgar she can give expression in English to the fact that some
cheese is perforated by asserting (4), but when speaking the regi-
mented language of philosophy she would assert the negation of (4).
More generally, she can assent to (19) when speaking loosely, and to
(20) when speaking strictly and literally:
(19) There are holes
(20) There are no holes.
This may be confusing to some people but the revolutionist will always
know when is when, and she will be happy to explain. Not so for the
hermeneuticist. If you are a hermeneuticist you do not have the same
leeway. To the extent that (19) is to be interpreted as (19'), (20) will
have to be interpreted as (20'):
(19') Something is perforated
(20') Nothing is perforated.
64 Individuals, Essence, and Identity
(21) It was cooler in the park after the sun had moved be-
hind the elms.
Would this be incoherent with the speaker's official view to the effect
that the sun does not move in the sky? Of course it would not. For our
astronomer would hasten to add that in uttering (21) she was speaking
loosely. If necessary, she could be more accurate and she would utter a
sentence that does not suggest that the sun has actually moved - for
example:
(or something much more awkward than this). She can do that and she
knows how to do it because she is a revolutionist; when she uses cer-
tain sentences of English she actually means something that goes be-
yond the literal or customary meaning of her words. (She means to ex-
press propositions that have a different form, if you like.) Now sup-
pose I utter a sentence such as (21). Am I to be taken to imply that the
sun has moved? Well - no. I suppose in this case you are entitled to
reinterpret my statement too in a way that makes it consistent with the
heliocentric theory: you may be charitable. ("Achille Varzi couldn't
possibly mean to say that the sun had moved! He must have been
Achille Varzi, Words and Objects 65
So much for this part of the story - the tool of linguistic analysis. The
only way I can make sense of it is as an honest revolutionary tool. But
revolutions cannot be improvised and I would not engage in one unless
I had already sorted out my views in advance. Linguistic analysis can
be useful as a tool to clarify what I mean when I use certain sentences
- or what we all should mean - but not what those sentences must
mean. Hence it supplies no shortcut to metaphysical investigation. In
particular, linguistic analysis is of little guidance when it comes to the
first important task involved in the drawing up of an inventory of the
world - that of figuring out the ontological categories under which the
items in the inventory should fall. Let me now move on to the second
task - that of figuring out a way of counting the items in each category.
As I mentioned, here it is customary to rely on the general tool pro-
vided by Leibniz's law, broadly understood as a principle of substitu-
tivity salva veri/ate. If we are smart, for some categories we may have
concocted some kind of identity criterion, but generally speaking the
principle of substitutivity supplies at least a negative test for identity: If
we hit upon a statement that holds true of something x but not of
something y, then we can be sure that x and y are distinct.
Also in this case, I am afraid I have mostly negative things to say.
It seems to me that the basic intuition behind this strategy is seriously
flawed and that it only succeeds in raising a dust that obstructs the real
difficulty involved in our metaphysical task. To make my case, let me
briefly review some concrete examples of how this strategy is typically
implemented.
Consider again this chair in front of me and the mereological fu-
sion of the molecules that constitute it. Are they the same thing? Well,
it seems plausible to suppose that the chair would survive the annihi-
lation of a single molecule. But - the argument goes - surely the
mereological fusion would not: That fusion of molecules must consist
of those very molecules, it must include them by definition.20 Hence
the chair and the fusion of its molecules have different properties (dif-
ferent modal properties) and should be distinguished by Leibniz's law.
For a second, standard example, consider a statue and the lump of clay
that constitutes it: two things or just one? Well- the argument goes-
Achille Varzi, Words and Objects 67
the artist made the statue this morning. But the clay was already there
yesterday. So the statue and the clay came into existence at different
times. Hence they have different properties (different temporal proper-
ties, in this case) and should be distinguished. The same line of argu-
ment is familiar also from the literature on events. Take Brutus's stab-
bing of Caesar. Is it the same event as Brutus's killing of Caesar?
Well, it seems reasonable to suppose that Caesar could have survived
the stabbing. But surely Caesar could not have survived his very kill-
ing. So, once again, an appeal to Leibniz's law would allow us to con-
clude that the stabbing and the killing are distinct events. And so on
and so forth. This line of argument is very popular and very pervasive
indeed?1 But is it legitimate?
Let us focus on one instance - say the chair and the mereological
fusion of the molecules that compose it, Xl ... X n . In that case the argu-
ment has the following structure: We come up with a statement which
is true of the chair but false of the fusion, and we conclude that the
statement must be about two things. Schematically:
(23) The chair in front of me could survive the annihilation
of molecule Xi.
(24) The fusion of molecules Xl ••. xn could not survive the
annihilation of molecule Xi.
Hence
(25) The chair in front of me is not the fusion of molecules
Xl .. · X n•
If the chair is not the same as the fusion of its molecules, then fine: We
are talking about two different entities and perhaps we can say that
(23") and (24") are both true. 22 Perhaps molecule Xi is an essential part
of the fusion but not of the chair. But this opposition would be prior to
Achille Varzi, Words and Objects 69
our thought experiment - it cannot be inferred from it and calls for in-
dependent grounds. How could we have de re intuitions about the chair
and about the fusion if we didn't even know whether they are one thing
or two? How could we compare their properties if we didn't know
what they are - if we didn't even know whether they are distinct? Be-
sides, why should we be able to settle identity issues in this world by
looking at what goes on in other worlds? Don't we need to know how
many passengers we are bringing along before we can embark in other-
worldly philosophical excursions? On the other hand, if the chair is the
same thing as the fusion of its molecules - and to rule that out would
be to beg the question - then that particular entity is the same in both
cases, so (23") and (24") cannot be both true. And which one of them is
false is a genuine metaphysical question: maybe molecule Xi is an es-
sential part of that entity, in which case (23") would be false; maybe it
is not an essential part, in which case it is (24") that would be false. (Of
course, it would then be awkward to assert the true sentence of which
(24") is the negation, i.e., to say that the fusion of Xl ... Xn could survive
the annihilation of Xi. It would be better to make the same statement by
asserting (23"). But that awkwardness is a heritage of our inclination to
oscillate between de dicto and de re readings: It is not a falsehood in-
dicator and it is only relevant from a pragmatic perspective.) It does
not matter now which premise is false. As I said, that would be a
genuine metaphysical issue. As far as the argument goes the point is
that we cannot simply assume that both premises are true on pain of
begging the question. We can have a priori reasons to accept both
premises only if we already have reasons to distinguish between the
chair and the fusion of its molecules in the first place, and that is sup-
posed to be the conclusion of the argument. As it stands, on a de re
reading (the only one that makes the argument valid) the argument is
either unsound or viciously circular. Hence it is useless.
The same diagnosis applies to the other non-identity arguments
mentioned abov<:(, as well as to other statements along the same lines.
In each case, I submit, the reasoning is either invalid (if read de dicto)
or question begging (if read de re). Thus, surely the tenns 'the statue'
and 'the lump of clay' have different senses; but it does not follow that
they have different referents. And if their referent is the same, if they
name the same thing, then either that thing was already there yesterday
70 Individuals, Essence, and Identity
or it was not. (If it was, of course it would have been awkward to refer
to it as a statue prior to this morning, when the artist actually shaped it
as a statue; but that awkwardness would be purely linguistic or ideo-
logical and would have no bearing on the ontological level.) Or take
the events, Surely the predicates 'stabbing' and 'killing' have different
intensions, which entails that the event descriptions 'Brutus's stabbing
of Caesar' and 'Brutus's killing of Caesar' have different senses. It
does not, however, follow that the two descriptions have different ref-
erents. And if their referents are the same, if they describe the same
event, then either that event could have been survived by Caesar or it
could not. (If it could, it would of course be awkward to refer to it as a
killing when counterfactualizing about Caesar's survival. We should
rather refer to it as a stabbing. But once again that awkwardness would
be purely linguistic, or ideological, and would have no bearing on the
ontological level.) And so on and so forth. To be sure, there is a differ-
ence between this last case - the event descriptions - and the cases of
the chair and the statue. Perhaps Brutus's stabbing of Caesar was his
killing of Caesar, but there are lots of other stabbings that are not kill-
ings. On the other hand, if this chair in front of me is the same as the
fusion of its molecules, then that chair next to you is also the same as
the fusion of its own molecules. Every chair must be the same as the
fusion of its own molecules, or else no chair is. Likewise, either every
statue is identical with the lump of matter that constitutes it, or else no
statue is. I take these to be important metaphysical tenets. But having
said this, the trouble with the argument is the same in all cases.
Let me stress also that the trouble concerns the form of the argu-
ment, not the conclusion itself: whether the entities in question are one
or two remains open. Moreover, the analysis is not quite neutral with
respect to the issue of contingent identity. If you think that a chair can
be identical with the fusion of its parts in some worlds but not in others
(or at some times but not at others), then the objection would not quite
apply.23 In that case we could speculate about the modal or temporal
properties of the chair and of the fusion, and perhaps we could dis-
cover that these properties are distinct without begging the question of
whether the chair and the fusion are in fact distinct. However in that
case it would remain to be shown how we can use Leibniz's law to go
from the observation that there are worlds in which the chair and the
Achille Varzi, Words and Objects 71
fusion have different properties to the observation that the actual chair
and the actual fusion have different properties - that is, different modal
properties. And unless we can do that the argument, though formally
non vicious, would still be pretty useless. Ditto in the other cases.
Now, I do not want to insist too much on this analysis here. 24 In its
general form it goes back to a point that Dummett made several years
ago in his first book on Frege, where he observed that whether or not
Leibniz's law can be used as a definition of identity, it cannot be made
to serve as a criterion for deciding the truth of identity statements.25
More recently, the same sort of consideration was put forward by Mi-
chael Della Rocca in the context of his discussion of essentialism and
by Stephen Neale in his analysis of event-referring descriptions. 26 It is
an obvious point, uncharitable as it may sound towards those philoso-
phers who do engage in non-identity arguments of this sort. In fact, I
do not even want to claim that Leibniz's law can never be used to pro-
vide grounds for a non-identity statement. I suppose that there are cir-
cumstances where one may find grounds for recognizing that a predi-
cate which is true of the bearer of one term is false of the bearer of an-
other term prior to any decision concerning whether the terms have the
same bearer. For example, I believe I have sufficient grounds for
making the statement
(26) Professor Bottani is sitting in this room.
Now we hear a scream coming from the room next door and I feel en-
titled to assert:
(27) The person who just screamed is not sitting in this
room.
From these two statements, together with the assumption that people
cannot be in two places at the same time, I can conclude that Professor
Bottani is not the person who screamed. Likewise, the truth of (26) to-
gether with that of
(28) Professor Bottani is not wearing a hat
is sufficient ground for me to conclude that, given the axioms of set
theory, the set of people sitting in this room is not the same as the set
72 Individuals, Essence, and Identity
of people wearing hats. All of this is fine. My point is, rather, that the
sort of evidence we need to rely on in cases such as these, where the
application of Leibniz's law is not question-begging, stems from a
complex web of factors. Among other things, the evidence builds on a
background theory about people and about sets. It certainly does not lie
in linguistic intuitions about the truth values of the relevant English
sentences, and perhaps that is why the non-identities that we can infer
come as no surprise. By contrast, it is typically linguistic intuitions that
seem to underlie ordinary non-identity arguments about such things as
chairs, mereological fusions, statues, and the like. And these intuitions
- I submit - are not warranted except on a de dicto reading. Hence they
are not warranted if we intend to apply Leibniz's law.
Some will reply that this is precisely the point to be stressed:
somehow there is a theory behind the claim that the premises of a non-
identity argument are true. It is not just linguistic intuitions. It is intui-
tions correlating the sortal terms that we use to talk about the entities at
issue - the sortals 'chair', 'statue', 'lump of clay', etc. - and the iden-
tity conditions that the theory associates with the entities falling under
those sortals, exactly as with people and sets. On this view, every sor-
tal comes with its identity and persistence conditions built in. And it is
such identity and persistence conditions that we rely on when it comes
to assessing the premises of a non-identity argument.
I do not intend to deny this. To the contrary I find this view per-
fectly reasonable. But it seems to me that the view is only acceptable
provided that the background theory is taken for what it is - a genuine
piece of metaphysics. If we want to say that a chair and a fusion of
molecules have different persistence conditions - that there are differ-
ent kinds of change that can and cannot be survived by the chair and by
the fusion, respectively - then we must do this properly. We must say
that it is a matter of what kind of thing a chair is, and what kind of
thing a mereological fusion is. We must do that before resorting to
Leibniz's law. And we can hardly do that simply by looking at our lin-
guistic practices and intuitions concerning the sortal terms 'chair' and
'fusion of molecules'. So, for example, David Wiggins and Jonathan
Lowe have famously argued that every individual is necessarily an in-
dividual of a kind, or sort, and that the kind or sort of thing that an in-
dividual is determines its identity and persistence conditions. Moreo-
Achille Varzi, Words and Objects 73
ver, one fully grasps the nature of a given kind or sort of thing, and
hence the sense of a sortal tenn designating it (or even the sense of a
singular tenn that can be used to refer to a specimen of it), only when
one grasps the associated criterion of identity. In the cases under dis-
cussion, these philosophers would say for instance that the sortal tenns
'chair' and 'fusion of molecules'
have different criteria of identity associated with them, and [ ... ] no indi-
vidual of a sort <jl can intelligibly be said also to belong to a sort \jI if <jl and
\jI have different criteria of identity. (Lowe [1989], p.70)27
Very well. I have no objections against this. But that is precisely be-
cause I do not think that our linguistic practices and intuitions con-
cerning the sortal tenns 'chair' and 'fusion of molecules' provide any
evidence for or against the idea that these tenns have different criteria
of identity associated with them. Some philosophers think they do;
other philosophers do not.
In the tenninology introduced earlier, we can also say that the link
between a sortal tenn and its identity criterion - if we want to insist on
it - can only be viewed as part of a revolutionary analysis, not as a
piece of natural language henneneutics. Or do we really think that
there is something about our use of the word 'chair' that detennines
whether this word picks out entities that differ from the fusions of their
constitutive molecules? Do we really think that there is something
about our use of the word 'statue' that detennines whether this word
picks out entities that differ from the lumps of stuff that constitute
them (or whether it picks out three-dimensional rather than four-
dimensional entities, for that matter)? Answering in the affinnative
would, I think, amount to assigning the words of our language a meta-
physical strength that they do not and cannot have. Exactly as with the
case of logical fonn, rather than directly theorizing our ideas about
what there is and how things are we would surreptitiously attribute our
ideas to the language we share with others. We would surreptitiously
attribute them to all the speakers of our language even if our ontologi-
cal intuitions are very different from those of others. And this is the
step that I find illegitimate.
74 Individuals, Essence, and Identity
6. CONCLUSION
I realize that the picture that I have presented is mostly a negative one.
If we are interested in the tools of metaphysics - the tools that can help
us draw up an inventory of the world - then I have only offered skepti-
cal arguments against some common practices. Still, I don't intend this
to be exclusively a negative picture. To the contrary, all of this sug-
gests that we have to take metaphysics seriously, since we cannot hope
to derive it from our linguistic practices. And to me this is a positive
thing. This is the beginning of the book, so I thought I could take the
liberty of looking at the dark side first. By the end of the book this fog
of negativity will have dispelled and most of the problems - most of
the genuinely metaphysical problems - will (I am sure) have been
solved.
Notes
1 Burgess and Rosen [1997], pp.222-233.
2 Ducasse [1942], p.233.
3 Melia [1995], p.224.
12 I'm going along with the received doctrine here, but see Oliver [1999].
13 Though I am in fact skeptical: see Casati and Varzi [1994], Chapters 3 and
12.
14 See Horgan [1978] for a representative statement.
15 The point is made in Bennett [1988], p.176.
16 See Parsons [1987-88].
Enrico Berti
According to van Inwagen, "Ryle has made no case for the thesis that
existence is equivocal". And he adds - but it is not clear whether refer-
ring to Ryle or in general - "I know no argument for this thesis that is
even faintly plausible". This enables him to say: "We must therefore
conclude that existence is univocal".
In fact, Ryle was not the only philosopher who admitted different
senses of being. Before him John L. Austin, the first who introduced
Aristotle in the analytical Oxford philosophy, in his famous article en-
titled "The Meaning of a Word" (1940) claimed that" 'exist' is used
paronymously", i.e. with a "primary nuclear sense" and other senses
dependent on it, just as 'healthy' in Aristotle (Austin [1970], p.71). In
Sense and Sensibilia he wrote:
'real' (the translation of the Greek, 'on', i.e. 'being') is not a normal word
at all, but is highly exceptional; exceptional in this respect that, unlike
Enrico Berti, Being and Essence in Interpretations ofAristotle 81
'yellow' or 'horse' or 'walk', it does not have one single specifiable, al-
ways-the-same meaning. (Even Aristotle saw through this idea). Nor does
it have a large number of different meanings - it is not ambiguous, even
'systematically'. (Austin [1962], p.64, italics in the text)
As everybody knows, the doctrine that being, and perhaps also exis-
tence, is at least not univocal, if not equivocal, is distinctive of Aris-
totle. He refers to it several times. 2 Admittedly, Aristotle never brings
any argument in defence of this doctrine. By so doing, he gives the
impression of considering this doctrine perfectly evident, though he
was clearly persuaded that he was the first philosopher who discovered
this truth. He blames in fact not only Parmenides, Zeno and Melissus,
i.e. the Eleatics, for having conceived being as univocal, but also De-
mocritus, and even Plato. 3 In the whole Corpus Aristotelicum there is-
as far as I know - only one passage where Aristotle makes an attempt
to prove that being is not univocal, at Metaphysics B 3, 998b22-27.
This is an astonishing situation, but just for this reason the passage is
worthy of some attention, greater than that which is usually reserved to
it even by Aristotle's interpreters.
Even van Inwagen's commentator Gary Rosenkrantz fails to men-
tion this passage. On the one hand, he admits that, because of the inti-
mate connection between the 'is' of existence and the 'is' of predica-
tion, or the 'is' of existence and the 'is' of identity, "until van Inwagen
provides some reason to think that predication and identity are univo-
cal, his argument against the equivocacy of 'exist' is not clearly valid"
(Rosenkrantz [1998]). On the other hand, he argues, this time against
Ryle, that there are species (not genera or different modes) of exis-
tence, because the system of the ontological categories "has entity (or
entityhood) as its summum genus". But this is exactly what Aristotle
denies in the quoted passage. Before analysing this passage, however,
let us say something about the relation between being and existence in
Aristotle.
about the ambiguity of verbs for being like the Greek einai", according
to which these verbs express either identity, or predication, or exis-
tence, or finally subsumption, but also that Aristotle consciously con-
sidered this thesis, and rejected it (Hintikka [1999], p.782). This would
be documented in Metaph. r 2, l003 b22-30, where Aristotle says that
"to be and to be one are the same [... ] since 'one man' and 'man' as
well as 'existent man' and 'man' are the same thing, in that the redu-
plication in the statement 'he is a man and an existent man' yields no
fresh meaning". Aristotle was operating with an unitarian concept of
being, says Hintikka, because "in the fullest sense, einai had to com-
prise all of the first three Frege-Russell senses of being, that is predi-
cation, existence, and identity".
This does not mean, Hintikka argues, that Aristotle did not intro-
duce distinctions between different uses, and even different senses, of
being other than the Frege-Russell ones. "The most prominent of them
are the distinction between being in different categories as well as the
contrast between potential being and actual being". In this way Hin-
tikka shows that it is impossible to isolate, in Aristotle, the existential
from the predicative sense of being, and that, since the latter sense is
multiple, inasmuch as there are many kinds of predicates, i.e. the
"categories", the existential sense is also multiple.
The same result had been reached five years before by Lesley
Brown. By focusing on the use of the verb 'to be' in Greek philoso-
phy, she had concluded that, while distinguishing 'to be something',
i.e. the predicative use of being, from 'to be haplos', i.e. its existential
use, Aristotle thinks that they are closely connected. So, when he
claims that 'is' is said in many ways, it is impossible to decide whether
he is analysing existential or predicative uses. Lesley Brown says
Aristotle insists on the inter-relations of the question 'Is Xl' and 'What is
Xl'. [ ... ] The distinctions he does consider philosophicalJy important -
chiefly that between essential and accidental being, and the different ways
in which, as he puts it, 'being is said' which correspond to the different
categories - cut across the syntactic distinction between complete and in-
complete, and do not correspond to the semantic distinction between 'ex-
ists' and the copula. (Brown [1994], pp.233-236).
Enrico Berti, Being and Essence in Interpretations ofAristotle 83
Owen proposed to indicate this sense of 'is' by two asterisks, i.e. is**,
while the first can be indicated by one asterisk, i.e. is*. "Then we can
say that while Arrowby is in existence there is** at least one man still
in existence, but if Arrowby dies there is** one man who is* no
more". This second sense of existence corresponds, as we can see, to
the existence that the Quinean tradition treats as univocal. According
to Owen, Aristotle nowhere distinguishes these two uses of the verb.
So it is impossible to say whether his analysis of the different predica-
tive senses of 'exist' applies to being* or to being**. We will see that
the argument against the univocity of being brought in Metaph. B ap-
plies not only to being*, but also to being**.
For the sake of completeness I would like to mention an article by
Paul Grice, who maintains that the passages quoted by Owen show
that the semantic multiplicity attributed by Aristotle to the verb 'to be'
concerns its existential use, but it is possible to find in his works other
passages, one of which is for instance Metaph. !1 7, 1017a23-31 (there
is no difference between 'man walks, or flourishes' and 'man is walk-
ing or flourishing'), from which it can be seen that this semantic mul-
tiplicity concerns the copulative use of being, i.e. the copula (Grice
[1988]t It is surprising that none of these interpreters, and of the
many others who treated the problem of semantic multiplicity of being
in Aristotle, engaged in a close analysis of the passage of Metaph. B,
where Aristotle tries to establish the reasons of this multiplicity. Only
recently, in a book entirely dedicated to homonymy, has Christopher
Shields provided a careful examination of that passage, and arrived at
the conclusion that it does not reach its aim, i.e. it does not demon-
strate that being has a multiplicity of meanings (Shields [1999]).
serving that "it is not possible for either One or Being to be a genus of
things" (992b22).6 We shall examine Aristotle's reasons for this thesis
later on. For the moment I would like to point out that not only Being,
which includes, as we will see, existence, but also One, i.e. the notion
to which existence is reduced in the article of van Inwagen, are not
univocal. This emerges from a passage of the Topics, where Aristotle
provides the following rule:
Look also at the genera of the predicates signified by the term, and see if
they are the same in all cases. For if they are not the same, then clearly the
term is homonymous. (I 15, 107a3-5)
The sentence "if they are not the same" genus means, I suppose, not
only that the predicate, i.e. the property, signified by the term in ques-
tion, is instantiated many times, i.e. belongs to different things, but
also that its many instantiations do not fall within one and the same
genus. This situation occurs when a property can be associated with
different categories, as we will see in the example brought by Aris-
totle. The categories, in fact, cannot be reduced to a single genus.
Resting on this, Aristotle may formulate the general rule, following
which every predicate, whose instantiations do not fall within the same
genus, is homonymous.
According to Cat. 1, the 'homonyma' are the things which "have
only a name in common and the definition of being which corresponds
to the name is different" (1 al_2). From the passage of the Topics it ap-
pears that Aristotle calls homonymous also the term which signifies
homonymous things. In particular there is no doubt, and all interpreters
agree on this, that a homonymous term has many meanings, i.e. is not
univocal.
The homonymous term Aristotle considers in the passage of the
Topics is "good" (agathon).
Good in the case offood is what is productive of pleasure, and in the case
of medicine what is productive of health, whereas as applied to the soul it
is to be a certain quality, e.g. temperate or courageous or just; and like-
wise also, as applied to a man. Sometimes it signifies what happens at a
certain time, as (e.g.) what happens at the right time; for what happens at
the right time is called good. Often it signifies what is of a certain quan-
86 Individuals, Essence, and Identity
tity, e.g. as applied to proper amount; for the proper amount too is called
good. (107"5-12)7
species".12 But Aristotle is not saying that the same genus 'animal'
would be predicated 'many times' (pollachis) of the species, but rather
that 'many animals' (polla zoia) would be predicated of it. Aristotle
does not seem to be thinking that it is impossible to predicate the same
genus of many differentiae, but rather that it is impossible to predicate
the genus even of a single differentia, because in this way many gen-
era would be predicated of the same species.
Tricot ([1950]) explains: "If the genus were attributed to the
differentia, the genus would be many times over an attributes of the
species [ ... ]. Indeed, animal is said once of the human species, once of
the rational differentia, which would surely constitute a new species,
relative to which the same difficulty would arise anew". But in this
way too we should read 'many times' instead of 'many animals'. Be-
sides, the consequence of reducing the differentia to a species belongs,
as we will see, to the second argument and it cannot be attributed, at
least immediately, also to the first. Zadro ([1974], pA85), on his part,
interprets: "Given that the differentia is in its turn predicated of the
species, this would also means that zoion would be predicated of
'man', with the result tha zoion would be a property both of the e idos
and of the dia/ora, and the definition by way of the genus would be
redundant". I agree that, if the genus could be predicated of the differ-
entia, it would be predicated twice of the species, but the consequence
indicated by Zadro, that in this way the definition would become re-
dundant, does not seem to me impossible, as also Aristotle himself
thinks.
In his book entirely dedicated to homonymy in Aristotle, Shields
provides a close analysis of the passage in Metaph. B 3. He argues that
Aristotle is here concerned with being in its existential sense. Shields
focuses also on the passage in Top. VI 6, and criticises the other inter-
preters (e.g. Grice) for not having made use of it. But he follows Waitz
and Ross in interpreting polla zoia as pollachis to zoion, and argues
that Aristotle's argument, thus understood, "seems to appeal implicitly
to homonymy". According to Shields, if the difference, e.g. 'being
two-footed', counts as being an animal, the notion of 'animal' em-
ployed in saying 'being two-footed is an animal' must be significantly
unlike the notion employed in 'Callias is an animal'. Therefore, "if one
were to predicate 'animal' of its various differentiae, then one would
Enrico Berti, Being and Essence in Interpretations of Aristotle 89
the first argument of Aristotle would converge with the second, and
the interpretation of Tricot would be justified.
Even in the case in which 'animal' would be predicated of 'ra-
tional' without being its genus, but simply as an accident of it, it would
be predicated also of 'man', i.e. of the species of which the difference
is predicated, in virtue of the logical rule according to which the predi-
cate of a predicate is a predicate of the subject (nota notae, nota rei).
In any case, Aristotle's arguments ultimately rest on the doctrine of the
definition of a species by its genus and its differentia, where the genus
expresses what every species has in common with the other species
within the same genus, and the difference expresses what distinguishes
one species from the others species of the same genus. If the genus
could be predicated of the difference, either as its genus or as an acci-
dent, then the definition would only indicate the common aspects of
the species, losing what enables it to distinguish them from one an-
other. If Aristotle's arguments are valid, as I have tried to show, we
can conclude that his attempt to demonstrate the homonymy of being,
or at least that being is not univocal, is successful. Metaph. B makes it
clear that the non-univocity holds both for being conceived as exis-
tence as well as for being conceived as unity.
The 'univocacy of number', invoked by van Inwagen, does not
hold among objects belonging to different genera, as it is the case of
beings. We cannot, in fact, count the objects contained in a room, if
they belong to different genera. We can count, for instance, persons,
tables, chairs, books. In the Aristotelian language they are substances.
But we cannot count, together with them, the colours of the tables, the
weight of the books, the actions or the feelings of the persons, though
we must admit that each of these things does exist and is at least one
instantiation of its class. In particular, Aristotle said that "to be one
[ ... ] is specially to be the first measure of a kind"l\ and "the measure
is always homogeneous with the thing measured: the measure of spa-
tial magnitudes is a spatial magnitude, and in particular that of length
is a length, that of breadth a breadth, that of articulated sounds an ar-
ticulate sound, that of weight a weight, that of units a unity".15 This
doctrine was endorsed by modem analytical philosophers such as P.T.
Geach, M. Dummett and C. Wright and it was attributed by them to
one of the founders of this tradition, i.e. G. Frege. 16
Enrico Berti, Being and Essence in Interpretations ofAristotle 93
Everything which is healthy is related to health, one thing in the sense that
it preserves health, another in the sense that it produces health, another in
the sense that it is a symptom of health, another because it is capable of it.
And that which is medical is relative to the medical art, one thing in the
sense that it possesses it, another in the sense that it is naturally adapted to
it, another in the sense that it is a function of the medical art. (Barnes (ed.)
[1985], 1053"24-27)
This happens also with being, where everything that it is said to be ei-
ther is a substance (ousia) or it is relative to substance:
some things are said to be because they are substances, others because
they are affections of substance, others because they are processes to-
wards substances, or destructions or privations or qualities of substance,
or productive or generative of substance, or of things that are relative to
substance, or negations of some of these things or of substance itself.
(1003"33-b l0)
The same situation holds - said Austin - also for the word 'exist', i.e.
being.!S Now, it seems to me that, if the Aristotelian doctrine of the
multiplicity of meanings relative to one were interpreted in this way,
i.e. if substance (ousia) was taken as the nuclear meaning, "contained
as a part" in all the other meanings, and "common" to all of them, as
suggested by Austin, substance would become the genus of being and
the other meanings would be only specifications of this genus. They
would specify the genus without modifying it. The genus is in fact the
common part of the definition, and it is specified by the suitable dif-
ferences. If therefore the definitions of different meanings have a part
in common, and this part is always the same, this part would be neces-
Enrico Berti, Being and Essence in Interpretations ofAristotle 95
sarily the genus. Moreover, given that the genus is part of the essence,
substance, conceived as a genus, would be nothing but an essence, the
essence of being.
But the examples given by Aristotle do not suggest this idea. The
different meanings of 'healthy' are in different relations with health.
Apparently, the relations Aristotle is thinking of are those of produc-
tion, preservation and manifestation of health. They are proper to
things such as a healthy medicine, a healthy climate, a healthy com-
plexion, which do not belong to the same genus and do not have the
same essence. We cannot say that health is the essence of the healthy
medicine, or the healthy climate, or the healthy complexion. We
should say that health is the product of the healthy medicine, or the
thing preserved by the healthy climate, or finally the thing manifested
by the healthy complexion. In all these cases there is surely a relation
between the single healthy thing and health. This relation is neverthe-
less different in each case. The different healthy things cannot be
therefore considered as mere modifications, or qualifications, of
health. This is also the case of being, at least in my opinion. Each
meaning of being, i.e. each category of being, substance, quality,
quantity, relation, time, place, etc., is in a certain relation with the sub-
stance and it is said to be in virtue of this relation. But substance is not
the essence of quality, quantity, relation, time, place, etc.. Nor are
these latter mere modifications, or qualifications, of substance in the
sense that they should be substances with some accidental qualifica-
tions: they are other kinds of being, different from the substance,
though dependent on it.
In the Eudemian Ethics, illustrating the different kinds of friend-
ship, Aristotle himself explains that they are all relative to one, which
is the primary, just as in the case of the word 'medical', and adds:
"Everywere, then, we seek for the primary. But because the universal
is primary, they also take the primary to be universal, and this is an
error" (Eth. Eud. VII 2, 1236a22-25). The target of his criticism are the
Platonists, who took the primary as a universal, i.e. a common aspect,
a common predicate, like the genus. According to Aristotle, this is a
mistake: the primary meaning of friendship is not the genus of which
the other meanings are the species. "The primary is - Aristotle says -
that of which the notion is present in us" (en hemin), not 'in the defi-
96 Individuals, Essence, and Identity
nition of all' (en pasin), as many interpreters believe. 19 This means that
the primary is only a term of reference, i.e. that to which the other are
in relation (pros), and it is common to all just for this reason, and not
because it is a universal, in conformity with which (kata) the other are
said.
Interpreters do not completely agree with one another on this
point. In a famous article entitled 'Aristotle and the Ambiguity of Am-
biguity', Hintikka distinguishes three cases: the case in which the
multiple applications of a term have totally different definitions
(homonymy), the case in which the multiple applications of a term
have totally identical definitions (synonymy), and finally the case in
which the multiple applications of a term have definitions which are
partially identical and partially different. According to Hintikka, this
last case is that of being. This is described by Aristotle sometimes by
the relation of its applications to one central point (pros hen), and
sometimes by the affirmation that their definitions are derivable from
each other "by adding to and taking away".20
There is a passage of Metaph. Z 4 which illustrates the latter
situation, and which Hintikka quotes in support of his interpretation. It
sounds:
Essence (to ti en einai) will belong, just as the 'what' (to ti estin) does,
primarily and in the simple sense to substance, and in a secondary way to
the other categories also, - not essence simply, but the essence of a quality
or of a quantity. For it must be either homonymously that we say that they
are, or by making qualifications or abstractions (prostithentas /wi aphai-
rountas) (in the way in which what is not known may be said to be
known), - the truth being (to ge orthon) that we use the word neither
homonymously nor in the same sense, but just as we apply the word
'medical' when there is a reference to one and the same thing, not mean-
ing one and the same thing, nor yet speaking homonymously; for a pa-
tient and an operation and an instrument are called medical neither homo-
nymously nor in virtue of one thing, but with reference to one thing (pros
hen). (1030"29-b 3)
In this passage it is not clear whether the use of the verb to be "by
making qualifications or abstractions" coincides or not with the use of
this verb with reference to one and the same thing. In the first case, the
Enrico Berti, Being and Essence in Interpretations of Aristotle 97
of beings (ousiai ton onton), and whether each of them, without being
something else, is being or one respectively, or whether it is necessary
to inquire what being and one really are, supposing that another nature
underlies them as a subject" (1001 a4_8). He attributes the first opinion
to some philosophers and the second to others. As supporters of this
opinion he gives the names of Plato and the Pythagoreans, whose doc-
trines are described as follows: "Plato and the Pythagoreans think that
neither being nor one is something different, but that this is what their
nature is, supposing that its ousia is to be one and being" (1001 a 9-
12). In Aristotle's language ousia means not only 'substance' but also
'essence', and this last meaning should be preferred when the term is
followed by the genitive (ousia of something). But if so, we must con-
clude that, for Aristotle, Plato and the Pythagoreans conceived of be-
ing and one as substances, whose essence was respectively to be being
and to be one.
Aristotle's criticism of the second opinion, which he ascribes to
Empedocles, ultimately rests on arguments which have sense only
from an Academic, i.e. Platonic, point of view. After this criticism,
Aristotle goes back to the first opinion, which he reformulates in the
following way: "If there exists some one itself (auto hen) and being,
then one and being are necessarily their substance (ousia). For nothing
different is predicated of them universally, but rather they themselves"
(l001a27-29). Aristotle normally uses the pronoun "itself' (auto) to
designate Platonic Forms.32 In the following line he applies it also to
being (auto on). Also in the light of this it seems that he is thinking of
Plato rather than the Pythagoreans, and that he is ascribing to Plato a
doctrine which considers Being itself, i.e. the Form of being, and One
itself, i.e. the Form of one (it is not clear whether they coincide or not),
as the substances, i.e. the formal causes, of all the beings. This corre-
sponds partially to the doctrine of the One and the indefinite Dyad as
principles of all things, ascribed to Plato in book A of Metaphysics.
What is most interesting is the criticism which Aristotle addresses
to this doctrine:
But on the other hand, if there is to be some being itself (auto on) and one
itself (auto hen), there is much aporia about how anything different will
exist alongside them: I mean, how beings will be more than one. For that
Enrico Berti, Being and Essence in Interpretations ofAristotle 105
which is different from being is not. So, in line with the argument of Par-
menides, the necessary consequence is that alI beings are one and that this
is being (to on). But either way it is difficult. (l 00 1"29_b1)
Notice that the sentence "that which is different from being is not" is
justified only "in line with the argument of Parmenides", i.e. only if
we admit that being has only one meaning, the doctrine that Aristotle
usually ascribes to Parmenides. 33 Ross thinks that here Aristotle is not
justified in ascribing to Plato the Parmenidean notion of being, be-
cause - he says - "for Parmenides it (to on) means 'what is', i.e. the
universe; for the Platonists it means 'being', i.e. the attribute of exis-
tence. It is this abstraction that they make a substance, and there is
nothing in this to prevent their recognizing of other substances".34
But Aristotle's interpretation of Plato's notion of being is not
without reasons. Elsewhere he says that the Platonists admitted two
principles, i.e. the One and the indefinite Dyad, because
they frame the difficulty in an old-fashioned way (aporesai archaikos),
for they thought that all things that are would be one - viz. Being itself
(auto to on), if one did not join issue with and refuted the saying of Par-
menides: 'For never will this be proved, that things that are not are'. They
thought it necessary to prove that that which is not is; for thus - of that
which is and something else - could the things that are be composed, if
they are many. (Metaph. N 2, 1089"1-6)
FOTIll,i.e. an Idea. And this was also the condition for conceiving be-
ing and one as the essence of a substance, i.e. being itself (ipsum esse
subsistens) and one itself (ipsum unum subsistens). In conclusion, if
the primary substance is the essence of being, being must be under-
stood univocally. If being has an essence, it is this essence. It cannot
be many essences. But this is impossible; because we see many things,
and their differences are existing and each of them is one. This is the
core of Aristotle's criticism of Plato as it is exposed in Metaph. B 4.
This criticism ultimately rests on the argument offered in Metaph. B 3,
and the view that Being and One cannot be genera.
Notes
Parts of this paper first appeared in the Proceedings of Aristotelian Society,
vol. CI part 2, 185-207; they are here reprinted by courtesy of the Editor of the Aris-
totelian Society © 2001.
I See van Inwagen [1998].
not possible for the genus to be predicated of the differentiae") and from the com-
mentary by Madigan ([1999], p.73).
!O This is the translation by Shields (ed.) [1999], pp.247-248.
12 Ross is clearly following Waitz (cf. Waitz (ed.)[1844-1846], II, p.500), who
interprets polla zoia as pollachis to zoion .
13 I am grateful to Stephen Menn for sending me, as yet unpubblished, his book
on The Aim and the Argument of Aristotle's Metaphysics, in which he shows how
this case would produce a sort of absurd infinite regress.
14 Aristot. Metaph. I I, 1052 b I6-18.
15 Ibid., 1053'24-27.
things said "in conformity with one" (kath 'hen) are distinguished from the things
said "in relation (pros) to one nature" in 1003 b I2-14.
24 This was noted by Leszl. On the interpretation of Owen see also Code [1996].
[1972].
31 More than twenty years ago I wrote an article on this subject, which, though
33 Cf. Madigan (ed.) [1999], p.113. Apparently, the Aristotelian criticism of the
existence of a being itself is shared by Owens [1973], p.21-35, esp. 33: "If the exis-
tence known in observable things is viewed as having a content within itself, it is left
open to Parmenides' relentless reasoning [ ... ] It leaves no room for anything else".
34 Cf. Ross (ed.) [1953], I, p.245.
SOME COMMENTS ON PROF. ENRICO
BERTI'S "BEING AND ESSENCE IN
CONTEMPORARY INTERPRETATIONS OF
ARISTOTLE"
David Charles
1. INTRODUCTION
(1) In Post. An. Bl and 2 (89 a33, 90b 12ff), there is talk of sub-
stances such as the moon, the sun, and the earth as 'being without
qualification (haplos)' which is contrasted with their 'being some-
thing', such as white, dark etc. This discussion appears, at first sight,
to mark the distinction as between 'A exists' and 'A is F', the existen-
tial and the predicative uses of the word 'to be'. At least, this is the
distinction that Cautious Aristotle is committed to maintaining. Else-
where, Aristotle seems to draw precisely this distinction when he dis-
cusses under what conditions 'a is F' entails that 'a is'. (Soph. El.
1.167a l-2, 180a36-8: see de Int. 21 a25-33). For the inference in ques-
tion seems to be one from the predicative to the existential sense of 'to
be'.
(3) Notwithstanding (2), for Cautious Aristotle, the verb 'to exist'
is used in its primary sense in the case of substances. If so, in his view,
the following will be true:
David Charles, Some Comments on Prof Berti 113
Many will say that, despite the evidence so far noted, Aristotle's views
cannot be those expressed by Cautious Aristotle. Indeed, it has fre-
quently been suggested, at least since G.E.L. Owen's influential paper
[1965], that for Aristotle to be is, in the final analysis, to be a sub-
David Charles, Some Comments on Prof Berti 115
'Does A exist?'
and answer it (in the optimal case) by finding a basic (or immediate)
middle term (e.g.: 'fire being quenched') which specifies the essence
of thunder. If he finds such a middle term, he will know that there is
one unified phenomenon, thunder, which exists. In this way one will
have established the existence of thunder as a kind.
The question
'Does noise belong on the clouds?'
Are there any objects which satisfy the linguistic expression 'noise
in the clouds' .
staying at the level of objects (and properties) and noting their inter-
connections.
Aristotle makes several moves to articulate the question:
'Does man exist?'
In Post. Anal. B 8, the question can be phrased using (e.g.) 'a specific
type of animal' or perhaps as elsewhere 'biped animal' (see Metaph. r
4, l006 a3lff), as the account of what 'man' signifies. In this case, to
establish that man exists is to establish that (e.g.) bipedality belongs to
animal by finding as a middle term one which specifies (in the optimal
case) the essence of man: e.g. being rational. In Metaph. Z.17, the ar-
ticulation of the question appears to be rather
'Why is matter of this type some definite thing?'
or more precisely
'Why is matter of this type arranged thus?'
to which the answer would be given in terms of what it is to be a man
(i.e. man's essence). For this explains why matter of this type is ar-
ranged as it is in this case.
Aristotle's aim is to regiment existence questions in such a way
that he can answer them using the resources of demonstration, and in
this way prove that (e.g.) man is indeed one properly unified kind. But
none of this requires him to say that the statement
'Man is without qualification/haplos'
must be elliptical for 'Man is an F'. It merely shows that the way in
which Aristotle seeks to prove (in his favoured demonstrative form)
that man exists rests on his complex accounts of what the term signi-
fies. For the latter introduce claims of the right form to be proved by
the presence of a middle term. In this way, scientific proof of the ex-
istence of a kind can depend on finding a middle term even though the
statement of existence is of the straightforward existential form:
'Man exists.'
David Charles, Some Comments on Prof Berti 121
Cautious Aristotle suggested that there are different, but focally re-
lated, senses of the existential verb 'to be' corresponding to different
types of categorial claims such as 'A is a substance', 'B is a quality,
etc .. In his view, there have to be different senses of 'exists' since
there are major differences in the ways in sentences such as 'A exist'
and 'B exist' are true. While he did not analyze 'A exists' as 'A is a
David Charles, Some Comments on Prof Berti 125
Notes
For discussion of this issue, see Williams (1996]
I
This point is well made by Barnes [1995], p.81 ff.
2
3 For further discussion of these issues, see Charles [2000].
4 I am indebted to Adam Beresford, Paolo Crivelli, Kit Fine and Peter van In-
wagen for their comments on an earlier version ofthis paper.
EXISTENCE, IDENTITY AND AN
ARISTOTELIAN TRADITION
Alex Orenstein
Quine opens "On What There Is" by endorsing the claim that every-
thing exists. The question:
What is there? [ ... ] can be answered [ ... ] in a word - 'Everything' - and
everyone wiJI accept this answer as true.
127
A. Bottani et aL (eds.), Individuals, Essence and Identity, 127-149.
© 2002 Kluwer Academic Publishers.
128 Individuals, Essence, and Identity
is inconsistent with what Quine and most of his readers take as an ax-
iom of identity theory, its total reflexivity:
(\ix) (x = x).
From this axiom we can derive '(\ix)(3y) (y = x)' which is an identity
rendering of the English 'Everything exists'. Doing this gives further
support to Quine's opening endorsement. Furthermore, (\ix)(3y) (y =
x) is equivalent to
-,(3x)-,(3y) (y = x)
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 129
and the latter is the denial of the conclusion (2'). The identity explica-
tion of the issues intensifies the problem.
In a word, the conclusion is a logical falsehood given the total re-
flexivity of identity and the standard rules of predicate logic. In using
identity we back up Quine's English language claim that everything
exists, and that the conclusion is a "contradiction in terms".
Much the same problem was presented earlier in Mathematical
Logic (p .150). (Identity renderings of the relevant claims inserted by
me appear in square brackets and I have used bold italics for a clause
that we will return to in Part 3.)
To say that something does not exist [(3x) ....{3y) (y = x)], or that there is
something which is not, is clearly a contradiction in terms; hence '(Vx) (x
exists), [(Vx)(3y) (y = x)] must be true. Moreover we should expect leave
to put any primitive name of our language for the'x' of any matrix' .. .x ... ',
and to infer the resulting singular statement from '(Vx)(x exists)'; it is dif-
ficult to contemplate any alternative logical rule for reasoning with
names. But this rule of inference leads from the truth '(Vx) (x exists), [... ]
to the controversial conclusion 'God exists' [(3y) (y = God)] and the false
conclusion 'Pegasus exists' [(3y)(y = Pegasus)] [... ]. The atheist seems
called upon to repudiate the very name 'God', thus depriving himself of
vocabulary in which to affirm his atheism, and those of us who disbelieve
in Pegasus would seem to be in a similar position.
For Quine, the Plato's Beard problem arises when we allow for
empty/vacuous names in predicate logic with identity and acknowledge
that some denials of singular existentials express contingent truths. In
his paper "The Kant-Frege-Russell View of Existence: towards the
Rehabilitation of the Second-Level View"2, David Wiggins shows how
the problem arises even when one does not allow empty names or true
denials of singular existentials.
Wiggins presents a revisionist Fregean account of singular exis-
tentials. He disagrees with Frege and Russell, who went to some pains
to argue that singular existentials, such as 'Julius Caesar exists', are
130 Individuals, Essence, and Identity
meaningless. Wiggins does not say why he disagrees with Frege and
Russell. I assume that unlike the climate of opinion surrounding the
acceptance of Frege and Russell's views, today Wiggins and most of
us prefer accounts of logical form which are more accommodating of
natural language locutions. It is simply too high handed to dismiss sin-
gular existentials as being meaningless. 3 Aside from this revision,
Wiggins wishes to remain true to Frege's central insights. While
holding that 'J.e. exists' expresses a contingent truth, central Fregean
themes are adhered to.
He treats singular existentials in terms of Frege's view of exis-
tence and of quantification as second level property/concept instantia-
tion. Wiggins quasi-fuses the identity sign '=' with the singular term
'J.C.', and treats '[= J.e.]' as representing a Haecceitian property or
individual concept. The English 'J.e. exists' is then accommodated as
the Fregean higher level property instantiation (3x) x[= J.e.]. Saying
that Julius Caesar exists is to be thought of as saying that the property
of being identical with Julius Caesar is instantiated. Wiggins is uni-
formly employing Frege's views of existence and existential quantifi-
cation as property instantiation. For Frege, to state that horses exist is
to hold that the property or concept of being a horse is instantiated.
Wiggins is merely extending this property instantiation view of exis-
tence and quantification to the Haecceity or quasi-Quinian like prop-
erty of being identical with J.C. 4
So far so good. Wiggins has made his case for accommodating
singular existentials as second level Fregean property instantiation. But
questions remain concerning those denials of singular existentials
Quine and others consider. Wiggins examines the contingently false 'It
is not the case that lC. exists'. But he does not in the same vein con-
sider prima facie contingently true denials of singular existentials such
as: 'Vulcan/Pegasus does not exist'.
He takes an object dependent approach to empty names. Object
dependent accounts are not so much a solution to Quine'S original
puzzle for empty names as a way of avoiding it by not allowing any
significant role to empty names. On these views, empty names do not
play any role in logic and argumentation. All the customary principles
of logic stay in place. Empty names simply do not enter the picture.
They and the sentences containing them do not occasion revising stan-
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 131
dard logic. The sentence 'Vulcan does not exist' does not furnish a
truth vehicle let alone a truth. s Bivalence is not in question. There is
nothing to be true, false, or have some other value.
As I see it, object dependence comes in several forms. For Evans
and Salmon (though for different reasons), sentences with empty
names are meaningless and do not provide truth vehicles. For Straw-
son, they are meaningful but do not provide us with a truth vehicle (are
not used to make statements). Wiggins holds that the sentence 'Vulcan
does not exist' fails to express a truth. 6
By restricting himself to something like statements as truth vehi-
cles, the strings 'Vulcan does not exist' and its identity predicate logic
correlate '-{3x) (x = Vulcan)' never have to be taken account of.
There is no problem of empty names since the principles of logic do
not apply to sentences containing them. The situation is reminiscent of
the net with spaces between the netting large enough to have no effect
on fish of a smaller size.
While Wiggins does not have to worry about reasoning from 'Eve-
rything exists' to 'Vulcan exists', he does deal with a related problem.
As we have seen, if one accepts the total reflexivity of identity, '(Vx)
(x = x)" as a logical truth, then (given standard predicate logic and
representing 'a exists' as '(3y) (y = a)') the following are theorems and
in that sense would also be considered logical truths:
(1) (Vx)(3y) (y =x), i.e., Everything exists.
(2) (3y) (y = Julius Caesar), i.e., Julius Caesar exists.
The principle of necessitation sanctions a transition from non-modal to
modal logic. It states that
If A is logically true, then it is necessary that A.
problem does not arise. Wiggins, though, allows that 'Julius Caesar
exists' expresses a contingent truth. He presents its contingent status
modally:
Julius Caesar might not have existed: Pos--.(3y) (y = Julius Cae-
sar),
from
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 135
Fa
alone (as on the customary statement of the rule). An additional prem-
ise is required to the effect that a exists. They also reject the ordinary
rule of universal instantiation. In much of what follows what I say ap-
plies to these free logicians who state the additional premise in the
form of the identity: (3x) (x = a). Their revised universal instantiation
and existential/particular generalization rules appear as:
(Vx) (-:X-)
(3x) (x = a) i.e., a exists
therefore ~
and
~
(3x) (x = a)
Everything is in space.
Identity is the source of the problem and not the background logic of
quantifiers.
(1) How a Revisionist Fregean can concede that Vulcan Does Not
Exist
Let us adapt to our own purposes a view that Frege put forward in "On
Sense and Reference." He proposed there that a sentence containing
the vacuous name 'Odysseus' expresses a proposition. Frege assumed
that it did not have a truth value and that it should not be construed as
part of our serious truth telling use of language, but only as figuring in
stories, novels, etc .. The proposition/thought expressed presumably is
composed of an individual property/concept for 'Odysseus' and of a
property for the predicate 'while deeply asleep was disembarked at
Ithaca'. The individual property/concept for 'Odysseus' can be taken in
the spirit of Wiggins' suggestion. Just as Wiggins formed the individ-
ual concept for 'Julius Caesar': [= J.C.], one can form an individual
concept for the empty name 'Odysseus': [= Odysseus]. Empty individ-
ual concepts/properties were not a problem for Frege. He already al-
lowed for empty general properties, such as the property of being a
unicorn, or of being phlogiston.
We follow Frege and recognize a role for such propositions in sto-
ries and novels (where they might be governed by an implicit "in the
story" or "pretense" operator, or playa part in a game of make be-
lieve). But, unlike Frege, I propose that these propositions also playa
role in serious discourse. Such propositions/sentences can serve as
truth vehicles.
As a further step, I take it that in these serious non-fictional con-
texts, these sentences/propositions are false. This last step, treating
vacuous singular sentences/propositions as false, has precedents. It is
the view taken in the Lesniewskian tradition. It is also found in Kant,
and further back in Terminist logicians such as Ockham and Buridan.
Moreover, the English sentences, 'Vulcan is a planet', 'Vulcan does
exist' will receive the same truth value of falsity as they would on Rus-
sell's theory of descriptions. For Russell and those following him, such
sentences having the surface form of a singular sentence are translated
using his theory of descriptions into "existential" generalizations that
are false when the existence component is not satisfied.
Following Lambert's and Bencivenga's terminology, call this "the
negative solution." Such sentences-propositions are special cases of
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 139
ist' records the fact that the property of being a unicorn is not instanti-
ated in anything. The property of being Vulcan is like the property of
being a unicorn in not being instantiated, and '-,vulcan exists' is like
'-.,Unicorns exist' in being true.
A major difference between Frege and the older tradition concerns
a second use of 'instance'. A second sense of 'instance' and 'instanti-
ate' occurs in saying that the sentence/proposition stating that
Socrates is in space
is an instance of the sentence/proposition stating that
Everything is in space.
We speak of universal "instantiation." Similarly, the sentence or
proposition stating that
Socrates is human
is said to be an instance that is the basis for so called "existentially"
generalizing to the sentence or proposition stating that
Something is human.
In this second sense a sentence, a proposition, or some other truth ve-
hicle is an instance and not an individual object as in the first "property
instantiation" sense. This second sense is the "generalization to or
from an instance" notion of an instance.
The sentence
-.,Vulcan exists
records a fact concerning the absence of property instantiation (in-
stance in sense one) and this sentence, the denial, can serve as the ba-
sis, an instance in the second sense, for generalizing to the generaliza-
tion
Something does not exist.
When we generalize from the denial of property instantiation (sense
one), to the generalization that something is not a unicorn, we are not
committed to the existence of anything. From not instantiating (sense
142 Individuals, Essence, and Identity
one) and so not existing, nothing follows about what exists; e.g., from
being a unicorn is not instantiated, nothing follows about what exists
(what is instantiated), even though 'Something is not a unicorn' does
follow. Since the second - the generalization sense of an instance -
allows for generalizations on not having a property (not instantiating),
the generalizing (quantity/quantifier) sense of instantiation does not
have the force of existence. The "generalization to or from an in-
stance" sense of 'instance' is in a way broader than the having a prop-
erty sense of 'instance'. Generalizations represent non-existence as
well as existence.
On the Fregean view, existence is tied to both senses of instantia-
tion. I regard this as a mistake. On the older view, negations of singu-
lar sentences do not require existents in order to be true. Neither such
negations nor any generalizations (or truth functional constructions)
derived from them require existents. So quantification, so called
"existential" quantification, has existential import only when it is
grounded in singular/atomic sentences - genuine property instantia-
tion. Quantification does not have existential force when it is grounded
in negations of singular sentences (a property not being instantiated).
The same view is found in traditions that make serious use of the
copula such as in Kant and in Lesniewski. Kant thinks of existence in
terms of the copula. It is the copula tie between singular noun/term and
the traditional predicate noun/term that is linked with existential im-
port and not the quantifier 'some'. "Logically, it [being] is merely the
copula of a judgment." (Kant, Critique of Pure Reason, Smith (ed.)
[1953] p.504) This copula account comes to the same as the property
instantiation one, if we are allowed to put the copula account of the
truth of singular judgements along the following lines. A singular
judgement is true when the individual the singular term/concept refers
to instantiateslhas the property that is represented by the predi-
cate/noun. Kant uses language employing notions of positing. A sin-
gular judgement is true when the object the subject concept posits is
posited by the predicate concept. When no object is posited the judge-
ment is false (the negative solution), (cf. Orenstein [1978], Ch. 4).
The above points are encapsulated in the maxim of traditional
logic that quality, not quantity, determines existential import (see
Thompson). Instantiating a property (existence) is recorded in atomic
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 143
sentences. Such sentences are affirmative, and the property being in-
stantiated is recorded in the predicate being applied (syntactically) to
the singular term. The sentence is true when there is an individual the
singular term represents and it has/instantiates that property. The sen-
tence is false and its negation true, when the individual does not in-
stantiate the property. This can be when there is an individual and it
does not have the property, or when there is no such individual (e.g.
Vulcan). Non-existence, as failure of a property to be instantiated, is a
sufficient condition for the falsity of an atomic sentence and the truth
of its negation. Quality determines existential import, in the sense that
any sentence whose truth follows from the negation of an atomic sen-
tence inherits that lack of existential import. Only sentences whose
truth requires the truth of an atomic sentence (property instantiation)
have existential import.
(3) Non-existents are not self-identical
which clashes with the contingent truth that Vulcan does not exist:
-,(3x) (x = Vulcan).
which the same thing is predicated of itself'. I say "in effect" because
the issues surrounding total reflexivity were discussed under the topic
of self-predication. For the Terminists this was the crucial test case for
existential import. They dealt with it in connection with the sophism
A chimera is a chimera.
Their discussion assumes that generalizations are explained in terms of
a doctrine of descent to the singulars. Universal and particular gener-
alizations are true or false when conjunctive and disjunctive expan-
sions of suitable singular sentences are true. So
All/Some men are mammals
requires the truth of
This man is this mammal and/or that man is that mammal and/or
etc.
A singular sentence figuring as a conjunct/disjunct underlying the
sophism would look like
This chimera is this chimera.
Such a singular sentence is false when the subject term is vacuous.
Anachronistically speaking Gumping from Ockham to identity in
predicate logic) such false singular sentences are counterexamples to
the total reflexivity of identity.
The views put forward in section 4 can be summarized for first order
predicate logic with identity:
1. Allow atomic/singular sentences (propositions, or your favorite
truth vehicle) of the form 'Fa' where the 'a' position can be
taken by empty names. Allow for truth functional compositions
and generalizations.
2. Inspired by the correspondence conception of truth such atomic
sentences are true provided the subject denotes one of the ob-
jects the predicate applies to. When the subject term is empty
the sentence is not true. Assuming bivalence, it is false. Its ne-
gation is true.
146 Individuals, Essence, and Identity
The truth condition for basic canonical instances such as 'Fa', in non-
vacuous cases is similar to the ones found in most logic texts for non-
vacuous atomic sentences. 'Fa' will be true when there is an a and it is
an F (a is a member of the set 'F' has as its semantic value). While dif-
ferent accounts can be given of vacuous singular sentences, in this pa-
per I adopt the negative solution and treat them as false. Like Terminist
logicians and Lesniewskians, 'Fa' is false when either 'a' is vacuous
or there is an object that is not part of the semantic value of 'F'. Such
singular sentences are false and their negations true when they contain
a vacuous term. I assume the usual conditions for negation, conjunc-
tion and the other connectives.
The truth conditions for generalizations are an adaptation of Ben-
son Mates method of beta-variants. Mates provided an account of un-
restricted quantifiers of first order logic which relies on taking a given
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 147
val (3x) -:X'- is true iff val-a-- is true under at least one beta-
variant.
Consider the following examples of how these truth conditions are ap-
plied. A sentence such as
Something exists, i.e., (3x)(3y) (x = y)
is true, since at least one instance is true on at least one beta-variant,
e.g., (3y) (a = y) on at least one interpretation 'a' is assigned an object
e.g. Kant.
Something does not exist, i.e., (3x)---,(3y) (x = y)
is true. Consider the case where the substitution instance for 'x' is
'Vulcan'. '---,(3y) (Vulcan = y)' is true given the falsity of the sentence
it negates, i.e. '(3y) (Vulcan = y)'. There is no atomic sentence 'Vulcan
= a' which is true; no matter what object is assigned to 'a'. Since there
is no beta-variant on which 'Vulcan = a' is true, i.e., its negation, i.e.,
'---,(3y) (Vulcan = y)' is true. As true it serves as the interpretation that
accounts for the truth of '(3x)---,(3y) (x = y)'. Recall that '(3x) -:X'-' is
true if there is an instance of it that is true on a given interpretation.
Notes
I There are several ways of dealing with these problems. In Orenstein [2000] I
examine the problem without appealing to identity. In Orenstein [1990] I take up
Quine'S statement of the problem in Mathematical Logic (see Quine [1940]).
2 Cf. Wiggins [1994]. There are many varieties of the slogan that exists is not a
real predicate. It is a mistake to cite Kant as in the Frege-Russell-Quine tradition of
construing existence in terms of quantification. Kant in his general/formal logic is in
the Aristotelian Terminist-Lesniewskian tradition of construing existentials in terms
ofthe copula. See the last section of this paper and Orenstein [1978].
3 Wiggins on p.94 cites two strings which are translations into English of Ger-
man sentences judged meaningless by Frege 'There is Julius Caesar' and 'Julius Cae-
sar exists'. I believe most who disagree with Frege would say that the second makes
sense and is meaningful, but the first does not appear to be well formed and to con-
stitute a sentence by itself. The first appears more well formed when we insert an
article, i.e., 'There is a Julius Caesar'.
4 Wiggins suggestion differs from Quine's in that for Quine '= a' is a fused
predicate and has no logical structure so that one cannot existentially generalize with
Alex Orenstein, Existence, Identity, and an Aristotelian Tradition 149
conditions for identifying the referent of a name (see Sainsbury [1990] and Wiggins
[1994]).
7 Numerous questions remain about the 'really' operator:
(a) Wiggins quite rightly extends Evans thoughts on 'really' from story telling to
hypothesizing. While this is all to the good, what it indicates is the ubiquity of
'really' claims. We can add 'really' to just about any claim. Instead of saying the
battIe of Hastings took place on such and such a date we can say that it really took
place on that date. As for fiction and elsewhere, we are not limited in using 'really'
in connection with exists. Instead of saying that Sherlock Holmes did not live on
Baker street we can say that Holmes did not really live on Baker street. See Wiggins
own example of actors - it is not an existential.
(b) 'Really' brings background assumptions such as beliefs or imaginings into
the foreground, Within the background assumptions we are calling attention to their
status as background assumptions.
(c) The Spice Girls iteration is derived from a common form of speech: What do
you want? What do you really want? What do you really really want? 'Really' has a
force not mentioned by Evans and Wiggins: that of emphasis of some sort or of
forcing one to look at the reasons for making such a claim.
(d) In their deployment of the 'really' operator in connection with existentials,
Wiggins and Evans are being ad hoc. They are arbitrarily singling out one sort of
case where 'really' may be used without saying how it has a special use here. They
are leaving out the use of 'really' to express surprise: I really never thought he would
- , to cast doubt: Did you really think -?, to simply reiteratc, as used for contrasting
in general not just for fiction/pretense or hypothesizing e.g., to ask for speaker's justi-
fication in making the non-really assertion or by the speaker of elliptically offering or
claiming assurance as to the justification.
(e) Besides, if 'really' has scope, then, given compositionality, we need an ex-
planation of the sub-sentence, i.e., 'Vulcan does not exist'.
8 It has also been used for issues concerning the empty domain though this as-
Mauro Mariani
The problem Orenstein is faced with can be set out in a form which has
a venerable past in the history oflogic - that of an inconsistent triad. In
other words, among the following three statements
(I) The language of first-order logic contains empty individual
terms.
(II) Negative sentences whose subject is an empty individual term
are true, while the affirmative ones are false.
(III) The quantifiers are interpreted referentially (using Tarski's
truth conditions).
at least one must be discarded. Otherwise, the logical law
(V'X) A(x) ~ A(t)
red' is true, on the basis of the fact that 'The white lady' (a character
in a Walter Scott novel) is obviously not red. I believe that the standard
reaction would be that this person is joking, or else that her use of
quantifiers deviates from its normal use in English. In other words the
difficulty lies in derivation of a sentence, that we expect will tell us
something about the real world, simply from the fact that a certain
name, let us suppose 't', is not denoting (when 't' is without denota-
tion the only condition of truth of 't is not B' is indeed the lack of de-
notation of 't'). From Frege's perspective this truth condition would
not be totally uninformative, because it would mean that to a given
sense no denotation corresponds (for example that 'the most divergent
series' is a non-denoting name is a truth of the analysis); but from a
Mill-Kripke perspective, in which the proper nouns do not have sense,
that a proper noun does not denote is a purely linguistic fact.
I am not of course arguing here that this interpretation of quantifi-
ers is inconsistent, or that it is not possible to construct first-order logic
in this way. I am simply saying that this interpretation does not con-
stitute an adequate formalization of the standard use of quantifiers in
natural languages. What Orenstein seems to suggest is that in natural
language quantifiers do not have existential import in themselves, and
that they receive this only from a faulty interpretation of the first-order
logic: we could go further and say that according to Orenstein, it would
be perfectly natural for us to say that there are things that do not exist,
if it were not that a Tarskian interpretation had been given to quantifi-
ers.
Orenstein rightly claims that his interpretation of quantifiers is not
substitutional, and he levels against such an interpretation the criticism
of not having enough names. This is by no means the only problem it
runs into: for example, as Dunn and Belnap have pointed ouf, the
strong completeness theorem would be no longer valid. If we employ,
however, the Henkin "trick" of adding to our language a denumerable
set of new names, we are able to meet the above objections. So, to my
mind, the substitutional interpretation of quantifiers, if emended along
these lines, is equivalent to Orenstein's (of course if we stipulate, un-
like R. Barcan Marcus 3, that A(t) is always false when A is a predica-
tive letter and t is an empty name).
154 Individuals, Essence, and Identity
deny that from 'Homer is a poet' should follow 'Homer is', on the ba-
sis of the fact that in the first of these sentences 'to be' is predicate by
accident. According to some modem interpreters the reason for this is
that in 'Homer is a poet' the predicate expresses an accidental feature
of the subject, whereas 'to be' for Homer means 'to be a man' (or
something of that kind), in other words to enjoy a certain essential
property: thus from the simple fact that Homer is a poet 'Homer is'
does not follow, not because this latter sentence is false, but because
from the predication of an accident the predication of an essence does
not follow. However, medieval tradition is nearly unanimous in giving
another, in my opinion more correct, interpretation to this passage. In
brief, in the context of De Int. 11 predication by accident means predi-
cation depending on another, and the fallacy discussed by Aristotle is
analogous to that of deriving 'Socrates is good' from 'Socrates is a
good cobbler': just as Socrates is good only insofar as he is a cobbler,
but not in an absolute sense, so Homer 'is' only insofar as something is
predicated of him, but not in an absolute sense. That this is the correct
interpretation is confirmed by the fact that immediately after this ex-
ample, Aristotle observes that in an analogous way, from 'the not be-
ing is thinkable' it does not follow that 'the not being is', where it is
not a question of predicating essences or accidents. If this is correct,
then in De Int. 11 Aristotle is questioning the fact that from the truth of
an affirmative sentence the existence of what is denoted by the subject
term (in brief, of its subject) always follows: i.e. there can be true af-
firmative sentences with a non-existent subject, and therefore, since
the Principle of Excluded Middle is not called into question in this
case, false negative sentences with a non-existent subject. In addition,
in De Int. 10, 19b 14-19 Aristotle introduces, among the primary forms
of affirmation and negation, "strange" sentences of the kind 'Every
man is' and 'Every man is not'. In my opinion the best way to make
sense of these sentences is to suppose that a universal can be predi-
cated truly also of subjects that do not at present exist. If also 'Homer
is a man' is true on the same grounds as 'Homer is a poet' was true,
while 'Homer is' is false, then 'Every man is' is a meaningful false
sentence. I shall return to this point in a short while.
In one medieval tradition there were at least two solutions (in
practice equivalent) to this type of exception to the rule that from 'A is
156 Individuals, Essence, and Identity
B' it follows that 'A is'. The first was based on the theory of the am-
plialio and the restrictio of the term subject, or on the fact that the
suppositio personalis5 of the term A could be amplified until it in-
cluded both the denotation of A in the past and (perhaps) in the future 6 ,
and the objects of thought, whereas 'is' (in its use de secundo adi-
acenle) continued to co-signifY the time and therefore to refer only to
time in the present. The second was based on the distinction between
internal predication and external predication. The sentence 'Homer is a
poet' constitutes an example of internal predication, because being a
poet connotes Homer (in Frege's terms it is part of the sense that al-
most everyone connects to the proper name 'Homer'), and therefore it
is true even though 'Homer' supponit for something that is not present;
whereas 'Homer is sick' represents an example of external predication
(the 'normal' predication), and can be true only if 'Homer' supponil
for something that is present. In other words the rule that from 'A is B'
follows 'A is' admits of exceptions only if the premise 'A is B' consti-
tutes a case of internal predication.
Now if we apply the theory of ampliatio also to 'Aristotelian'
quantifiers we can suppose that they range or do not range over the
amplified supposilio of the term 'man'. In the second case 'Every man
is' is logically true (at least if we admit that men exist), while in the
first it can be, or rather certainly is, false. If we translate this distinc-
tion in terms of "Fregean" quantifiers, and read 'U(x)' as un instance
of internal predication, a sentence of the type '(::Ix) (U(x) 1\ -'E!(x))' -
where 'E!' corresponds to 'is' de secundo adiacente7 - can be true.
This apparently contradicts what I previously said about the meaning
of quantifiers in natural language, but only apparently. The domain of
quantifiers can include or not include the amplified suppositiones of
the terms, but the interpretation of the quantifiers remains in any case
referential. If the domain does not include them we have the normal
use of the quantifiers, and if it includes them we have a special use that
remains anyway referential.
This Aristotelian tradition (represented among others by Peter of
Spain), notwithstanding the fact that it too admits the possibility of
quantifYing over the non-existent, is deeply different from the one
Orenstein is referring to. First of all the truth of a sentence with non-
denoting subject does not depend only on the quality of the sentence
Mauro Mariani, Orenstein on Existence and Identity 157
(ii) the subject term supponit for something for which the predi-
cate does not supponit (when i.e. (i) holds);
or
Orenstein on the other hand considers only condition (i), which he lays
out, however, in a slighty different form:
The difference between (iv) and (i) is that, since in (iv) 'this A' and
similar expressions are proper names, it is a sufficient condition of (iv)
that all these names are empty, i.e. that A supponit pro nihilo; whereas,
if this latter circumstance occurs, (i) cannot even be formulated and it
is necessary to use the clause (iii). As a consequence the interpretation
of quantifiers in the Ockhamist tradition is on the whole nearer to that
of a referential type, even if the negative sentences with non-denoting
terms are considered true.
2.2. Kant
According to a consolidated interpretative tradition (not without its ex-
ceptions, for example Hintikka lO), Frege shares with Kant the thesis
that existence, properly understood, is a second order predicate. It is
therefore a bit surprising that Orenstein attributes to Kant the thesis
that so-called existential quantification actually has existential signifi-
cance only when it is based on an affirmative sentence, but not "when
it is grounded in negations of singular sentences".
In actual fact Orenstein justifies his interpretation through a certain
number of quotations, that, taken out of context, end up by meaning
the opposite of what Kant really wanted to say. For example
"Logically, being is merely the copula of a judgment" is understood as
if it meant that the existential significance of an affirmative sentence
lay, from a logical point of view, in the copula. On the contrary, in the
context from which this quotation was taken (Critique of Pure Reason,
Transcendental Dialectic, Book II, Chapter ill, Section IV) Kant op-
poses (a) the copulative use of being, in which 'is' is limited to putting
Mauro Mariani, Orenstein on Existence and Identity 159
All this seems in contradiction with what I said before, that affirmative
sentences with non-existent subject can be true, but the contradiction is
only apparent. Indeed all the Kantian applications of the principle Non
entis nulla sunt predicanda concern synthetic sentences 12, whereas for
the analytical ones this principle seems not to apply: so in Kant too we
find, at least implicitly, the distinction between internal and external
Mauro Mariani, Orenstein on Existence and Identity 161
Notes
1 To avoid complications 1 assume that A is a predicative letter, to be able to
limit (II) to atomic sentences.
2 Cf. Dunn and Belnap [1968], pp.I77-85.
3 Cf. Marcus [1962], pp.252-59.
4 Cf. Kaplan [1969], pp.178-14.
S Actually there are many varieties of suppositio, but in this paper 1 am con-
cerned only with suppositio personalis: so by suppositio 1 will mean henceforth sup-
positio personalis.
6 1 say "perhaps" because in many cases, especially as far as individuals are con-
cerned, the future existence is uncertain. It is not by chance that medieval people of-
ten used as an example Antichrist, whose future existence was guaranteed by the
Book of Revelation.
7 In this case 'Every A is' would always be true, since its standard translation in
the logic of predicates would be (Vx) (U(x) ~ E!(x)).
8 For example if x is man then x is Peter or x is John, etc. (the 'is' of identity is
not distinct from 'is' of predication)
9 Cf. Matthews [1973], pp.13-24.
10 Cf. Hintikka [1986], pp.249-67.
11 The same thing goes for Leibniz of course, and it is probable that Kant him-
self depends on Leibniz.
12 In particular the cosmological antinomies. Kant takes great care to demon-
strate that this is not a dialectical game nor a questioning of the Principle of the Ex-
cluded Middle, and observes that the antinomies are made up of pairs of sentences of
the form 'A is B' and 'A is not B', whose joint falseness demonstrates only that A
does not exist (I do not intend here to go into the question of what Kant means by the
worlds not existing).
13 Of which 1 have spoken at some length concerning one ofthe Aristotelian tra-
ditions.
ABSTRACT OBJECTS: A CASE STUDY
Stephen Yablo
1. NECESSITY
My relations to other concrete objects are almost all accidental. But the
number II's relations to other abstract objects (especially other num-
bers) would seem to be essential.
The most striking differences, though, have to do with existence.
Concrete objects (with the possible exception of "the world" on one
construal of that phrase) are one and all contingent. But the null set and
the number 11 are thought to exist in every possible world. To add to
the mystery, one normally supposes that existence is inversely related
to essence: the bigger x's essence, the "harder" it is for x to exist, and
so the fewer worlds it inhabits. And yet here is a class of objects ex-
tremely well endowed in the essence department, and yet incapable of
not existing.
You would have to be in a coma not to wonder what is going on
here. Why is it that so much about abstract objects is essential to them?
What is it about numbers et al. that makes it so hard for them not to
exist? And should not objects that turn up under all possible conditions
have impoverished essences as a result?
It may be that I have overstated the phenomenon. Not everyone
agrees that numbers even exist, so it is certainly not agreed that they
exist necessarily. There would be more agreement if we changed the
hypothesis to: numbers exist necessarily provided they can exist, that
is, unless they are impossible2 . And still more if we made it: numbers
exist necessarily provided they do exist. But these are nuances and de-
tails. I think that it is fair to say that everyone, even those who opt in
the end for a different view, has trouble with the idea that 11 could go
missing.
So I ask: Why should a numberless world seem impossible (al-
lowing that the appearance may be only prima facie)? Why should it
seem impossible for numbers to have had different intrinsic properties,
or different relational properties vis a vis other abstract objects? Why
should numbers seem to be absolutely modally inflexible?
2. APRIORITY
priori. That 3+5 = 8 is a fact that could be known on the basis of expe-
rience - experience of counting, say, or of being told that 3+5 = 8. But
the same is true of most things we know a priori. It is enough for a
priority that experience does not have to figure in our justification.
And this seems true of many arithmetical claims. One can determine
that 3+5 = 8 just by thinking about the matter.
Like the felt necessity of arithmetic, its felt apriority is puzzling
and in need of explanation. It is a thesis of arithmetic after all that
there are these things called numbers. And it is hard to see how one
could be in a position to know a priori that things like that really ex-
isted.
It helps here to remember the two main existence-proofs philoso-
phers have attempted. The ontological argument tries to deduce God's
existence from God's definition, or the concept of God. The knock
against this has been the same ever since Kant; from the conditions a
thing would have to satisfy to be X, nothing existential follows unless
you have reason to think that the conditions are in fact satisfied. Then
there is Descartes's cogito. This could hardly be expected to give us
much guidance about how to argue a priori for numbers! In any case,
the argument does not seem to be a priori. You need to know that you
think, and that knowledge would seem to be based on your experience
of self.
I said that the ontological argument and the cogito were the two
best-known existence-proofs in philosophy. Running close behind is
Frege's attempted derivation of numbers themselves. If the Fregean
line is right, then numbers are guaranteed a priori by logic together
with definitions. Shouldn't that be enough to make it a priori that
numbers exist? Perhaps, if the logic involved were ontology-free. But
Frege's logic affirms the existence of all kinds of higher-type objects.
(Frege would not have wanted to call them objects because they are
not saturated; but there is little comfort in that.) The Fregean argument
cannot defeat doubts about a priori existence, because it presupposes
they have been defeated in presupposing the apriority of Fregean logic.
A different strategy for obtaining a priori knowledge of numbers
goes via the "consistency-truth principle": in mathematics, a consistent
theory is a true theory. If we can know a priori that theory T is consis-
166 Individuals, Essence, and Identity
3. ABSOLUTENESS
I take it as a given that mathematical truth does not feel relative in this
way. It feels as though 3+5 is just plain 8. It feels as though the power
set of a set is just plain bigger than the set itself.
One could argue that the notion of truth at work here is still at
bottom a relativistic one: it is truth according to standard math, that is,
truth according to whatever theories are accepted in the mathematical
community.
But this is a far cry from what we want, and act like we have. For
now the question becomes, why is this math standard and not that? It
would be nice if the answer adverted at some point to the theory's be-
ing true or correct in a way that logically coherent alternatives are not
correct. To run the point the other way around: If all PA can claim over
the competition is greater utility or naturalness given our projects and
Stephen Yablo, Abstract Objects: a Case Study 167
'true' because models that threaten to falsify it are shown the door;
they are not part of the theory's intended subject matter.
Call this the debunking explanation of why it seems wrong to deny
the standard axioms. I do not say that the debunking explanation is out
of the question; I do not want to rule it out that ZF serves in effect as a
reference-fixer for 'set' But, that certainly is not how it/eels. If some-
one wants to argue that Infinity is wrong - that the hereditarily finite
sets are the only ones there are - our response is not "save your breath!
deny Infinity and you're changing the subject". Our response is: "that
sounds unlikely, but let's hear the argument". No doubt we will end up
thinking that the Infinity-denier is wrong. The point is that what he is
wrong about is the sets. It has to be, for if he is not talking about the
sets, then we are not really in disagreement.
Suppose though the debunkers are right that ZF is true because it
sets the standard for what counts as a set. This still does not quite ex-
plain our sense that ZF is correct. Why should we be so obsessed with
the sets as opposed to the pseudo-sets? To the extent that ZF and 'sets'
are a pair, curiosity about why ZF seems so right is a lot like curiosity
about why the sets seem so right. It does not matter how the questions
individuate, as long as they are both in order. And so far nothing has
been said to cast doubt on this. So I ask again, why do ZF and the sets
seem so right?
Three puzzles, then: one about necessity, one about apriority, one
about absoluteness. Let us start with necessity; the other two puzzles
will be brought in shortly.
What accounts for 11 's tenacious grip on reality? One natural
thought is that there is something about abstractness that prevents a
thing from popping in and out of existence as we travel from world to
world. 3 Crispin Wright and Bob Hale observe that it is hard to think
what conditions favorable for the emergence of numbers would be, and
it is hard to think of conditions unfavorable for their emergence. It is
however easy to think of conditions favorable for the emergence of Mt.
Stephen Yablo, Abstract Objects: a Case Study 169
McKinley. The reason, one imagines, is that numbers are abstract and
Mt. McKinley is not.
But, granted that numbers do not show up only when conditions
are right, how does that bear on their necessity? Explanations come to
an end somewhere, and when they are gone we are left with the brute
facts. Why shouldn't the existence/nonexistence of numbers be a brute
fact? Traditionally existence has been the paradigm of a phenomenon
not always admitting of further explanation. So, granted that numbers
aren't contingent on anything, one still wants to know why they should
not be contingent full stop.
A second possible explanation is that it is part of the concept of an
abstract object (a "pure" abstract object, anyway) to exist necessarily if
at all. An object that made an appearance in this world, but was miss-
ing from others, would by that alone not be abstract.
Suppose that's right; an otherwise qualified object that does not
persist into all worlds does not make the cut. Let us think about these
contingent would-be abstracta some more. What sort of object are we
talking about here? The obvious thought is that they are exactly like
real abstracta except in the matter of necessary existence. But the ob-
vious thought is strange, and so let us ask explicitly: Could there be
"shmabstract" objects that are just like their abstract cousins except in
failing to persist into every world?
Fiddling with an object's persistence conditions is generally con-
sidered harmless. If I want to introduce, or call attention to, a kind of
entity that is just like a person except in its persistence-conditions - it
is missing (e.g.) from worlds where the corresponding person was born
in Latvia - then there would seem to be nothing to stop me. If you can
have shmersons alongside persons, why not shmumbers along with
numbers?
You may think that there is a principled answer to this: a princi-
pled reason why abstracta cannot be "refined" so as to exist in some
worlds but not all. If so, though, then you hold the view that we started
with: there's something about abstractness that precludes contingency.
What is it? Earlier we looked at the idea that where abstracta are con-
cerned, one world is as good as another. But although it is true that a
contingently existing number has no apparent rationale for turning up
here rather than there, why should that bother us? Why should the
170 Individuals, Essence, and Identity
Conservative theories are theories compatible with any story that might
be told about how things go physically, as long as that story is consis-
tent in itself. (I am going to skate lightly over the controversy over how
best to understand 'consistent' and 'compatible' here. The details are
not important for what follows.)
Stephen Yablo, Abstract Objects: a Case Study 171
From this it seems a small step to the suggestion that the only distinc-
tively modal intuition we have about mathematical objects is that the
theory of those objects is conservative. So construed, the intuition is
quite correct. And it is correct in a way that sits well with our feeling
that existence is never "automatic" - that nothing has such a strong
grip on reality that it is incapable of not showing up.
So: is our intuition of the necessity of '3+5 = 8' just a (confused)
intuition of quasi-necessity, that is, conservativeness?
172 Individuals, Essence, and Identity
I think this is very unlikely. Yes, every world has a physical dupli-
cate with numbers. But one could equally go in the opposite direction:
every world has a physical duplicate without them. If the permanent
possibility of adding the numbers in makes for an intuition of neces-
sity, then the permanent possibility of taking them out should make us
want to call numbers impossible. And the second intuition is largely
lacking. A premise that is symmetrical as regards mathematical exis-
tence cannot explain why numbers seem necessary as opposed to im-
possible.
A second reason why necessity is not well-modeled by conserva-
tiveness is this. Arithmetical statements strike us as individually neces-
sary. We say, "this has got to be true", not "this considered in the con-
text of such and such a larger theory has got to be true". But the latter
is what we should say if our intuition is really of conservativeness. For
conservativeness is a property of particular statements only seen as ex-
emplars of a surrounding theory. A statement that is conservative in
the context of one theory might fail to be conservative in the context of
another. (It might be inconsistent with the other.) Nothing like that
happens with necessity.
A third problem grows out of the discussion above of consistency
as sufficient for truth. Suppose that two theories contradict each other.
Then intuitively, they cannot both be necessary; indeed if one is neces-
sary then the other is impossible. But theories that contradict each
other can both be conservative.
Someone might reply that if contradictory means syntactically
contradictory, then contradictory theories can so be necessary. All we
have to do is think of them as describing different domains, e.g., dif-
ferent subsets of the universe of sets.
That is true in a technical sense. But the phenomenon to be ex-
plained - our intuition of necessity - occurs in a context where contra-
dictory theories are (the technical point notwithstanding) experienced
as incompatible. If I affirm Infinity and you deny it, we take ourselves
to be disagreeing. But both of us are saying something conservative
over physics.
When two statements contradict each other, they cannot both be
necessarily true. Or rather, they cannot unless the truth is relativized
somehow: to the background theory, a certain type of model, a certain
Stephen Yablo, Abstract Objects: a Case Study 173
6. FIGURALISM
The conservativeness gambit has many virtues, not the least of them
being its short way with abstract ontology. At the same time there are
grounds for complaint. One would have liked a theory that made
arithmetic "necessary" without making it in a correlative sense
"impossible". And one would have liked a theory less friendly to rela-
tivism. The best thing of course would be if we could hold onto the
advantages of the Field proposal without giving up on "real" necessity,
and without giving up on the intuition of absolute truth or correctness.
Is this possible? I think it just may be - or at least I have an idea to
this effect that is not immediately obviously hopeless. I can indicate
the intended direction by hazarding (what may strike you as) some ex-
tremely weird analogies:
(D) 'the number ofFs is large iff there are many Fs'
'your marital status changes iff you get married or ... '
174 Individuals, Essence, and Identity
Fourth, the vocabulary's utility for this purpose does not depend
on conceiving of its referential-looking elements as genuinely standing
for anything. It does not depend on conceiving its referential-looking
elements any other way, either. Those if any who take stomach-
butterflies, party affiliations, and numbers dead seriously derive the
exact same expressive benefit from them as those who think the first
group insane. And both groups derive the exact same expressive bene-
fit as the silent majority who have never given the matter the slightest
thought.
7. NECESSITY AS BACK-PROPAGATED
I said that all of the statements strike us as necessary, but I did not of-
fer an explanation of why. When it comes to the non-mathematical
statements, an explanation is quickly forthcoming.
Stomach-butterflies and the rest are representational aids. They
are "things" that we advert to not (not at first, anyway) out of any in-
terest in what they are like in themselves, but because of the help they
give us in describing other things. Their importance lies in the way
they boost the language's expressive power.
By making as if to assert that I have butterflies in my stomach, I
really assert something about how I feel - something that it is difficult
or inconvenient or perhaps just boring to put literally. The real content
of my utterance is the real-world condition that makes it sayable that S.
The real content of my utterance is that reality has feature BLAH: the
feature by which it fulfills its part of the S bargain.
The reason it seems contingent that Pat's marital status has
changed is that, at the level of real content, it is contingent: she could
have called the whole thing off. The reason it seems necessary that
Pat's prospects have improved iff it has become likelier that she will
be successful is that, at the level of real content, it is necessary, for the
two sides say the very same thing.
How does the world have to be to hold up its end of the 'the num-
ber of apostles is even' bargain? How does the world have to be to
make it sayable that the number of apostles is even, if we are going to
say there are numbers? There have to be evenly many apostles. So, the
176 Individuals, Essence, and Identity
real content of 'the number of apostles is even' is that there are evenly
many apostles.
Now, that there are evenly many apostles is something that need
not have been the case, and that it takes experience to confirm. At the
level of real content, then, 'the number of apostles is even' is epistemi-
cally and metaphysically contingent. But there might be other number-
involving sentences whose real contents are necessary. To the extent
that it is the real content of these sentences that we hear them as ex-
pressing, it would be natural for us to put those sentences on the list of
necessary truths.
This is why (some) number-involving sentences, e.g., 'the number
of Fs = the number of Gs iff the Fs and Gs are equinumerous' feel
necessary at the same time as we have trouble seeing how they could
be necessary. Our two reactions are to two different contents. The
sentence feels necessary because at the level of real content it is a tau-
tology: the Fs and Gs are equinumerous iff they are equinumerous.
And tautologies really are necessary.
The reason we have trouble crediting our first response is that the
sentence's literal content - that there is this object, a number, that be-
haves like so - is to the effect that something exists. And it is baffling
how anything could cling to existence so tightly as to be incapable of
not turning up.
Why do the two contents get mooshed together in this way? One
reason is this. A sentence's conventional content - what it is generally
understood to say - can be hard to tell apart from its literal content. It
takes work to remember that the literal meaning of 'he is not the
brightest guy in town' leaves it open that he is the second brightest. It
takes work to remember that (literally) pouring your heart out to your
beloved would involve considerable mess and a lengthy hospital stay,
not to mention the effect on your beloved. Since there is no reason for
us to do this work, it is not generally realized what the literal content in
fact is.
Consider now '7<11 '. To most (!) people it means that seven
things are fewer than eleven things. But the literal content is quite dif-
ferent. The literal content makes play with entities 7 and 11 that meas-
ure pluralities size-wise, and encode by their internal relations facts
about their outnumbering relations. Of course, these plurality-measures
Stephen Yablo, Abstract Objects: a Case Study 177
7 and 11 are no more on the speaker's mind than blood is on the mind
of someone pouring their heart out to their beloved. '7<11' is rarely
used to describe numbers as such, and so one forgets that the literal
content is about nothing else.
Someone might say that the literal contents of pure-mathematical
statements are quickly recovered, once we set our minds to it. It is the
real contents that need further explanation. I do not actually think that
the real contents are always the same, so there is a considerable
amount of exaggeration in what follows. But that having been said, the
claim will be that arithmetic is, at the level of real content, a body of
logical truths - specifically, logical truths about cardinality, while set
theory is, at the level of real content, a body of logical truths of a com-
binatorial nature.
8. ARITHMETIC
even if there are only finitely many concreta. 0 is the number of non-
self-identical things, and k+ 1 is the number the numbers <k.
Making as if there are numbers is a bit of a chore; why bother?
Numbers are there to expedite cardinality-talk. Saying '#Fs = 5' in-
stead of '3sx Fx' puts the numeral in a quantifiable position. And we
know the expressive advantages that quantification brings. Suppose
you want to get it across to your neighbor that there are more sheep in
the field than cows. Pre-eN) this takes (or would take) an infinite dis-
junction: there are no cows and one sheep or there are no cows and two
sheep or there is one cow and there are two sheep, and etc. post-eN) we
can say simply that the number of sheep, whatever it may be, exceeds
the number of cows. The real content of '#sheep > #cows' is the infi-
nite disjunction, expressed now in finite compass. s
But although this may clarify what the real contents of applied ar-
ithmetical statements are, statements of pure arithmetic are another
matter.
Take first quantifierless addition statements. What does the con-
crete world have to be like for it to be the case that, assuming numbers,
3+5 = 8? Assuming numbers is assuming that there is a number k
numbering the Fs if there are k Fs. But that is not all. One assumes too
that if no Fs are Gs, then the number of Fs and the number of Gs have
a sum = the number of things that are either F or G. All of that behind
us, the real-world condition that makes it OK to suppose that 3+5 = 8
is that
33X Fx /\ 3 sy Gy /\ Vx -'(Fx /\ Gx) ~ 3gZ (Fz v Gz)
This has already been seen for atomic and negated-atomic truths 6 •
These give us all the arithmetical truths (up to logical equivalence)
when closed under four operations: (1) conjunctions of truths are true;
(2) disjunctions with truths are true; (3) universal generalizations with
only true instances are true; (4) existential generalizations with any
true instances are true. It is easy to check that these operations preserve
the property of being logically true at the level of real content. We can
illustrate with case (4). Suppose that 3x cf>(x) has a true instance cf>(n).
By hypothesis of induction, cf>(n) is logically true at the level of real
content. But the real content of 3x cf>(x) is a disjunction with cf>(n) as a
disjunct. So the real content of3x cfJ(x) is logically true as well.
9. SET THEORY
Sets are nice for the same reason as numbers. They make possible
sentences whose real contents we really believe, but would otherwise
have trouble putting into words. One can imagine introducing set-talk
for this purpose in various ways, but the simplest is probably this. "In
the beginning" we speak a first-order language with quantifiers ranging
180 Individuals, Essence, and Identity
over concreta. These quantifiers can be singular or plural; one can say
'there is a rock such that it ... ' and also 'there are some rocks such
that they ... '. Now we adopt the following rule:
(S) if there are some things a, b, c ... ,
then *there is a set {a, b, c, ... }*.
Since the antecedent here states the real-world condition under which
we are to make as if a, b, c, ... form a set, a, b, c, ... are limited to con-
crete objects. But the reasons for collecting concreta into sets apply
just as much to the abstract objects introduced via (S). So (S) is
strengthened to
(8) if *there are some things a, b, c ... ,*
then *there is a set {a, b, c, ... }*.
This rule, like (N) in the last section, works recursively. On the first
go-round we get sets of concreta. On the second go-round we get sets
containing concreta and/or sets of concreta. On the third we get sets
containing concreta, sets of them, and sets of them. And so on through
all the finite ranks. Assuming that there are only finitely many con-
creta, our output so far is the hereditarily finite sets: the sets that in
addition to being themselves finite have finite sets as their members,
and so on until we reach the concrete objects that started us off.
What now? If we think of (8) as being applied at regular intervals,
say once a minute, then it will take all of eternity to obtain the heredi-
tarily finite sets. No time will be left to obtain anything else, for exam-
ple, the first infinite number 0).
The answer to this is that we are not supposed to think of (8) as
applied at regular intervals; we are not supposed to think of it as ap-
plied at all. (8) does not say that when we establish the pretense-
worthiness of 'there are these things', it becomes pretense-worthy that
'they form a set'. (8) says that if as a matter of fact (established or not)
*there are these things*, then *there is the set of them*. If in fact
*there are the hereditarily finite sets*, then *there are the von Neu-
mann integers (0 = 4>, n+l = {0,1, ... n} )*. And now (8) tells us that
*there is the set {O, 1,2,3, ... }*, in other words, *there is 0)*. I cannot
argue this here, but similar reasoning shows that we get all sets of rank
Stephen Yablo, Abstract Objects: a Case Study 181
<X for each ordinal <x. (S) yields in other words the full tower of sets:
the full cumulative hierarchy.
Now, to say that (S) yields the full cumulative hierarchy might
seem to suggest that (S) yields a certain fIXed bunch of sets, viz. all of
them. That is not the intention. There would be trouble if it were the
intention, for (S) leaves no room for a totality of all sets. To see why,
suppose for contradiction that *a, b, c, ... are all the sets*. (S) now tells
us that *all the sets form a set V*. This set V must for familiar reasons
be different from a, b, c, ... So the proposed totality is not all-
encompassing. (I will continue to say that (S) yields the full cumulative
hierarchy, on the understanding that the hierarchy is not a fixed bunch
of sets, since any fixed bunch you might mention leaves something
out. This does not prevent a truth-definition, and it does not prevent us
from holding that some sentences are true of the hierarchy and the rest
false?).
Conjuring up all these sets is a chore; why bother? The reason for
bothering with numbers had to do with cardinality-type logical truths.
Some of these truths are infinitely complicated, but with numbers you
can sum them up in a single finite sentence. Something like that is the
rationale for sets as well. The difference is that sets help us to deal
with combinatorial logical truths - truths about what you get when you
combine objects in various ways.
An example will give the flavor. It is a theorem of set theory that if
x = y, then {x, u} = {y, v} iff u = v. What combinatorial fact if any does
this theorem encode? Start with '{x, u} = {y, v}'. The real content of
that is that theyxu are themyv - or, to dispense with the plurals, that (x =
y v x = v) /\ (u = Y v u = v) /\ (y = X V Y = v) /\ (v = X V V = u). So what
our theorem is really saying is that
Ifx = y, then
([(x = y v x = v) /\ (u = Y v u = v) /\ (y = X V Y = v) /\
(v = X V V = u)] iff u = v).
This is pretty simple as logical truths go. Even so it is not really com-
prehensible; I at least would have trouble explaining to someone else
what it says. If truths as simple as this induce combinatorial boggle-
ment, it should not be surprising that the set-theoretic formulations are
found useful and eventually indispensable.
182 Individuals, Essence, and Identity
(AI) r(x E z) is
Note that the first line simplifies to v YEZ x = y; that is in practice what I
will take the translation to be. (The reason for the quantified version is
that it extends better to the case where z is the empty set.) The third
Stephen Yablo, Abstract Objects: a Case Study 183
line marks the one place where E is not eliminated. If z is not a set,
then it is (literally) false to say that x belongs to it, which is the result
we want. The rule for identity-statements is
(A2) rex = y) is
The real content of cf> is found by repeatedly applying r until you reach
a fixed point, that is, a statement cf>* such that r( cf>*) = cf>*. This fixed
point is a truth-functional combination of "ordinary" statements true or
false for concrete (non-mathematical) reasons. These ordinary state-
ments are to the effect that x = y, where x and yare concrete, or x = y,
where one is concrete and the other is not, or x E z, where z is con-
crete. 1I
How do we know that a fixed point will be reached? If cf> is a gen-
eralization, the (R;)s tum it into a truth-functional combination of at-
oms 'If. If 'If is an atom talking about sets, then the (A;)s tum it into a
generalization about sets of a lower rank, and/or non-sets. Now we ap-
ply the (R;)s again. Given that cf> contains only finitely many quantifi-
ers, and all the sets are of finite rank, the process must eventually bot-
tom out. 12 The question is how it bottoms out, that is, the character of
the sentence cf>* that gives cf>'s real content.
I claim that if cf> is a set-theoretic truth, then cf>* is, not quite a logi-
cal truth, but a logical consequence of basic facts about concreta:
184 Individuals, Essence, and Identity
Basis Step
Recursion Step
(a) Ifx andy are nonempty sets, then r(x:= y) is (I\UEX VVEY u:=
v) 1\ (I\VEY V UEXV:= u). (al) If it is true that x:= y, then r(x:=
y) is a conjunction of disjunctions, each of which has a true
disjunct u := v. By hypothesis of induction, these true dis-
juncts have logically truecc real contents. So rex := y) has a
logically truecc real content. And the real content of rex = y)
is also that of x := y. (a2) If it is true that x :t:. y, then rex :t:. y)
is a disjunction of conjunctions, each of which is built out
of true conjuncts. By hypothesis of induction, these true
conjuncts are logically truecc at the level of real content. So
rex :t:. y) has a logically true cc real content. And the real
content of rex :t:. y)is also that of x :t:. y.
(b) If z is a nonempty set, then rex E z) is vYEZ X = Y (bl) If it is
true that x E z, this has a true disjunct x = y. By hypothesis
of induction, x = y has a logically truecc real content. But
then rex E z) is logically true cc at the level of real content,
whence so is x E z. (b2) If the truth is rather that x (i!: z, then
rex (i!: z) is a conjunction of true conjuncts. By hypothesis of
induction, these conjuncts are logically truecc at the level of
real content. So rex (i!: y) has a logically truecc real content,
whence so also does x (i!: y.
10. SUMMING UP
The view that is emerging takes something from Frege and something
from Kant. One might call it "Kantian logicism". The view is Kantian
because it sees mathematics as arising out of our representations.
Numbers and sets are "there" because they are inscribed on the specta-
cles through which we see other things. It is logicist because the facts
that we see through our numerical spectacles are facts of first-order
logic.
And yet the view is in another way the opposite of Kantian. Kant
thinks necessity is imposed by our representations, and I am saying that
necessity is imposed on our representations by the logical truths they
186 Individuals, Essence, and Identity
- the idea that when you have got a detenninate bunch of things, you
are entitled to the set of those things - then that person arguably does
not mean the same thing by 'set' as we do. l3 '
Notes
I Although on the Frege-Russell definition of number, there is, arguably, intrin-
sic change. The empty set can change too, if as Lewis suggests it is definable as the
sum of all concreta. But I am talking about what we intuitively expect, and no one
would call these definitions intuitive.
2 Hale and Wright suggest in "Nominalism and the Contingency of Abstract
Objects" that Field might not accept even that much. Field does say that numbers are
conceptually contingent. But it would be hard to pin a metaphysical contingency the-
sis on him, for two reasons. (1) He is on record as having not much use for the notion
of metaphysical necessity. (2) To the extent that he tolerates it, he understands it as
conceptual entailment by contextually salient metaphysical truths. If salient truths
include the fact that everything is concrete, then (assuming they are not concrete)
numbers will come out metaphysically impossible
3 Impure abstracta like singleton-Socrates are not thought to be necessarily ex-
istent. So really I should be talking about pure-abstractness. I will stick to "abstract"
and leave the qualification to be understood. (Thanks here to Marian David.)
4 30X Fx =df 'ix (Fx ~ x"* x) and 3n+1X Fx =df 3y (Fy /\ 3"x (Fx /\ x"* y))
5 There is an analogy here with Hartry Field's views on "the reason" for having
a truth-predicate, in the absence of any corresponding property.
6 Mario Gomez-Torrente pointed out that some atomic truths have not been pro-
vided out with real contents, a fortiori not with logically true real contents. An exam-
ple is (3+2)+1 = 6. This had me worried, until he pointed that these overlooked
atomic truths were logically equivalent to non-atomic truths that had not been over-
looked. For instance, (3+2)+ 1 = 6 is equivalent 3y «3+2 = y) /\ (y+ 1 = 6)). A quick
and dirty fix is to think of overlooked sentences as inheriting real content from their
188 Individuals, Essence, and Identity
not overlooked logical equivalents. A cleaner fIx would be desirable, but Mario has
not provided one yet.
7 See the last few pages of Putnam [1967] and "Putnam Sematics" in Hellman
[1989].
8 Alternatively, there are some things all of which are Fs, and some things not
the same as the fIrst things all of which are Fs, and etc. (Say there are two Fs. You
can pick both of them, either taken alone, or neither of them. Note that 'all the Fs'
and 'none of them' are treated here as limiting cases of 'some of the Fs').
9 You could do it with plural quantifIcation over ordered pairs
10 The idea is that XEZ describes x as one of the things satisfying the condition of
on whether x is indeed identical to y. Statements of the second and third types are
necessarily false, since concreta cannot be sets or have members.
12 The same argument would seem to work with sets of infInite rank; there are
(the sets) that are too many to form a set. This widespread view is at odds with (S)
only if it is supposed that there is some defInite bunch of things including all and only
the sets. If the sets are a defInite bunch of things, it is very hard to understand what
could be wrong with gathering them together into a further set. I agree with Putnam
when he says that "no concrete model [of Zermelo set theory] could be maximal -
nor any non-concrete model either, as far as that goes. Even God could not make a
model for Zermelo set theory that it would be mathematically impossible to extend,
and no matter what 'stuff He might use. [ ... ] it is not necessary to think of sets as
one system of objects [... ] in order to follow assertions about all sets" (Putnam
[1967], p.21).
14 I am grateful to a number of people for criticism and advice; thanks in par-
ticular to Gideon Rosen, Kit Fine, Gil Harman, Mario Gomez-Torrente, Marian
David, Ted Sider, Paul Horwich, and Stephen Schiffer.
KINDS, ESSENCE, AND NATURAL
NECESSITY
E. Jonathan Lowe
other hand, this very fact suggests that it is at least misleading to speak
of that alleged relation as one of "necessitation" and this may help to
explain Armstrong's tendency to downplay such talk in his more re-
cent writings.
Let us assume, in what follows, that laws of nature do indeed con-
cern universals rather than particulars. Scientific essentialists may be
expected to agree, because one of the primary claims often made in
support of their view is that natural kinds and properties depend for
their very identity upon the laws into which they enter.20 Thus, one
contention is that a property owes its identity to the contribution it
makes to the (perhaps conditional) causal powers of physical objects
possessing that property, as determined by the causal laws governing
interactions involving such objects. For example, if we ask what prop-
erty mass is, we may be told that mass is that property in a body which,
amongst other things, makes it necessary for a force to be applied to
the body in order to accelerate it, in accordance with Newton's Second
Law of Motion, F = Ma. Again, we may be told that sphericity is that
property which, amongst other things, makes a body which possesses it
liable to roll down an inclined plane (provided that the body is also
rigid and subject to a gravitational force), in accordance with various
natural laws. In support of this view, it may be pointed out that we can
only detect the properties of physical objects by interacting with them
or with other objects affected by their activities. So it seems reasonable
to adopt something like the following criterion of identity for (physi-
cal) properties: the property of being F is identical with the property of
being G if and only if F and G make the same contribution to the
causal powers of physical objects possessing them. Then, given that
the causal powers of objects are determined by the natural laws gov-
erning the properties (universals) which those objects possess, the
foregoing criterion seems to reduce to this: the property of being F is
identical with the property of being G if and only if F and G enter into
the same laws in the same ways.
Thus, consider all the laws into which the property of mass, M,
enters, such as the law that force equals the product of mass and accel-
eration (F = Ma) and the law (Newton's Law of Gravitation) that the
gravitational force between two bodies is proportional to the product of
their masses and inversely proportional to the square of the distance
E. Jonathan Lowe, Kinds, Essence, and Natural Necessity 199
this does not detract from my point that, so far, we have been given no
good reason to adopt this position, nor the corresponding position re-
garding the transworld identity of properties. Quite generally, one can-
not advance directly from an intraworld criterion of identity to an in-
terworld criterion of identity. As a reductio ad absurdum of the suppo-
sition that one can, consider the claim that because two material ob-
jects of the same kind cannot occupy the same place at the same time -
so that spatiotemporallocation provides a criterion of identity for ma-
terial objects of any given kind - it therefore follows that no material
object of a given kind could have occupied a spacetime location differ-
ent from its actual spacetime location - that this very chair, say, could
not have been located where that chair is now. Plainly, no such conclu-
sion does follow.
At this point, the scientific essentialist may reply that properties
are relevantly different from particular objects and events where these
issues of identity are concerned. 24 Objects and events are non-
repeatable entities with unique spatiotemporallocations, so that we can
identify them in this world in ways which differ from the ways in
which we identify them "across" worlds. He may try to press home the
point by urging that it simply does not make sense to suppose that the
very same property - mass, let us say - might, in another possible
world, make the same contribution to the laws of that world that, let us
say, electrical charge does in this, the actual world. Or, to take another
example, he may urge that it simply doesn't make sense to suppose
that there might be a world in which the kind electron exists, but in
which the law is that electrons have unit positive charge instead of unit
negative charge. These otherworldly "electrons" would surely just be
positrons.
However, one can oppose scientific essentialism without being
driven to accept these extreme consequences. If we consider less ex-
treme variations in the laws, the scientific essentialist's case seems
much less plausible. Must we say, for instance, that a world in which
the law of 'gravitation' is an inverse cube law, or a law making the
force inversely proportional to a non-integral power of the distance
very slightly different from 2, is a world in which the property of mass
does not exist? What about a world in which the universal constant of
gravitation, G, has a slightly different value from its value in this world
E. Jonathan Lowe, Kinds, Essence, and Natural Necessity 201
laws they enter into in every possible world (including, of course, the
actual world)? If it is part of the very essence of electronhood and of
unit negative charge that electrons have unit negative charge, because
this law involving them obtains in every possible world, how do we
know that our world is one in which these very universals do indeed
exist and are exemplified by particular entities, as opposed to being a
world in which the law in question is merely vacuously true and the
entities that we experiment upon exemplify quite different universals?
The scientific essentialist may be tempted at this point to appeal to
something like Hilary Putnam's notion of a 'stereotype'?9 According
to Putnam, a natural kind term, such as 'water' or 'lemon', has associ-
ated with it a bundle of stereotypical properties - in the case of
'lemon', the properties, say, of being yellow, juicy and acidic - which
do not necessarily belong to every individual exemplar of the natural
kind in question but which do belong to "typical" exemplars and which
can therefore be used, albeit only defeasibly, to identify something as
being an exemplar of that kind. However, in the first place, if these
stereotypical properties are indeed genuine properties - universals -
possessed by individual objects, then the scientific essentialist ought,
in all consistency, to say that these properties too owe their very iden-
tity to the laws into which they enter, in which case it becomes equally
mysterious how we can ever know empirically that anything exempli-
fies them. Secondly, when we are concerned with fundamental kinds
such as the kind electron, it appears that we can no longer suppose that
there are stereotypical properties associated with the kind but which
not every individual exemplar of the kind necessarily possesses: all
electrons, it seems, are necessarily exactly alike in respect of their in-
trinsic properties. Hence, the proposed model for our ability to identify
exemplars of natural kinds empirically, even if it served the scientific
essentialist's purposes in some cases - which I do not think it does -
would not serve them in the case of fundamental natural kinds. I con-
clude that the scientific essentialist has an undischarged burden of ex-
plaining how, in general, our empirical knowledge of natural laws can
be possible if, as he maintains, the very essence of a property resides in
its lawlike connections with other properties. 30
E. Jonathan Lowe, Kinds, Essence, and Natural Necessity 205
Notes
1 There is controversy amongst commentators as to whether Hume was an
antirealist or a sceptical realist concerning causal necessity: see Strawson [1989]. I
adopt no particular position in this paper on this historical issue.
2 See, especially, Shoemaker [1998], Fales [1993], and Ellis [1999]. See also
[1993], p.l72 and p.229, and also Armstrong, [1997], p.223ff (he endorses the
'making happen' idiom at pp.210-11).
11 See further Lowe [1989], Chapter 8.
12 For Armstrong's dismissive view of kinds, see Armstrong [1997], p.65ff.
13 Armstrong maintains that non-causal laws are supervenient: see Armstrong,
[ 1995].
17For Armstrong's deployment of a similar argument on behalf of his own view
oflaws, see his reply to Fales in Bacon et al. (eds.), [1993], p.144ff.
18 But see further Lowe [1987], for reasons to modify this proposal in certain
ways which, however, do not undermine the argument of the present paper. For more
on the distinction between "relative" and "absolute" necessity, see Lowe [1998],
p.18ff.
19 Of course, Armstrong himself is not a realist concerning possible worlds: see
Armstrong [1989] and Armstrong [1997], p.l72ff. For my own view of the ontologi-
cal status of possible worlds, see Lowe [1998], p.256ff. Nothing I say in the present
paper in the language of possible worlds is intended to imply a commitment to their
reality.
20 See the papers cited in note 2 above.
21 See further Ellis [1999], p.30.
22 See Davidson [1980]. He abandons the proposal in Davidson [1985].
[1998], p.43.
206 Individuals, Essence, and Identity
24 One essentialist who does at least discuss the matter - though not, to my
[ 1982].
28 I do not mean to suggest that I endorse this intraworId criterion of identity for
properties myself: in fact, I reject it on the grounds that it is circular. For my own
account ofthe identity-conditions of properties (universals), see Lowe [1999].
29 See Putnam [1975].
30 I am very grateful for comments received when this paper was presented at
the Bergamo Conference on 'Individuals, Essence, and Identity: Themes of Analytic
Metaphysics', and especially grateful to Dr. Kathe Trettin, who replied to the paper. I
am also grateful to an anonymous referee for some criticisms which prompted me to
clarify certain aspects of the paper.
KINDS OF NECESSITY: A COMMENTARY
ON E.l. LOWE'S PAPER "KINDS, ESSENCE,
AND NATURAL NECESSITY"
Kathe Trettin
chapter of his book Kinds of Being (1989): "[ ... ] the notion of an 'indi-
vidual' and of a 'sort' or 'kind' are opposite sides of a single concep-
tual coin: each is only understandable in tenns of the other. Individuals
are necessarily individuals ofa kind, and kinds are necessarily kinds of
individuals". Therefore, "a realism with regard to particulars or indi-
viduals [... ] implies realism with regard to sorts or kinds" (Lowe
[1989], p.4f.). Although this quasi-Hegelian fonnulation sounds a fa-
miliar note to my Gennan ear and therefore I have no trouble under-
standing it, I would argue that this interesting mutual dependence
should be spelled out not only in conceptual but in ontological tenns.
What is ontological dependence and in what respect does it affect the
dependence between individuals and kinds? Although Jonathan Lowe
points to that very deep and interesting question throughout this book,
I cannot see an entirely satisfactory answer. Are kinds primary enti-
ties? On Lowe's view, they are at least indispensable, whenever one
accepts individuals in one's ontology. But if kinds have to be instanti-
ated or exemplified by individuals in order to count as real entities,
why should individual qualities not be able to build up those kinds, for
example, by constituting resemblance classes? In what sense, then, are
kinds ontologically stronger than resemblance classes of tropes? On
the other hand, if one holds, as our Opponent does, that individuals
and kinds necessarily depend on each other and that kinds are the very
tenns of lawlike statements, why should one not characterise this de-
pendence as some sort of metaphysical or ontological necessity? Al-
though this is implied by Lowe's own proposal, one might have
wished a more explicit assertion in order to make sure that metaphysi-
cal necessity is not to be identified with a brand of causal-cum-
possible-worlds-realism.
There is, of course, a good reason for Jonathan Lowe not to reduce
kinds. For ifhe did, he would eventually either fall back to the camp of
David Hume, or he would eventually find himself slapped on the back
by one of the Necessitarians with a cordial welcome to their club. In
order to avoid unacceptable company, Lowe is quite right to fortify his
favourite entity, the Kind, and make it as strong as possible. The new
label is: 'substantial universal'. This is, at least to my understanding, a
clear announcement to the effect that not any old universal can serve
as a supporter of laws. Moreover, the distinction between substantial
Kathe Trettin, Kinds of Necessity: a Comment on E.J Lowe 213
Identity
ON THE NOTION OF IDENTITY IN
ARISTOTLE
Mario Mignucci
1.
(for those things are said to be the same in species which fall under the
same species). Similarly, those are the same in genus which fall under the
same genus, as horse and man. (Top. I 7, 103'7-14, translation by Robin
Smith slightly modified, see Smith (ed.) [1997])
This is not the only place where an examination of the word 'tauton' is
offered,s or where this tripartition of the uses of 'sameness' is men-
tioned. 6
Numerical identity is probably the only kind of identity that one
would expect to find under the heading of sameness, so much so that,
below, when referring to numerical identity, we shall simply call it
'identity' without any further qualification, unless this is required by
the obvious necessity offollowing Aristotle's terminology.
Aristotle claims that 1..1 and ~ are (numerically) the same if they
have different names denoting one thing. To formalise his characteri-
sation of identity and the ways in which, according to him, 'tauton' is
used, we must make use of a language slightly richer than the usual
one adopted in a standard first order calculus with identity. We have to
introduce not only 'a', 'b', 'c', ... , as arbitrary names for individuals
and 'x', 'y', 'z', ... , as variables ranging over a domain of individuals,
but also '1..1', 'v', '~', ... , as arbitrary names for general or individual
entities, and 'p', '0', 'T', ... , as variables ranging over a domain of
general or individual entities. Then, if we take Aristotle's words as a
sort of definition of identity, we can say that 1..1 and ~ are the same, i.e.
fact, they are not the same: 'cloak' is a different name from 'coat'. It is
what is denoted by 'cloak' that is the same as what is denoted by
'coat'. Therefore, identity holds between things, pragmata, and not
linguistic entities. As we shall see, this observation is crucial for the
development of our analysis. 7
There are, however, at least two disturbing aspects of this way of
putting things. The first depends on the example of numerical identity
offered by Aristotle in the text. We would expect him to quote the case
of an individual with two names, as for instance 'Tullius' and 'Cicero'
for Cicero. But he mentions 'cloak' and 'coat', which are general
terms. For those of us raised in the protective shadow of Frege it may
be shocking to accept an identity relation between the denotata of non-
individual terms. The impression that Aristotle is not interested in dis-
tinguishing the case of sameness among individuals and sameness
among general or abstract entities is confirmed by the fact that else-
where he states a proposition expressing sameness of a particular with
itself as an example of (numerical) identity. 8 I will not discuss here this
apparent anomaly of Aristotle's approach to identity. Let me only ob-
serve that his view does not seem to be conditioned by the context in
which the analysis of identity is carried out in the Topics. In the Meta-
physics, where he seems very concerned to avoid assigning an onto-
logical import to the denotata of general terms, we find the same free
attitude towards the bearers of the identity relation: they can be not
only individuals but also what is referred to by universal terms. 9
Let us concentrate on the second awkward feature of Aristotle's
way of characterising identity in text (A). The linguistic aspect of it
should not by any means go unnoticed. Identity is said to occur when
the names of the entities involved by the relation refer to one and the
same object. As Paolo Crivelli in his comments has acutely observed,
this approach to identity implies that things to which identity applies
must have a name, and this assumption is not at all obvious. Worse
than that, identity seems to apply only to things which have at least
two possible names, as condition (iii) in (1) suggests. Apart from the
fact that it is not at all clear that we can give a name to a grain of sand
in the sea, although we can claim that it is self-identical, Aristotle is
well aware that names cannot match things, since the former are finite
and the latter infinite in number. 10 Even if we include definite de scrip-
220 Individuals, Essence, and Identity
(8) The same has several meanings: we sometimes mean the same nu-
merically; again we call a thing the same if it is one both in definition and
in number, e.g. you are one with yourself both in form and in matter; and
again if the definition of its primary substance is one, e.g. equal straight
lines are the same, and so are equal and equal-angled quadrilaterals -
there are many such, but in these equality constitutes unity. (Metaph. I 3,
1054a32-b3, Oxford revised translation modified, see Barnes (ed.) [1985])
But the textual evidence for such a move is not very strong,17 and one
might wonder what is to be gained by such a characterisation of iden-
tity. Surely (1 *) is not a real definition of identity since the notion of
sharing the same attributes involves the notion of identity.
(Numerical) sameness is contrasted with specific and generic
identity. We can say that two individuals a and b are specifically iden-
tical, namely Spec (a, b), if the species to which a belongs is the same
as the species to which b belongs, i.e.
(2) Spec (a, b) if S (a) = S (b)
It should be clear that' Sex), in (2) stands for 'the species of x'. Here
we can safely use, as Aristotle does, identity to define specific identity
because we can draw a clear-cut line between the two notions. To
speak in the Aristotelian way, if a and b are specifically (but not nu-
merically) identical they still count as two, which does not happen
with (numerical) identity. However, specific identity is based upon
(numerical) identity, because specific identity occurs only when the
species of a is (numerically) the same as the species of b. Moreover,
numerical identity implies specific identity, in the sense that
(3) a = b ~ Spec (a, b)
things which enjoy specific identity enjoy also generic identity, be-
cause individuals which fall under the same species fall also under the
same genus. 18
The problem that these definitions raise concerns the role that we
have to assign to them. Are they introducing distinctions in the notion
of identity in such a way that, for instance, we are allowed to equate
specific and generic sameness to something like relative identity en-
dorsed by Peter Geach and criticised by David Wiggins?19 Or are they
specifying a generic notion of identity in three different kinds of
sameness? I believe that we must strongly resist these views. The no-
tion of identity which is implied in the tripartition of the meanings of
'tauton' is always the same and it is simple identity. In the case of the
so-called 'numerical sameness', it is identity of what is denoted by two
names for objects, while in the case of specific and generic identity it
is identity concerning the species or the genus of two objects. I am in-
clined to think that Aristotle is not classifying kinds of sameness but
taking into account linguistic uses of 'same' in connection with propo-
sitions such as 'Coriscus and Dion are the same in species' or 'man
and horse are the same in genus'. Consider for instance the proposition
'Coriscus and Dion are the same in species'. Aristotle's analysis is
probably intended to warn anyone who is dealing with this proposition
that it must be taken as asserting not that Coriscus and Dion are the
same thing, but that they share the same species. No new notion of
identity is introduced, and Aristotle's distinctions are simply meant to
make it clear what identity refers to.
2.
In the same chapter from which we quoted text (A) there is a puzzling
passage concerning (numerical) identity, which is worth considering.
Aristotle says:
two-footed terrestrial animal for man. The second case is when it is indi-
cated by means of a proprium, e.g. capable of knowledge for man or car-
ried upwards by nature for fire. The third case is when it is indicated with
an accident (apo tou sumbebekotos), e.g. the one sitting or the musical one
for Socrates. All these case are intended to signify what is one in number.
(Top. I 7, 103'23-31, Robin Smith's translation modified)
This text is meant not only to present cases in which (numerical) iden-
tity or oneness is expressed but also offer a sort of hierarchy of them.
Aiming to have a better understanding of what is behind this strange
passage, let us reverse the order of cases and start from the last, i.e. the
case of identity concerning individuals. If we put Aristotle's examples
in a proper sentential fonn we get propositions asserting a relation of
identity in which one and the same particular is picked out by a proper
name, 'Socrates', and by what looks like a definite description, 'the
one sitting' or 'the musical one'. If this view is correct we can fonnu-
late one of these examples in a semi-fonnal way by stating:
(5) Socrates = the one sitting
Aristotle says that in such a case the definite description is taken "from
an accident", surely because 'being sitting' is an accident of Socrates.
Since in (5) identity is expressed by means of an accident one can
confidently infer that (5) and all proposition of this kind are contingent
statements. By making a further step, one might be tempted to con-
clude that (5) and the other similar propositions express contingent
identity. We say that a = b expresses identity contingently if its truth
does not rule out the possibility of a f. b. The argument in favour of
this further conclusion might be as follows. We can generalise (5) by
introducing the 'I' operator used by Russell to represent definite de-
scriptions, so that we can express the fonn of a definite description by
'lxF(x)', which means 'the unique x which F_S'.20 In this way, the logi-
cal fonn of Aristotle's example is
(6) a = IxF(x)
Moreover, the accident which plays a role in (5) is 'being sitting' and
it is obviously an accident of Socrates. Therefore, the condition is
added that F(a) is an accidental predication. What is characteristic of
Mario Mignucci, On the Notion of Identity in Aristotle 225
accidental predicates is that they need not hold of the subjects to which
they are attributed. Therefore, if F(a) is constituted by an accidental
predication we can state F(a) and O-F(a). Now, we can easily derive
F(a) from a = IxF(x), so that the following implication
3.
say that here 'cloak' stands for the class of the individuals which are
cloaks, or one might imagine that 'cloak' and 'coat' are names of con-
cepts. Whatever the ontology of these things may be, it is hard to reject
the idea that 'cloak' and 'coat' refer to special entities, being names, or
rather, proper names for them.
Suppose now that these names are not directly referring, but must
be expanded into descriptions, 'the object called 'cloak", or 'the ob-
ject called 'coat", in the same way we imagined that 'Tullius' and
'Cicero', taken as descriptions, had to be expanded into 'the one called
'Tullius", 'the one called 'Cicero". It would be plausible to maintain
that in 'the object called 'cloak' (or 'coat')' the property of being
called 'cloak' ('coat') is accidental with respect to the object referred
to in the description. After all, according to Aristotle, names are as-
signed by convention and not by nature. 26 Therefore, if 'Tullius ==
Cicero' is a contingent statement of identity under the assumption that
'Tullius' and 'Cicero' stand for definite descriptions, since these de-
scriptions involve an accidental property of the thing designated, the
same should be true for 'cloak = coat' because also in this case de-
scriptions referring to accidental properties are used. But we have seen
that 'cloak == coat' cannot be taken as a contingent statement of iden-
tity since it is put on the same footing as definitional identity.
To avoid this conclusion, we must give up either the assumption
that (i) 'cloak' and 'coat' are proper names, or that (ii) "being called
'cloak' ('coat')" is an accidental attribute for the object referred to, or
finally that (iii) 'cloak' and 'coat' function as definite descriptions. It
is hard to dismiss the first assumption, because 'cloak' and 'coat' are
called by Aristotle onomata, "names", in text (A) and they are said to
be names for one and the same pragma, "object". Therefore, they do
not play the role of kategoroumena, "predicates", of a subject, their
main function being to point to something.
One might challenge the second assumption along the following
lines. It is true that in some sense "being called 'cloak' ('coat')" is an
accidental attribute of the object so described, since one might imagine
a different name or group of names for this object. But still a differ-
ence remains with the case of 'Tullius' and 'Cicero'. Synonymous and
homonymous terms are given in a language. Therefore, we must pre-
liminarily fix a language in order to decide about identity involving
230 Individuals, Essence, and Identity
to a certain extent true, since we cannot find for the Principle of Iden-
tity assertions as explicit as those concerning the related Principle of
Non-Contradiction, which is said to be one of the most reliable axioms
of philosophf8 and a necessary truth. 29 However, first of all, there is
evidence to maintain that for Aristotle a = a (or 1-1 = 1-1) is a sound
statement. In Metaph. !!. 9, 10 lSa S-9 that a thing is the same as itself is
quoted as an obvious case of numerical identity and in I 3, 1054a35
"you are the same as you" is offered as an evident example of numeri-
cal oneness. Moreover, in a passage of the Sophistical Refutations the
Principle of Identity is used to build up a sophistical argument con-
cluding that one and the same thing is the same as itself and different
from itself,30 and the explanation of the paradox does not involve any
restriction on the Principle itself. Finally, the idea that a proposition
such as 1-1 = 1-1 constitutes a philosophical principle is suggested by a
passage in the Metaphysics where Aristotle says:
CD) To inquire why a thing is itself is not a real inquiry, since for it the
fact or that the thing is must already be evident (e.g. that the moon is
eclipsed), but the fact that a thing is itself is the only reason and the only
cause to be given for all such questions as why man is man or musical is
musical; unless one were to say that each thing is indistinguishable from
itself and its being one just meant this; but this is common to all things
and is a short and easy way with the question. (Metaph. Z 17, 1041 "14-20
Oxford revised translation modified)
4.
We are in big trouble. On the one hand, we have claimed that identity
can be contingent in the sense that a = IxF(x) is contingent and, on the
other, we have concluded that a = b is a statement of necessary iden-
tity. A contradiction arises. Since a = b is a necessary statement we are
allowed to write
(l1)a=b~Da=b
But why should one take an identity statement like (5) as implying a
necessary proposition in the sense of (16)1 When we say that Socrates
is the one sitting, we are not thereby committed to the claim that nec-
essarily Socrates is the only sitting object. It might have been that Soc-
rates is not sitting or that he is not the only one to sit. We can more
plausibly take (5) to mean that one and the same person, Socrates, is
described as the only sitter and addressed as 'Socrates'. It may be that
the description is based on an accident of Socrates but Socrates, how-
ever described, is always the same Socrates as the one called 'Socra-
tes', and this is a necessary fact. In this perspective if we would like to
make explicit the kind of necessity involved by the use of identity in a
statement such as (5) when (5) is interpreted as (13) and identity is
supposed to satisfy (11), we should state something like
5,
Let us come back to text (C). Various cases of identity are mentioned
and a hierarchy among them is established. We have already consid-
ered two of these cases. One occurs when identity is expressed by at-
tributing a definite description to an individual; the other is the case of
two synonyms for the same general entity. It seems legitimate to
equate to the latter the case of two synonyms for the same individual,
i.e. the case of identity expressed by propositions such as 'Tullius =
Cicero'. After all, in a passage parallel to text (C) Aristotle mentions
the proposition 'you are you' as an example of the strongest case of
identity, and this fact shows that he does not distinguish between the
situation of general entities and individuals. 34
We must briefly explore now the two remaining cases, i.e. the case
of the attribution of a proprium to its subject and the case of definition.
Here, once again, Aristotle is applying identity to entities denoted by
general terms, taking the latter as proper names for the former. In this
perspective, we must consider the dejiniendum which appears as a
member of definitional identity as expressing the name of a class or
concept, no more and no less than 'Tullius' is the name for an individ-
ual and a way to indicate it. If we take this point of view seriously, we
can think of a dejiniens as a special kind of definite description. As we
have seen, 'the one sitting' is a definite description of Socrates apo tou
sumbebekotos. In the same way, we can consider 'two-footed terres-
trial animal' as a definite description of what is designated by the
name 'man'. The only difference with respect to Socrates and 'the one
sitting' is that 'two-footed terrestrial animal' is not apo tou sumbebe-
kotos with respect to man, being not based on an accidental attribute of
man, but it expresses what answers the question 'ti esti anthropos;',
'what is a man?', i.e. it picks out the essence of man.
We can interpret the case of proprium along the same lines of ap-
proach. Basically, a proprium is an attribute which applies to a subject
and no other. In this sense, 'capable of knowledge' is a proprium of
man because only men are capable of knowledge. Consider for in-
stance the statement 'man = capable of knowledge' and, as before,
take 'man' as a name of the entity designated by 'man'. Then 'capable
of knowledge' can be considered as a definite description of this en-
Mario Mignucci, On the Notion of Identity in Aristotle 235
What is interesting is that for Aristotle the two different ways of con-
sidering propria are equivalent and interchangeable, even if the onto-
logical commitment of the two approaches is quite different. But I
would not like to be trapped by this difficult question.
It is more appropriate to point out that we are now in a position to
explain what the cases of identity considered by Aristotle are. It should
be clear that it is not a difference in identity: a and b cannot be more
(or less) identical than c and d are. If they are identical they are simply
the same. Nor can it be a question of "strong" versus "weak" identity,
i.e. necessary and contingent identity. As we have seen, identity in
Aristotle's view is always "strong" in the sense that it is necessary. We
must take not only 'man = two-footed terrestrial animal' as a necessary
proposition but also 'Socrates = the one sitting' as involving necessity
in the sense we have explained.
I think that the cases described by Aristotle in text (C) refer to dif-
ferent ways in which identity can be expressed. We can express iden-
tity by picking two names for the same object, or we can use a definite
description. As we have seen, definitions, propria and definite de-
scriptions for individuals all fall in the latter case, although they mani-
fest differences. In the case of a definition a description is offered
which reveals the essence of the definiendum. In the case of a pro-
prium we have a description based on a necessary but not essential at-
236 Individuals, Essence, and Identity
Notes
I Top. I 5, 102"7-9.
2 Top. I 5, 1028 36 ff.
3 See e.g. Maier [1969-1970], II b, p.78; Hambruch, [1904], p.3 ff.; Solmsen
[1929], p.194; Ross [1939], pp.251-272; Bochenski [1968], pp.22-24; During [1966],
p.69 ff.; Barnes [1981], pp.17-59.
4 I use 'sameness' and 'identity' as synonyms. Therefore, by attributing a theory
of sameness to Aristotle we intend to ascribe to him a theory of identity, pace White
[1971], pp.I77-197. See also Miller [1973], pp.483-490.
5 E.g. Metaph. 119; 13, 10548 32 ff.
6 Texts in Bonitz [1955 2], 125"57 ff.
7 I am grateful to Professor Enrico Martino for having made me aware of the
II I am once more grateful to Paolo Crivelli for his useful comments, which
have greatly helped me to correct my previous view and better understand Aristotle's
position.
12 Alex., In Metaph. 615.20 ff.
13 The same view in Ross [1953], II, p.287).
14 See e.g. Metaph. !l. 9, 1018"4-9.
15 Metaph. !l. 9, 1018 8 7-9.
currences of a in -A).
(LS) is notoriously at odds with systems of contingent identity. However, also these
systems normally adopt a non-modal version of it, in the sense that (LS) is supposed
to hold if -Aand IJdo not contain modal operators. We have respected this condition
in using (LS) since (6) contains no modal operator (see Hughes and Cresswell [1996]
pp.332-334).
23 E.g. Metaph. Z 15, 1039b 27 ff.
24 A.Po. II 13, 96"24 ff.
25 The fact that proper names are supposed to have a primary referential func-
tion does not necessarily imply that they do not possess a sense. On this question see
D. Wiggins [1976], pp. 221-255
26 Int. 2, 16"19-20
27 See e.g. Berti [1987], pp.212-213.
28 See e.g. Metaph. r 3, 1005 b 5 ff.
29 E.g. Metaph. r 3,1005 b I9-22.
30 SE 30, 181 b 13 ff.
31 See e.g. Ross [1953], II, pp. 206-207, Reale [1993], III, pp.396-397, Frede
and Patzig [1988], II, pp.310-312.
32 Russell was aware of the ambiguity involved by the use of the ',' operator
and to avoid it he introduced the notion of scope of a description (Principia Mathe-
matica, I, p. 173). The ambiguity of ',' increases in a modal context (see Hughes and
Cresswell [1996], pp.324-325).
33 I am aware of the fact that using formulas with arbitrary names in a modal
context can raise difficulties (on them see e.g. Hughes and Cresswell [1996], p.274
ff.). However, I do not think that I need take care of them at the level of analysis on
which I am moving.
34 Metaph. I 3, 1054"34-35.
35 For the definition of proprium see Top. I 5, 102"18-24.
36 This idea is implied by Metaph. 11 9, 1018"7-9. See also L. Wittgenstein
[1922], 5.5303: "Beiliiufig gesprochen: von zwei Dingen zu sagen, sie seien
identisch, ist ein Unsinn, und von Einem zu sagen, es sei identisch mit sich selbst,
sagt gar nichts. "
37 Metaph. 119, 1018"7-9; Metaph. I 3, 1054"33-35.
SAMENESS IN ARISTOTLE'S TOPICS
Paolo Crivelli
their definitions; for when we have shown that they are not the same we
shall have demolished the definition. But the converse of this last state-
ment does not hold: for to show that they are the same is not enough to
establish a definition. To show, however, that they are not the same is
enough of itself to overthrow it. (Top. I 5. 102"6--17)9
Notes
J Cf. Alex. Aphr. in Top. 57, 23-4; 60, 11-2; Pacius [1597], p.355; Rolfes
[1919], p.205; Brunschwig (ed.) [1967], p.124; Smith (ed.) [1997], p.68.
2 I shall freely (and somewhat sloppily) use "kinds of sameness" as equivalent to
"uses of the expression 'the same'''.
3 Cf. VII 1. 152b 30-3; Ph. VII 1. 242a32-~; de An. 15. 411 b20-1; HA II 1.
497b9-13. The classifications of the uses of 'the same' offered by Aristotle in the
Metaphysics (89. 1017b27-1018 a9; 13. 1054 a32-b3) are different from the one we
find in Topics I 7 (the main difference is that they involve the concept of matter,
which at the time of the Topics was not yet a philosophical tool of Aristotle's).
4 Cf. Alex. Aphr. in Top. 60, 17-61,5; Pacius [1597], p.355).
5 For 'priority in definition', see Metaph. Z 1. 1028 a31-6; M 2. 1077b2-4 (cf. Ph.
VIII 9. 265"22-4; Metaph. 8 11. 1018b30-6; Z 13. 1038b27-8; e 8. 1049b I2-4).
6 Aristotle explicitly recognises that numerical sameness is the most fundamental
David Wiggins
1.
Bredon Hill and then another possible world may be simply specified
by saying that there, in that second world, the very same boy and girl,
namely Jack Adams and Jill Jones, walk up Muswell Hill. Why is that
rejected? Why can things not be as simple as that?
Is the thought that, strictly, the identities of the objects in a world
are not to be stipulated in this way but must be founded, however indi-
rectly, in their other properties and relations? Is the thought that, in the
two would-be world-specifications just indicated, the names, 'Jack
Adams' and 'Jill Jones' are really shorthand in the constructor's vo-
cabulary for descriptions on the basis of which more elaborate identifi-
cations of the sort that Hintikka envisages might have been effected;
that otherwise - is this the idea? - these vocables are only the names
that two persons who are yet to be identified bear in the world under
construction?
If nobody has any inclination to say something analogous to this
about the ordinary properties and relations, then why is the claim so
perennially attractive where identity is concerned? Why is identity seen
as a poor relation of other properties and relations? Why is it problem-
atic to find Jill Jones in world one and in world two but unproblematic
to find the colour blue in world one and in world two? Why are we so
tempted to elide the difference between the reasonable claim that we
are not free to stipulate in a possible world whatever we like and the
questionable claim that the identities in a possible world need a foun-
dation? That is the question I begin from.
2.
Perhaps the relevant thought is that the identity of objects does not an-
nounce itself; that identity itself is unobservable, and unobservable
precisely because it is somehow consequential on other things. But if
so, then the linguistic philosopher within me wants to protest: "Do you
really mean that? Can one not see straight off and entirely directly
(however fallibly) that it is one's brother, not one's uncle, who has
come to call on one?" But, at best, that is a straw in the wind.
David Wiggins. Identity and Supervenience 249
Perhaps the relevant thought is that, even allowing for the point
just registered in my quietist cum language-centred protest, identity is
inscrutable except as resemblance; that identity is the vanishing point
of resemblance, and resemblance can only be a matter of properties
and relations.
If that were the thought, the quietist response (and my response)
would be that it hardly takes deep philosophy to reflect that resem-
blance itself is neither sufficient for identity, nor yet necessary. It is
only necessary to this extent: at any given time, x must exactly resem-
ble x. "Identity is all right", the quietist says, "leave it alone. Don't
spoil it. Try not to paint the lily."
3.
and so on to '(z)(w) (if Fxzw then Fyzw)" etc. The conjunction of all these
formulae is coextensive with 'x = y' if any formula constructible from the
given vocabulary is; and otherwise we can without conflict adopt that
conjunction as our version of identity. In so doing we impose a certain
identification of indiscemibles, but only in a mild way. (Quine [1960],
p.230)
Wherever such eliminations are possible (the thought may run), there
is a clear sense in which the predicate 'is the same as' is indeed re-
ducible to other predicates and relation words. For 'x is the same as y'
is replaceable by another and longer open sentence with the same free
variables x and y, an open sentence that will have to have the same
extension. How better or more satisfyingly could the whole mystery be
dissipated of identity and our grasp of identity? The identity of x and y
comes down to the reflexivity and congruence of some relation in
which x stands to y. That is all there is to identity, or so it may be said.
4.
The first thing to remark is that the recipe Quine has given for the con-
fection of this first level surrogate of identity is indeed a sound recipe.
Its technical correctness is beyond doubt. The only question is what the
recipe shows about the relation of identity. Nor is there any doubt that
Quine furnishes a single and unitary recipe. One of the questions that
remains open, however, is whether the results of applying this recipe
do justice to the evident univocity of the sentence form 'x is the same
asy'.
Suppose that, for each of several languages unlike English in hav-
ing a clearly delimited set of predicates, each different from all the
others in respect of its basic vocabulary, we frame by exhaustion its
surrogate identity predicate, and then we replace the extant '=' predi-
cate of the language by the surrogate predicate. What do the surrogates
framed for these various languages have in common? Each expresses a
certain congruence with respect to the predicates of the particular lan-
guage into which it is introduced. But it may trouble us that, so long as
we restrict ourselves carefully to that which can be expressed in first-
David Wiggins, Identity and Supervenience 251
level terms and we are confined to that which can register within
Quine's various first-level versions of 'x = y', we cannot express what
it is that constitutes each surrogate an identity or congruence predicate.
For Quine's method works by surreptitious allusion to and silent de-
pendence upon an idea which is essentially second level. The second
level idea is that identity is the relation whose holding between a and b
ensures that every property of a is a property of b and vice versa. The
trouble with such a second-level account of identity, however, is that it
involves quantifying over all first-level properties, including identity
(unreconstructed identity) itself and the countless other properties that
latently involve identity.
It might be replied that none of this matters for the philosophical
purposes we are pursuing so long as there is for each language some
surrogate identity predicate.
Such a defence would raise difficult questions about how it is that
speakers with different active vocabularies that determine different
surrogates understand one another - unless on the basis of a second
level conception of identity. But let us not dwell on this here because
there is ground for a more specific reservation. It relates to ontology
and it can be put in Quine's words:
If the universe is taken as that of persons, and the predicates are inter-
preted in ways depending on nothing but people's incomes then the pro-
posed manner of defining 'x = y' will equate any persons who have equal
incomes; so here indeed is an unfavourable case where 'x y' [defined in
=
the manner indicated here, as in 'Reply to Professor Marcus'] does not
come out with the sense of genuine identity. (Quine [1963], § 1, p.15)
5.
No doubt there are purposes for which Quine's remedy is fully ade-
quate. But would it give everything he needed to the kind of empiricist
philosopher I have undertaken to answer, the one (I mean) who worries
about the empirical credentials of umeconstructed identity and thinks
he sees what he is looking for in the idea that one can dismantle the
identity relation into reflexiveness and congruence?
Suppose one believes that the proper interpretation of a living lan-
guage depends on the interpreter's taking the language in its full prac-
tical context and explaining the speakers' use of the predicates of the
language by reference to its speakers' constant commerce with one an-
other and with various objects that are manifest in the portion of the
world to which they are seen by the interpreter as responding reasona-
bly in perception, in thought and feeling, and in action. If that is the
picture one has of interpretation (and this might as well be the empiri-
cist's picture too), then one will want to question whether absolutely
everything that grounds interpretation for a given language must be
such as to register in the explicit predicative repertoire of the language
itself. Surely it can be practically manifest to the interpreter that the
speaker is concerned with people not groups, for instance, and practi-
cally manifest in ways that outrun explicit speech. If that is right, one
concludes, then Quine'S reconstrual of ontology cannot fill the bill.
All right, some defender of the surrogate may say. Let me allow
that an interpreter needs to interact with the objects his subjects inter-
act with. Let me allow that one who interprets has to deploy practical
understanding in tracking his subjects' practical understanding. But I
still side with Quine. There must be some process (the defender may
say) by which an interpreter's implicit or practical understanding of the
role of language in the life and conduct of its speakers can be made
more and more explicit. Let the interpreter draw on his own predica-
tive repertoire. And then, where necessary, let him (however notion-
ally) fill all the gaps in the language he interprets. That will sort out the
people/income groups problem. It will sort out any other ontological
difference that is a real difference.
Yes, I reply, but now one must ask whether this reply will seem
satisfactory so soon as one reflects that the process of explication need
David Wiggins. Identity and Supervenience 253
not terminate. In so far as the process does not terminate, it seems that,
even if the predicative resources and repertoire of the language are
constantly extended towards semantic saturation, the intended defini-
tion by means of exhaustion of predicates may never be available. Al-
ways the explicit predicative repertoire may be expected to lag behind
the demonstrative cum practical capacities of subjects and interpreters.
6.
There are questions I do not want to try to resolve here, not least about
Quinean doctrines of indeterminacy or unscrutability of reference. It is
more important for present purposes to gain a fuller appreciation of the
scale of the difficulty we are concerned with. A further consideration is
this. John Wallace points out that Quine's method of reducing '='
yields rather stranger results than any that Quine himself has explicitly
countenanced. 3 Applied to a quantificational language with just three
unanalysed predicates 'x is a forest', 'x is a tree' and 'x grows in y',
Quine'S proposal forces the truth-value false upon the sentence 'In
every forest there grow many trees' .
The defender of surrogate identity can claim that such troubles as
this will disappear so soon as we consider first-order theories with an
expressive power more closely approximating to that of English.
Wallace suggests that the claim of eliminability would need to be re-
phrased as follows: if a theory has a finite number of unanalysed predi-
cates, then a finite number of predicates can be added to the theory, so
that, in the resulting expanded theory, what we usually intend by iden-
tity is eliminable by the prescribed method. But now - at least for the
purposes of the evaluation of the idea that identity can be reduced or
explained in this way - everything is seen to depend on what predi-
cates are added in order to force the desired truth-value upon sentences
involving the constructed predicate that is to be introduced in lieu of
'='. Suppose that the appearances can only be saved if monadic and
polyadic predicates presupposing identity or place-, time- or thing-
individuation are supplied. If so, it is upon the presence of these that
the success of the elimination recipe will depend. But in that case, no
254 Individuals, Essence, and Identity
real elimination has been achieved of the kind that that empiricist I be-
gan with ought to have an interest in. The whole case remains to be
made for the philosophical claim that the presence in a language of the
identity predicate boils down to some predicable's introducing a re-
flexive relation that is a congruence relation with respect to the prop-
erties imported by the predicates of the language.
7.
8.
(8) The dog at Colonel Ross's stable, a dog who did not bark
and did not awaken the stable lads on the night of Silver
Blazes's disappearance.
Holmes reflects that character (9) the abductor, must have been famil-
iar to (8) the dog, Otherwise the dog would have barked when the ab-
ductor led Silver Blaze out onto the moor. So (2) John Straker = (9),
The knife suggests that Straker meant to lame Silver Blaze. If so, he
must have intended to lay money against Silver Blaze. Who then is
(10)? Scarcely (6), who was abroad with a leaded stick but had no mo-
tive. The only other possibility is the horse himself. In fright and reac-
tion to the prick of the knife, he kicked and trampled John Straker. So
(1) = (10). In which case character (6) is innocent. Our cast of ten
characters is really a cast of eight.
Is this sort of thing, which is utterly familiar to us, the sort of thing
that the empiricist has in mind? If it is, then the obvious objection
seems to be that what is here in question is deductive reconstruction
from an incomplete narrative of a fuller narrative; and this is the re-
covery of a narrative in which things have an identity already deter-
mined. Epistemically speaking, there is here a triumph, a triumph of
"deduction". But there is no model here for the constitutive determina-
tion of identity out of properties and relations that do not include iden-
tity.
Let this obvious finding be the occasion to reflect that the idea of
supervenience is often said by those who set store by it to be this: that
properties from a certain range of higher level properties should super-
vene on some other range of more explanatorily basic properties. But,
if supervenience is a relationship between classes of properties, then is
not the identity itself of the subjects of such properties to be treated as
something taken already for granted? The supervenientists seem to
suggest this themselves when they say (e.g.) that "it is not possible that
two things should be indiscernible in respect of their lower level prop-
erties without also being indiscernible in respect of their upper level
properties". On this account of the matter, identity itself lies outside
the intended area of supervenience.
To defuse this objection, the empiricist might want to explore the
following suggestion: facts about the identity of objects supervene on
facts about the exemplification of properties and relations other than
the identity of objects. When the claim is put in this way, it is left open
what the exemplification of the other properties and relations comes to.
Maybe these exemplifications are given in feature placing or pointil-
David Wiggins, Identity and Supervenience 257
lO.
There are things I do not understand here. But even from this be-
nighted condition, I am moved to mention an apparent difficulty, a dif-
ficulty long since familiar from discussions of Leibniz's Principle ea-
dem sunt quorum unum alteri substitui potest salva veritate. The Iden-
tity of Indiscernibles, on one standard interpretation of Leibniz, comes
to this: x is identical with y if and only if, for all pure properties cP, x
has cP if and only if y has cP; where pure properties are those that in-
volve or presuppose neither identity itself nor place-, time-, nor indi-
viduation, nor thing-individuation.
This idea of a pure property is right for Leibniz and it is equally
right for our purposes. But then, as Wittgenstein noted (Tractatus
Logico-Philosophicus 5.5302), any such principle will rule out two
objects' having absolutely all their properties in common. So, as Max
Black pointed out, it rules out the logical possibility of a universe con-
sisting of two qualitatively indistinguishable spheres. The wild implau-
sibility of this exclusion seems unfavourable to any supervenience the-
sis. Following on from where Wittgenstein and Black left off, P.F.
Strawson then pointed out, in similar opposition to Leibnizian mona-
dology and the Identity of Indiscernibles, that
When faced with such examples, some people (once upon a time I was
one of them) may not flinch from the thought that it is a presupposition
of our individuative practices that we do not live in such worlds as
Strawson and Wittgenstein have described. Because the presupposition
is global, they believe it can be lived with - pragmatically, say, and/or
as a "framework principle".
If this were the worst of it (I reply) then maybe it would be bear-
able. But matters are even worse than this represents. Is it also a pre-
supposition of our individuative practices, or a framework principle for
them, that there are no fully symmetrical objects? (A symmetrical ob-
ject with exactly matching top half and bottom half has parts that are
indiscernible. It is effectively reduced to half of itself by the Identity of
Indiscernibles as strictly construed. Moreover, an object that is sym-
metrical about all planes which bisect it seems to be precluded alto-
gether. For not only would the top half of such an object be indiscerni-
ble from its bottom half. The left side of the bottom half would be in-
discernible from the right side of the bottom half. And as regards the
residual eighth ... Eventually, we are left with nothing but a line and
then, at the limit, a geometrical point.) Are we to believe that it is a
"framework principle" that no object is symmetrical about all planes
which bisect it?
Strawson's example was devised to embarrass the Leibnizian use
of the Identity of Indiscernibles. But more generally, the thing it seems
to show is that the facts about identity in a given set-up need not su-
pervene on the qualitative and structural properties of the array that
constitutes the set-up. Rather, we may need facts about identity in or-
der to characterize the array itself. If that is right, then no interesting
thesis of supervenience seems defensible. Of course, it is a crucially
important fact about identity, it is of the essence of identity, that it rubs
shoulders with other properties and relations. But that is not enough
for supervenience. Identity is just identity. It cannot be confected or
fixed from other properties and relations.
11.
That is the conclusion that I am led to. But the empiricist's response
may be to say that, with a symmetrical object, we can make the dis-
David Wiggins, Identity and Supervenience 259
tinction between one half and the other deictically. A similar response
might have been offered to the WittgensteiniMax Black set-up.
If the empiricist says that, then I entirely agree. But in agreeing I
am not restoring any presumption in favour of the thesis that identity is
some sort of resultant from other properties and relations. Rather the
point that is agreed about deixis seems to me to remove the very need
for that presumption. For it undermines the main reason that philoso-
phers might give themselves for adhering to the Identity of Indiscerni-
bles. Is it not the whole answer to the empiricist's worries about iden-
tity for him to come to see our grasp of identity (and our grasp of
shape and array) as a by-product of our grasp of reference to objects
and the singling out of objects? Identity is not definable. It is as fun-
damental as any notion can be. But that's all right. It's what you would
expect. For even on these terms, the notion of identity is tractable.
12.
13.
So much for the primitive case and the elucidation of what is involved
in that case. The larger class of primary judgments of identity one as-
sembles by reaching beyond the primitive cases and counting into this
larger class judgments that are answerable to the very same dialectic of
David Wiggins, Identity and Supervenience 261
14.
First, if we look again at the primitive case based on singlings out and
prolongations of singlings out, then it is worth noting something about
congruence. Simply by virtue of what he finds on the basis of a sin-
gling out, the prolongation of that singling out and the renewal under
changing or changeable conditions of that act of singling out, a thinker
who does these acts in the way that I have described can then declare -
no matter what property l/J is and no matter whether the question of a's
or b's instantiating l/J figured within any inquiry of his into the spatio-
temporal paths of horse a and horse b - that object a has l/J if and only
if object b has l/J. For simply to determine correctly the answer to the
continuity question about the horse, the question (that is) about the
traceability through their life-histories of the horse that scratched itself
and the horse now in the comer, precisely is to settle it that, no matter
what property l/J is, horse a has l/J if and only ifb has l/J.
In epistemological reflection of this criteriological point, the hy-
pothesis now presents itself that it is impossible even in theory to con-
ceive of some way independent of the prior discovery that a = b by
which to establish that a and b have all and only the same properties.
For suppose, simply seeing identity as reflexivity plus congruence, one
were to renounce all elucidations of identity other than those given in
terms of a's and b's complete community of properties. Then how
would one think about the non-permanent properties enjoyed by an
individual a identified with respect to the past and the properties en-
joyed by an individual b identified with respect to the present? One is
only justified in pooling the non-permanent properties of a and b if
there is some other basis for the identity of a and b than their having
all their properties in common. The only basis on which such pooling
is possible (I suggest) is the notion of a certain kind of continuant with
respect to which one can ask 'what is it for an f to persist?' It is this
that makes room for the idea of a sequential history of a thing's doings
and undergoings. Without this idea, little sense will be made of very
much that we actually do with the concept of identity. As Leibniz puts
the point that I too have wanted to insist upon, "By itself continuity no
more constitutes substance than does multitude or number.... Some-
thing is necessary to be numbered, repeated and continued" (Gerhardt
David Wiggins, Identity and Supervenience 263
15.
My second closing point is that the claim I have just made about con-
gruence is not limited to concrete continuants. Consider natural num-
bers, for instance. Suppose that, in making a tally of the fs, Edward
first uses the vocable 'one' then two more vocables. His tally of theft
he gives as 'three'. Suppose Kallias, making a tally of the gs, first de-
ploys 'hen' then uses two more vocables of which the second is 'tria'.
Then on this basis alone, which suffices to secure the numerical iden-
tity, we can say that whatever is true of Edward's tally number is true
of Kallias' tally number. (Or so anyone will say who agrees to make
sense of natural numbers as objects.) No more is needed. In practice
we don't start with congruence, we reach congruence last.
16.
One concluding point. A fair summary of a great deal that I have said
could be condensed into this.
(1) a=b~(3f)(a~b)
There is no question of supplanting the left hand side by the right hand
side and achieving thereby a philosophical analysis. But the right hand
side casts light on the left hand side because it indicates (or reminds
us) how, in practice, we exercise our understanding of identity. We
exercise it by deploying among the other understandings that we have,
our understanding of some determinate or determination of the deter-
minable notion entity of some particular/determinate kind. In deploy-
ing that in the world of objects we exercise the clear indistinct idea we
have of that particular determination. The possibility of such determi-
264 Individuals, Essence, and Identity
Notes
1 I make this claim in Chapter Six of Sameness and Substance Renewed (Wig-
gins [2001]). Several extended passages of the present article appear in another form
in Sameness and Substance Renewed. They appear here by kind permission of Cam-
bridge University Press.
2 As is well known, Saul Kripke has spoken out eloquently against rulings of this
kind. See Kripke [1980]. If his efforts had eradicated completely the underlying
philosophical conceptions that have seemed to sustain such a ruling, then I should not
have begun my speech here.
3 See John Wallace [1964]. See pp. 80ff. John Wallace's original sentence was
"in every forest there grows more than one tree". Timothy Williamson points out to
me that the sentence it would be better for Wallace to consider is "In every forest
there grow many trees". Unlike "more than one", "many" does not reimport identity.
Then (Williamson suggests), in order to make Wallace's point properly secure, let
the quantificational language be treated as possessing a "many" quantifier as primi-
tive. The philosophical significance of Quine's elimination recipe ought not to de-
pend on the question whether "many" is admissible as a logical constant.
COMMENTS ON WIGGINS'S "IDENTITY
AND SUPERVENIENCE"
Edmund Runggaldier
Timothy Williamson
273
A. Bottani et al. (elis.), Individuals, Essence and Identity, 273-303.
© 2002 Kluwer Academic Publishers.
274 Individuals, Essence, and Identity
1.
with the fundamental principles governing the notions of truth and fal-
sity (Williamson [1994], pp.187-98; see Andjelkovic and Williamson
[2000] for further issues). However, for the sake of argument we may
bracket this general concern in order to investigate some more specific
problems for many-valued logic as a framework for an account of
vague identity.
The relevant systems of many-valued logic do more than postulate
a neutral status. They treat it as another truth-value, an alternative to
truth and falsity. The two most salient systems are three-valued logic
and fuzzy logic (continuum-valued logic). In three-valued logic, the
truth-values are just truth, falsity and neutrality. Fuzzy logic, by con-
trast, postulates a continuum of degrees of truth from perfect truth to
perfect falsity, with neutrality in the middle; these degrees are typically
identified with the real numbers in the closed interval [0,1], with 1 as
perfect truth, 0 as perfect falsity and 0.5 as perfect neutrality.
The most distinctive feature of many-valued logic is that by defi-
nition it generalizes the truth-tables of two-valued Boolean logic by
postulating a principle of generalized truth-functionality: it computes
the truth-value of a complex sentence formed by applying operators
such as negation, conjunction, disjunction or the conditional to simpler
sentences as a function of the truth-values of those constituent sen-
tences. For example, given that neutrality is a truth-value and that the
negation of the neutral 'SI = S2' is neutral too, generalized truth-
functionality implies that the negation of any neutral sentence is also
neutral. Granted that negation permutes truth and falsity, that is enough
to fix the truth-table for negation in three-valued logic. The easiest
generalization to continuum-valued logic is that if the degree of truth
of a sentence a is x, then the degree of truth of its negation -u is I-x,
just as the probability of """'a is I-x if the probability of a is x. The
phrase "many-valued logic" will be restricted here to systems with
generalized truth-functionality.
Let us focus on the binary connective of conjunction. Since the
conjunction a /\ 13 entails each of a and 13, but entails no more than is
needed to have those two entailments, it is natural in continuum-valued
logic to assign a /\ 13 a degree of truth no higher than the degree of
truth of a and no higher than the degree of truth of 13, but no lower
than is needed to meet those two constraints. Thus the degree of truth
Timothy Williamson, Vagueness, Identity, and Leibniz's Law 277
ing it as not tall; we have no idea how to decide the question. On the
many-valued approach to vagueness, we must therefore evaluate the
sentence 'The east tower is tall' as neutral. Since the west tower is ex-
actly the same height as the west tower, we must evaluate the sentence
'The west tower is tall' as neutral for the same reason. Since the nega-
tion of a neutral sentence is itself neutral, we must also evaluate the
sentence 'The west tower is not tall' as neutral. But then (i) is the
conjunction of two neutral sentences, so (as noted above) the many-
valued approach forces us to evaluate (i) as neutral too. But that is to
misevaluate (i). It is not neutrally poised between truth and falsity;
rather, in the circumstances one should clearly prefer denying (i) (as-
serting its negation) to asserting it. Thus the many-valued approach
bungles the semantics of complex sentences. The objection does not
depend on the peculiarities of any particular many-valued system; it is
endemic to the approach, for the crucial assumptions were just gener-
alized truth-functionality and the possibility of neutrality.
Could a many-valued logician reply that (i) merely sounds false
because, by speaking asymmetrically of the height of the east tower
and the height of the west tower, the speaker generates a false conver-
sational implicature that they are different? No. For someone who de-
nies (i) speaks equally asymmetrically of the height of the east tower
and the height of the west tower, but generates no false conversational
implicature that they are different. The negation of (i) is wholly ac-
ceptable. What is wrong with (i) is that it semantically implies a differ-
ence in height between the two towers, not that there is a difference in
what it semantically implies about their height.
This kind of objection to many-valued logic is neither new nor
esoteric. Many-valued logicians have never answered it in any plausi-
ble way. It is therefore disappointing to find that, without even men-
tioning such criticisms, some philosophers continue to invoke many-
valued logic as though it provided an appropriate framework for an
account of vague identity. Yet we can easily check that the many-
valued mishandling of (i) extends to examples involving identity.
Suppose that I point to the west tower and say:
(ii) That tower = the west tower.
Timothy Williamson, Vagueness, Identity, and Leibniz's Law 279
present earlier and the ship present later respectively, without pre-
judging their identity or distinctness. We suppose that it is vague
whether SI and S2 are the same ship. Now make Hobbes' addition to
the story: someone collects the old timbers and, once he has them all,
puts them together in the original arrangement. Call that ship 'S3'. We
may suppose that it is equally vague whether S I and S3 are the same
ship. But it is quite clear that S2 and S3 are distinct ships; they are at
different places at the same time. Hence this holds:
(iv) S2 is not the same ship as S3.
One might suppose that (iv) would suffice to falsify this conjunction:
For under the same interpretation (A4) has a neutral premise and a
false conclusion. But the example provides no intuitive grounds for
denying the validity of (A4). Identity is a one-one relation.
Other cases make the same points about (A3) and (A4). Many
friends of vague identity hold that some ordinary objects have vague
boundaries, while other objects have precise ones. For example, it
might be vague whether the mountain Ml coincides in its boundaries
with the precisely bounded object M2 and equally vague whether Ml
coincides in its boundaries with the slightly different precisely
bounded object M3. While the identity claim 'M2 = M3' is clearly
false, some friends of vague identity regard both 'Ml = M2' and 'Ml
= M3' as neutral. The case has the same logical structure as that of the
ships.
These problems give us good reason to reject any account of vague
identity that uses many-valued logic on those grounds alone, independ-
ently of how it treats the Evans argument.
2.
than definite falsity (0) when the denotations of a and b are straight-
forwardly distinct (Priest [1998], p.333). These cases are especially
blatant because the authors have formulated their model theories with
care. Other attempts to defend the coherence of vague identity make
their unfaithfulness to the intended reading of = less blatant by de-
scribing the models more sketchily.
Such models may come close to the intended meaning of =, but
that is not good enough. A single model in which a single formula is
assigned a truth-value inconsistent with its intended reading is enough
to misclassify as invalid a principle that is valid on its intended read-
ing. Logic is an exact science; the model theory serves its purpose only
if it is exactly faithful to the intended reading of =. Otherwise, it con-
stitutes no objection at all to logical arguments against vague identity.
Of course, that is not yet to say that such arguments are valid.
The unfaithfulness to intended readings is not an accidental feature
of these non-classical models. Rather, it is essential to the robustness
of the results. The theorists want to prove that there are non-classical
models of vague identity. They fail to prove that conclusion if one can
prove, using classical assumptions in the metalanguage, that their non-
classical models are not models of vague identity - for they have not
shown that those classical assumptions are false. Given classical as-
sumptions in the metalanguage, a model is faithful to the intended
readings of the symbols only if it is in effect a classical model, so a
faithful model of vague identity is a classical model of vague identity.
But the theorists constructed non-classical models of vague identity
just because they conceded that it could not have classical models.
More precisely, on classical metalogical assumptions, either the deno-
tations of a and b in a model M are identical or they are not. If they are
identical and M is faithful to the intended reading of = then a = b is
true in M; if the denotations are not identical and M is faithful then a =
b is false in M. Either way, a = b is bivalent in M. To avoid this result,
the non-classical models treat = unfaithfully.
The preceding argument assumes that the singular terms a and b
have unique denotations in the model. Of course we may expect
something more complicated if the model treats a and b as vague. But
vagueness in the singular terms is not our topic, for it does not consti-
tute vagueness in identity. The question is whether it is vague of some
284 Individuals, Essence, and Identity
3.
Brutus"* Caesar.
Surely (A5) is valid. But imagine a sceptic who raises doubts along
these lines:
Is there really a property P of having stabbed Caesar such that 'Brutus
stabbed Caesar' is true if and only if Brutus has P and 'Caesar did not stab
Caesar' is true if and only if Caesar does not have P? It would be a rela-
tional property; can we be sure that there really are relational properties?
Furthermore, the notion of stabbing is not used in fundamental natural sci-
ence; do non-fundamental notions really stand for properties?
Of course, some people claim that vagueness shows that not all
trivial theorems of classical mathematics are true. That claim is usually
premised on many-valued logic, which has already been rejected. Of
course, it might have some other premise. Nevertheless, the aim in the
remainder of this paper will be to explore issues of vague identity un-
der the assumption that classical reasoning may be used in the meta-
language.
Let us apply (CLL) to the claim that it is sometimes neither true
nor false to say of given objects that they are identical. Suppose that
the assignment A assigns the object 0 to the variable x and the object
0* to the variable y. If it is neither true nor false to say of 0 and 0* that
they are identical, then the formula x = y is neither true nor false rela-
tive to A, when = means identity. Let A * be the assignment that differs
from A at most in assigning 0 to y; since A * assigns 0 to x too, x = y is
true relative to A *. Since x = y is by hypothesis not true relative to A,
by (CLL) 0 and 0* are not identical. But then if= really means identity,
x = y should be false relative to A, contrary to hypothesis. Thus the
supposition that it is neither true nor false to say of 0 and 0* that they
are identical is incoherent.
Someone might object that in a context in which we are contem-
plating truth-value gaps for identity statements with uniquely denoting
terms, 'not true' in (CLL) must be replaced by 'false':
(CLL-) Let an assignment A assign an object 0 to a variable v, an
assignment A * assign an object 0* to v, and A * be exactly
like A in every other way. Suppose that a sentence ex is
true relative to A and false relative to A *. Then 0 and 0*
are not identical.
But this modification gets things back to front. Entertain for a moment
the dialetheist idea that a sentence can be both true and false relative to
the very same assignment A#. Let A# assign the object 0# to a variable
v. When we take both A and A* to be A#, and both 0 and 0* to be 0#,
(CLL-) delivers the absurd result that 0# is distinct from itself. If one
accepts (CLL-), one does so on the basis of the anti-dialetheist as-
sumption that if ex is false relative to A * then it is not also true relative
to A *. That is, the reason for accepting (CLL-) is that it follows by the
mutual exclusiveness of truth and falsity from (CLL), which is a spe-
Timothy Williamson, Vagueness, identity, and Leibniz 's Law 291
inadequate, for it would make the formula ,~ex. 1\ ,~,ex. true when-
ever the truth-value of the proposition expressed by ex. could not be
known; but our current interest is not in cases in which the source of
the unknowability has nothing to do with vagueness (undecidable
propositions of mathematics may provide examples; see Williamson
[1995]). The intended epistemicist reading of ~ prescinds from such
irrelevant obstacles to knowledge. The relevant obstacle is, very
roughly, that a vague sentence might express a true proposition on its
actual interpretation I while expressing a false proposition on some
close counterfactual interpretation 1*, where speakers' insensitivity to
the difference between I and 1* makes any belief they express by ut-
tering the sentence too unreliable to constitute knowledge (Williamson
[1994], pp.230-4 has more details). In such circumstances we may de-
scribe 1* as indiscriminable from I. The natural proposal is therefore
that ~ex. is true on an interpretation I if ex. is true on every interpretation
1* indiscriminable from l On such a semantics, what kind of context
does ~ create? We can answer the question by examining the simplest
interesting case.
Consider a propositional language with just two singular terms, the
constants a and b, in which the only atomic sentences are a = a, a = b,
b = a and b = b. There are just three primitive sentence operators, ~, '
and 1\. Other operators are introduced as metalinguistic abbreviations.
A model is an ordered pair of a class of interpretations and a reflexive
symmetric relation of indiscriminability defined over them. A sentence
is valid in a model if and only if it is true on every interpretation in the
model. For most purposes reference to the model can be left tacit. An
interpretation I assigns a denotation to each constant. The sentence 11 =
12 is true on I if and only if I assigns the same denotation to the con-
stants 11 and {z. As usual, ,ex. is true on I if and only if ex. is not true on
I, and ex. 1\ ~ is true on I if and only if ex. is true on I and ~ is true on l
Finally, ~ex. is true on I if and only if ex. is true on every interpretation
indiscriminable from l
On this semantics, = is treated as a logical constant; it is always
interpreted as identity, just as , and 1\ are always interpreted as nega-
tion and conjunction respectively. The logical constants are treated as
precise. In the case of identity, this assumption has been challenged
(Hirsch [1999]). However, identity is a logically salient relation, the
Timothy Williamson, Vagueness, Identity, and Leibniz's Law 297
Dorothy Edgington
1.
I refer to an object, say a ship, at a time, fl. Call the ship'S 1'. I can see
it, touch it, go for a sail in it. I know what I am talking and thinking
about. The same kind of thing happens at a later time, fl, when I refer
to a ship - call it 'S2'. But, because of transformations which have
taken place, it is, it seems, indeterminate whether or not S 1 is the same
ship as S2.
There are also synchronic cases of apparently indeterminate iden-
tity. I gesture towards a mountain and say 'That mountain is high'.
Some distance away, you gesture towards a mountain and say the same
of it. It can be unclear whether we speak of the same mountain. For we
are at opposite sides of a landmass with a dip in it of such a size and
shape that it is not clear whether the landmass consists of two moun-
tains, or one mountain with a dip in it.
A challenging argument by Gareth Evans [1978] seems to show
that there cannot be objects such that it is indeterminate whether or not
they are identical. Suppose it is indeterminate that SI = S2. It is not
indeterminate that S2 = S2. But now S 1 has a property which S2 lacks
- that of its being indeterminate that it is S2. By Leibniz's Law, it fol-
lows that SI is not S2. If they were identical, there could be no prop-
erty which S 1 has and S2 does not have.
305
A. Bottani et al. (eds.), IndiviLlllllls, Essence and Identity, 305-318.
© 2002 Kluwer Academic Publishers.
306 Individuals, Essence and Identity
What is the best account of cases like the above? How exactly
should Evans's (very brief) argument be formulated and is it
irresistible? These are the themes of Timothy Williamson's paper.
There are three kinds of approach to the problem, which I shall
call the semantic, the epistemic, and the realist views. According to the
semantic view, there are no objects such that it is indeterminate
whether they are identical. All material objects have perfectly exact
spatio-temporal bounds. It is our ways of referring to them - our words
and concepts - which are vague. We simply have not decided to which
amongst various overlapping candidate objects 'Sl' or 'that mountain'
refers. There are indeterminate identity statements in this sense: on
some permissible sharpenings of our terms'S 1 = S2' is true; on other
permissible sharpenings 'S 1 = S2' is false. This construal is immune
from Evans's challenge. No objects are being said to stand in the rela-
tion of indeterminate identity. What is indeterminate is to which ob-
jects our singular terms refer. (This point is made by David Lewis
[1988])
The semantic view is set aside at the outset of Williamson's paper.
It is more natural, he says, to hold that the vagueness concerns the ob-
jects themselves, and not merely their names.
When the name 'Sl' was introduced, it was clear which ship was being
pointed at; likewise when the name 'S2' was introduced. Thus there are
objects (the ships Sl and S2) and a relation (identity) such that it is vague
whether those objects stand in that relation.
And:
similar identity puzzles arise for natural objects, such as planets, many of
which [ ... ] would have had the same existence and nature even if beings
capable of forming representations had never evolved (p.274).
The aim of his paper is to find a way to make sense of the phenomenon
of vague identity in terms of a relation between objects.
The epistemic view (which Williamson favours), like the semantic
view, holds that all objects have determinate boundaries. There is a
determinate fact of the matter about when or where S 1, or that moun-
tain, ends. On the epistemic view, unlike the semantic view, I do refer
to one determinate object by 'Sl' or 'that mountain'. But in the cases
Dorothy Edgington, Williamson on Vagueness. Identity. Leibniz's Law 307
at issue, I cannot know what their exact bounds are (and this ignorance
has a special kind of explanation). The quick response to Evans's ar-
gument, from an epistemic perspective, would be that epistemic con-
texts are opaque. From 'It is not known whether a = b' and 'It is
known that b = b' it surely does not follow that a and b are not identi-
cal. However, orthodoxy has it that where there is opacity, the sen-
tences cannot be construed as expressing a relation between the ob-
jects, a and b. Williamson goes on to show how to define genuine re-
lations between the objects to which our terms refer, '01 and 02 are not
presentable as either clearly identical or clearly distinct', etc.. I return
to the details in §4 below.
The realist view, unlike the other two, holds that objects really do
have indeterminate, or fuzzy, boundaries. In the typical case, a material
object - an animal or plant, river or mountain, house or chair - comes
into existence, and goes out of existence, more or less gradually. There
is no precise moment at which it begins or ends. In between, its spatial
boundaries are not completely determinate. In most cases the regions
of indeterminacy are small and untroublesome. In most cases, this in-
determinacy does not give rise to indeterminacy of identity. That an
object fades a little fuzzily into its surroundings does not entail that it
is indeterminate which object it is. Call these 'ordinary fuzzy objects'.
Neither the semantic nor the epistemic view can allow that there
are such objects. For even if they do not give rise to vague identities,
they do give rise to vague statements - about their spatial or temporal
extent, and other things consequent upon that. For instance, 'Ben Ne-
vis has a surface area of at least 20 square kilometres' may be neither
clearly true nor clearly false, as may 'There are spruce trees on Ben
Nevis' if the only spruce trees are on its fringe. The semantic account
of this vagueness is that we have left undecided to which precisely
bounded object 'Ben Nevis' refers. On the epistemic view, these vague
statements do have determinate truth values, but we cannot know
which: 'Ben Nevis' does refer to an object with perfectly exact
boundaries, but we do not know what these exact boundaries are. Nei-
ther of these is as natural as the realist view. This provides some mo-
tive for trying to find an acceptable realist account.
It is what I shall call peculiar fuzzy objects which are threatened
by Evans's argument: cases in which the fuzziness of objects' bounda-
308 Individuals, Essence and Identity
ries make it unclear whether, in a given stretch of space and time, there
is one object or two. Consider our mountain(s). Abbreviate 'that
mountain' said from the right, to 'a', and 'that mountain' said from the
left, to 'b'. The only trees are on the right. 'Determinately, a has trees
on it' is true. 'Determinately, b has trees on it' is false. For it is inde-
terminate whether b stretches as far as the trees. Construing the state-
ments realistically, something is true of a which is not true of b. So, by
Evans's argument, a and b are distinct mountains.
Suppose our original ship, S 1, is being gradually transformed into
another ship, S3, in such a way that throughout the process, there is a
ship there. (Someone offered to buy the original ship, which is of his-
toric importance. The sale was agreed on the conditions that it be re-
placed by a replica, and the replacement be continuous and gradual).
S2 is the ship I refer to around the middle of the process. Is it still S l,
or has it become S3? The realist wants to say: it is indeterminate
whether S2 is the old ship or the new. By Evans's argument, it follows
that it is not the old ship, and not the new ship. So it must be a third
ship. Sl, S2 and S3 are all distinct. Very well, there is a third ship.
When did it begin? Is the ship we have about a quarter of the way
through still Sl, or has it become S2? If we try to say 'it's indetermi-
nate which it is', from Evans's argument it follows that it is neither. So
there is a fourth ship. And so on. If realism about peculiar fuzzy ob-
jects is defensible, fault must be found with Evans's argument.
2.
is not how these model theorists see matters: for them, there are ob-
jects in the model such that it is indeterminate whether they are identi-
cal. But I think Williamson is right that these models can have no sua-
sive force for those who are not already, independently, convinced that
objects can stand in this relation.
In § 1 Williamson criticises attempts to deal with vague identity in
terms of a many-valued logic, whether three-valued or continuum-
many-valued, which operates under the assumption of generalized
truth-functionality: the value to be assigned to (e.g.) a conjunction is a
function of the values of the conjuncts. Call clear truth 1 and clear
falsehood o. On the continuum-many-valued approach a proposition
gets a value the closer to I the closer it is to clearly true, 0.5 when it is
evenly balanced between the two poles. On the three-valued approach,
if we call the third value - neither clearly true nor clearly false - 0.5,
Williamson's criticisms can be stated in a uniform way. On both ap-
proaches, the value to be assigned to the negation of a proposition A is
one minus the value to be assigned to A; and the value to be assigned
to (A /\ B) is the minimum of the values to be assigned to A and to B.
The negation rule causes no problems, but the conjunction rule does.
Suppose Cicero is a borderline case of tall-let v(Cicero is tall) = 0.5.
As Tully is no more or less tall than Cicero, v(Tully is tall) = 0.5. So
v(Tully is not tall) = 0.5; and v(Cicero is tall and Tully is not tall) =
0.5; and v(It's not the case that (Cicero is tall and Tully is not tall» =
0.5. But 'Cicero is Tully' is clearly true. The inference 'Cicero is
Tully; so it is not the case that (Cicero is tall and Tully is not tall)' has
a clearly true premise and a conclusion with value 0.5. No valid argu-
ment should take us from a clearly true premise to a conclusion which
is not clearly true: the argument, on this account, is invalid. Yet the
argument is a trivial variant of Leibniz's Law; and it is intuitively
compelling. Similarly, 'Cicero is tall and Tully is not tall; therefore
Cicero is not Tully' has a premise with value 0.5 and a clearly false
conclusion. Surely, a one-premise valid argument with a clearly false
conclusion must have a clearly false premise. Again, the argument ap-
pears to be invalid on this account; again the argument is intuitively
compelling. Our elementary ways of reasoning with identities breaks
down.
310 Individuals, Essence and Identity
extends as far as the trees. So v(-,(There are trees on b)) = 0.5. It fol-
lows that v(a =F- b) is at least 0.5, hence v(a = b) is at most 0.5.
This framework is non-revisionary about identity. Identity is re-
flexive, transitive and symmetric, and Leibniz's Law, and its contrapo-
sitive, are valid. We still have to face up to Evans's argument. But
there is a many-valued theory of vagueness which is immune to the
objections raised against the generalized-truth-functional variety.
3.
clude that they are distinct emeralds, whether or not the predicate
'grue' picks out 'real' property. Williamson's own proposal is equiva-
lent to the metaphysical principle on a wide reading of property, such
that every predicate does pick out a property. He construes Leibniz's
Law as a metalinguistic principle, about objects satisfying predicates or
open sentences; equivalently, predicates or open sentences being true
of objects; equivalently, assignments of objects to the variables in open
sentences yielding closed sentences, true, or not, as the case may be.
Thus his (CLL):
Let an assignment A assign an obj ect 0 to a variable v, an assignment A *
assign an object 0* to v, and A * be exactly like A in every other way.
Suppose that a sentence a is true relative to A and not true relative to A *.
Then 0 and 0* are not identical. (p.289)
In terms of (CLL), Evans's argument would run like this: consider the
open sentence 'Determinately x = y'. It is true for an assignment of
objects <0, 0> to its variables. It is not true for an assignment <0, 0*>
to its variables. So 0 "::j:. 0*.
As Williamson says, (CLL) is a consequence of an elementary
theorem of the mathematical theory of functions. I am willing to accept
that it provides the most compelling version of Evans's argument.
Nevertheless, it seems to me arguable that it is irresistible only given
the assumption that the objects in question are all either clearly identi-
cal or clearly distinct. These are, after all, the objects with which
mathematics deals (and with which we all deal, almost all of the time).
You cannot determinately count peculiar fuzzy objects!
Consider the open sentence 'There are trees on x', as it is treated in
the framework I sketched in the last section. The assignment of a to x
yields a sentence with value 1. The assignment of b to x yields a sen-
tence with value 0.5 ~ a sentence which does not get the value 1. Ap-
plying the thinking behind (CLL), it would follow that a and b are dis-
tinct. But all that does follow, in that framework, is that the value to be
assigned to a = b is at most 0.5. It does follow that a and b are not
clearly identical. It does not follow that they are distinct.
So we have a choice. Accept (CLL) and discard the framework as
defective. Or accept the framework; claim that (CLL) is valid when all
the objects involved are either clearly identical or clearly distinct, and
314 Individuals, Essence and Identity
I have tried to show that we need not lose our grip on the notion of
identity. We can preserve all the classical laws governing identity
Dorothy Edgington, Williamson on Vagueness, Identity, Leibniz's Law 315
where peculiar fuzzy objects are concerned, except for one small
amendment: within the scope of the 'determinately' operator, from
/).(cfJ(a» and -,/).(cfJ(b» it follows only that -,/).(a = b). If it is determinate
that x and yare the very same thing, then it is determinate that what-
ever applies to x applies to y. If something determinately applies to x
but does not determinately apply to y, it is not determinate that what-
ever applies to x applies to y. So it is not determinate that x and yare
the very same thing.
4.
Graeme Forbes
319
A. Bottani et al. (eds.), Individuals, Essence and Identity, 319-340.
© 2002 Kluwer Academic Publishers.
320 Individuals, Essence, and Identity
(1) Ifu =J. v, and wt and wt are each distinct from u and v, then ifat
wt, a satisfies x@u, and at wt, b satisfies x@v, then a = b.
3. ROBERTSON'S CRITIQUE
5. EXTRA-STRENGTH HAECCEITISM
This criterion makes being the first m-c propagule extrinsic to any p
that has it in w, using (3b): in w, if p is not the first m-c propagule, we
can contrive an x-t contraction of w in which p is the first m-c propa-
gule simply by deleting p's m-c predecessors from w. (3) also makes
being the Jcth m-c propagule, k ~ 2, extrinsic to any p that has it in w,
using part (a) and the same x-t contraction ofw. 21
However, matters are less clear cut if we focus on m, which per-
sists, if in distributed form, through various episodes of constituting
propagules. (3) makes it intrinsic to p to be initially constituted of m,
and if it is also intrinsic to m not to have c-constituted any propagule
before p, we have good reason to count it intrinsic to p to be the first
m-c propagule. 22 Might it be intrinsic to m not to have c-constituted any
propagule before p? The negation of this property is unstable under x-t
contraction so long as we are allowed to contract away the life of m
prior to t «3) does not allow this). But there is some plausibility in the
thought that if at t it was intrinsic to x to be F then at a later t' it should
330 Individuals. Essence. and Identity
the whole point of appealing to (3) or (4) was to settle a lemma used en
route to inferring that it is essential.
The general difficulty with approaches like (3) and (4), that essen-
tial properties are counted as intrinsic just because they are essential,
motivates the very different approach of Francescotti [1999]. He pro-
poses that a property is extrinsic if it is based on a relation, except, in
effect, if the relation is one of Humberstone's interior ones. So the
problem is to filter these relations out of the group that will give rise to
extrinsic properties. To do this, Francescotti introduces the idea of a d-
relational property of x, which is a property of x consisting in a rela-
tion that is borne to a distinct thing y. P is an impure d-relational prop-
erty of x if there are Rand y, y I- x, such that having P consists in being
in R to y; P is a pure d-relational property if there is a relation R, a sec-
ond-order property X and a quantifier Q such that having P consists in
bearing R to Q of the X's, and furthermore, possibly one of the X's is
not identical to X.24
Being the first m-c propagule would be extrinsic on this account if
it consists in being the earliest of all the m-c propagules, and being a
subsequent m-c propagule would be extrinsic if it consists in being the
later of some two of the m-c propagules, or in originating after the first.
The question is whether there is anything that will be counted as in-
trinsic on this approach. Why not say that squareness is extrinsic, since
to be square consists in being one of the square things? Simply ex-
cluding identity and non-identity from the possible values of R will not
help, since in each case there will be other relations and/or properties
that will do the job, for example, in the present case, being same-
shaped with?5 Nor is it sufficient to exclude being R to one of the X's
from consideration in evaluating X for intrinsicness (France scotti
[1999], p.603), since even if Y I- X, a thing may be R to one of the X's
iff it is R to one of the Y'S.26 It seems that a much greater burden must
fall on the notion of consisting in, allowing us to decide between the
following two accounts of being the first m-c propagule: that it con-
sists in being the earliest of all the m-c propagules, or rather that it
consists in being the first propagule that matter m constitutes in con-
figuration c. The latter, as we noted above, muddies the picture in view
of the intrinsic properties of m. 27
332 Individuals, Essence, and Identity
For those who think the number of m-c predecessors which a propa-
gule has is extrinsic to it, regardless of whether there is an otherwise
successful account of intrinsic/extrinsic which clearly says so, the ap-
propriate question is whether the number of such predecessors might
be relevant to the identity of such a propagule. We do not need to give
a fonnal definition of relevant to the identity of x; we only need to
make it plausible that certain features count, so that any definition in-
curs an obligation either to so count them, or to explain why the ap-
pearance of plausibility is an illusion. Nor does this move weaken the
case for origin essentialism: principle (1) can play the role it does in
defending EBO just as well if we read the notation x@u to include all
the identity-relevant features x possesses at u, as well as its intrinsic
features. So: are there other cases where number of predecessors, in
some relevant sense, plays an important or central role in detennining
identity?28
One example is the natural numbers. If we have a conception of
them according to which they have no internal structure, then what
distinguishes one natural number from another is its position in the
standard ordering. Position in the ordering consists just in number of
predecessors. It might be objected that this case is of no relevance to
ours, since the composite entities with which we are concerned differ
from numbers precisely in that they do have internal structure, their
intrinsic nature. But of course, the special feature of our crossworld
identity puzzles is that intrinsic nature (along with everything else ex-
cept identity) is factored out by being held constant across the entities
whose identity is to be settled. The only plausible candidate for an es-
Graeme Forbes, Origins and Identities 333
only when the item a in WI satisfying x@u and the item b in W2 satis-
fying x@v originate close enough in time. Close enough is of course
vague, and there are fundamentally two different ways one might ac-
commodate the vagueness, either through counterpart theory or a
vague accessibility relation. However, in the present context, what is
important is that qualifying (1) in this way concedes nothing to extra-
strength haecceitism. We still have a substantial and interesting suffi-
ciency condition for crossworld identity which, if the reasoning of sec-
tion 2 is cogent, turns out to support substantial and interesting neces-
sary conditions for crossworld identity.
would go for a certain tree, since each such world unfolds past t into
streams that are the same vis a vis the branch-growth process for the
tree in question.
Could defenders of (EBR) reply that this argument also betrays
bias, since it is only future contingents that are indeterminate, and it is
precisely their view that whether or not a tree grows b is not a contin-
gent matter? Perhaps it is now determinate that b will grow on the tree
in the same way that it is now determinate that all future samples of
water will have chemical composition H20.
Suppose for definiteness that there are exactly two possible trees
which have the same origin as a certain actual tree and are of the same
biological kind, but one of these trees, A, grows b, and the other, E,
does not. Might it be determinate at t and earlier times that the actual
tree is A rather than E, even though nothing has occurred by t that re-
quires or rules out b's growing (nothing that does not prejudge the
tree's identity)? The answer to this question has to be no, if we think
the future is open in any respect. For if we ask how it could be deter-
minate at t that the tree is A, before any branches grow on it, the only
non-question-begging answer is that it is because the tree grows b at
some time after t. But if the present can acquire determinacy in this
respect from the future, why not in every other respect? (Note that such
backwards acquisition of determinacy is not suggested in the chemical
composition case - we do not think it is now determinate that all future
water will have chemical composition H20 just because in the future,
all water has chemical composition H20.)
A broad range of deviant essentialisms seem to be in tension with
the openness of the future as we have understood it. The only essen-
tialist principles about composite objects that sit well with it and that
are consistent with (1) are principles that focus exclusively on the ini-
tial states of those objects and their ancestry. For any allusion at all to
subsequent states will generate the puzzles about determinacy of iden-
tity that we have just described. 31
Notes
I Assume that x is restricted to organisms which originate from a single entity.
Graeme Forbes, Origins and Identities 337
2 See, for example, Mackie [1974], McGinn [1976] and Salmon [1982], Kripke
himself gave "something like a proof' of a related thesis about the matter of which a
table is composed, in endnote 56 of Kripke [1972], where it is printed in
"inexplicably garbled" form (Kripke [1980], p,l), An erratic reader of endnotes, I
was unaware of it until it appeared, corrected, in Salmon [1979], By then I had al-
ready devised a similar argument about acorns and oak trees, which appeared in print
in Forbes [1980a,b],
3 For a defense of such primitive identities, see Chisholm [1970],
a and b principle (Wiggins [1980], p, 96): "[ ... ] if identity is what we want to eluci-
date, [we need] a criterion which will stipulate that for a relation R to be constitutive
of the identity of a and b, a's having R to b must be such that objects distinct from a
or b are irrelevant to whether a has R to b." So R could not include an "absence of a
better candidate" provision. Wiggins and I would both disagree with Mr. Justice Ot-
ton of the Scottish High Court, in the celebrated case of Middlebridge Scimitar Ltd
versus Edward Hubbard. "Mr. Hubbard [ ... ] was granted a court order enforcing an
agreement under which Middlebridge [ ... ] agreed to buy [the Bentley racing car Old
No.1 from him] [ ... ] for £6.8 million [ ... ]. The case centred on whether Mr. Hub-
bard's car was the one which sped the diamond heir Capt Wolf Barnato to victory at
Le Mans in 1929 and 1930 or whether it had undergone so much rebuilding it was no
longer the genuine article. Middlebridge [ ... ] said it had been promised the Le Mans
winner - and the [Hubbard] Bentley was not that car because it had been completely
rebuilt by a master mechanic" (The Scotsman, 28 July 1990). The crucial considera-
tion in his finding against Middlebridge, according to Otton, was that "there is no
other Bentley, extinct or extant, which could legitimately lay claim to the title of Old
No 1 [ ... ]" And they say analytic metaphysics has no practical application.
S Humberstone refers to intrinsic properties in this wider sense as interior. See
Humberstone [1996], p.239-40 for discussion of this sense, attributed to Dunn
[1990], and the whole paper more generally for an instructive discussion of the in-
trinsic/extrinsic distinction.
6 Fine [1994] identifies the asymmetry in this case with essential/accidental, and
rejects the modal account of essential property on the grounds that it does not dis-
criminate DAz.zE {x}(x) and DAz.XEZ({X}). But if the fundamental asymmetry is
intrinsic/extrinsic, we need a further argument that all essential properties must be
intrinsic before this difference can overturn the modal definition.
7Yablo ([1999], p.486) says that "on almost anybody's account," the zygote Z
from which he (YabJo) developed stopped existing before he started, so descending
from Z is extrinsic to him. Perhaps it is well to separate persons and their bodies, in
which case we can still say that it is intrinsic to Yablo's body to develop from z, since
Yablo's body came into existence with z, even ifit took a while for Yablo to occupy
it. Yablo goes on to say that since part of what it takes to be Yablo is to descend from
z, being Yablo is extrinsic to him as well. This is a sense of extrinsic on which I have
no secure grip (though Schoenberg once said that, since no-one else wanted the job
of being Schoenberg, he had to take it on).
338 Individuals, Essence, and Identity
an organism originating from a propagule p must causally interact with one originat-
ing from a different propagule q. It is a further problem to give a precise account of
causal isolation that is of use in harder cases. Also, something with a certain origin
cannot be made the best candidate for identity with a certain entity simply by throw-
ing in some causal interaction with its rivals: the causal interaction would have to be
somehow in the nature of the case.
10 Robertson's actual example is essentiality of leaf-color. However, if we
choose a non-exclusive property, insisting on its essentiality as an alternative to
(EBO) will not block every counterexample to (1). For example, we can suppose that
the colors of a, x and y in the argument for (EBO) are all the same; we still get a
counterexample to (1). (I assume that growing b is exclusive.)
II Monovular twinhood is not a counterexample to this assumption. Pace Rob-
ertson ([1998], p.735, n.ll) I would say that the propagules from which identical
twins originate are the two daughter cells resulting from the non-standard mitotic
division of the zygote.
12 Despite the obvious echoes of the Ship of Theseus, I think that organisms
contrast with artifacts in important relevant ways. An organism can persist through a
complete change of its matter. But while a ship may undergo repairs at certain times,
so that ultimately there is a ship whose matter is entirely different from the original
ship's matter, I have never seen a good reason to hold that a single ship persists
through such a process, Justice Otton notwithstanding. Fear of vagueness is often the
main motivation; see the discussion of the Mac of Forbes in Forbes [1987].
13 However, it is not so clear that an appropriate causal isolation condition is
met in this case (cf. note 9). So some might try to defend this kind of extrinsic deter-
mination, as is familiar from the Ship-of-Theseus literature; see, e.g., Garrett [1988],
and the response in Mackie [1989].
14 For example, it seems possible that I might have been an identical twin. But
reflecting on the symmetry of mitotic division, the hypothesis that there is a world
where I am one and not the other of a pair of twins seems no better than the hypothe-
sis that 0 is identical to Lefty or else to Righty. See (Forbes [1980a], pp.353-5) for
further discussion of twinning.
15 Is it also crucial that the zygote the Queen could not originate from is some-
one else's, the Trumans' child's? Even if it were a merely possible zygote, I doubt
Graeme Forbes, Origins and Identities 339
that that weakens the pull of Kripke's claims. Hawthorne and Gendler [2000] offer an
origin essentialism "lite" (their (21)) which says that there is no world where the ac-
tual Queen comes from the actual Trumans' daughter's actual zygote and the actual
Trumans' daughter comes from the actual Queen's actual zygote. But this is very
much weaker than the intuition which Kripke's discussion promotes, at least in me.
16 Mackie [1987] endorses extra-strength haecceitism, though without the bene-
fit of supporting examples like McKay's. She seems to agree with my verdict about
the Lefty/Righty case, but argues that there is no reason to insist on parallel treat-
ments of transtemporal and transworld identity (Ibid pp.l97-8). But I would say that
identity is identity. If the thing which is F is identical to the thing which was G
(wide-scope tense) requires grounds, then the thing which is F is identical to the
thing which would have been G if. .. (wide-scope modal) must also require grounds.
17 Exactly one is too strong, since one bare difference can give rise to others, if
the primitively distinct entities are parts of other entities. I ignore this complication.
18 Hawthorne and Gendler [2000] raise the interesting and complicated question
of what happens to the defense of (EBO) in a counterpart-theoretic framework, where
it seems that a two-candidate world would just be a world with two counterparts of
some actual entity, which is relatively unproblematic. But the new argument for
(EBO) just given does not use two-candidate worlds, and in a counterpart-theoretic
framework, shows that the counterpart relation would have to hold in some instances
and fail in others even though there is no difference between these instances with
respect to the factors that ground or determine (degree of) counterparthood. This is
no improvement on ungrounded identity. I hope to pursue these issues, including the
Faith-Hope-CharitylPeter-Paul-Mary case from (Hawthorne and Gendler [2000],
p.293), in another paper.
19 See further Forbes [1994a]. Another proposal is that in certain cases there is
no fact of the matter about transworld identity. But this position does not seem to
change the issues in any significant way (though it does complicate the possible-
worlds semantics).
20 Hawthorne and Gendler ([2000], p.293) argue that "[ ... ] the intuitive strength
of the necessity of origins thesis surpasses that of [predecessor essentialism], so if the
project is to generate arguments in favor of the former, it seems best not to invoke the
latter". This might be so if we were trying to explain why (EBO) is intuitive and
thought that a successful non-debunking explanation would have to access explicit
reasons for holding (EBO) and portray (EBO) as inheriting its intuitiveness from
those reasons. But in general, explaining why something plausible is true may require
us to call upon non-obvious lemmas.
21 I interpret not the first to mean the second or later, excluding not at all. Cer-
tainly, if p is not an m-c propagule, contraction will not tum it into one.
22 Here I am assuming that at least for a range of intrinsic properties, if the con-
stituting matter of p has them, so does p (the primary exceptions would be properties
involving p itself).
23 Vallentyne bites the bullet on this issue ([1997], pp.216-7). Yablo modifies
(4) to get round the problem. According to (4) it is the truth-value of P-xw ~ Pxw'
that is criterial; in the revised version (Yablo [1999], p.492) it is the truth-value of
340 Individuals, Essence, and [dentity
Pxw ~ Px'w' that is criterial, where x' is whatever is constituted in w' by the basic
elements of w that make up x (in w). But this makes the criterion harder to apply. For
example, the new notion of part (p.491) allows w to be a part ofw' so long as there is
some concept of sum such that w' is the sum of certain basic elements and w is the
sum of a subset of those elements. It is not obvious that this will keep the shape of x
the same in wand w' unless we make an ad hoc stipulation that only those notions of
sum that do not allow basic elements to arrange themselves into a different shape and
still be the same sum are to be used. Since Yablo wants shape to be intrinsic (p. 480),
such stipulations are apparently needed. And constituting matter m will still come out
intrinsic to the entity e it constitutes, contra Yablo's intentions, unless the same basic
elements can configure themselves as they are in e without thereby forming m. We
will need a special notion of sum to justify this.
24 This is (c*) of (Francescotti [1999], p.604) except that I have used a second-
order variable X for a function from worlds to sets in place of the rigid class-term 'C'
in (c*), which renders the possibly pointless.
25 Francescotti's final version of his criterion says that P is intrinsic to x iff there
are non-d-relational properties such that x's having P consists in its having those
properties. This does not affect the overgeneration problem.
26 It will not help to restrict Y to those values such that necessarily, any Y is an X
and vice versa. Some would say this means Y = X anyway. And it takes us quite far
from the original intuition behind d-relationality.
27 There is a problem with Francescotti's account of consists in (p. 599), which
makes it symmetric, a view perhaps associated with Hegel: a nation consists in its
people, and vice versa.
28 Excluding infinite regresses, essentialism about number of predecessors de-
termines identity in a recycling sequence so long as we are not given two primitively
different starting entities.
29 An exception is the rather special case of spatio-temporal point of origin. My
most recent discussion of this is in (Forbes [1999], §2). I think this special case has to
be ruled out by independent considerations.
30 There are ways of disputing this, but these workarounds are costs of the view
under discussion.
31 This paper includes some parts of a lecture given at the 22nd International
Penelope Mackie
from the other, and any properties that distinguish p from q will be ac-
cidental, not essential. If so, any way that p could have been is a way
that q could have been, and this is a counterexample to the sufficiency
principle (1). Further, if the identities of zygotes are not appropriately
grounded, it is plausible to say that the identities of the organisms that
come from those zygotes cannot be appropriately grounded in the
identities of the zygotes. The possibility of recycling, far-fetched
though it is, thus poses a serious problem for Forbes. It seems futile to
suggest that (EBO) respects his grounding principle and his sufficiency
principle unless this problem can be solved.
To meet this difficulty, it seems that Forbes must deny that the zy-
gotes p and q are distinguished only by their accidental properties. An
obvious move is to reject the 'independence thesis' mentioned above,
and attribute to q, as an essential property, the property of being pre-
ceded by p.8 Since the problem can be generalised to include recycling
sequences of any length, a comprehensive solution along these lines
requires what Forbes calls "predecessor essentialism": that "an entity's
predecessors in a recycling sequence are its predecessors in every
world in which it exists" (p.328). Since any essential property of x is a
component of every sufficient condition for being x, predecessor es-
sentialism prevents recycling from generating "two-candidate" worlds.
Of course (if it is to do the work required of it), predecessor essential-
ism must be taken to imply not only that it is essential to the first zy-
gote in a recycling sequence that it have no predecessor, but also that
in the normal case, where there is just one zygote and no recycling, this
zygote's having no predecessor is one of its essential properties.
One difficulty concerning the appeal to predecessor essentialism is
its apparent conflict with the principle that identities and non-identities
must be grounded in intrinsic properties (§6). However, since Forbes
has already conceded that identities may be grounded in properties that
are not, strictly speaking, intrinsic, he takes it that the principal chal-
lenge he has to face is, not that the number of predecessors an item has
is not one of its intrinsic properties (for he concedes that it may not
be), but that it is not even "identity-relevant" (§7).
To illustrate this challenge, Forbes considers examples involving a
sequence of ships each of which is originally constructed of the same
particular matter m according to the same plan I. Call each such ship an
Penelope Mackie, Forbes on Origins and Identities 347
Suppose that in the actual world (WI) there is just one m-/ ship, SJ,
which comes into existence in the year 1900. Then, given a possible
world W2 in which there are two m-/ ships, one (S2) which comes into
existence in 1895, and a recycled m-/ ship (S3) which comes into exis-
tence in 1905, Forbes's principles imply that it is the first ship in W2
that is identical with SI. \3 If we suppose, in addition, a possible world
W3 in which there are two m-l ships, S4 and ss, one of which (S4) comes
into existence over a thousand years before 1900 (e.g., in 850 A.D.),
and a recycled m-l ship (ss) that comes into existence in 1905, we may
(according to Forbes) identify the second ship with the actual SJ, in
spite of the fact that it lacks the property that SI actually has of being
the first m-l ship.
We thus have two possible worlds, W2 and W3, each of which con-
tains an m-/ ship that originates in 1905. These ships (S3 and S5), in
spite of their similarities, are not the same ship, since one is identical
with SI and the other is not. But now we must ask: what, according to
Forbes, grounds the non-identity of S3 with ss? The only differentiating
feature to which Forbes can appeal is that S3 's predecessor in W2 was,
while S5'S predecessor in W3 was not, built at approximately the same
time as SI was built in WI. However, to treat this feature as what
grounds the non-identity is, on the face of it, a paradigm of the type of
"best-candidate" reasoning that Forbes must reject if his argument for
the necessity of origin based on the grounding principle is not to be
undermined. 14
The only possible defence to this objection is that this version of a
"best-candidate" theory of the identities of composite objects is supe-
rior to the rival versions that Forbes must reject. In the final section of
this paper I consider whether this defence can be sustained.
overall than does the 1905 ship in w*. With what right can Forbes in-
sist that, in spite of this, the 1895 ship in Case 2 is not a genuine
"competitor", and that, if either w* ship in Case 2 is identical with s"
it can only be the 1905 ship?
Perhaps Forbes might respond that in Case 1, the "competitors"
are causally connected with one another, because they share their
original matter m, whereas the putative "competitors" in Case 2 need
have no causal connection with one another at all. However, there are
two problems with this response. First, Forbes has provided no argu-
ment for the claim that an appeal to causally isolated competitors is
illegitimate in a way that that an appeal to causally connected com-
petitors is not (cf. §2 above). Secondly, even if it is accepted that the
distinction between causal isolation and causal connection is relevant
here 17, it is not true that the "competitors" in Case 2 must be causally
isolated from one another. For example, we may fill out the details of
Case 2 in such a way that the 1905 ship is deliberately constructed by
its designer to be a replica of the 1895 ship, with the result that the
ships are causally connected although they do not share their original
matter.
I have claimed that if he concedes that the identities of composite
objects need not be grounded in exclusive essential properties, Forbes
has, in effect, undermined his original argument for necessity of origin
principles such as (EBO). However, he appears to suggest that its place
may be taken by a "best-candidate" account of the grounding of the
identities of composite objects of which the necessity of origin is a
consequence. It remains to be seen whether a justification for the de-
tails of this "best-candidate" account can be provided that does not beg
the question in favour of the necessity of origin principles that Forbes
seeks to defend. I think that this represents a formidable challenge to
this new argument for the necessity of origin, although, of course, I
have not shown that the challenge cannot be met
Notes
I "Origins and Identities". All references are to this paper, unless otherwise indi-
cated.
Penelope Mackie, Forbes on Origins and Identities 351
2 See, in particular, Forbes (1980a] and [1985], Chapter 6. See also Forbes
[1986] and [1994a].
3 See, for example, Robertson [1998], p.741, Yablo [1988], and my [1987],
p.186. Cf. my [1998], pp.64-5.
4 From now on I shall usually drop the explicit qualification 'non-trivially', tak-
are relevant here, since Forbes's principle (EBO) is alleged to depend on a more gen-
eral principle concerning 'composite objects', of which ships provide an example.
10 The example is due to Hawthorne and Gendler [2000], p.293. One could ac-
cept predecessor essentialism without accepting the associated sufficiency claim: i. e.,
the claim that if an m-/ ship x has a certain number of predecessors, then being an m-/
ship with that number of predecessors is sufficient for identity with x. However, if
predecessor essentialism is to do the work that Forbes requires of it, the additional
sufficiency claim is required.
II As I shall use the expression 'position in a sequence', a thing's position in a
sequence is independent of how many successors it has. For example, a ship that is
the first and only m-/ ship has the same position in a sequence (in the relevant sense)
as a ship that is the first in a sequence of 100 recycled m-/ ships.
12 Some of Forbes's remarks may suggest that he has in mind a different modifi-
cation, according to which predecessor essentialism is retained, but a ship's approxi-
mate time of origin is also one of its essential properties. However, this would not
accommodate Forbes's agreement with Hawthorne and Gendler's claim that there can
be a case where the second m-l ship in a recycling sequence could have been the first
and only m-/ ship (the case of Z; Forbes, p.333).
13 This assumes that a five-year gap is sufficiently small to make the times of
origin "close enough" in the sense Forbes intends (§7, p.334). If not, just rerun the
example with a smaller temporal gap.
14 Cf. §2 above. The best-candidate reasoning is also, of course, congenial to the
IS Evidently, if this account works for ships, it may be adapted, mutatis mutan-
dis, to provide a basis for necessity of origin principles for other composite objects
that does not require that those objects have exclusive essential properties.
16 To talk of 'competition' for identity is to run the risk of apparent conflict with
the principle of the necessity of identity. However, I hope that my use of this lan-
guage can be interpreted in a way that avoids any genuine conflict with this principle.
17 Elsewhere (Mackie [1987], §7, and Mackie [1989], I have suggested that it
may be more plausible to claim that an identity can depend on the presence or ab-
sence of a causally connected individual than to claim that it can depend on the pres-
ence or absence ofa causally isolated individual. Cf. Forbes, note 9.
Part Four
Anthony Savile
apart, and hence together lack precisely that unbreakable unity which
alone would entitle a good Leibnizian to account the original compos-
ite living thing a true substantial unity. Setting aside the soul or the
dominant monad, the body itself is surely no more than a conventional
or phenomenal unity, just like the block of ice or the flock of sheep or
the wormy cheese. If this is so, it is bound to strike us that Leibniz's
determination to pick out certainly men and, he is inclined to think,
many other naturally occurring things as truly substantial composites
that are metaphysically quite distinct from aggregative phenomena like
the cheese is under threat. To put it starkly, how could there possibly
be such real things whose high grade, substantial, reality depends on
their possessing a body whose reality itself is no more than low grade
and phenomenal?
When Leibniz considers the example of the cadaver he is quite ex-
plicit that there we do have to do with a phenomenon and not a com-
plex substance. And given that, it may strike us as astute enough on his
part, but ultimately, surely, quite hopeless, to say that living things are
indeed composite substances and that they never truly die. For even if
that might serve to mark out a possible relation between dominant mo-
nads and bodies that is distinct from the relation between the self and
the body that becomes a corpse, it is all too likely that we shall say on
the basis of straightforward observation that such a relation is never
realised, and hence that neither cabbages, nor ants nor even men are
instances of it. The introduction of the everlasting body constantly
conjoined with the indissoluble simple soul like substance is likely to
seem nothing more than a kind of metaphysical wishful thinking. In-
tellectually speaking, it looks quite inapplicable to the actual world.
Such an accusation is mistaken, I think, and it does scant justice to
the subtlety of Leibniz's thought. To set things right we need only re-
flect a little on the idea of a monad's organic body. In our perceptual
engagement with the world we are sensitive to far off events through
the way in which mechanically they impress themselves or leave traces
on other bodies, which in turn transmit that information to yet other
bodies (Monadology §61). A chain of such transmitted impressions
terminates at the point at which a simple soullike monad is immedi-
ately sensitive to information that is mechanically inscribed on a par-
ticular body of matter (Monadology §63). The body of which the mo-
358 Individuals. Essence. and Identity
stantial whole would also cease to exist, indeed in the case of a rational
intelligence not only would the self have ceased to be, but with that
both the man and the man's body would also have come to grief. What
is left over, the cadaver, is as we know now distinct from the man's
body, so with the disappearance of the one we have the disappearance
of the three (the two elements and the whole).
What makes for the natural indestructibility of simples is precisely
their having no parts, and what makes for the dependence of com-
plexes on simples is the thought that were things otherwise all sem-
blance of genuine unity would be lost. So if it could be put to Leibniz
that it is available to him to think of nature as containing genuine re-
silient (possibly even unbreakable) unities without thereby needing to
fall back on unextended immaterial simples, the way would be open
for him to abandon his claim that nature's compound substances are in
their way no less everlasting than the simples that he finds himself
obliged to posit. The natural eternity of the embodied cabbage or ant or
man could then be quietly dropped, with the clear additional advantage
that we might thereby disengage Leibniz from the taxing task of ac-
counting satisfactorily for the existence of material things in terms of
the aggregation of immaterial elements, a topic with which I have stu-
diously avoided all engagement.
In treating of Leibniz's understanding of bodily identity through
time I have drawn on the idea of a certain material mass serving a cer-
tain function. What makes a particular mass of monads a human body
is that that mass is what the dominant monad needs to draw on in order
to realise with the particular clarity that distinguishes it from other
monads its peculiarly human perceptions and its more or less pressing
human appetitions. 4 What we draw on in determining that the organic
body that is made up of this particular mass of monads is the same as
the organic body made up at some different time of that other mass of
monads is whether those two masses serve the needs of the same
dominant monad in its psychic economy. If the same mass of monads
comes to serve a different master, then it will constitute different bod-
ies at those different times. No mass of monads can immediately and
directly serve two or more masters at once, though there is no impedi-
ment to the material mass that makes up the body of one complex sub-
364 Individuals, Essence, and Identity
stance being for a while part of the organic body of another such com-
plex substance.
To my mind, what is suggestive about this is that since that way of
thinking about the body underlies the overall theory, even though
Leibniz never did so it must be open to him to ask whether it might not
be equally well deployed in thinking about those fundamental activities
of the simple substances themselves that in his eyes constitute their
essential properties. That is, if the body is what the dominant monad
needs to enable it to function as a perceiving and appetitive entity,
what is there to stop us, or indeed Leibniz himself, from entertaining
the thought that perception and desire are not so much activities of
immaterial simples as complicated manners of functioning of com-
pounds? If the functioning of the dominant self requires a material
mass of monads to work together in particular ways for it to be able to
perceive the world and entertain and gratify its desires, what is there to
stop us from supposing that perception and appetition themselves
should be understood in terms of the ways in which certain complex
material compounds behave? Developing the very train of thought that
leads Leibniz to the introduction of organic bodies itself raises the
question whether the existence of monadic simples is forced on him at
all.
It is of course plain what Leibniz in fact thinks stops him from
adopting any such stance, and we shall come to that in a minute. But
what I should emphasize here is that his resistance to the idea precisely
does not take the form of saying that to adopt such a course would be
to paint ourselves into an ontological corner from which there is no
escape. That is, it is precisely not to say that abandoning an ontology of
simples would ipso facto be to abandon complexes as well. For if we
think of men and ants and cabbages just as material substances of dif-
ferent sorts, and different in kind by reason of the sorts of activity that
they standardly and systematically go in for, then just as before we can
think of them as unbreakable unities in that as long as they are able to
perform a sufficiently wide range of their species-specific activities no
loss of bodily parts will count as impugning the existence of the
wholes that they are. Moving away from thinking of Wittgenstein as a
particular dominant monad together with his organic body and replac-
ing that with a conception of him as a particular embodiment of a
Anthony Savile, Leibniz and the Persistence of Living Things 365
Notes
I Draft ofletter to Arnauld (8 December, 1686), in Parkinson (ed.) [1973], p.64.
2 Cf. Nouveaux Essais II. xxvii. 5 : "Si les vegetables et les brutes n 'ant point
d'ame, leur identite n 'est qu 'apparente; mais s'its en ant, l'identite individuelle y est
veritable a la rigeur, quoyque leur corps organises n 'en gardent point", see
Gerhardt (ed.) [1978], V, p.215). Cf. Remnant and Bennett (eds.) [1981], p.232.
3 At New Essays Il.xxvii.6 Leibniz writes: "In fact, an organic body does not
remain the same [«n'est pas de meme»] for more than a moment, it only remains
equivalent. And ifno reference is made to the soul there will not be the same life, nor
a vital entity either. So the identity in that case would be merely apparent." (Remnant
and Bennett (eds.) [1981], p.232)
4 Some hint ofthis is found in Leibniz's saying at New Essays II.xxvii.8 (Rem-
nant and Bennett (eds.) [1981], p.235): "Indeed it does seem to me that we have to
add something about the shape and constitution of the body to the definition of man
when he is said to be a rational animal; otherwise according to my views, Spirits
would also be men." Again in "A Specimen of Discoveries about Marvellous Se-
crets" we find: "So it was sufficient that souls should be given to the lower animals,
especially as their bodies are not made for reasoning, but destined to various func-
tions - the silkworm to weave, the bee to make honey, and the others to the other
functions by which the universe is distinguished" (Parkinson (ed.) [1973], p.84).
5 It is noteworthy that when Leibniz first used the term 'monad' (in correspon-
dence with Johann Bernouilli in 1690) it designated composites such as men or
sheep. (see Gerhardt (ed.) [1978], III, p.542). What drove him to introduce the meta-
physical simples that later went under that name was the thought that only if such
Anthony Savile, Leibniz and the Persistence of Living Things 367
simple things existed could the unity ofthese composites be accounted for. Under my
proposed revision Leibniz could have retained his original usage and bypassed the
introduction of immaterial monads altogether.
ON NATURALISING LEIBNIZ
(A REPLY TO ANTHONY SAVILE)
Richard Glauser
plying. The principles would just apply to things that are different from
those to which they apply in Leibniz. For instance, one would pre-
sumably have to take complete notions to be individual essences of
organic bodies. This means that, in the case of an animal or a person, a
complete notion will have to contain both physical and mental predi-
cates, because one and the same individual organic entity instantiating
the complete notion will have both physical and mental accidents.
Furthermore, one might also go as far as to say that naturalised
Leibniz could retain the general features of Leibniz's account of hu-
man freedom. He is a compatibilist of sorts, inasmuch as he holds that
freedom is compatible both with divine foreknowledge and with pre-
determination. According to the Theodicy, human freedom rests
mainly on three features: contingency, spontaneity and intelligence. 2 If
Leibniz were to adopt Savile's suggestion, he could maintain roughly
the same compatibilist position. I say 'roughly' because, although there
would be no difficulty in retaining contingency and intelligence as they
are discussed by Leibniz himself, his metaphysical account of sponta-
neity would probably have to be abandoned, for reasons to be adduced
below, in favour of our everyday notions of spontaneity and absence of
constraint, which are the ones that count, and suffice, for ordinary
moral and legal purposes. To abandon his metaphysical account of
spontaneity would imply that he give up the claim that "we [i.e. per-
sons] are constrained only in appearance and that in metaphysical
strictness we are in a state of perfect independence as concerns the in-
fluence of all the other created beings".3 (I shall return to this below.)
The real Leibniz can say this as long as persons are dominant monads,
between which there is no real transeunt causation. 4 But it would be
strange for Savile's Leibniz to say that persons are merely bodies with
mental functions and, also, that in metaphysical strictness no person is
ever constrained by any other. Nevertheless, one can grant Savile that
naturalised Leibniz might say, more plausibly, that, although people
qua thinking bodies sometimes are constrained, whenever they are not
and they do act spontaneously they satisfy one of the three necessary
conditions for acting freely. Given that the other two conditions can
also be satisfied, such an adjustment concerning spontaneity would not
jeopardize the essentials of Leibniz's theory of human freedom insofar
as it is relevant to everyday moral purposes.
372 Individuals, Essence, and Identity
Savile's revision also runs against what one may roughly call
Leibniz's platonism, i.e. the idea that the difference between bodies
and rational minds simultaneously covers at least three contrasts: be-
tween the sensible and the intelligible, the material and the immaterial,
the continually fluctuating and the immutable. With the abandonment
of Leibniz's versions of idealism and platonism the general form and
structure of his system would scarcely be recognisable. 6
The question of the relation between mind and body, just men-
tioned in connection with Leibniz's versions of idealism and plato-
nism, naturally leads to the question of pre-established mind-body
harmony. Leibniz has both a posteriori reasons and a priori reasons
for defending pre-established mind-body harmony.7 His a priori rea-
sons are related to his claim that pre-established mind-body harmony is
a special case of universal mind-mind harmony. To cut a long story
short in an admittedly rough and ready way, bodies are made up of
minds or soul-like substances, so that whatever pre-established har-
mony is to be found between a mind (dominant monad) and its body
will depend on the harmony to be found between the mind (dominant
monad) and the soul-like monads that make up its body. Once minds
are reduced to biological and psychic functions of bodies in accordance
with Savile's reform, any pre-established mind-body harmony that re-
mains will no longer be a special case of pre-established mind-mind
harmony. It will be a special case of body-body harmony, a harmony
between two aspects - mental and physical- of one and the same body.
In sum, mind-mind harmony will be a special case of body-body har-
mony, where the harmony between one mind and another will be taken
as obtaining between two bodies considered together with their psychic
functions. And mind-body harmony will also, but in a different way, be
a special case of body-body harmony, where the harmony between a
mind and its body will be taken as obtaining between two aspects -
psychic and physical - of one and the same body. In both cases, the
rejection of Leibniz's idealism makes body-body harmony come out as
basic, and pre-established harmony is set on its head.
But under Savile's proposed revision how well will such a con-
ception of pre-established mind-body harmony work? I think it will not
work well. In order to show why, let me begin by making two, rather
long preliminary remarks.
374 Individuals, Essence, and Identity
Let us go along with that, and try to see what happens if naturalised
Leibniz attempts to retain pre-established mind-body harmony includ-
ing, therefore, the causal asymmetry mentioned above.
The metaphor of the two clocks will no longer suggest a harmony
between the mental accidents of a dominant monad and the accidents
of its organic body. Presumably it will suggest a harmony to be found
between the series of mental accidents and the series of physical acci-
dents of one and the same organism. Now, harmony is not identity. So,
if naturalised Leibniz wants to retain some form of pre-established
mind-body harmony, he will probably have to say that the physical ac-
cidents and the mental accidents of one and the same organism con-
stitute two distinct, parallel series, where the members of one series are
neither type-identical nor token-identical with members of the other
series. It seems that we have accident-dualism of one and the same
body. Also, since pre-established mind-body harmony implies causal
asymmetry in the sense introduced above, real transeunt causation is
ruled out. So, naturalised Leibniz will presumably say that the two dis-
tinct, parallel series of accidents of one and the same organic body will
be causally independent of each another. Physical accidents will cause
other physical accidents of the same body; mental accidents will cause
other mental accidents of the same body. But, strictly speaking, mental
accidents cannot cause physical accidents, and vice versa. This is even
stranger than epiphenomenalism for a philosopher bent on some form
of mind-body naturalism. Even if naturalised Leibniz were to hold on
to Leibniz's ideal notion oftranseunt causality, that would not hide the
fact that the mental and physical accidents of an organic body will be
causally inert with respect to each other. Such a position is perhaps not
logically impossible. However, if one really is justified in believing
that matter thinks, such an attempt to salvage pre-established mind-
body harmony is going to seem queer indeed.
Where does the queerness stem from? I suggest it stems from the
assumption that, if naturalised Leibniz really is justified in believing
that matter thinks, then the kind of justification he has for believing
that makes it implausible to hold that the mental and physical accidents
of one and the same body have no real causal relations at all between
each other. For, ask such a naturalistic philosopher, say Hume or
Searle: are mind-body and body-mind causality metaphysically real?
378 Individuals, Essence, and Identity
He will reply that they are just as real as any causality is ever likely to
be in this world. That is the point with naturalism: minds and their
states are natural phenomena; so whatever kind of causal relations ob-
tain between physical phenomena also obtain between physical phe-
nomena and mental states, and vice versa, and between mental phe-
nomena. Of course, Hume would add, as naturalised Leibniz would,
that such causality depends on nomological regularity. Fine, but if
mind-body and body-mind causality (based on pre-established regular-
ity) are just as real as any other causality, then Leibniz's causal asym-
metry has been lost. Thus, naturalised Leibniz will not hold that im-
manent causality is metaphysically real whereas transeunt causality is
not. Even if he accepts a distinction between immanent and transeunt
causality, he will put them on a par with each other; in fact he will not
grant the distinction any metaphysical significance, just as Hume does
not. But, again, once causal asymmetry is lost, so is the distinctive
feature of pre-established mind-body harmony. At this point, I submit,
pre-established mind-body harmony will have to give way. And, so,
naturalised Leibniz is not a revised Leibniz but the author of an alto-
gether different philosophy.
Related to this is the question of the real causal source of the dou-
ble series of a body's accidents. Leibniz's pre-established harmony de-
pends on monads' containing within themselves the real causal source
of the entire series of accidents they will ever have. So, if naturalised
Leibniz still wishes to preserve pre-established harmony, he will
probably say that an organic body contains entirely within itself an
autonomous source of causation capable of spontaneously producing
the two parallel series of all physical and mental accidents that are ever
to affect the body during its finite lifetime. But where will such a
causal source lie? Savile does not say. It cannot be attributed to a
dominant monad and to the monads that compose the body, for they
have been rejected. So, would such a causal source be discoverable
empirically, by scientific means, within the body? That does seem un-
likely. If one replies that such a causal source of the accidents of an
organic body lies within the smaller organic bodies of which it is com-
posed, and so on ad infinitum for the immanent causal source of each
of the smaller organic bodies, then it must be shown that this kind of
reply is not an asylum ignorantiae. In any case, is it even remotely
Richard Glauser, On Naturalising Leibniz 379
plausible to hold that an organic body has within itself such a source of
causal power? If naturalised Leibniz is to be credited with the kind of
scientifically based naturalistic outlook that makes it plausible to be-
lieve that minds are bodily functions, the same outlook seems to make
it utterly implausible to say that within an organic body lies a source of
immanent causal activity that spontaneously produces the complete,
double series of physical and mental accidents that are to affect the
body during its entire lifetime. Finally, if it is objected that such a po-
sition is tenable for the mere reason that it is not logically impossible,
the reply is that whatever plausibility may have been gained by making
animals' and persons' life spans what we ordinarily take them to be
has been lost by the proposed revision.
things lead to grace by means of the very ways of nature and that this
globe, for example, must be destroyed and repaired by natural ways at
those times which the government of spirits demands for the punishment
of some and the reward of others"; "[ ... ] sins must therefore carry their
punishment with them by the order of nature, and even by virtue of the
mechanical structure of things; and that noble actions, similarly, attain
their rewards through ways that are mechanical in relation to bodies".16
Let us set aside the issue of the future destruction of the world, and fo-
cus merely on the claim that in the best possible world virtue will be
rewarded and vice punished by entirely natural ways, whatever the
natural ways are. But let us also keep in mind that all rational minds,
which are to be rewarded or punished in due course, are perpetual
members of the City of God, an everlasting moral state comprising all
380 Individuals. Essence. and Identity
human minds, angels, and God as prince and legislator. Now, Leibniz
is clear (1) that citizenship in the City of God requires that every finite
rational mind be naturally immortal, and (2) that the natural system of
reward and punishment governing the City of God requires not only
that citizens be naturally immortal, but also that they retain their con-
sciousness, memory and personal identity after what we call their natu-
ral deaths.I7 It is important to recall this, because, even though Leibniz
does go along to a certain extent with the Spinoza-Shaftesbury view
according to which virtue is its own reward, and vice its own punish-
ment, and though he believes, too, that during what we call this life
vice generally causes some unhappiness to the vicious person, and
virtue causes some happiness to the virtuous person, nevertheless he
also believes that, all said and done, by the end of what we call this life
the vicious sometimes end up being happy, and the virtuous unhappy.
That this is a fact shown by experience according to Leibniz (as it will
be later for Kant) is made clear in the Theodicy18 and elsewhere. 19 And
this fact, if it were not "corrected", would be "contrary to order"; it
would violate the pre-established harmony between nature and grace,
and be incompatible with the perfect government of the City of God.
So, how can such a disorderly situation be corrected? Only if rational
minds are naturally immortal, and if unrewarded virtue and unpunished
vice in this life are duly rewarded and punished in what we call an af-
ter-life. That is why Leibniz believes that natural immortality is neces-
sary for due reward and punishment to be naturally, and completely
administered. Although they are not always naturally administered in
full during this life, reason and natural religion justify the belief that
things will be set right in what we take to be an after-life. And even
natural immortality is not enough; permanence of memory, conscious-
ness and personal identity are required, toO.20 Hence, natural immortal-
ity and permanence of memory, etc. are essential to Leibniz's pre-
established harmony between nature and grace and to his theory of the
City of God. (It would be an ad hoc manoeuvre to reply that it is open
to naturalised Leibniz to assert flatly, contrary to the real Leibniz, that
all good and evil deeds are in fact entirely, adequately and naturally
rewarded and punished during this life. In any case, what reason would
naturalised Leibniz have to make such an assertion, since this does not
Richard Glauser, On Naturalising Leibniz 381
follow from the best possible world theory, nor from the a priori prin-
ciples of his system?)
Leibniz's conception of natural religion, too, depends on natural
immortality. He rejects the idea that religion should be founded on
revelation alone, and is determined to establish it on reason and nature
as well. Now, the cornerstone of Leibniz's natural religion is the natu-
ral immortality of souls, which can be known by reason. 21 And minds
are naturally immortal, according to Leibniz, if, and only if, they are
immateriaP2 On Savile's proposal, however, minds are bodily func-
tions and so can hardly be immaterial and naturally immortal. On these
four counts - natural immortality, natural religion, pre-established
harmony between nature and grace, and the theory of the City of God -
naturalised Leibniz's position is incompatible with the real Leibniz's
deep convictions.
As I have tried to show, naturalised Leibniz has to give up two
parts of the general theory of pre-established harmony, i.e. mind-body
harmony and the harmony between nature and grace. If so, the general
theory as Leibniz conceived it is crippled. So, then, is a proof of the
existence of God that Leibniz considers one of his most original and
most powerful. 23
But, to return to the question of immortality, one might ask: So
what? Can we not have naturalised Leibniz plus miraculous immortal-
ity and miraculous preservation of personal identity in an after-life?
But that would amount to turning that which Leibniz describes as natu-
ral into something miraculous in naturalised Leibniz. Also, the claim
to a natural religion would be lost, since the belief in immortality could
no longer be justified by reason. Religion would be confined to revela-
tion, which is precisely what the real Leibniz sought to avoid. 24 How,
in any case, could he claim to know that a miracle takes place after
what would be the natural death of rational minds?
Finally, let us reflect that Leibniz's conception of the best possible
world depends on his dispensing with the constant invocation of the
idea of God's miraculous intervention. Leibniz brings this about, not
by rejecting miracles altogether, but by fitting whatever miracles there
are into the general order of the best possible world. 25 If miracles are to
be incorporated in such a general order, they are bound to be few and
far between. Indeed, one of the reasons for which Leibniz was proud of
382 Individuals, Essence, and Identity
how could Leibniz's theodicy survive the loss of the complete accom-
plishment of divine justice?
Savile says: "there is nothing in Leibniz's philosophy that forbids
him from widening the range of explanatory devices that apply to the
material world beyond the purely mechanical as soon as we find our-
selves able to draw upon them with success", in order to claim that
minds are just bodily functions. In the two last sections I have argued,
to the contrary, that if Leibniz were well justified in accepting that
minds are bodily functions, that they are neither immaterial nor natu-
rally immortal, then ripple effects would undermine enough of the es-
sentials of his system to enforce abandonment of it. 28
Notes
1 Rescher [1993], p.159.
2 Cf. Theodicy, §288, in Gerhardt (ed.) [1978], VI, p.288.
3 A New System of the Nature and the Communication of Substances, § 16, in
that in the case of a substance being said to act on other substances, the substance
said to act goes through a real change whereby it attains a higher degree of perfection
than it previously had, whereas the other substances endure a real change leading to a
lesser degree of perfection. Cf. Discourse on Metaphysics, § 15, Loemker (ed.)
[1976], p.313. Of course, these real changes are brought about by immanent causa-
tion.
10 I use the expression 'causal asymmetry' in this sense only, i.e. to express the
contrast at the level of metaphysics between real immanent and, in a sense, unreal
transeunt causation. I do not use the term to express the fact that causal relations have
a fixed direction. I take it for granted that in Leibniz (and naturalised Leibniz) all
causal relations have a fixed direction.
384 Individuals, Essence, and Identity
11 Lest there by any doubt on this point, consider that one must distinguish two
claims in Leibniz: (1) the mental states of a Leibnizian substance follow each other in
an orderly series pre-established by God, and explainable by the substance's com-
plete notion, and (2) they do so in fact by being really produced by a source of im-
manent spontaneous causal activity within the substance. The two claims are distinct.
Yet (1) without (2) would make no sense to Leibniz. It would be like trying to say
that all the predicates that can be truly attributed to an individual subject (when the
truth is contingent) are contained in the subject's complete notion, yet it is not the
case that the substance instantiating the complete notion is the metaphysically real
cause of (all) its accidents instantiating the predicates; i.e. it is not the case that (all
of) our thoughts are really produced by us by way of a kind of emanation analogous
to the way by which God produces and preserves us. But (1) and (2) lead to the
causal asymmetry mentioned above, essential to the theory of pre-established mind-
body harmony.
12 Indeed, in his review of Shaftesbury's Characteristics, Leibniz praises
Shaftesbury for his world "all of one piece" and for its "universal harmony". But a
few lines further he adds that it lacks pre-established harmony (Loemker (ed.) [1976],
p.633). At first sight this seems strange, for Leibniz often uses the two expressions as
roughly equivalent. (Cf. Finster, Hunter et alii [1988], pp.136-138). Also, Leibniz
knows, as does any reader of the Characteristics, that Shaftesbury's world is just as
well regulated by causal laws as most other philosophers' worlds are. And he also
knows that all the law-like regularities in Shaftesbury's world have been ordained by
God, i.e. they are pre-established. So, why does it lack pre-established harmony? I
suggest it is because it does not conform to Leibniz's theory of pre-established har-
mony, i.e. it lacks the conjunction of (1) and (2), mentioned above, and therefore
lacks the contrast (asymmetry) between immanent and transeunt causation.
\3 Nouveaux Essais sur l'entendement humain, Preface, Gerhardt (ed.) [1978],
V, p.59.
14 An Essay Concerning Human Understanding II, 27, §25, Nidditch (ed.)
[1975], p.345.
15 Cf., for instance, Monadology, §87, Loemker (ed.) [1976], p.652; Theodicy,
p.125.
18 Cf. Gerhardt (ed.) [1978], VI, p.lIO-lll.
19 "Since experience shows us that God, for reasons unknown to us but surely
very wise and based on a greater good, permits many evil persons to be happy in this
life and many good persons to be unhappy, a fact which would not conform to the
rules of a perfect government such as God's if it had not been corrected, it follows
necessarily that there will be another life and that souls will not perish with the visi-
ble bodies. Otherwise there would be crimes unpunished and good deeds unre-
warded, which is contrary to order" (Reflections on the Common Concept of Justice,
Loemker (ed.) [1976], p.564). Cf. Opinion on the Principles of Pufendorf, in Riley
Richard Glauser, On Naturalising Leibniz 385
(ed.) [1988], p.66-68. On this whole issue I refer the reader to Grua [1953], Chapter
15.
20 Cf. letter to Malebranche, Gerhardt (ed.) [1978], IV, p.300.
21 Nouveaux Essais sur {'entendement humain, Preface, Gerhardt (ed.) [1978],
V, pp.60-61.
22 Cf. Jolley [1984], p.21.
23 Cf. letter XXII to Arnauld, Gerhardt (ed.) [1978], II, p.115. Cf. also VI,
p.541.
24 Cf. Gerhardt (ed.) [1978], V, pp.60-61.
2S Cf. Discourse on Metaphysics, §7, Loemker (ed.) [1976], pp.306-307.
26 Cf., for instance, Gerhardt (ed.) [1978], VI, p.541.
27 Not to mention that it would hardly befit Leibniz's morally perfect and wise
legislator, a citizen among others in the City, although the most prominent one, to be
obliged to perform miracles in order to see to it that some of his citizens are punished
and others rewarded.
28 I thank David Wiggins for his comments on an earlier version of this paper.
Any shortcomings are my own.
TEMPORAL PARTS AND IDENTITY
ACROSS TIME
1.
then by linking this up with all other slices (or parts) of the relevant sort
which stand in a certain causal relation to the slice (or part) existing at (or
around) time t. (Tooley [1977], pp.97-98)
of) the table' and, similarly, 'the book on the chair at t*' means 'the
four-dimensional book the t*-slice of which is on (the t*-slice of) the
chair'. And, of course, no particular problems are raised by the asser-
tion that two such descriptions as these might be names for a single
object.
Both these theories about the way in which time-involving definite
descriptions are related to the world can be represented pictorially. The
first theory can be represented like this:
the book on
the chair at t*
o
_____________ 1 _______ 1 ______________ _
o
past t* future
Figure 1
It is used merely to display the fact that certain phrases denote certain
objects.)
The second theory, Tooley's theory, may be represented by this pic-
ture:
o
_____________ 1 _______ L ______________ _
Figure 2
In this picture, the time-axis and the books mean what they meant in
Figure 1. The rectangle represents the boundary of the four-
dimensional object that is what the book really is. The description-
referent relation is again represented by labels, but the labels are fixed
to the "whole" book and not to slices of the book. Moreover, each of
the cords attaching the labels to the "whole" book passes through a
book-slice - the same slice it is attached to in Figure 1 - on its way to
its point of attachment to the book. This feature of the picture is in-
tended to represent Tooley's idea that a description like 'the book on
the chair at t' gets "linked up with" the book via the t-slice of the book.
(The points of attachment of the labels have no significance; I have to
draw them as attached somewhere.)
In my view, these pictures embody grave illusions about the nature
of enduring objects and about the way in which time-involving de-
scriptions apply to their referents and about the kind of facts expressed
by sentences formed by flanking the identity sign with time-involving
descriptions. (Let us call such sentences 'temporal identity-sentences'.)
I believe it is a grave illusion to suppose that there are four-
Peter van Inwagen, Temporal Parts and Identity across Time 391
2.
What are temporal parts supposed to be? - or, if you like, What is
'temporal part' supposed to mean? Many philosophers find these
things, and this phrase, wholly unproblematic. They construct elegant
solutions to various philosophical problems by appealing to temporal
parts, and they seem to assume that these objects enjoy the same meth-
odological rights as numbers or sets: although there are philosophical
problems that could be raised in connection with them (1 dare say they
will be willing to concede that much), temporal parts are well enough
understood that philosophers can appeal to them without incurring any
obligation to interrupt their discussions of personal identity - or what-
ever - to explain them. (The most celebrated example of this sort of
392 Individuals, Essence, and Identity
~b'
su stantIve ~b'
attn utIve ~l
copu a ~d'
pre lcate
adjective ../ '-- adjective..J
'-----------...V y
subject predicate
the subjects of the verbs they in fact modify, are guilty of a fallacy I
like to call adverb-pasting. If the adverb-pasters were given free rein,
all manner of fascinating philosophical problems would be created.
Consider, for example, the sentence
Alice, viewed full-face, is strikingly beautiful, but Alice, viewed in
profile, is aesthetically unremarkable.
Here we have the germ of the problem of cross-perspectival identity:
what is the relation between the strikingly beautiful Alice-viewed-full-
face and the aesthetically unremarkable Alice-viewed-in-profile? And
there is the problem of cross-evaluational identity: what is the relation
between the brilliant Hume-according-to-Professor-A and the doltish
Hume-according-to-Professor-B?
More or less the same points (to revert to the case of temporal ad-
verbs) apply to more complicated sentences, like our 'St. Paul's in
1850 was dingier than st. Paul's last year'. The grammatical structure
of this sentence may be compared with the grammatical structure of
'Condorcet as representative figure of the Enlightenment is more inter-
esting than Condorcet as original thinker'. ("Condorcet is interesting."
"How? In what respects?" "Well, as a representative figure of the En-
lightenment; less so as an original thinker." "St. Paul's was dingy."
"When? At what times?" "Well, in 1850; less so last year.") I trust that
no one will want to say that 'Condorcet as a representative figure of
the Enlightenment' is a name of a certain temporal part of Condorcet, a
part that comprises just those moments at which he was engaged in
representing the Enlightenment, while 'Condorcet as original thinker'
is a name of the part of Condorcet that comprises those moments at
which he was engaged in original thought.
A similar account applies to 'Philip drunk'. As a first approxima-
tion to a correct account of the sentences in which this phrase figures,
we may say that a sentence like 'Philip drunk is rash' is just a fancy
way of saying 'Philip, when he is drunk, is rash', or - even less fancy-
'Philip is rash when he is drunk'. (We may compare 'Philip drunk is
rash, but Philip sober is crafty' to 'Philip the man is rash, but Philip the
politician is crafty'. Lest someone argue that 'Philip the politician' is a
name for a temporal part of Philip, the part comprising just those mo-
ments at which he was engaged in politics, we may stipulate that all
394 Individuals, Essence, and Identity
not suppose the doubters will accept my premises, but it will be instructive
to find out which they choose to deny.
First: it is possible that a person-stage might exist. Suppose it to ap-
pear out of thin air, then vanish again. Never mind whether it is a stage of
any person (though in fact I think it is). My point is that it is the right sort
ofthing.
Second: it is possible that two person-stages might exist in succes-
sion, one right after the other but without overlap. Further, the qualities
and location of the second at its appearance might exactly match those of
the first at its disappearance. Here I rely on a patchwork principle for pos-
sibility: if it is possible that X happen intrinsically in a spatiotemporal re-
gion, and if it is likewise possible that Y happen in a region, then also it is
possible that both X and Y happen in two distinct but adjacent regions.
There are no necessary incompatibilities between distinct existences.
Anything can follow anything.
Third: extending the previous point; it is possible that there might be
a world of stages that is exactly like our own world in its point-by-point
distribution of intrinsic local qualities over space and time.
Fourth: further, such a world of stages might also be exactly like our
own in its causal relations between local matters of particular fact. For
nothing but the distribution of local qualities constrains the pattern of
causal relations. (It would be simpler to say that the causal relations su-
pervene on the distribution of local qualities, but I am not as confident of
that as I am of the weaker premise.)
Fifth: then such a world of stages would be exactly like our own sim-
pliciter. There are no features of our world except those that supervene on
the distribution oflocal qualities and their causal relations.
Sixth: then our own world is a world of stages. In particular, person
stages exist.
Seventh: but persons exist too, and persons (in most cases) are not
person-stages. They last too long. Yet persons and person-stages, like ta-
bles and table-legs, do not occupy spatiotemporal regions twice over. That
can only be because they are not distinct. They are part-identical; in other
words, the person-stages are parts of the persons.
More exactly, I do not see how it could be that the first half of the two-
year-man's career could be the career of anything, and I do not see how
it could be that the second half of the two-year-man's career could be
the career of anything. When I examine the story of the creation and
annihilation of the two-year-man, I do not find anything in it that
comes to the end of its existence after one year: the only thing "there"
(as I see matters), the two-year-man, will not come to an end after one
year; he will, rather, continue to exist for another year. And, in the
same way, when I examine the story, I do not find anything in it that
begins to exist halfway through the story.
These remarks are not meant to be a refutation of Lewis's argu-
ment. They are meant only to identify the point in the argument at
which one philosopher, myself, parts company with Lewis. In identi-
fying this point, I am merely accepting Lewis's invitation: "I do not
suppose the doubters will accept my premises, but it will be instructive
to find out which they choose to deny."
To recapitulate: in virtue of Lewis's explanation of what temporal
parts are, I understand the term 'temporal part', but I do not see how
(in the sense of 'temporal parts' Lewis's explanation has supplied) a
thing could have temporal proper parts. And this is not the end of my
difficulties with proper parts - my difficulties, that is, with under-
standing how temporal parts (understood as "Lewis parts") could have
the features that those who appeal to temporal parts in their philo-
sophical work suppose them to have. When I look at temporal parts
through the lens Lewis's explanation supplies, I see things that seem to
me obviously to have modal properties at variance with the modal
properties that are commonly ascribed to temporal parts. It seems to
me to be obvious that the one temporal part of a thing must be
"modally ductile" and "modally compressible." Consider, for example,
Descartes and his one temporal part, himself. Descartes's one temporal
part - Descartes - existed from 1596 to 1650. The object that is called
both 'Descartes' and 'Descartes's only temporal part' might have ex-
isted for twice as long as it did (it is modally ductile), and it might
have existed for only half as long (it is modally compressible). It seems
obvious to me, moreover, that this object might have had entirely dif-
ferent momentary properties at the corresponding points in its career.
Suppose, for example, that Descartes had been stolen by Gypsies
Peter van Inwagen, Temporal Parts and Identity across Time 399
shortly after he was born; if that had happened, then the object that is
in actuality Descartes's one temporal part would still have existed but
might well never have acquired the property "is able to speak French."
These modal propositions about Descartes's single temporal part seem
to be inconsistent with the modal properties that are ascribed to tempo-
ral parts by those philosophers who believe that there is useful philo-
sophical work for temporal parts to do, for these philosophers, or most
of them, seem to treat temporal parts as things that have their
"temporal extensions" and their careers essentially. Now I may be
wrong about this. I am doing no more than recording an impression. I
cannot point to any passage in which a philosopher has said in so many
words that temporal parts have either their temporal extensions or their
careers essentially. My point is only this: if there are philosophers who
think that temporal parts have their temporal extensions and their ca-
reers essentially, I cannot see how what they believe could be true.
Let us now return to the topic of identity across time. If there are,
as I believe, no temporal parts, or if each enduring thing has only one,
where does this leave us with respect to this notion? Surely it is tempo-
ral parts (or stages, or phases, or whatever) that are the terms of the
cross-time identity relation? If a thing has no temporal parts, or has
only one, what can be meant by the assertion that it exists at different
times?
3.
on the table at t
Figure 3
cases, the verb in such descriptions can be dropped: 'The book on the
table at noon was red', is good English, though, I would point out, a
'that was' is present "in spirit" in this sentence, even if it is unpro-
nounced, for the adverbial phrase 'at noon' modifies an understood
'was'. (Philosophers who suppose that such phrases as 'at noon' and
'in 1850' are adjectives describing the location of temporal slices or
whatnot are, as we have seen, mistaken.) In more complicated cases,
there is no possibility of wholly eliminating tensed verbs from time-
involving descriptions, as is shown by 'the car that used to be owned
by the man who will marry the woman who had been the first woman
President' .
There is one sort of temporal identity sentence that cannot be rep-
resented in any very straightforward way by a picture in the style of
Figure 3. I have in mind temporal-identity sentences containing so-
called "phase-sortals." (A phase-sortal is a count-noun such that a
given object may fall within its extension at one time but not at an-
other.) Consider, for example
The surgeon who removed the tumor from my brain is the boy
who once shined my shoes.
Obviously (one might argue) we cannot represent a fact of the sort this
sentence purports to express by a picture of someone twice labeled, for
no one is simultaneously a boy and a surgeon. And it will do the
friends of Figure 3 no good to protest that the 'is' in this sentence is a
rhetorical conceit, the strictly correct copula being 'was', since pictures
in the style of Figure 3 are unable to represent the use ofa tensed iden-
tity-sign.
There are various ways to deal with this problem. One way would
be to say that this sentence is not really an identity-sentence at all, but
a predication. That is, to say that what the speaker of this sentence is
doing is saying of a certain surgeon that he (the surgeon) is now such
that as a boy he shined his (the speaker's) shoes and was, moreover,
the only boy to do so. I shall not explore this avenue. I shall instead
investigate a way of treating this sentence as a real identity-sentence.
But what I am going to say is no more than a proposal for dealing with
phase-sortal identity-sentences. The "predication" analysis, or some
other analysis entirely, may tum out to be more fruitful.
Peter van Inwagen, Temporal Parts and Identity across Time 403
4.
is probably not a phase-sortal, but dogs, like most things, can fall un-
der various phase-sortals (like 'tailless dog') and a picture can avoid
representing a dog as falling under a given phase-sortal only by es-
chewing detail; and a picture that eschews all detail is not a picture at
all, but, like the Bellman's map, a perfect and absolute blank.
This problem, the problem of what exactly the picture is to show,
is not a problem that the friends of Figure 3 face only when they are
attempting to generalize the device exemplified by Figure 3 to enable it
to depict those facts of temporal identity that involve objects that no
longer exist. Figure 3 itself presents them with this problem. Suppose
that Tooley's book had got a stain on its cover between t and t*. Shall
our diagrammatic representation of a book include a stain or not? We
can, of course, depict the book so sketchily that we do not have to de-
cide about that. But this tactic will not work in all cases: suppose we
are concerned with a temporal identity sentence about a human being
who is at one time a frail four-year-old girl and at another time a
grossly obese, bearded, six-foot-tall (surgical) male who has lost a leg?
Even a stick-figure has to have a definite number of legs. (And, of
course, someone who, like me, thinks that Danton was at one time a
fetus and at another time a severed head will sometimes find it even
more difficult to draw sufficiently sketchy pictures.) If we considered
only cases like that of the stained book and the girl-man, we should
probably be tempted to say that obviously, the picture should represent
the book (or whatever) as it is at the time the picture is scheduled to be
displayed. Thus, if the book was unstained at t, stained at t*, and now
once again unstained, the picture, if it is to be shown now, should rep-
resent the book as unstained. But we have already seen why this will
not work: there are temporal identity sentences whose terms denote
objects that no longer exist; and, of course, an object that does not now
exist cannot be depicted as it is now. (For that matter, there are iden-
tity-sentences that are not temporal identity-sentences and whose terms
no longer exist. 'The horse Caligula made a consul was the only horse
to hold political office', for example.)
The solution to this problem is a simple one: it does not make any
difference what the picture shows. Remember, the picture is only a
picture. Showing one thing twice-tagged is simply a device for graphi-
cally representing the fact that the descriptions inscribed on the tags
406 Individuals, Essence, and Identity
have the same referent. If we want to represent graphically the fact that
'the most famous teacher of Aristotle' and 'the most famous pupil of
Socrates' denote the same man, we have only to draw a man labeled
with tags bearing these phrases. We are no more constrained to draw
him at some particular age or in some particular condition or circum-
stances, than the author of an illustrated history of philosophy is con-
strained to choose a picture of Plato that shows Plato as being of some
particular age or as being in some particular circumstances. The author
of the illustrated history of philosophy knows that he may print a pic-
ture (a detail from the School of Athens, say), that represents Plato as
being - how else? - of a certain determinate age and in certain par-
ticular surroundings, and so on, and that it will be perfectly correct to
label it simply 'Plato'. (This label, by the way, will not be a description
of the picture; or not in the way 'Raphael, 1509-12, The Vatican' is. It
will be a description of the intentional content of the picture, like 'the
mechanism ofa watch' or 'the structure of RNA'. We should think of
the word 'Plato' printed under a picture as applying to the figure in the
picture and not to the picture.) That is, the label 'Plato' is correct tout
court and is not an abbreviation for 'Plato in old age' or anything else.
But if a detail from a picture showing a young man conversing with
Socrates and a detail from a picture showing an old man conversing
with Aristotle can both correctly be labeled 'Plato', then they can both
be correctly labeled 'the philosopher who, in middle age, founded the
Academy', because these two phrases denote the same object.
One minor point about the labeling of pictures. Suppose the author
of an encyclopedia article on General MacArthur accompanied his ar-
ticle with a single photograph of MacArthur, one taken when its sub-
ject was an infant and labeled 'Douglas MacArthur'; suppose the arti-
cle contained no word of explanation of the fact that it was accompa-
nied by a picture of an infant. There would be a lot wrong with that,
but the picture would not be mislabeled. David Lewis has reminded us
that someone's use of a sentence can be faulted on lots of grounds
other than falsity; similarly, someone's use of a captioned picture can
be faulted on lots of grounds other than the incorrectness of the cap-
tion. When I say that it "makes no difference what the picture shows,"
I do not mean to deny the obvious truth that it would be a queer thing
to do to represent the fact expressed by 'The father of Charles II was
Peter van Inwagen, Temporal Parts and Identity across Time 407
not easily be dislodged from our minds by argument. I do not say this
is impossible. After all, I have tried to dislodge Figures 1 and 2 from
your minds by arguing that the temporal parts that figure essentially in
these pictures do not exist. But if your view of "identity across time" is
supplied by one or the other of these pictures, you will set out to find
some premise or inference in my arguments that you do not accept (if
you attend to these arguments at all). And, of course, you will succeed.
No attempt to refute a view that rests on powerful and appealing pic-
tures can hope to succeed unless it supplies a rival picture of its own.
And that is my only reason for asking you to consider Figure 3.
5.
about time itself and from wonder about special problems of identity
like those presented by the Resurrection of the Dead or the Ship of
Theseus. I would hazard a guess, however, that the root of the so-
called problem of identity through time has something to do with what
I have called "adverb pasting."
Here is a famous passage from Locke.
Wherein identity consists - Another occasion the mind often takes of
comparing, is the very being of things, when, considering anything as ex-
isting at any determined time and place, we compare it with itself existing
at another time, and thereon form the ideas of identity and diversity. (Es-
say, Book I, Ch. 27)
I have a hard time resisting the impression that Locke thought that 'it-
self existing at another time' was a name. Or perhaps, since the point I
want to make has nothing to do with pronouns, I should say that I have
a hard time resisting the impression that Locke thought that phrases
like 'Mary existing in 1689' and 'Mary existing in 1690' are names,
and, moreover, names for things that are in some sense two, even if
they are also in some sense one. (If Locke did think this, however, it
does not seem to have done his investigations of substantive problems
about vegetable, animal, and personal identity any harm. My purpose is
not to criticize Locke's whole treatment of his subject.) And it seems
even clearer to me that, if Locke did not accept this thesis, neither did
he reject it. Perhaps the most accurate thing to say is that the idea of
"the two Marys" touched the fringes of his thought so delicately as to
give him no occasion to ask himself what he thought about it. This
judgment of mine is a matter of "feel" and is probably one I, who am
no very experienced reader of seventeenth-century English prose, have
no business making. But I am made uneasy by 'itself existing at an-
other time'; why not 'itself as it had been at another time'? If someone
repeatedly makes judgments like 'Mary was sadder in 1690 than she
was in 1689' and 'Mary was wiser in 1685 than she was in 1680',
there is nothing really wrong with saying of him, "He's always com-
paring Mary as she was at one time with Mary as she was at another."
(But it would be better to say, " ... comparing the way Mary was at one
time with the way she was at another.") If, however, someone says, " ..
. comparing Mary existing at one time with Mary existing at another,"
410 Individuals, Essence, and Identity
Notes
1 The theory I am ascribing to Tooley is very like the theory I called "Theory 2"
in van Inwagen [1990a], pp.245-55.
2 This is the theory I called "Theory 1" in van Inwagen [1990a].
3 Quine [1953], pp.65-79.
4 Achille Varzi has asked me several questions about my analysis of the role of
'in 1850' in the sentence 'St. Paul's in 1850 was dingy'. They could be summed up in
this question: "Granted, 'in 1850' is not an adjective modifying 'St. Paul's'; but is the
only alternative that it is an adverb modifying 'was'? - are there not other possibili-
ties?" According to the traditional grammar I was taught in school, the copula 'was'
is "the verb of' the sentence, 'St. Paul's in 1850 was dingy' and 'in 1850' modifies it.
But a more up-to-date grammar might tell us that when 'to be' functions as a copula,
it does not belong to the grammatical category "verb" (despite the fact that 'to be',
whatever its function, displays the grammatical accidence traditionally definitive of
the category "verb": tense, voice, mood, aspect), but rather to the category "takes an
adjective and makes a verb." According to this view of the matter, the simplest verb
Peter van Inwagen, Temporal Parts and Identity across Time 411
in 'St. Paul's in 1850 was dingy' is 'was dingy' - the past tense of the verb 'to be
dingy' - and that 'in 1850' modifies 'was dingy' (applies to it to produce the com-
plex verb 'was dingy in 1850'). Another up-to-date view of the function of ' in 1850'
in this sentence is this: 'in 1850' is not an adverb (a "takes a verb and makes a verb")
at all, but a sentence-modifier, a representative of the grammatical category "takes a
sentence and makes a sentence"; in the present case, it modifies 'St. Paul's was
dingy'. Which of these three accounts of the function of ' in 1850' is correct- ifthere
is indeed a fact of the grammatical matter - makes no difference for our purposes, for
each account has the consequence that 'St. Paul's in 1850' is not a noun-phrase and
hence does not represent itself as denoting an object. In the text, I presuppose the
traditional view, but the correctness of the traditional view is in no way essential to
my arguments. My arguments could easily be expressed in the terms provided by
either of the "up-to-date" accounts of the function of 'in 1850'.
5 A momentary property is a property an object could have at one time and lack
at another - like being seated and being Socrates' widow, and unlike being de-
scended/rom King David.
6 This definition leaves open the question whether there may be parts of a per-
son's career that are "topologically unsuited" to being the careers of objects. Sup-
pose, by suitable correlation of numbers with moments of time, we associate the do-
main of Descartes's career with the real numbers 0 through 1, inclusive (we should
be able to do this if there was both a first and a last moment of Descartes's exis-
tence). Could the part of Descartes's career whose domain is the rational numbers
between 0 and 1 be the career of an object? A part whose domain is a set that has no
Lebesgue measure? How about some relatively well-behaved (topologically speak-
ing) but non-connected set? - say, one corresponding to March 1610 and Good Fri-
day, 1633? These are questions that we can leave to the friends of temporal parts.
How they are answered is irrelevant to our argument.
7 I ignore Kripkean scruples about whether what was apparently a human being
that was created ex nihilo would really be a human being.
8 This paper first appeared in The Monist, 83 (2000), p. 437-459 and then as
Chapter 8 of van Inwagen [2001]. Some of the early parts of this essay (and a bit to-
ward the end) are taken from my essay "Plantinga on Trans-World Identity," in Tom-
berlin and van Inwagen (eds.) [1985]. Reprinted by premission: copyright © 2000,
The Monist, Peru, Illinois, 61354.
VAN INWAGEN ON TEMPORAL PARTS
AND IDENTITY ACROSS TIME
Andrea Bottani
I shall not argue against the two main theses of van Inwagen's paper:
the idea that persons, cats and trees are three-dimensional entities
lacking temporal parts; and the attack on the 'fallacy of adverb past-
ing', adverb pasting being a certain reading of the semantic role tem-
poral qualifications play in predicative sentences, i.e. the tendency to
attach them to singular terms rather than to predicates. Even if I shall
not discuss these central issues, I shall deny their purported relation-
ship, the idea that four-dimensionalism can be in some way "entailed",
"produced" or "generated" by the fallacy of adverb pasting. My thesis
will be that four-dimensionalism and adverb pasting are quite inde-
pendent of one another. In addition, I shall say something about the
notion of a temporal part. For a major aim of van Inwagen's paper is to
define a clear, neutral concept of a temporal part in terms of which the
thesis that ordinary objects have temporal proper parts can be under-
stood, and so asserted or denied. I will claim that van Inwagen's defi-
nition of 'temporal part' is not acceptable, nor does it make easier for
the enemy of temporal parts to state her case. In my commentary, the
latter point comes first, the former later.
413
A. Bottani et al. (eds.), Individuals, Essence and Identity, 413-426.
© 2002 Kluwer Academic Publishers.
414 Individuals, Essence, and Identity
Temporal parts and identity across time are matters of heavy philo-
sophical disagreement Some philosophers find it quite easy to grasp
the idea of a temporal part but extremely difficult to understand the
notion of identity across time (at least, the notion of a genuine, nu-
merical identity between things wholly existing at different times).
Other philosophers find it quite easy to grasp the notion of identity
across time but extremely difficult to understand the idea of a temporal
part (at least, the idea of a temporal proper part of an ordinary object
such as a person, a cat or a building). There is a stubborn temptation,
mirrored perhaps in the use of the words 'understand' and 'grasp' in
the above account, to see the disagreement at issue as one concerning
what can or cannot be conceived, rather than what can or cannot be
believed. In particular the enemies of temporal parts have often
seemed inclined to treat the notion of a temporal proper part of an or-
dinary object like a person, a cat or a tree as simply unintelligible -
which, at first sight, would deprive the thesis that ordinary objects
have temporal proper parts of any clear semantic content Neverthe-
less, most enemies of temporal parts firmly believe that events and
processes, such as wars and tennis matches, have proper temporal
parts. And most of them firmly believe that everything - continuant or
event - has at least one improper temporal part (i.e. itself). In order to
have such beliefs, one has to understand the general notion of being a
temporal part of something, and this clearly shows that what is at issue
here is not the intelligibility of some notion or other but the possibility
that some thesis is true: when the enemy of temporal parts says that the
notion of a temporal proper part of an ordinary object is unintelligible,
she is better interpreted as simply saying that it is impossible that per-
sons, cats and buildings have temporal proper parts. Obviously this is
not an argument, nor does it make easier for the enemy of temporal
parts to state her case. For ifit is difficult to prove that a given thesis is
false, it cannot be easier to prove that it is necessarily false.
Peter van Inwagen firmly resists the temptation to settle the debate
on temporal parts by relying to idiosyncratic semantic intuitions. In his
paper, he defines a neutral notion of temporal part in terms of which
the thesis that ordinary objects have temporal proper parts can be un-
Andrea Bottani, Van Inwagen on Temporal Parts and Identity across Time 415
property P, it may happen that an object could have P at one time and
lack P at another time while another object could not. In that case, we
should say that P is momentary for the first object but non-momentary
for the second, The property of being seated, for example, is momen-
tary for me but it is non-momentary for trees, for at no moment can a
tree be seated. And it is non-momentary for any stage of me whose ca-
reer has in its domain just one moment (for obviously no such tempo-
rally flat thing could have the property of being seated at one moment
and lack it at another). So, one possible interpretation of van In-
wagen's definition of 'career of an object' is that the career of an ob-
ject x assigns to any time t at which x exists a set of properties such
that: (i) they are momentary for x (i.e. x could have them at one time
and lack them at another); (ii) x has them at t. But this interpretation is
uncharitable, for it gives rise to several unpleasant consequences.
Suppose I am seated at a moment t. And imagine for the sake of
argument that there is a stage of myself in Lewis's sense whose career
has just t in its domain (call it L t , for short). My career would assign to
t a set containing the property of being seated, for that property is mo-
mentary for me, and I have it at t. But the property of being seated
would not be momentary for L t , whatever it may be, for L t could not
have that property at one time and lack it at another. I (Incidentally, this
would be true of any property of L t , so that no property instantiated by
Lt at t would be temporary for Lt , and Lt's career would assign to t an
empty set of properties). Then, under the uncharitable interpretation,
the career of L t would not assign to t a set containing the property of
being seated, and so Lt's career and my career would assign to t differ-
ent sets of properties. Hence, by definition of 'part of a career', Lt's
career would not be part of my career. And Lt. whatever it may be,
would not be a Lewis-part of me in van Inwagen's sense. For the same
reasons, mutatis mutandis, L t would not be a proper Lewis-part of any
object.
I conclude that, under the above uncharitable interpretation of 'ca-
reer of an object', no object could have a proper Lewis-part whose ca-
reer has in its domain just one moment. So, it would be possible for
any object x to have a Lewis-part y such that x ;;f:. y (i.e. y is a proper
part of x) and there is no z such that x = z+y (this would happen when-
ever y's career has in its domain all the moments which are in the do-
Andrea Bottani, Van Inwagen on Temporal Parts and Identity across Time 417
main of x's career except one). Far from being temporal parts in
Lewis's sense, Lewis-parts in this uncharitable sense would not even
be parts in the usual mereological sense.
All these consequences can be avoided by taking the expression
'momentary property' occurring in van Inwagen's definition of 'career
of an object' in an absolute rather than a relative sense. It is easy to
derive an absolute notion of momentary property from a relative one,
in one of the usual ways in which absolute notions can be derived from
relative ones: a property is momentary simpliciter if and only if it is
momentary for at least one object (i.e. at least one object can have that
property at one time and lack it at another). Accordingly, the career of
an object x should be conceived as a function which assigns to any
moment t at which x exists just the set of properties that (i) are mo-
mentary simpliciter (i.e. at least one object - not necessarily x - can
have them at one time and lack them at another); (ii) x has them at t. 2 I
assume this is the intended sense of van Inwagen's expression 'career
of an object'. The intended notion of Lewis-part has to be defined ac-
cordingly.
Even in this charitable interpretation of 'Lewis-part', however,
there is reason to doubt that temporal parts in Lewis's sense are just
Lewis-parts in van Inwagen's sense. Strictly speaking, this is only an
exegetical point, but it may be a good way of approaching some non-
exegetical problems. Imagine for the sake of argument that there is a
Lewis-part of me whose career has in its domain the moment t (for
brevity, call it L). And suppose I have at t the momentary property of
being a seated person. Then, L has to have at t the property of being a
seated person too, for L has at t just the momentary properties that L's
career assigns to t. And L's career, being part of my career, has to as-
sign to t just the set of momentary properties that I have at t (among
which, the property of being a seated person). But no stage of me in
Lewis's sense can have that property at any moment. For, according to
Lewis, person stages are not persons, and so they can never be seated
persons (at best, they can be seated person stages). Hence L, though
undoubtedly a Lewis-part of me in van Inwagen's sense, is not a stage
of me in Lewis's sense (it cannot be such a stage, because it comes out
with the wrong properties). Since the same, mutatis mutandis, is true
of any Lewis-part of anything, the point can be easily generalized as
418 Individuals, Essence, and Identity
2.
pasting. In fact, it can come to the surface only after we have discarded
adverb pasting, i.e. only after we have decided to attach the temporal
phrase 'in 1850' to the predicate 'is dingy' rather than to the subject
'St. Paul's'.
lf this is not enough, consider the thesis that processes like foot-
ball matches have temporal parts. I assume most of us agree on that: it
might be that football matches do not exist at all but, if they exist, then
they must have temporal parts (having temporal parts is the only way
they can exist at all). Well, suppose someone comes in and says: "that
matches have temporal parts is a fallacy springing from adverb past-
ing; we have only to realise that the subject of the sentence 'the foot-
ball match was enjoyable during its first ten minutes' is 'the football
match' - and not 'the football match during its first ten minutes' - to
see that football matches do not have temporal parts". Well, I think
this would simply miss the point. And the point is that a football match
can have the property of being enjoyable during its first ten minutes
only if the first ten minutes of the football match have the property of
being enjoyable. Still, this point can be made only after having at-
tached the temporal phrase 'during its first ten minutes' to the predi-
cate 'is enjoyable' rather than to the subject 'the match' - and so, only
after having discarded adverb pasting.
My conclusion· is that we have no need to subscribe to adverb
pasting in order to think that ordinary persisting objects have temporal
parts. We can believe in four-dimensionalism - and think that there is
an important and meaningful "problem of identity across time" - while
keeping firmly away from adverb pasting. In a nutshell: four-
dimensionalism does not entail adverb pasting.
So, I come to the other direction of the implication: does adverb
pasting entail four-dimensionalism? Does the idea that the sentence
'St. Paul's in 1850 was dingy' is about st. Paul's in 1850 entail that St.
Paul's has temporal parts? One could be tempted to answer negatively,
arguing as follows: that the sentence is about St. Paul's in 1850 does
not exclude that it is about S1. Paul's, for St. Paul's in 1850 might sim-
ply be St. Paul's. After all, the temporal part of St Paul's which occu-
pies the year 1850 might well be an improper part of St. Paul's. If so,
the temporal phrase 'in 1850' would be attached to the name 'St.
Paul's', but nevertheless St. Paul's would have no proper temporal
424 Individuals, Essence, and Identity
parts, for it would have only one improper part - St. Paul's itself - and
four-dimensionalism would be false.
Well, I think this would be a bad argument. For, what about the
sentence 'St. Paul's in 1999 was not dingy'? IfSt. Paul's has only one
temporal part - itself - then 'St. Paul's in 1999' denotes exactly what
'St. Paul's in 1850' denotes, i.e. St. Paul's itself. But then, how can
this object be at once dingy and not dingy? If the temporal phrases 'in
1850' and 'in 1999' attach to the name 'St. Paul's', then the terms 'St.
Paul's in 1850' and 'St. Paul's in 1999' cannot be co-referential, for
otherwise the truth value of the sentence 'St. Paul's in 1850 was
dingy' would be insensitive to the replacement of 'in 1850' with 'in
1999' - and we know that it is not so insensitive. That is the reason
why even Tooley's theory does not seem to work: for, according to
Tooley, 'the book on the table at t' and 'the book on the table at t*'
denote the same book (the same set of temporal parts of a book), and
so it is not easy to see how both the sentence 'the book on the table at t
is stained' and the sentence 'the book on the table at t* is not stained'
could be true (how could the same book be both stained and not
stained?).
So, we might conclude: if we attach temporal phrases to subjects
and not to predicates, then we are compelled to think that an expres-
sion resulting from attaching a temporal phrase to a subject denotes a
temporal proper part of what the subject denotes. For otherwise we
could not explain, for example, why 'St. Paul's in 1850 was dingy' is
true and 'St. Paul's in 1999 was dingy' is false. So, the adverb paster
cannot keep away from four-dimensionalism, understood as the thesis
that ordinary persisting objects have temporal proper parts.
Still, however, I think this would be an over-hasty conclusion, for
the temporal phrase 'in 1850' might be attached both to the subject
and to the predicate. This, I assume, is what happens when 'in 1850' is
applied to the whole sentence 'St. Paul's is dingy' (likewise, when we
say that in a possible world W John is rich, what we seem to say is that
the individual that John is in W is rich in W). So, the adverb paster can
keep away from four-dimensionalism if he is careful enough to attach
temporal phrases also to predicates and not only to singular terms. His
position would be as follows.
Andrea Bottani, Van Inwagen on Temporal Parts and Identity across Time 425
1. 'St. Paul's in 1850 was dingy' has the following structure: 'St.
Paul's in 1850' is the subject, 'is dingy in 1850' is the predicate.
2. 'St. Paul's in 1999 was dingy' has the following structure: 'St.
Paul's in 1999' is the subject, 'is dingy in 1999' is the predicate.
3. The subjects of the two sentences are co-referential: both denote
St. Paul's.
4. The first sentence is true, the second sentence is false.
5. The reason why the two sentences differ in truth value is that they
apply different predicates to co-referential terms.
Notes
1 It might be objected that what is a temporally flat stage of me in the actual
world could be a temporalIy thick object in some other possible worlds. If so, LI
could lose and gain the property of being seated, even if it does not do so in the ac-
tual world. I have no clear intuitions about that matter, particularly if the background
modal semantics is counterpart theory. For the relation of counterparthood is incon-
stant, and I am not sure there cannot be a sense of 'counterpart of in which a tempo-
rally flat stage of me has in some worlds one or more temporally thick counterparts
(though that would seem to me to be very strange). What seems to me to be certain is
the general point that, if objects have stages in Lewis's sense, then there are proper-
ties which are temporary for some objects but not for some of their stages in Lewis's
sense. Take a biological organism of the species homo sapiens and the zygote it was.
Lewis would say that the zygote is a stage of the organism. But the property of being
seated, though temporary for the biological organism, is not temporary for the zy-
426 Individuals, Essence, and Identity
gote, for no zygote can be seated at any time (not even, I would dare to say, in coun-
terpart theory). Since the same holds for the property of not being seated, the zy-
gote's career could not be part of the organism's career, and so the zygote could not
be a Lewis-part ofthe organism.
2 May be that all qualitative properties, even dog or animal, are momentary in
this weakest sense. For, if we observed a dog gradually developing into a lycaon, we
would not say that one living organism is dead and another has taken its place (the
same, mutatis mutandis, if we observed a plant gradually developing into an animal
in virtue of its metabolism). If so, the charitable interpretation of 'Lewis-part' would
not be so charitable. For the property of being a person would be momentary sim-
pliciter and so it would be in the set of properties that the career of any person as-
signs to any moment at which that person exists. But the property of being a person
would never be in the set of properties the career of a stage of a person assigns to a
moment, for no person stage is a person at any time. Hence, no stage of a person (no
matter whether person stages exist or not) could be a Lewis-part of any person. After
all, however, there might be at least some qualitative properties which are not mo-
mentary simpliciter (i.e. no object can have them at one moment and lack them at
another), in particular all last sortal properties, such as living organism, and perhaps
person.
3 At any rate, even if the problem could in some way be resolved, the restriction
to simple temporary properties would not solve the problem it is intended to solve.
For consider the property of being old. It does not seem to be a complex property,
and it is undoubtedly temporary (objects can gain it). But, if an object x is old at a
time t, certainly there are many stages of x in Lewis's sense that are not old at t
(some of them may even be born just at t, though all are stages of something that is
old at t). On the contrary, no Lewis-part of x in van Inwagen's sense can exist at t
without having at t the property of being old, for Lewis-parts have to inherit, at any
moment t of their existence, all the temporary properties that the objects they are
Lewis-parts of have at t. Once again, the conclusion is that many Lewis-parts in van
Inwagen's sense are not temporal parts in Lewis's sense. Moreover, in the previous
sentence, 'many' can be replaced by 'all' on the following grounds. Suppose that, for
any precise age an object can be, there is a property X such that any object has X iff
it is exactly that age. And suppose that, for any precise length of time an object will
still persist, there is a property Y such that any object has Y iff it will persist exactly
for that length. If there are such kinds of properties, at any moment any object has to
have exactly one property of the first kind and one property of the second kind. But,
for any property of the first kind and any property of the second kind an object x has
at t, no y can have the same two properties at t and be a proper temporal part of x.
For a temporal part of an object x can be proper only if it is shorter than x. Hence, no
object can have a proper temporal part whose career is part of its career, and no
proper temporal part of any object can be a Lewis-part ofthat object.
4 I have argued above that, when the enemy of temporal parts claims that the
notion of a temporal part is unintelligible, she is better interpreted as simply saying
that it's impossible that ordinary things have any temporal parts. If so, there is no
notion to be in need of explanation here.
CHANGE AND CHANGE-ERSATZ
UweMeixner
Like other sciences, metaphysics must save the phenomena. But, usu-
ally, this can be done in a variety of ways. The problem is to give good
reasons for preferring one way over the other. Take such a familiar
phenomenon as change. Here we find a metaphysical position that, on
the face of it, seems to deny that there is any change at all; I shall call
it supereternalism. Yet, philosophers on the side of supereternalism
have worked hard to make it save the phenomenon of change - with
truly remarkable success. The feeling remains that, nevertheless, a
non-supereternalistic conception of change is more appropriate. This
paper will give reasons for believing this. But before coming to su-
pereternalism and change, I want to discuss something else.
tative being and numerical being are vastly different concepts, as can
be seen from the two principles that govern their use:
(Bl) For every object x and every property fof objects: x is quali-
tatively fiffx hasl
(B2) For every object x and object y: x is numerically y iff x is
identical with y.
Given this, (AI) simply says: if an object had different properties from
those it has in fact, it would have different properties from those it has
infact. Nothing could be more certain than this. But (A2), on the other
hand, means: if an object had different properties from those it has in
fact, it would not be identical with the object it in fact is numerically
identical with. This is rather doubtful.
Suppose we nevertheless accept (A2), undaunted by the charge
that we are apparently confusing qualitative and numerical being, the
concepts which Plato laboured to distinguish in the Sophist. How can
we then preserve our intuitions of contingency?
The answer of counterpart theory is very well known: although, in
a certain sense, no object can have other properties than it has in fact,
there is a sense, and it is alleged to be the only truly relevant sense, in
which an object can have other properties than it has in fact. This
sense is given in the following two contingency-descriptions:
(Cl) Let x be an object that does not in fact have the property I
But it is possible that x has f (and it is therefore contingent
that x does not have f), since there is a possible world wand
a counterpart y of x in w such that y has fin w.
(C2) Let x be an object that in fact has the property I But it is pos-
sible that x does not havef(and it is therefore contingent that
x has f), since there is a possible world wand a counterpart y
ofx in w such thaty does not havefin w.
The immediate impression one obtains from (Cl) and (C2) is that the
contingency that counterpart theory is able to provide is nothing but
counterfeit contingency, at best an ersatz for genuine contingency. For
a possibility that is a possibility for an object x only in virtue of its be-
ing an otherworldly fact for another object y, different from x, al-
430 Individuals, Essence, and Identity
then the truth of (D) is doubtful indeed. For in view of the fact that no
object is ever numerically different from what it is numerically now
(namely, itself), (D2) is nothing more or less than a more elaborate
way of stating that no object ever has other properties than it has now,
which is a claim of supereternalism (or perhaps better, but certainly
longer: supersempiternalism).
Superetemalism flies even more in the face of intuition than does
superessentialism, for superetemalism implies that at least in a sense,
but prima facie not at all in a marginal sense, there is no change, and
432 Individuals, Essence, and Identity
erties than they have now, are, in point of fact, taken by the supereter-
nalists to have these other properties merely by analogy, as the owner
of a dog may be said, in a sense, to have a tail, since his dog has a tail.
Undoubtedly, the relationship between the temporal counterparts of an
object and the object itself is in some sense closer than the relationship
between a dog-owner and his dog. This has the consequence that the
supposedly analogous having of a property may be said to be much
more similar to properly having (or possessing) a property than the
analogous having of a tail is similar to properly having a tail. Never-
theless, just as one can comment that the owner of the dog does not,
properly speaking, have a tail, one can also comment, that an object, if
supereternalism is correct, never ever has, properly speaking, other
properties than it has now; it has other properties only by analogy.
Let me briefly indicate how supereternalists can deal with a spe-
cial kind of change, namely coming to be and passing away. This kind
of change poses no special problem for them, but is subsumed under
(El) and (E2), if existence is taken to be a property of objects that is
appropriate for change. Yet, to regard existence as a property of ob-
jects with respect to which they change, immediately implies, even
under (E1) and (E2), that there are objects that at some time do not
exist, that is: it implies temporal possibilism. Therefore, if supereter-
nalists deny temporal possibilism and affirm that all objects exist at all
times and nevertheless do not want to deny coming to be and passing
away, they must explicate coming to be and passing away in a differ-
ent way than by subsuming them under (El) and (E2). Under temporal
actualism, existence, if a property at all, is quite obviously not a prop-
erty that is appropriate for change.
How can they do it? Thus:
(Fl) An object x will come to be if there is no temporal counter-
part of x now, but also a future time at which there is a tem-
poral counterpart ofx. 6
Let me state some principles for temporal counterpart theory - the the-
ory which is the basis of a superetemalistic conception of change:
(Gl) For every object x there is precisely one time t such that x is
at t a temporal counterpart of x. (Uniqueness of lemporallo-
calion.)
(G2) For every object x and time t there is at t at most one tempo-
ral counterpart of x. (Uniqueness ofa temporal counterpart.)
(G3) For every object x, object y and time I: if y is at t a temporal
counterpart of x and x is different from y, then x is at t not a
temporal counterpart of y. (Restricted asymmetry of the tem-
poral counterpart relation.)
Note that the majority of these principles could not be justified by de-
fining temporal counterparthood as a similarity concept. One might
think of the following definition: x is at ( a temporal counterpart of y
=def X is a t-Iocated object which is (among all the t-located objects)
maximally similar to y, presupposing that for every object x there is
precisely one time t such that x is a t-Iocated object (which presuppo-
sition makes it possible to derive (Gl)). But this definition does not
help to establish (G2): why should not an object have at a time ( two or
Uwe Meixner, Change and Change-Ersatz 435
more t-Iocated objects that are all maximally similar to it? Nor does it
help to justify (G5): although y is a t'-located object that is maximally
similar to x, and x is a t-Iocated object (that is maximally similar to x),
that does not mean that x is a t-Iocated object that is maximally similar
to y; there may be a t-Iocated object that is more similar than x to y.
And quite clearly, the definition does also not help to establish (G6).
Here are the two main theorems that follow from the stated princi-
ples:
GTl: 'x is an object and y is an object and 3t' (t' is a time and x is
at t' a temporal counterpart of y)" or in other words 'x is
simpliciter a temporal counterpart of y' expresses an
equivalence relation over objects.
Proof
DEFl: R(x, y) =def. X is an object and y is ail object and 3t' (t' is a
time and x is at t' a temporal counterpart of y) [x is sim-
pliciter a temporal counterpart ofy].
Then: (1) For all objects x: R(x, x) (according to (G 1). (2) For all x
and y: R(y, x) ~ R(x, y): Assume: R(y, x); we have: x is at a time t a
temporal counterpart of x (according to (Gl)); hence according to
(G5): R(x, y). (3) For all x, y and z: R(x, y) and R(y, z) ~ R(x, z): As-
sume R(x, y), R(y, z); hence according to (G6): R(x, z).
GT2: If time is linearly ordered, then the set of all (at some time)
temporal counterparts of any given object is timewise line-
arly ordered.
Proof
Let r be any object. Consider the predicate 'Beforer(y, z)' which is
defined as follows:
DEF2: Beforer (y, z) =def R(y, r) and R(z, r) and the time at which y
is a temporal counterpart of itself is before the time at
which z is a temporal counterpart of itself.
Assume that time is linearly ordered, that is, that the before-relation
between times is transitive, irreflexive and linear over times (the latter
436 Individuals, Essence, and Identity
means that for all times t and t': t before t', or (' before t, or t = t'). It
immediately follows on the basis of DEFI, DEF2 and (GI): (I) For
every object x: not Beforer (x, x). (2) For every object x, y and z: Be-
fore r (x, y) and Beforer (y, z) :::> Beforer (x, z). What remains to be
proved is: (3) For every object y and z such that R(y, r) and R(z, r): Be-
fore r (y, z) or Beforer (z, y) or y = z.
Assume then: R(y, r) and R(z, r). Assume moreover: not Beforer(Y,
z), not Beforer (z, y). Hence by DEF2, (GI) and the linearity of time:
the time at which y is a temporal counterpart of itself is the time at
which z is a temporal counterpart of itself. By DEFI and the assump-
tions: there is a time t' at which y is a temporal counterpart of r, and
there is a time til at which z is a temporal counterpart of r. Hence by
(G4): y is at t' a temporal counterpart of itself, and z is at til a temporal
counterpart of itself. Hence by (G I) and what has already been de-
duced: t' = til. Hence y is at t' a temporal counterpart of r, and z is at t' a
temporal counterpart of r. Therefore according to (G2): y = z (which
was to be shown).
One may add another principle:
(G7) For all times t and t' and objects x: if there is at t a temporal
counterpart of x and at t' also a temporal counterpart of x,
then there are at all times between t and t' temporal counter-
parts ofx. (Density o/temporal counterparthood.)
(G7) has the consequence that for any object r the equivalence set {x:
R(x, r)} ordered by Beforer (y, z) is isomorphic to a certain time inter-
val ordered by the before-relation between times (restricted to that in-
terval). This is esthetically pleasing, but one may well wonder whether
it is not too restrictive to postulate (G7V
ordinary material things? And, really, could the sense described above
in which there is change in the superetemalist's world be even called
'ersatz change'? Hardly. Even the usual superetemalist would admit
that much - and would proceed to making his world somewhat less
strange: to finding a fairly acceptable ersatz for the normal change of
normal objects by considering higher order objects, namely certain
sets of his basic objects. What are considered to be 'normal objects' - I
shall call them 'Aristotelian objects', since they correspond to Aris-
totle'sfirst substances - can be absorbed by these higher order objects
in the following manner.
What I have hitherto simply called 'objects,' I shall from now on
call 'O-objects' (think of the word "object" as replaced by the word '0-
object' in the above principles and definitions of temporal counterpart
theory); certain sets of O-objects I will call' I-objects'. In the supere-
temalist's eyes, I, Uwe Meixner, and other Aristotelian objects are of
course not O-objects, but rather i-objects: each Aristotelian object, ac-
cording to the superetemalist, is a set of precisely the O-objects related
by the (simple, two-place) temporal counterpart relation to a certain 0-
object; it is, in other words (according to GTl), an equivalence set of
that relation: a i-object. The O-objects in the sets that are I-objects are
precisely the (momentary) temporal stages of the I-objects.
Since I-objects are set-theoretic constructions out of O-objects,
one will define the I-objects' having of properties in terms of the
having of properties ofO-objects. As follows:
DEF3: Let x be a I-object, fa property of objects, t a time: x has f
at t =def. there is an element y of x [O-object y in x] such that
y is at t a counterpart of y [y is located at t], and y has fat t
[and this means, since we are talking about a O-object, at
all times].
Proof
Let M be the set of the momentary temporal stages of an Aristote-
lian object, say, k. Consider some momentary temporal stage m of k
(there must be such stage, otherwise k would not be an Aristotelian
object). Since m is a momentary temporal stage of an Aristotelian ob-
ject, we have by (G8): m is a O-object.
(1) Assume Z is simpliciter a temporal counterpart of m; hence by
(G9): z is a momentary temporal stage of k, hence zEM.
(2) Assume zEM, hence Z is a momentary temporal stage of k; m is
also a momentary temporal stage of k; hence by (G I 0): Z is simpliciter
a temporal counterpart of m.
But GT3, the superetemalist continues, has the corollary
Proof
Assume M is the set of the momentary temporal stages of an Ar-
istotelian object. Hence by GT3: there is a O-objecty such that: Vz (z is
simpliciter a temporal counterpart of y iff ZEM). Hence there is a 0-
objecty such that {z: z is simpliciter a temporal counterpart ofy} = M.
Hence M is a I-object (according to the definition of I-objects).
Clearly, it is an immediate consequence of GT4 that Aristotelian
objects are I-objects if Aristotelian objects are each identified with the
set of their respective momentary temporal stages (and the supereter-
440 Individuals, Essence, and Identity
nalist has already argued that they can be identified with their respec-
tive sets of momentary temporal stages).
5. RESISTING REDUCTION?
this relation be governed by the ten principles given above (with the
possible exception of (G7». Let I-objects be accepted - these set-
theoretical constructions out of O-objects - and let Aristotelian objects
correspond one-to-one to certain I-objects. In a word, let us accept the
whole basis for reducing Aristotelian objects to I-objects. Are there
reasons, then, why we should resist reducing them to I-objects never-
theless, and reasons which are independent of considerations of the
possible ontological primacy of Aristotelian objects, an ontological
primacy which so many philosophers nowadays are unwilling to be-
lieve in?
If such reasons are not to be sought in considerations of the onto-
logical primacy of Aristotelian objects, then such reasons can only be
found in considering some important function Aristotelian objects can
perform, but not their reductive counterparts, the corresponding 1-
objects. At this point, it is again helpful to look at the modal analogy.
6. MODAL CONTINUANTS
that are not reducible to properties those objects have within or at pos-
sible worlds.
Consider the relation of realizing part of a possible world. This
relation can be either taken causally or cognitively; if it is taken caus-
ally, it means as much as making real or making actual; if it is taken
cognitively, it means as much as cognizing as real or cognizing as ac-
tual. Of course, one can say that a human person x, for example, real-
izes at possible world w part y of w; but the point is that one can say
this only because, and in the conceptual order after, one can simply
say that x realizes part y of w; the latter form of expression is the pri-
mary form, the former merely secondary and, as it were, epiphenome-
nal. Suppose, then, that person x realizes part y of world w; from this,
there follows not merely the triviality that y is actual at w, but that y is
actual simpliciter. Quite clearly, the following relational property of
person x, the property of realizing part y of w, is irreducible (1) to the
properties x has (properly speaking) at possible worlds other than w.
But it is also irreducible (2) to the properties x has (properly speaking)
at w, the reason being that what properties x has at w does not deter-
mine, not even partially, which of those properties are simpliciter ac-
tual or simpliciter real properties of x, while x's partially realizing w
does indeed partially determine this,16 no matter even whether we un-
derstand realizing in the sense of causally realizing or in the sense of
cognitively realizing.
The force of this argument is that possible objects that are true
(namely, literally reality-making) agents and true (namely, literally
reality-cognizing) cognizers have to be modal continuants. Not all ob-
jects, of course, can be plausibly held to be true agents and cognizers,
but persons, certainly, are traditionally conceived to be such. They do
their deeds and they gain factual knowledge, and thereby contribute to
determining what is simpliciter real. Therefore, what is in this world
or in another cannot completely determine what they are, since it does
not even contribute to determining what is simpliciter real. Persons
are, to put it in a word, world-transcendent. 17 And this means, in par-
ticular, that persons are over and above the properties they have at
possible worlds: they are modal continuants.
Hence, unless we want to give up our traditional way of under-
standing ourselves, a way that does not allow that doers and knowers
Uwe Meixner, Change and Change-Ersatz 445
be absorbed entirely by the field that their doing and knowing is di-
rected at, we have good reason to believe that there are modal contin-
uants, namely ourselves. I am, of course, aware of the fact that the
general philosophical climate in recent years is inimical to the tradi-
tional conception; what I would deny is that endorsing this climate, the
climate of naturalism, is the only rational option.
7. TEMPORAL CONTINUANTS
8. A METAPHYSICAL PICTURE
Notes
1 There also is a deeper reason, and certainly one that is more credibly a reason
oJLeibniz or Lewis, for their position. See section 6 below.
2 To my mind, Lewis' argument against literal trans-world identity, and mutatis
mutandis against literal trans-time identity, literal identity over time, (see Lewis
[1987], p.199ff, p.202ff, p.2l0) are far from convincing. The arguments have the
following two contestable presuppositions: (1) If literally the same object x existed
wholly at two worlds, respectively: two times, then, unacceptably, it would be a part
of both worlds, respectively times (or in other words: then both worlds, and both
times, would overlap in x). (2) If literally the same objecl x had, as a whole, an in-
trinsic property Jat a world w [time I], and not at a world w' [time 1'], thenJwould,
448 Individuals, Essence, and Identity
15 This idea is clearly visible in the following passage from an outstanding su-
pereternalist: "[O]ur ideas of bodies are nothing but collections formed by the mind
of the ideas of the several distinct sensible qualities, of which objects are composed,
and which we find to have a constant union with each other. But however these
qualities may in themselves be entirely distinct, it is certain we commonly regard the
compound, which they form, as one thing, and as continuing the same under very
considerable alterations. [... ] The smooth and uninterrupted progress of the thought,
Uwe Meixner, Change and Change-Ersatz 449
being alike in both cases [that of a "succession of related qualities," and that of "one
continued object, existing without any variation"], readily deceives the mind, and
makes us ascribe an identity to the changeable succession of connected qualities."
(David Hume, A Treatise of Human Nature, Book I, Part III, Section III, p.270 of
MacNabb (ed.) [1987].) In the quoted passage, we also find, very apparent, the
blending of numerical with qualitative identity (over time), which is a consequence
of the mentioned idea and which makes Hume implicitly subscribe to the following
superetemalistic principle: if object x has at moment t other qualities than object y
has at moment t', then x*- y. (Note that if 't" is replaced by 't', we have a principle
that is entirely uncontroversial.)
16 To make this more palpable, consider a typical way x qua human person is
involved with world w. For example: x lifts his hand in w. Now, x's partially (caus-
ally) realizing w may well tum this into: x (simpliciter really) lifts his hand (and
thus: lifting his hand is a simpliciter actual property of x). If it does so, then x is
causally responsible (in an absolute sense) for the (simpliciter real) lifting of his
hand; if it does not, x is not causally responsible for it. (The talk of responsibility
here must not blind one to the fact that the crucial point in the argument is not con-
tingency, or freedom of the will, but the imparting of reality or actuality in an abso-
lute sense: world-bound individuals are not up to this.)
17 The world-transcendence of persons and other substances is argued for in
much greater detail in Meixner [1997a].
18 It would be implausible to hold that all Aristotelian objects are irreducible. In
fact, there is no reason to hold that inanimate Aristotelian objects are irreducible.
19 Note the contrast in content between the sentences 'tl is present at II' and 'II
is present' (the first sentence is true at all time, the second only at td. It is also im-
portant for the argument to realize that the latter sentence cannot be synonymous
with the sentence 'II is the time (or belongs to the time) of Ihis utterance'. The rea-
sons for this are: (1) The moment of time th having no extension, cannot be the time
of an utterance (and if it belongs to the time of an ongoing utterance of 'II belongs to
the time of this utterance' or of' II is present', it may nevertheless not be presenl, but,
instead, a later or earlier moment of time). (2) There might not be any utterance at
(incorporating) II at all, while II is nevertheless presenl.
20 In discussion, Kit Fine offered the following illuminating analogy: the build-
ers of a house (realizers of a world) cannot be parts of it. I would merely add: nor can
they be parts of the totality of all possible houses that might be built.
STARTING OVER
Christopher Hughes
1.
2.
Olson holds that (i) I am essentially this animal; and (ii) this animal
essentially has this brainstem. (Compare this to the more familiar
Cartesian view that I am essentially this thinker, and this thinker es-
sentially has this mind). Given (i) and (ii), when my brainstem is de-
stroyed, I am destroyed with it; when a new (duplicate) brainstem is
put into my corpse, the resulting person cannot be me, but only my
duplicate. Doubts could be raised about both (i) and (ii), but I shall not
pursue them here. 6 Instead, I shall modify Olson's story slightly.
Suppose a surgeon is carrying out more or less the procedure de-
scribed by Olson. First, she removes my original brain stem. Next, she
gets the duplicate brainstem out of storage, and prepares to hook it up
to the rest of my organs. Unfortunately, in attempting to get the dupli-
cate brainstem from its storage compartment to my body, she drops it,
damaging it severely. Duplicate brain stems are expensive, and the
surgeon does not have a "backup" one. Luckily, she still has my origi-
nal brain stem, which has not yet become damaged. She immediately
puts it back into me, and normal metabolic service is resumed. Be-
cause my body-except-for-my-brainstem was a corpse for such a short
454 Individuals, Essence, and Identity
Is this consistent with (NTB)? I imagine Olson would say it is. Ol-
son (somewhat tentatively) suggests that if an animal has its head sev-
ered, the animal would not go on existing as a headless body, but
would go on existing (for a short time, at least) as a severed head. As
he sees it, this is because the severed head contains the parts of the
brain that direct and coordinate vital functions, and does not, upon
being severed, immediately lose its capacity to direct and coordinate
those functions. 7 I gather from Olson that the parts of the brain that co-
ordinate vital functions are in the brain stem. So, I take it, Olson would
think that if my brain stem is removed from the rest of my body, I go
on existing as a brain stem, for at least as long as the brain stem retains
its capacity to direct and coordinate vital functions. Thus Olson could
say the identity of the person at the end of the botched operation with
the person at the beginning of that operation is compatible with (NTB).
Before the operation, I was a normal human animal with a whole
brain, head, arms, legs, etc. When my brain stem is separated from the
rest of my body, I become a mutilated animal, all of whose parts are
parts of my brain stem. When the brain stem is reattached to what it
was separated from, I become a normal human animal again, with a
normal set of organs.
I have various reservations about this. To start with, supposing
that an animal could survive being "pared down to" the dimensions
(and constitution) of a brainstem, seems rather like supposing that the
central heating system in our house could survive being "pared down
to" the dimensions (and constitution) of a thermostat, or supposing that
a car could be pared down to the dimensions (and constitution) of a
motor. If there is nothing more to a thing than a brainstem, it is hard to
believe that thing could be a (complete, albeit mutilated) animal, rather
than just an (ex) part of an animal. s If (as Olson supposes) I am essen-
tially a (human) animal, and nothing with the constitution of a brain-
stem is a (human) animal, then I cannot survive the separation of my
brainstem from the other parts of me as a brainstem.
But suppose we grant that, when my brainstem is separated from
the rest of me, I am nothing over above my separated brainstem9 • It
seems that we can still tell the story in such a way that I am no longer
living at some time between separation and reattachment. Suppose that
after the brainstem is removed, it momentarily ceases to function, and
456 Individuals, Essence, and Identity
cal level" at all times between its removal and its reattachment? Why
suppose any that any processes at all need to go on throughout that in-
terval? Suppose that, the time between the removal of my brainstem
and its reattachment is one year and a few minutes. During the period
of a year, my quadrant of the universe undergoes what Sydney Shoe-
maker calls a "local freeze", in which all processes (even subchemical
and subatomic ones) stop. Local freezes are, I take it, physically im-
possible. But my concern is with (NTB), which says that nothing that
exists or might have existed could come back from non-existence. If
local freezes are (metaphysically) possible, and if it is necessary that
whatever ceases to live, ceases to be, then (NTB) is false.
Alternatively, suppose that medical technicians could "take apart"
my brainstem, breaking it down to individual cells. And suppose that
before the individual cells died, the technicians could put all the cells
back together in exactly the way they were put together before the
brainstem was taken apart. I find it very doubtful that my life would go
on during the period that each of the cells that had composed my
brainstem were living their own separate lives. (It has often been sug-
gested that my life was not going on before the primitive streak stage
of my fetal development, precisely because the cells in the clump of
cells whence I came were not then cooperating in a single life, but
living their own separate lives). Still, if the technicians could separate
my brain stem from the rest of me, take it apart, put it back together,
and reattach it to the rest of my body quickly enough, the person ex-
isting at the end of the procedure would be me. Suppose that doctors
knew that I had a small number of cancerous cells in my brain stem,
and that the only way they could discover which cells were cancerous
was to take my brainstem completely apart, and examine the cells that
had constituted it one by one, destroying the (few) cancerous cells,
"reassembling" the others, and then reattaching the reassembled brain
stem to the rest of my body (before it had succumbed in any signifi-
cant way to entropy). If the procedure were medically feasible, I would
not refuse it on the grounds that it could not save my life, because the
person existing at the end of it would be a mere duplicate of me.
If this is right, then someone who wants to endorse both the Aris-
totelian principle that nothing could go on longer than its life and
(NTB) is committed to counterintuitive judgments about what sorts of
458 Individuals, Essence, and Identity
longer living animal, and finally as a reassembled and once again living
animal.
3.
Suppose I drop a cup on a hard floor, and it breaks into many pieces. If
I painstakingly glue the pieces together (in the right way), I'll end up
with a cup. Moreover, it seems that I'll end up with the cup I dropped
and broke. Suppose that last year you lent me the cup I broke into
many pieces yesterday, and glued back together today. You knock on
my door today, and say, "I lent you a blue cup last year. Could I have
it back?" I could hand it to you, and say, truthfully, "Here it is. I'm
very sorry; I dropped it on the kitchen floor yesterday, and I glued the
pieces back together as best I could". I have returned your cup, though
not in the condition in which you loaned it to me.
Suppose, though, that instead of glueing the pieces back together;
I had simply collected them, put them in a box, and handed them to
you. Then, I'm inclined to say, although you lent me a cup, I didn't
return it; I only returned the (ex) bits of it. True, you might say: "I
loaned him a beautiful cup, and he returned it in a thousand pieces."
But, then, in a horror movie, a character might say: "Jones is in the
dustbin in a thousand pieces". What the dustbin contains is surely not
Jones, but things that used to be parts of Jones. Similarly, I lean to-
wards saying, what the box you handed me contains is not my cup, but
bits of china that once were and will again be parts of a cup.
Here is a different case, about which I have somewhat stronger
intuitions. Christ predicted the destruction of the Temple, and it came
460 Individuals, Essence, and Identity
to pass. I take it that when the Romans destroyed the Temple, it went
out of existence: it didn't survive its destruction. In this context, there
is some interest in the etymology of the Latin verb destruere, whence
the English terms 'destroy' and 'destruction' come. Destruere means
"un-pile", just as construere means "pile together"; the destruction of a
building consists in putting asunder the stones it is made of, just as the
construction of a building consists in putting together the stones it is
made of. Even in Latin, though, it had something like the wider sense
that "destroy" has in English (one can destroy something by causing it
to cease to exist in all sorts of ways - e.g. by burning it). I presume
that destruere went from meaning something like "demolish" to
meaning something like "destroy" in the current sense because de-
molishing a house (or a temple) destroys it.
When the Romans destroyed the temple, they did not do anything
as drastic as vaporizing each of the stones it was made of (a good bit
of Temple wall is still intact today). Suppose that after the Romans had
knocked down most of the Temple stones, and broken some of those
stones into fragments, the Emperor had had a vision in which God told
him that the destruction of the temple was an abomination in His sight,
and that amends had to be made. Suppose that the emperor had imme-
diately sent a team of his best builders to Jerusalem: their job would be
to find the (now partially dispersed) Temple stones, "glue" them back
together if necessary, and then put them back together in the way they
had been put together before the Romans destroyed the Temple. That
way the Temple the Romans had destroyed would be rebuilt -
"com 'era, dov'era", as Venetians would later say in a different con-
text. If all this had happened, then the building made of the re-
assembled stones would be the building that the Romans destroyed,
just as the cup made of the re-assembled fragments of china is the cup
I dropped and broke.
Examples of this sort could be multiplied. They trade on the fact
that in thinking about the persistence and identity of composita we
have two tendencies. First, we have a tendency to think of separating a
thing's parts from each other as prejudicial to its continued existence. 1I
Of course, it is not a hard and fast rule that separating a thing's parts
causes it to go out of existence: whether or not a certain sort of parts-
separation will result in a thing's non-existence will depend both on
Christopher Hughes, Starting Over 461
what kind of thing it is, and on what kind of parts they are. When you
take a tent apart (in the usual way), and put the canvas on one shelf,
and the poles on another, it seems that you end up with a ("partially
scattered") disassembled tent, rather than a bunch of ex parts-of-a-tent.
(Suppose that you loaned me your tent in an assembled state, driving it
over on the back of a pickup truck. If you ask for your tent back, I
comply with your request, even if I give it back in a disassembled
state). On the other hand, if you separate the parts of a 500 Lire coin
from each other (in a throughgoing way), you end up with with some
bits of metal that used to be parts of a coin, rather than a
"disassembled coin". Suppose, though, you "take a tent apart" by cut-
ting the tent canvas into many small pieces. Then, I think, the tent does
not survive. (It is similar to a case in which I take an old T-shirt, and
cut it up into lots of small strips to use as cleaning rags. That surely is
enough to spell the demise of my shirt). Like a coin, a tent is not ame-
nable to certain sorts of parts-separation. (The only sorts of things I
can think of that are amenable to any sort of parts-separation are ag-
gregates of simples).
We also have a tendency to think that if you put the same parts
back together in the way they were put together when they constituted
a K, you get a K back. Again, I do not claim that this is hard and fast
rule (it may "go soft" in certain ship of Theseus cases). But we tend to
think that ceteris paribus if you put the original parts back together the
right way, you get the original engine back; if you put the original bits
of china back together the right way, you get the original teacup back,
and so on. Another example of this way of thinking is found in early
Christian thinking about the afterlife.
Early Christians appear to have thought of the afterlife (often - I
do not mean exclusively) in terms of a bodily resurrection, conceived
as the resuscitation of a continuously existing but temporarily ex-
animate corpse. But they were aware that things could not always
work that way: corpses were sometimes devoured by animals, or
burned, or the like. How could those whose corpses had been de-
stroyed rise again? The Church Fathers (often) answered by appealing
to God's ability to "re-assemble" the continuously existing bits that
had composed a person's living body, and subsequently his corpse,
before the destruction of the latter. Thus Irenaeus said that if God
462 Individuals, Essence, and Identity
could create the first man and his body, He would have no trouble re-
making our bodies, by putting back together the bits of the body that
had decomposed. And Athenagoras argued that since God knows
where all the particles that used to compose a person's body have
gone, and knows just how to put them back together in order to a get a
person's body back, there is no difficulty about God's resurrecting
those whose corpses have decomposed. This conception of a bodily
resurrection via the reassembly of surviving parts of a decomposed
corpse appears to have antecedents in various (Jewish and Christian)
apocalypses. 12 In both cases, we have the assumption that if God puts
together (even very small) ex parts-of-our-bodies together in the right
way, his reassembling those parts is also his reassembling our bodies.
Of course, someone might say that early Christians and the Church
fathers had an axe to grind: they needed to make room for a bodily
resurrection on the last day, and saw no way of doing so without sup-
posing (counterintuitively) that reassembly could get someone's body
back. It is interesting to note, though, that Lucretius also appears to
have thought that reassembly of a person's long dispersed atoms could
get that person (and his body) back (cf. De Rerum Natura, Book III,
847-851). Given that Lucretius wants his readers to stop worrying
about death and what might lie beyond it, it is rather awkward for Lu-
cretius to grant that (the right sort of) reassembly of atoms might hap-
pen, and, if it did happen, would bring me back.
Some philosophers - among them van Inwagen and Olson -
would be unhappy with the idea that our resurrection could be under-
stood simply in terms of God's reassembling all the dispersed bits of
me in the right way on the last day. Van Inwagen and Olson accept the
Lockean idea that if an organism exists at a certain moment, it exists
whenever and wherever - and only when and only where - its life is
occurring.13 And they think that, whether or not a suspended life can
resume, a life that has been disrupted cannot. 14 So, they would say,
unless my life goes on (possibly in a "squeezed down" state) or is non-
disruptively suspended, God could not get me back on the last day, be-
cause He could not start my life again, whatever He did in the way of
reassembly of the things that used to be parts of me. At best, He could
produce a replica of me.
I do not see this. The teacup that was shattered into ever so many
Christopher Hughes, Starting Over 463
fragments is the teacup that has now been painstakingly glued back
together. Suppose a dog steps on a powerful mine. Small bits of its
body fly out at high speed in all directions. Then God works a miracle,
putting all the bits of the dog's body (and the dog) back together in just
the way they were just before he stepped on the mine. (If you were
watching the miracle, it would look like what you would see if you ran
a video-tape of an explosion backwards). Why is not the dog whose
parts were reassembled the reassembled dog, just as the cup whose
parts were reassembled is the reassembled cup? The reassemblies are
very different, but does that matter? After all, suppose that it had been
a teacup, rather than a dog, that had been exploded, and God had mi-
raculously put its little bits back together (in the right way). Why
wouldn't we have a miraculously reassembled teacup? Similarly, why
wouldn't we have a miraculously reassembled dog? A defender of van
Inwagen's view might protest that it just is intuitively plausible that,
even if the suspension of an organism's life does not mean that organ-
ism is gone for good, the disruption of its life does. Again, I do not see
this. We have a continuum of cases in the organism ceases to live a
normal life, and it gets harder and harder to get the parts of the organ-
ism that took part in that life back in the same state they were in before
the organism stopped living a normal life. At the easy end of the spec-
trum, we have van Inwagen's frozen organisms, that either go on liv-
ing a "squeezed down" subchemical life, or have their life suspended.
If we start "disassembling" the organism - removing its brainstem, say
- it gets harder to get the parts back in the state they were in just be-
fore disassembly, but it may still be feasible. If we thoroughly disas-
semble the organism - right down to individual cells - I imagine it is
infeasible to get the parts in the state they were in just before disas-
sembly (at least so long as the organism is as complicated as a full-
grown dog). But it may be feasible someday, and if you could do it,
the organism you had at the end of the process of cellular reassembly
would be the organism you had before disassembly. If the organism is
broken up into small enough (non-functional) parts, as happens in the
case of the dog who steps on the mine, then it would take a miracle to
get the dog's parts back in the state they were in just before he stepped
on the bomb. But if God worked that miracle, I do not see why the dog
reassembled from exploded dog parts wouldn't be the dog whose life
464 Individuals, Essence, and Identity
4.
usual way? Why not suppose that we are counting by identity, and that
when the parts of a bicycle are dispersed in a certain way - when they
are, as EJ. Lowe puts it, "appropriated" by other bicycles - that bicy-
cle ceases to exist?
A rather different strategy for defending (NTB) from the alleged
counterexample would turn on the idea that 'bicycle' is a phase-sortal
for a kind of thing that ceases to be a bicycle, but does not cease to be,
when its parts are dispersed and appropriated. On one way of devel-
oping this suggestion, 'bicycle' would be a phase-sortal for a set of
bicycle parts. The idea would be that there are four (overlapping) sets
of bicycle parts -A, B, C, and D - in our story. Anyone of those sets
of bicycle parts is a bicycle when and only when its parts are related to
each other in the right way. 18 Nothing I do to the parts of A, B, C, and
D causes anything to go out of or come into existence. It merely causes
this or that individual to become or cease to be a bicycle. I have only
two bicycles at the beginning, in the middle, and at the end of the
story; and the bicycles I have at the beginning and at the end (A and B)
are different from the bicycles I have in the middle of the story (C and
D). But none of this is a threat to (NTB).
Again, I would not want to reject this suggestion out of hand. But I
do not see how to make it work. I doubt that 'bicycle' can be a phase-
sortal for a set of bicycle parts: if it were, bicycles could not survive
the replacement of most or all of their parts. But, waiving this objec-
tion, champions of this way of defending (NTB) are going to have
treat a whole of sortals as phase-sortals. For we can tell the sort of
story we told about bicycles about lots of other things. Suppose I have
a set of fifty cups. All of them become slightly damaged, losing a
(small) chip here or a (small) chip there. I collect all the chips and glue
them together, so that I now have fifty-one cups. Then I break the
fifty-first cup into the fragments I made it from, and glue each frag-
ment back onto the cup it originally belonged to. I end up with only
fifty cups. On the strategy for defending (NTB) under consideration, at
the end of the story my fifty-first cup still exists: although it has ceased
to be a cup, it hasn't ceased to be. I find this very difficult to believe.
Suppose a visitor says to me: "Last time I was here, I drank tea out of
a really strange looking cup. Where is it?". I do not think I could prop-
erly say: "Here, and here, and here .... " [pointing to all the bits in my
Christopher Hughes, Starting Over 467
fifty cups]. What I am pointing to is not it (that cup you drank tea out
of), but a bunch of bits of china that used to but no longer constitute it.
(Compare this to the case in which a visitor says to you: "Last time I
was here [ten years ago] I met this really difficult teenager. Where is
she?" I could with perfect propriety say "Here she is", pointing to my
- no longer teenage - daughter).
Again, suppose a frog has a billion cells, and I have got a billion
frogs. I take one cell from each frog (making sure to pick the right
kind of cells), and then put them together in such a way that they are
taking part in the same frog-life. I now have a billion and one frogs.
Suppose I subsequently take apart the cells I put together, and put each
of them back in the frog it originally came from, leaving me with just
one billion frogs. On the envisaged defense of (NTB), we would deny
that my going back to the status quo ex ante caused any frog to go out
of existence: it just caused a frog to cease to be a frog. Again, though,
after all the billion cells have been put back in their original frogs, is it
really true that the frog different from those billion frogs is still around
- though it is no longer a frog - in the same way that a sapling planted
by the Thames twenty years ago is still around, although it's no longer
a sapling? I do not think so. In sum: if we block bicycle-type counter-
examples to (NTB) by appeal to the idea that the relevant sortals are
phase-sortals we will end up making it too hard for things (cups, frogs,
and so on) to go out of existence. 19
5.
is not the body you have now. Moreover, none of the cases we have
discussed so far are genuine cases of a thing's going out of existence
and subsequently coming back. (If the thing x (frog, cup, temple, bicy-
cle ... ) existing at { went out of existence, and subsequently came back
into existence at {' , there would not be any state S that the thing was in
at that later time whose causal history was x-involving from { right
through to I').
An argument for (NTB) from the causality-persistence principle
just formulated is no more compelling than the principle itself. And the
principle does not look immediately compelling to me. Upon consid-
ering it, my first thought is that it is neither obviously true nor obvi-
ously false; my second thought is to test it against my intuitions con-
cerning various hypothetical cases (e.g. those involving taking things
apart and putting them back together); and my third thought, after
having done that, is to reckon that it is false. More generally, my sus-
picion is that attempts to argue for (NTB) will either depend on princi-
ples too weak to support (NTB), or depend on principles too strong for
us to put any confidence in.
But the main point I want to make here about the causality-
persistence principle under consideration is one lowe to Dean Zim-
merman. Zimmerman points out that someone might accept that a
thing can persist only if the stages of its career are held together by the
glue of (what he calls) immanent causality relations, and still insist on
the possibility of interrupted existence (for living or non-living things):
On the face of it, immanent-causal relatedness among stages of a thing
would seem to be compatible with its making discontinuous spatiotempo-
ral jumps, or even being "temporally gappy." If it is possible for an object
to persist through temporal gaps during which it has no stages, then there
must be suitable immanent-causal relations which cross the temporal gap
between earlier and later stages. But, given that the kind of immanent-
causal connections that normally preserve a Life could cross spatial and
temporal gaps, there is no reason to think that one and the same Life could
not contain spatial jumps or temporal gaps. (Zimmerman [1998], p.384)
6.
7.
Is it true that, as the Scholastics would put it, nihil polesl redire
(nothing can come back)? Is something rather like it true? Perhaps; the
Christopher Hughes, Starting Over 473
Notes
I Aquinas attributes the view to Aristotle, and accepts a somwehat weakened
version of it himself (see his Commentary on the Sentences, IV, 1, 1, 1). I have often
heard the view attributed to Locke, though I have not been able to find a passage
where Locke explicitly endorses it. I first began to think about the principle years
ago when Anil Gupta suggested to me that it was one of the relatively few Lockean
principles that was true.
2 Fred Feldman has vigorously defended this view in his Confrontations with
the Reaper (cf. Feldman [1992]).
3 See my forthcoming "On the Real (?) Distinction between Persons and Their
Bodies".
4 Eric Olson expresses sympathy for and attempts to motivate this (Aristotelian)
view in Olson [1997], pp.150-52.
5 Olson says that in the Cartesian demon case, "the interruption of your biologi-
cal life is brief." I take it he cannot mean this. For he seems to think that a life cannot
outlast the organism living it (p.l37), in which case the demon who annihilates me
and replaces me with a perfect duplicate a thousandth of a second later terminates
my life, rather than interrupting it.
6 For more on animalism, and Olson's particular version of animalism, see my
[2001].
7 Cf. Olson [1997], p.133.
8 Olson's views appear to imply that if you separate my brainstem from the rest
of me (without destroying either), and separate Daniele Giaretta's brain stem from
the rest of him (without destroying either), and then transplant my brainstem into the
rest of Daniele, and Daniele's brainstem into the rest of me, at the end of the process,
the animal with almost of all my body and my mind will be not me but Daniele, and
the animal with almost all of Daniele's body and his mind will be not Daniele but
me. I find this very hard to believe.
9 I say, "I am nothing over and above my brainstem" rather than "I become my
brainstem", because, after the separation, one could truly say about me, but not of
my brainstem, "that's something that used to weigh about one hundred and forty five
pounds."
10 Van Inwagen suggests that the life of a frozen cat might be "squeezed into"
11 Of course, if you separate just a tiny part of something from most of its parts,
the thing usually goes on existing without the bit separated from it (e.g. as a chipped
cup); it's a throughgoing separation of part from part that is prejudicial to a thing's
continued existence.
12 See, for example, the Apocalypse of Peter 4: 3-4, "He will command the
beasts and the birds; He will command that they give back all the flesh they have
eaten, because He requires humans to make their appearance." (see Bauckman
[1988], p.272).
13 See van Inwagen [1990], p.145.
14 Ibid., p.147: "We may be confident that the life of an organism which has
been blown to bits by a bomb or which has died naturally and has been subject to the
normal, "room-temperature" processes of biological decay for, say, fifteen minutes
has been disrupted. [ ... ] Ifa life has been disrupted, it can never begin again; any life
that is going on after its disruption is not that life."
15 I briefly discuss the case I am describing in Hughes [1997].
16 See Lewis [1976].
he calls immanent causality between earlier and later states of the persisting object, I
take it he is endorsing a principle in the neighborhood of the one formulated here.
See his remarks on the persistence of a body: "To say that immanent causal connec-
tions are required for the persistence of a body is to say that later states of the body
must be causally dependent, at least in part, on its earlier states. But not just any sort
of causal dependence seems sufficient to give us the kind of immanent causation that
is crucial to the persistence of a body. It is not enough [... ] that the way my body
was at death serve as a blueprint for God's creating a new one at the general resur-
rection. That is causal contribution of a sort; but here the causal chain passes through
God's mind; it does not remain "immanent" with respect to processes going on
within a living human body." (Zimmerman [1998]). As we shall see, Zimmerman
does not argue from the necessity of immanent causality for persistence, to the im-
possibility of uninterrupted existence.
21 Shoemaker [1979].
22 For an attack on the view that identity across time could not be primitive, see
Saul Kripke's unpublished lectures, "Time and Identity."
Christopher Hughes, Starting Over 475
French, P., Uehling, T., and Wettstein, H. (eds.), 1986, Midwest Stud-
ies in Philosophy XI: Studies in Essentialism, Minneapolis: Uni-
versity of Minnesota Press.
Francescotti, R., 1999, "How to Define Intrinsic Properties", Noils 33,
590-609.
Freddoso, M. and Schuurman R. (eds.), 1980, Ockham, W., Ockham's
Theory of Propositions, Notre Dame (IN): University of Notre
dame Press.
Gaifinan, H., 1975, "Ontology and Conceptual Frameworks", Erk-
enntnis 9, 329-353.
Gallois, A., 1998, Occasions of Identity. The MetaphysiCS of Persis-
tence, Change, and Sameness, Oxford: Clarendon Press.
Garrett, B., 1988, "Best-Candidate Theories and Identity", Inquiry, 79-
85.
Geach, P., 1972, "Some Problems about Time" in Geach [1972], 302-
317.
Geach P.T., 1972, Logic Matters, Oxford: Blackwell
Geach, P.T., 1980, Reference and Generality. An Examination of some
Medieval and Modern Theories, Ithaca: Cornell University Press.
Gerhardt, C.I. (ed.), 1978, Die Philosophischen Schriften von G. W
Leibniz~ Olms: Hildesheim.
Gerhardt C.L (ed.), 1962, Leibnizen's Mathematische Schriften,
Hildesheim: Olms.
Gibbard, A., 1975, "Contingent Identity", Journal of Philosophical
Logic 4, 187-221.
Goldman, A. L., 1971, "The Individuation of Action", Journal of Phi-
losophy 68, 761-774.
Grice, P., 1988, "Aristotle on the MUltiplicity of Being", Pacific
Philosophical Quarterly 69, 175-200.
Grua G., 1953, Jurisprudence universe lie et theodicee selon Leibniz,
Paris: PUF.
Hacker, P.M.S., 1982, "Events, Ontology and Grammar", Philosophy
57,477-486.
Hale, B. and Wright C., 1996, "Nominalism and the Contingency of
Abstract Objects", in M. Schirn, (ed.) Frege: Importance and
Legacy, Berlin: de-Gruyter.
486 Individuals, Essence, and Identity
Jolley, N., 1984, Leibniz and Locke: A Study of the «New Essays on
Human Understanding», Oxford: Clarendon Press.
Kamp, H., 1975, "Two Theories about adjectives", in E. Keenan (ed.),
Formal Semantics of Natural Language, Cambridge: Cambridge
University Press, 123-155.
Kaplan D., 1969, "Quantifying in", in D. Davidson and J. Hintikka
(eds.), Words and Objections: Essays on the Work of wv.o.
Quine, Dordrecht: Reidel, 178-214.
Kim, J., 1993, Supervenience and Mind, Cambridge: Cambridge Uni-
versity Press.
Kripke, S., 1972, "Naming and Necessity", in D. Davidson and G.
Harman (eds.), Semantics of Natural Language, Dordrecht: Rei-
del,252-355.
Kripke, S., 1980, Naming and Necessity, Oxford: Blackwell.
Laks A. and Most G. (eds.), 1993, Theophrastus, Metaphysique, Paris:
Les Belles Lettres.
Lambert, K., 1983, Meinong and the Principle of Independence, Cam-
bridge: Cambridge University Press.
Lejewski, c., 1954, "Logic and Existence", British Journal for the
Philosophy of Science 5, 104-119.
Lejewski, C., 1958, "On Lesniewski's Ontology", Ratio 1, 150-176.
Leszl, W., 1970, Logic and MetaphysiCS in Aristotle, Padua: Ante-
nore.
Lewis, D. K., and Lewis, S. R., 1970, "Holes", Australasian Journal of
Philosophy 48,206-212 (as reprinted in Lewis [1983a], 3-9).
Lewis D., 1976, "Survival and Identity", in A. Oksenberg Rorty (ed.),
The Identities of Persons, Berkeley: California University Press,
1976, 17-40.
Lewis, D., 1983, "In Defense of Stages", appendix b to Lewis [1976]
in Lewis [1983a], 76-77.
Lewis, D., 1983a, Philosophical Papers, Vol. 1, New York: Oxford
University Press.
Lewis, D., 1986, On the Plurality of Worlds, Oxford: Blackwell.
Lewis, D., 1986a, "Events", in Philosophical Papers, vol. 2, Oxford:
Oxford University Press, 241-269.
Lewis, D., 1988, "Vague identity: Evans misunderstood", Analysis 48,
128-130.
488 Individuals, Essence, and Identity
Marconi, D., 1979, "Le Ambigue VirtU della Forma Logica", in Tempo
verbale e strutture quantificate in forma logica, Atti del Semina-
rio, Firenze: Accademia della Crusca, 265-284.
Marcus, R.B., 1962, "Interpreting Quantification", in Inquiry 5, 252-
259.
Martin, C.B., 1993, "Power for Realists", in Bacon et al. (eds.) [1993]
175-86.
Mates, B., 1971, "Leibniz on Possible Worlds", in Frankfurt H. (ed.),
Leibniz, New York: Dobulbeday.
Mates, B., 1972, Elementary Logic, New York: Oxford University
Press.
Matthews, G.B., 1973, "Suppositio and Quantification in Ockham",
Nous 7, 13-24.
McGee, V., and McLaughlin, B., 1995, "Distinctions without a differ-
ence", Southern Journal ofPhilosophy 33, 203-251.
McGinn, C., 1976, "On the Necessity of Origin", The Journal of Phi-
losophy 73, 127-35.
McKay, T. J., 1986, "Against Constitutional Sufficiency Principles",
in French, Uehling and Wettstein (eds.) [1986], 295-304.
Meixner, U., 1997, Axiomatic Formal Ontology, Dordrecht: Kluwer.
Meixner, U., 1997a, Ereignis und Substanz, Paderborn: SchOningh.
Melia, J., 1995, "On What There's Not", Analysis 55, 223-229.
Miller, F.D.M., 1973, "Did Aristotle Have the Concept of Identity?",
The Philosophical Review 82, 483-90.
Moravcsik, J.M.E. (ed.) 1968, Aristotle. A Collection of Critical Es-
says, London-Melbourne: Macmillan.
Munitz, M. (ed.), 1973, Logic and Ontology, New York: New York
University Press.
Neale, S., 1990, Descriptions, Cambridge (MA): MIT Press.
Nidditch, P. (ed.), 1975, John Locke, An Essay Concerning Human
Understanding, Oxford: Clarendon Press.
Oliver, A., 1999, "A Few More Remarks on Logical Form", Proceed-
ings of the Aristotelian Society 99,247-272.
Olson, E., 1997, The Human Animal, Oxford: Oxford University Press.
Orenstein, A., 1978, Existence and the Particular Quantifier, Phila-
delphia: Temple University Press.
490 Individuals, Essence, and Identity