You are on page 1of 41

Substance, Identity and Time

Author(s): E. J. Lowe and Harold W. Noonan


Source: Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 62 (1988), pp.
61-100
Published by: Wiley on behalf of The Aristotelian Society
Stable URL: http://www.jstor.org/stable/4106761 .
Accessed: 28/10/2014 17:31
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley and The Aristotelian Society are collaborating with JSTOR to digitize, preserve and extend access to
Proceedings of the Aristotelian Society, Supplementary Volumes.

http://www.jstor.org

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME


E. J. Lowe and Harold W. Noonan
I-E.

J. Lowe
I

A tomato is sitting on the table. It has been sitting there for the
past five minutes. But what makes the tomato that is now sitting
there the same tomato as the one that was sitting there five
minutes ago? A crazy-sounding question! One is inclined to
reply: 'Nothing makesit the same tomato: it just is the same, and
there's an end on it'. If pressed a little further, however, one
might continue: 'Look, five minutes ago a tomato was sitting on
the table. It persistedthere, undisturbed, for five minutes. At no
time was the tomato removed from the table and replaced by
another. And that is why the very same tomato is still there now,
five minutes later'. But in virtue of what did that original tomato
persist- what kept it, that very same tomato, in being? Again a
crazy-sounding question. Surely one is not called upon to
explain why something like a tomato should continue to exist
from one moment to the next? One may indeed be called upon
to explain the coming-to-be
or the ceasing-to-be
of a tomato, but
Isn't the request for an explanation
surely not its continuing-to-be.
of the tomato's persistence rather like a request for an
explanation of an object's continuing to move with a uniform
velocity when not acted upon by any force?Perhaps we might
speak by analogy of a 'law of existential inertia'.' I think there is
something sound in this no-nonsense response-but I also think
that quite a lot of work needs to be done to earn a right to use it.
(And even then, I do not consider that the response, in unqualified form, is appropriate in the case of things like tomatoes.)
So far a number of subtleties have been glossed over. For one
thing, there is a distinction to be made between explaininga
tomato's persistence and saying what that persistence-the
tomato's 'diachronic identity'-consists in. For another, it won't
do just to say that the persistenceof something like a tomato calls
In fact, modern physicsdoesembrace what seems to amount to just such a law, in the
form of the law of the conservation of mass/energy (though, obviously, one could hardly
appeal to this law directlyin the case of something as complex as a tomato).

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

62

I-E. J. LOWE

for no explanation, without saying in some detail what exactly it


takes to be something like this. For, of course, if we change the
example to one concerning a very differentsort of object-say, a
trumpet-blast--then the questions which earlier sounded crazy
transforminto quite sensible ones. But, then, what exactly is the
relevant differencebetween a tomato and a trumpet-blast?The
tomato, one might want to say, is a continuant--or,if one is oldfashioned, a substance-while the trumpet-blast is aprocess.Well
and good, but this isjust fancy labelling. What is it forsomething
to be a 'substance'? A quick response would be: it is to be
something whose coming-to-be or ceasing-to-be calls for an
explanation (these being 'substantial changes'), but whose
continuing-to-be does not. But this responsewould be tooquick.
For it is far from clear yet that by this definition a tomato-or
indeed anything-will qualify as a substance.
Let us return to the distinction between explaining a thing's
persistence and saying what its persistence, or diachronic
identity, 'consists in'. I take it that what is involved here is the
quite general distinctionbetween providinga causal explanation
of the occurrenceof a phenomenon and saying in some revealing
way what that phenomenon really is-disclosing its 'real
essence'. (Comparethe distinctionbetween sayingwhy lightning
occurs and saying that it is an electrical discharge.) However,
even granting this important distinction, it seems clear that the
two sorts of concern will be intimately related. In particular, it
may well be urged that if no explanation of a thing's persistence
could be forthcoming this might be precisely because its
persistence could not be revealed to consist in anything
independently understandable. And, indeed, I strongly suspect
that this is how matters stand with respect to someof the things
we are inclined to call substances--though not, I think, with
respect to tomatoes. However, in what follows I shall leave aside
questions concerning the explanation of persistencein favour of
questions concerning the 'essence' of persistence.
To be in a position to say in what a thing's persistenceconsists
is, in more familiar terminology, just to be able to supply a
criterion(or 'principle')of diachronic identity for that thing, and,
more generally, for things of its sort. (In saying this I presuppose
that a 'criterion' in the presentcontext is not to be understoodas
an evidentialor heuristicprinciple, but rather as a metaphysical-

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

63

cum-semantic one.) In these terms, then, what I wish to


maintain is that there are some sorts of things (not,however,
including tomatoes) for which no such criterion of identity can
be supplied. In the case of things of these sorts, identity through
time is primitiveor ungrounded
(as I shall put it). I believe, for
reasons I shall try to make clear, that there mustbeungrounded
identities-ultimately because the notion of such identities
underpinsour very conception of time itself. (The thought is not
altogether novel, having strong Kantian echoes: but I hope that
in its detailed development I may have something new to offer.)
To this it may be objected-before I even begin-that no
to howthings
legitimate advance may be made fromourconceptions
mustbe.In a sense, I grant this: but if it can be established that we
cannotbut conceive things to be thus-and-so, little enough of
interest can be made of the thought that things might
nonetheless not be thus-and-so-for, ex hypothesi,no positive
content can possibly be conferred upon such a thought.
II
In what termsmight one hope to supply a criterionof diachronic
identity for something like a tomato? I think that there are three
general approachesone might take, which I shall call theproperty
instantiationapproach,the temporalparts approachand the substantial

constituents
approachrespectively. I shall argue that the first and
second of these approaches are inadequate while the third,
which is adequate, demands the existence of ungrounded
identities. Hence, if no other approach is forthcoming, this may
be taken as establishingthe credentialsof ungroundedidentities.
Later I shall advance a positive argument in their favour.
According to the property instantiation approach, the
diachronic identity of a tomato is grounded in some spatiotemporal-cum-causal condition on the instantiation of tomatohood-the crudest versionof the theory being that the identity is
grounded simply in the spatiotemporalcontinuity of such in-

stantiation.2That is to say, what supposedly makes it the case


' It
as opposed to that
may be wondered why I speak of the instantiation oftomatohood,
of some cluster of non-sortal properties(suchas size, shape, colour and so on). My reason
is that I do not believe that sortalsare definable in terms of such properties:but anyone
who thinks otherwise is at liberty to take my use of the term 'tomatohood' as merely
abbreviatory.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

64

I-E. J. LOWE

that the tomato now sitting on the table is the same tomato as the
tomato sitting on the table five minutes ago is that there is a
spatiotemporally continuous sequence of place-times stretching
from the place-time occupied by the tomato on the table five
minutes ago to the place-time occupied by the tomato on the
table now, such that tomatohood is fully instantiated at each
place-time in this sequence. (We have to say 'fullyinstantiated'
so as to exclude, for instance, a case in which a tomato is largely
but not wholly destroyed and subsequently miraculously
regenerates:for in such a case one would be reluctant to identify
the later tomato with the earlier one.) Obviously, it is not
important to this account whether or not our tomato is removed
from its table and then returned during the five minute
interval-which is as it should be.
Now, one objection which might be raised against the
foregoing account is that it rules out a priorithe possibility of
interrupted or intermittent existence for something like a
tomato. I think that this sort of objection is probably valid, farfetched though it may seem. But it would appear that it is not
fatal to the property instantiation approach in general.For
instance, one might, consistently with this approach, loosen the
requirement on spatiotemporal continuity while at the same
time adding a causalcondition to distinguish between cases of
interrupted existence of the same tomato and cases of the
annihilation of one tomato and its later replacementby another
created ex nihilo.The condition would be something to the effect
that in order for later instantiations of tomatohood to ground
the identity of the sametomato as earlier instantiations,the later
instantiations would have to be causallydependent
on the earlier
instantiations in certain appropriate ways. (Such a condition
will arguably be needed in any case, so that its invocation here
should not be seen as ad hoc.)
But there is I think a far more seriousobjection of principle to
the property instantiation approach. What could possibly be
meantby saying that tomatohood is fully instantiatedat a certain
place and time?Just this, surely:that a tomatoexists at that place
and time. In fact, matters are a little more complicated than
this, but not in a way that helps the property instantiation
approach. It is crucial to the chances of successof that approach
that the 'full' instantiation of tomatohood at a certain place and

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

65

time be construed in such a manner that multipleinstantiations


at the same place and time are excluded. For otherwise the
approach will not be able to handle, for instance, a case
involving a bunch of contiguous tomatoes maintaining or
changing their relative positions without losing contact: it will
apparently be incapable of saying whichof the tomatoes at a later
time is to be identified with whichof the tomatoes originally
present. In short, 'full' instantiation should be interpreted as
'full and unique'instantiation at a place and time. But this only
serves to bring out more clearly than ever that what this talk of
the ('full and unique') instantiation of tomatohood at a certain
place and time amounts to isjust that there should exist at that
place and time exactly one whole tomato. However, such a
proposition obviously presupposes,and hence cannot help to
provide, an account of the identity-conditions of tomatoes. For
to speak of onetomato isjust to speak of one andthesametomato.
To this it may be replied that all that is presupposed is an
account of the synchronic
identity-conditionsof tomatoes,whereas
what is now at issue is the question of their diachronic
identityconditions. My response would be to put pressure on the
assumption that the synchronic and diachronic identityconditions of things like tomatoes are independently intelligible.
Clearly, it is not an inessentialproperty of tomatoes that they are
things of a sort that persist through time (even if we can make
sense of the thought of this or thatparticulartomato having only a
very short-lived existence). And so a synchronic identitycriterion for tomatoes which failed to reflect this fact could not
properly be representedas a criterionfor the synchronic identity
of tomatoes,as opposed say to qualitatively similar objects of a
more ephemeral sort (such as, perhaps, the temporalparts of
tomatoes with which we shall shortly be concerned). A
synchronic identity-criterion for tomatoes should tell us under
what conditions we have to do with one and the same as opposed
to two distinct tomatoesat a certain time: and this cannot in
general be a matter untouched by considerations of prior
existence, given the persistent nature of things that are of the
tomato kind.
This might however be challenged on the following grounds.
A necessaryand sufficientcondition for the diversityof tomatoes
at a given time, it may be said, is the diversity of their locationsat

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

66

I-E. J. LOWE

that time, because no two tomatoes can occupy precisely the


same place (or indeed even partially overlapping places) at the
same time. And this, it might seem, is a condition whose
obtaining is not affected by considerations of prior existence.
However, while of course accepting the principle that tomatoes
exclude one another from the same place at the same time,3I do
not accept that it does not rest upon assumptionsconcerning the
diachronic identity of tomatoes, nor do I accept that it can be
directly appealed to as providing a criterion of synchronic
identity for tomatoes-for, on the contrary, it seemsto me that it
is a principle which must if anything be seen to emergefromsuch
a criterion rather than to constitute it. I shall take these two
points in turn.
First, then, we should appreciate that the capacity of an
object to exclude another from the same place is one that can
only be exercisedin the course of a finite period of time, not just
instantaneously: so that the ascription of such a capacity to an
that it is a persisting sort of thing. Secondly,
object presupposes
however, we need to ask just what it is about tomatoes that
confers upon them this special power of mutual placeexclusion-a power not possessedby objectsof many other sorts,
such as shadows and beams of light. We can after all easily
imagine two objects looking like tomatoes approaching one
another and merging together: though the very fact of such a
merger would disqualify these objects from counting as tomatoes.
Perhaps the most tempting answer is to say that what is
distinctive about tomatoes is that they are materialobjects:
different tomatoes are composed at any given time of different
portions of matter, and different portions of matter quite
generally,it may be said, exclude one another from the same
place at the same time. Now, this answer will certainly notdo as
it stands, because it fails adequately to accommodate such
mundane facts as that a quantity of water may seep through a
portion of porous clay pot. Perhaps indeed one may refine what
is meant by 'existence in the same place' so as to discount such
3To accept this is not, obviously, to deny that a place (region of space) may contain a
bunchof tomatoes, but only to imply that in such a case the place in question must be
divisible into disjoint sub-regionseach of which contains no more than one tomato. (It
was, of course, in this uncontentious sense that I spoke earlier of the multiple
instantiation of tomatohood at the same place and time.)

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

67

cases-stipulating, for example, that existence in the same place


implies co-existence in every spatial location within that place
(though this threatens to run into the opposite danger of
discounting toomanycases by allowing, for instance, two portions
of gold to coincide provided that their atoms do not). But, what
is more to the point, the answer already presupposessome grasp
of the notion of the identity and diversity of portionsof matter,a
notion from which the general exclusion principle must
somehow be seen to emerge if it is finally to be endorsedat all. A
of our putative
grasp of that principle cannot itself be constitutive
of
the
of
understanding
identity and diversity portionsofmatter,
because it is a mutualexclusion principle and hence, in the
absence of a prior grasp of what qualifies as a single and distinct
portionof matter,only tells us that something, we know not what,
excludes another such thing from the same place at the same
time-in short, it tells us virtually nothing. So if we do accept the
answer now being contemplated, we must clearly give up the
thought that the power of material objects like tomatoes to
exclude one another from the same place at the same time is
what ultimatelyunderpins their synchronic identity or diversity.
Now in fact I should say that I do not think that we ought to
accept this answer, because I do not believe that a quite nonspecificexclusion principle formatteringeneralis easily defensible.
But then it becomes clearer than ever that mutual exclusion
principles for specific kinds of material objects--including
tomatoes-must have the status of derivative truths relying for
their appeal at least partially upon a prior grasp of the specific
synchronic identity-criteria appropriate to objects of the kinds
in question.4
III

Having, I hope, exhausted for the time being the dubious


attractions of the property instantiation approach, let us turn
4The point may be highlighted by the familiar examples of cases in which, very
plausibly, we shouldsay that two distinct material objectsexist in precisely the same place
at the same time-e.g. a bronze statue and the lump of bronze of which it is made. What
I would emphasise is that such spatiotemporal coincidence is possible precisely because
statues and lumps of bronze have differentcriteria of identity-for this confirms that the
direction of explanation runs from identity-criteria to exclusion (or non-exclusion)
principles, rather than viceversa.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

68

I-E. J. LOWE

next to the temporalparts approach.(Later on I shall develop a


further argument inimical to both of these approaches.) This
second approach differs chiefly from the first only in replacing
talk of the (full and unique) instantiation of tomatohood at a
certain place and time by talk of the existence at a place and
time of a temporal part, or stage, or slice, of a tomato. A
diachronic identity-criterionfor tomatoes will then be framed in
terms of spatiotemporal-cum-causal conditions on sets or
sequences of such temporal parts. This approach has an
apparent advantage over the previous one in that temporal
parts of tomatoes plainly cannot be just the same sortsof things
as tomatoes,and hence no immediate circularity threatens when
the diachronic identity-criterion for tomatoes is stated by
referenceto such entities.5Whereas talk of the (full and unique)
instantiation of tomatohood at a certain place and time was
transparentlyjust an oblique way of speaking of the existence at
that place and time ofexactlyonewholetomato,talk of the existence
at a certain place and time of a temporal part or stage of a
tomato is not so obviously a mere verbal ploy.
But in another way the temporal parts approach seems
blatantlycircular. For how are the 'temporal parts' of tomatoes
(assuming indeed that we countenance the existence of such
things) to be individuated and identified save by referenceto the
very tomatoes of which they are parts?The expression'temporal
part of a tomato' is a theoretical term of art, unlike the term
'tomato' itself, so that it is not open to one just to leave the
question of their individuation to 'common sense' or 'intuition'.6
Perhaps however it will be useful to compare the notion of a
temporalpart of a tomato with that of a spatialpart, which is a
5Some philosophersdo, I concede, believe that a temporal part of an object of the sort
p may (particularlyif the part has quite an extended duration) itselfqualify as an object
of the sort(p:see, e.g., Anthony Quinton, TheNatureof Things(London: Routledge and
Kegan Paul, 1973), p. 70. But I can see no advantage for them in this, and only the
disadvantage of having to face the threat of immediate circularity which may otherwise
be avoided.
6 In his second postscript to 'Survival and Identity' (reprinted in his Philosophical
Papers,Vol.I (Oxford: Oxford University Press, 1983)), David Lewis writes: 'A personstage is a physical object, just as a person is ... It does many of the same things that a
person does: it talksand walksand thinks... It even has a temporal duration. But only a
brief one, for it does not last long' (p. 76). He then goes on to argue that person-stages
thus conceived do indeed exist and constitute the temporal parts of persons.The firststep
in his argument is this: 'First:it is possible that a person-stagemight exist. Suppose it to

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

69

good deal more familiar. Now the phrase 'spatial part of a


tomato' is arguably ambiguous in an important way. It may
either be taken to mean 'part (= component, constituent) of a
tomato which is spatiallyextended'-under which interpretation
one of a tomato's seeds, say, will count amongst its 'spatial
parts'; or alternatively it may be taken to mean something like
'object consisting at any given time of the matter enclosed by a
geometrically defined surface not extending beyond the outer
skin of a tomato'-under which interpretation a quarter-inchthick cross-section through the middle of a tomato will count
amongst its 'spatial parts'. However, it seems plain that the
notion of a temporal
part of a tomato, if it is to play any distinctive
theoretical role, can only be supposed to be modelled on the
second of these senses of 'spatial part'-a temporal part of a
tomato is supposedly a temporal 'cross-section' or 'slice' of a
tomato. For if by a 'temporal part' of a tomato one merely
meant, by analogy with the first sense of 'spatial part', a
constituent of a tomato which is temporally extended, then a
tomato's spatial and temporal parts in these senses would be
precisely the same sorts of things (things like its seeds). But
unfortunately it seems clear that only spatial partsof a tomato in
this first sense constitute objects which are individuable and
identifiable independently of the tomatoes of which they are
parts.
As against this last claim it may be urged that if we are
presented with a single slice (= spatial cross-section)of a tomato
on a plate, we can individuate and identify this slice without
being in any position to say from whichtomato it has been cut.
But this objection tradeson an ambiguity in the notion of a 'slice'
(or spatial 'cross-section').By a 'slice' of a tomato one may either
mean a particular type of spatial part of a tomato in the second
sense defined above orelseone may mean something like 'object
obtained by actuallycutting twice through a tomato in two
approximately parallel planes'. The spatial parts of a (whole)
appear out of thin air, then vanish again' (ibid.). But, I would contend, all that Lewis has
really succeeded in doing here is to introduce us to the fanciful notion of a veryshort-lived
person(i.e. to the idea that a personmight in some miraculous way be conjured into and
out of existence in a trice), and as such he has failed to introduce us to a category of
independently individuable entities in terms of which a criterion of diachronic identity
for persons might be non-circularly specified.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

70

I-E. J. LOWE

tomato in our second senseof'spatial part' do notinclude slicesof


it in this second sense of 'slice'. There is no possibility of
identifyingslices of the two different sorts-geometricalslices and
physicalslices, as we might respectivelycall them. To see this we
have only to consider the differentcapacities which the two sorts
of slice possessto undergo certain kinds of temporal change. For
instance, a physical slice of a tomato can clearly undergo all
manner of changes in shape, whereas the very being of a
geometrical slice is partly definedby its shape. (In the case of a
physical slice, shape only enters into the definition of how the
object is produced.)Again, if the contents of a tomato are
rearranged, the material contained in one of its geometrical
slices may well alter considerably (as various seeds, quantities of
juice and so on alter their locations within the tomato). No
comparable possibilities for changing its constituent material
arise in the case of a physical slice, however. But these and
related facts also serve to show that geometrical slices, unlike
physical ones, are not individuable independently of the whole
objects of which they are slices: thus a geometrical slice of a
tomato is partly individuated by referenceto its relativeposition
within the tomato, which it evidently cannot alter. Moreover, it
is clearly geometrical
slices rather than physical ones which must
provide the spatial model for temporalparts or 'slices' of objects
(since, apart from anything else, nothing very obviously
corresponds in the temporal case to the physical act of cutting
which creates a physical spatialslice).
Now, if I am right in saying that the spatial parts of a tomato
in our second sense of 'spatial part' are objects which are not
individuable or identifiable without reference to the tomato of
which they are parts, and also right in supposing that the notion
of a temporalpart of a tomato can only (at best) be seen as
modelled upon this second sense of 'spatial part', then it would
seem to follow that even if we do countenance such objects as the
temporal parts of tomatoes, they will not be fit entities in terms
of which to frame a non-circularcriterion of diachronic identity
for tomatoes.' In reply to this it might once again be urged that
7By now some readersmay have wanted to accuse me of taking too narrowa view of
the temporal parts approach, and in particular too narrow a view of what a temporal
part or stage would have to be. Thus Sydney Shoemaker, a prominent adherent of the
approach, has written: 'Person-stagescan be thought of as "temporal slices", not of

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

71

in fact the notion of a temporal part of a tomato only


presupposes a synchronicidentity-criterion for tomatoes, and
hence can non-circularly contribute to a diachronic criterion.
But again I would reply, as I did when a similar move was made
on behalf of the property instantiation approach, that one is not
entitled to presume that synchronic and diachronic identitycriteria for objects like tomatoes are independently intelligible.
The temporal parts approach seemedinitially to have an
advantage over the property instantiation approach precisely
on this score: but what we have seen is that in reality the two
approaches are in the same boat-and it is a sinking boat.

IV
I turn thirdly, then, to the final approach to diachronic identity
that I shall consider-the substantial constituents approach,
which I favour.8According to this approach, what underpins
the persistenceof something like a tomato is the persistenceof its
component parts-and by these I mean its 'spatial parts' in our
first sense of the term (i.e. things such as the seeds and skin of a
tomato). Actually, this is of course a slight oversimplication
because a tomato can undergo a certain amount of change in its
component parts without loss of identity (without, that is,
persons, but of the historiesor careersof persons.[Or] one might think of a momentary
stage as a set of property instantiations ... Or one can think of a momentary stage as an
ordered pair consisting of a thing and a time' ('Personal Identity: A Materialist's
Account', in Sydney Shoemaker and Richard Swinburne, PersonalIdentity(Oxford:
Blackwell, 1984), p. 75). However, the second of these suggestions would reduce the
temporal parts approach to the property instantiation approach, while the first and
third would transparently make the individuation of 'stages' parasitic upon that of the
continuant objects whose diachronic identity they were invoked to account for. Such
circularitydoes not, it is true, worryShoemaker,who elsewhereconcedes that by his own
account the persistence-conditionsofcontinuants cannot be non-circularlyspecified (see
'Identity, Properties, and Causality', in Sydney Shoemaker, Identity,Cause,and Mind
(Cambridge: Cambridge University Press, 1984)). But I cannot agree that a circular
specification of persistence-conditions, however non-trivial, may legitimately be
presented as an account of what persistence consistsin. (Colin McGinn also makes this
84 (1987), pp.
point in his review of Identity,Cause,andMind in TheJournalof Philosophy
227-232.) So my answer to the objection raised in this note is that I adopt the
interpretationof the temporal partsapproach which I do because it seems to me to be the
least unpromising on this score.
8 I argue directly in defence of this approach and against the temporal partsapproach
in my 'Lewis on Perdurance versus Endurance',Analysis47 (1987), pp. 152-154 and my
'The Problems of Intrinsic Change: Rejoinder to Lewis', Analysis48 (1988).

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

72

I-E. J. LOWE

ceasing to be). This is especially obvious during the time that a


tomato is still growing. Nonetheless, at any given time at which
it exists, a tomato must have component parts of certain
sorts-skin, seeds, quantities ofjuice and so on-and it is only in
virtue of the persistencefrom moment to moment of a sufficient
proportion of such components that the tomato as a whole
manages to persist. A further slight complication is that, as I
understand the term 'component part of a tomato', certain
objects which are materially included in a tomato will not
qualify as being amongstits component parts, although they will
qualify as standing to the tomato in the ancestralof the
component parthood relation-that is, they will be parts of
parts of parts . . of parts of a tomato. Thus an electron in an
atom in a molecule in ... in a cell in a seed of a tomato will not

qualify as a component part of a tomato in my sense. My reason


for making this stipulation is that I want to use the phrase
'component part of an object' in a way that makes it legitimate
to regard the sensitivity of an object's identity to changes in
its component parts as a largely conceptual matter. (Clearly, it
can be no largely conceptual matter whether changes in the
electronsin a tomato have a bearing on its diachronic identity,
whereas it is very much such a matter that changes in
its grosser constituents like its seeds and its skin have such
a bearing: and this is why I want to distinguish the seeds
and skin but not the electrons as 'component parts' of the
tomato.)
It is clear that the substantial constituents approach cannot
and does not even purport to offer an exhaustive account of the
persistenceof substances:it only offersa schemafor an account of
the persistence of any composite substance in terms of the
persistenceof its component parts. I say a 'schema' because the
approach does not presume to be able to tell us, in advance of
empirical scientific inquiry and theory-construction,just what
the 'component parts' at any point in the hierarchy of
composition will be. In this respect the approach is utterly
different from the previous two approaches, both of which are
thoroughly aprioristic in character. A clear consequence of the
substantial constituents approach is, however, its commitment
to the existence of ungrounded
identitiesat the base of the hierarchy
of composition, and on this issue the approach does take an a

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

73

priori stance. Some thing or things-be it primitive hyle or


quarks-must simply persist,without rhyme or reason, and in
this all higher-level material persistence must ultimately be
grounded.
However, here it may be felt that our third approach's
combination of empirical and a priori claims renders it
particularly vulnerable. For who is to say that physical science
will not in due course reveal a way to account for the persistence
of so-called 'fundamental' particles in terms of spatiotemporalcum-causal relations between non-persistent entities of some
sort (not indeed concocted pseudo-entities like 'time-slices', but
something the postulation of whose existence would be empirically well-motivated)? Certainly, in the presentstate of physical
science the substantial constituents approach looks to be
vindicated: for the fundamental (non-composite) particles of
modern physics are distinguishable into those that are and those
that are not 'stable', and the 'half-life'of such a particle (infinite
in the case of a stable particle like the electron) is apparently
taken to be an essential property of that particle which is not
further explicable (like also the chargeon an electron). But
perhaps this isjust a parochial featureof contemporaryphysical
theory?
In fact, however, I do not think that the commitment to
ungroundedidentitiesis as vulnerableto the futuredevelopments
of physics as it might appear to be. What undoubtedly is an
empirical issue open to future revision is the correctnessof our
contemporary theoryof matter.' What is not I think thus
vulnerable is some broad notionof matter as the ultimate and
itself ungrounded ground of all physical persistence.My reason
for thinking this is, as I indicated at the outset, that I believe that
the very concept of timecannot be divorced fromsuch a notion of
matter. For time, I consider, essentially involves change,and a
change can only be understoodby referenceto something which
persiststhrough that change. But to deny that there is anything
9It is worth emphasising that the substantial constituents approach does not demand
that the ultimate constitution of material things be particulate
in nature, as contemporary
physical theory would suggest it is (making due allowance, of course, for the
wave/particle duality of quantum phenomena). Thus the approach can readily allow,
for instance, that at its most fundamental level physical reality might have to be
described by means of a vocabulary of massnouns rather than by one of countnouns.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

74

I-E. J. LOWE

whose persistence is ungrounded is to imply that everything's


persistencemay ultimately be accounted for in terms of change,
which conflicts with the stated dependency of the notion of
change on the notion of something persisting through change.
Thus expressed, the argument is no doubt suspiciouslyabstract
and condensed-but I shall try in what remains of this paper to
spell it out in more convincing detail.
V
The first premise is that time essentially involves change-by
which I mean that time essentially involveshappenings
or events.'
I do not necessarilywant to imply by this that there could not be
time in a universe which was not qualitatively distinguishable
from one moment to the next (though nor do I particularlywish
to defend the suggestion that there could).All I am saying is that
it is partly constitutive of the notion of time that time should
embrace events or happenings, and that events or happenings
are to be understood as changes--although not necessarily
changes to something or in something (this would be tooeasy a
way to argue for the necessity of persisting substance). When a
change occurs, something begins to be the case which was
previously not the case-for instance, it begins to be the case
that a light is glowing or that there is noise in a certain room."
Change in this sensecouldconceivablystill occur in a qualitatively
"oIn 'Time without Change' (reprinted in his Identity,Cause,andMind,op.cit.), Sydney
Shoemakerargues that at least in some logically possibleworlds it could be reasonableto
hypothesise that a period of time had elapsed during which nothing whatever had
changed in any way. I am not persuaded by his argument, though I cannot discuss it
here. But, in any case, it is clear that the argument cannot be construed as establishing
that time might pass in the absence of any changewhateverwithout presupposing that
the persistenceof objects in such a world would notbe grounded in spatiotemporal-cumcausal conditions on sequences of momentary entities (suchas durationless'time-slices'),
since the continual comings-to-be and ceasings-to-be of such entities would precisely
constitute changesduring the supposedly changeless period of time (even though no
qualitativechange need be involved in such a case). So Shoemaker'sargument, even if it is
correct, cannot be used to any effect against me by adherents of the views of persistence
which I am attacking, at least as far as my first premise is concerned.
" It is customary when discussing change to distinguish between 'real changes' and
'mere Cambridge changes'-e.g. between Socrates'sdying and Xanthippe's becoming a
widow. But since the latter sort of change is arguably parasitic upon the former, it is not
crucial to my argument to restrictthe senseof'change' it invokesto that of'real' change.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

75

unvarying universe-for it might begin to be the case, say, that


elsequalitatively
somethingwas yellow where previouslysomething
indistinguishable was yellow.'2 (Whether this is really possible
depends on the status of the principle of the identity of
indiscernibles.)
The second premise of our argument is that a change can only
be understood by referenceto something which persiststhrough
that change. Again, it is not being urged that a change must be
understood as a change in or to something which remains the
same throughout that change. For some events are not changes
in or to things in any very obvious sense at all. All that is being
urged is that when something begins to be the case which was
previously not the case (i.e. when a change occurs), there must
exist at the time of the change something which also existed
prior to the change. For suppose on the contrary that nothing
existing at the time of the change existed prior to the change:
that would of course imply that everything
in existence at the time
of the supposed change had begun to exist at that very time.
However, what is it to say this but just to say that the time in
question was the beginning of the universe, and hence the
beginning of time? But the beginning of time cannot be the time
of any changein our sense, since we cannot meaningfully speakof
not
anything's beginning to be the case then whichwaspreviously
the case.'3Hence we have reduced our original supposition to
absurdity.
No doubt this reasoning may appear sophistical on firstsight,
but I believe that deeper examination will vindicate it. Suppose
I tell you that nothingthat has existed between now and five
minutes ago existed earlier than five minutes ago: what canyou
make of this but that I am saying that the entire universe began
to exist five minutes ago? The earth, the sun, the stars-all these
and everything else existing now or in the past five minutes
began to exist, I say, no earlier than five minutes ago: how could
12
Thus consider a universecontaining just two balls, one yellow and the other red but
otherwise qualitatively indistinguishable from one another: and then suppose that at a
certain moment the yellow ball turns red while the red ball turns yellow (and nothing
else changes). The states of such a universebefore and after the change would differonly
in respect of the identities of the ball which was red and the ball which was yellow.
13This means that if we want to call the beginning of the universe an event,we had
better make this an exception to the rule that events are changes.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

76

I-E. J. LOWE

you credit this and yet still make sense of the thought that there
were events happening longer than five minutes ago? Ex
no recordof any such events could possiblynow exist. So
hypothesi,
what would be the differencebetween talking of these supposed
events as havingoccurred
earlier than five minutes ago and talking
of them as belonging to an altogether different space-time
continuum--another 'possible world'?
But perhaps you will dispute the suggestion that no record of
these supposedly earlier events could now exist-urging that
this trades on an ambiguity in the term 'record'.A recordmay
either be an objectbearing a trace of some earlier event, or else it
may just be an effectof some earlier event. But, it would seem,
only recordsin thefirst sense could not exist in our hypothesised
case: there might still be statesof presentlyexisting objectswhich
could be attributed to causeswhich happened more than five
minutes ago. However, it seems to me that to argue in this
fashion is to beg the very question at issue. Our question is what
reason we could have, in the hypothesised case, to suppose that
any events occurredearlierthan five minutes ago: it is no answer
to say that we couldattribute various states of presently existing
objects to causes which occurred more than five minutes ago
without explaining with what justification we could suppose
causal relationships to be capableof embracing a timespan
exceeding five minutes in the past. After all, the putative causes
of these present states would themselves
precisely be eventsin the
I
disputedcategory.Furthermore, find it very hard to see how any
such justification could be forthcoming: for how could the
required causal influenceshave been propagated in the absence
of any objects surviving from the alleged earlier time into the
currentfive-minuteperiod?No photons, for instance,transmitted
from objects existing in the alleged earlier period could be
received by us-for all existing photons, being persistingobjects
themselves, would ex hypothesihave existed for no longer than
five minutes.
It remains now to be shown exactly how these considerations
lead to our declared conclusion. As I remarkedearlier, to deny
that there is anythingwhose persistenceis ungrounded is to imply
that everything's
persistence may ultimately be accounted for in
termsof change. This may clearly be seen by referenceto the two
approaches to persistence discussed previously-the property

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

77

instantiation approach and the temporal partsapproach-both


of which may be taken to deny the need to appeal to
ungrounded identities in giving an account of persistence.
According to each of these approaches, the persistence of an
object is to be accounted for by-that is, is to be regarded as
consistingin-the obtaining of certain spatiotemporal-cumcausal relations between non-persistingentities of some sort:
instantiations of appropriate properties at certain place-times
according to the first approach, and 'time-slices' of objects
according to the second (time-sliceswhich need to be instantaneous
if the theory is to be worth its salt-for if the slices haveto persist
primitively even only for a short duration, the theory cannot
pretend to offer an account of persistenceas always grounded).
However, this means that both approaches account for all
persistencein terms of change. On the temporal parts approach
this is evident, since each of the time-slicesof a persistingobject
has to comeintobeing,and each such coming-to-be constitutes a
changenecessary for the continued persistence of the object in
question. But it is quite as evident also on the property
instantiation approach. Clearly, this latter approach must be
committed to denying that the instantiationof, say, tomatohood
at a certain place-time is something essentially having a
duration extending beyond that time. The theory must presume
that tomatohood isfreshlyinstantiated at each place-time in the
history of a single tomato, since if it were compelled instead to
say that no such fresh instantiation is required from time to
time, then it would become transparent that its proposal
would differ only verbally from the claim that tomatoes
just
persist. In short, instantiations must be treated as eventsor
changes.

So according to both of these approaches, everything's


persistence is ultimately to be accounted for by references to
changesof some sort-either the momentary instantiations of
certain properties or the comings-to-be of durationless timeslices. But the very notion of a change, we have argued,
that of the persistence of at least somethingthrough
presupposes
that change and so cannot be appealed to in giving a perfectly
general account of what the persistence of objects consists in.
The persistence of at least somesorts of things must, then, be
primitive or ungrounded, in that it can consist neitherin

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

78

I-E. J. LOWE

relationshipsbetween non-persistingthings norin the persistence


of other sortsof things. This must, I have argued, be presumed in
order to make senseof time at all, given the necessaryconnection
between the concepts of time and change."

4 This paper was largely completed before the publication of an exchange on the
subject of persistencebetween MarkJohnston and Graeme Forbes in The Aristotelian
VolumeLXI (1987). It will be clear to any readerof that exchange
Society's Supplementary
that my sympathies lie rather more withJohnston's position than with Forbes's,but that
there are also considerable differencesbetween my position and Johnston's. (It will be
equally evident that my position has some strong affinitieswith the one that Saul Kripke
has defended in his celebrated lectures on identity over time.) I am grateful to Susan
Lowe and to David Over for helpful discussionsof an earlier draft.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME


E. J. Lowe and Harold W. Noonan
IIH-HaroldW. Noonan
I
Jonathan Lowe argues that there must be some things whose
identity over time is primitive or ungrounded, ultimately
because the notion of such ungrounded identities underpins the
idea of time itself. For such things, he claims, no criterion of
diachronic identity can be supplied.
But what is 'a criterion of diachronic identity'? Lowe
emphasisesthat what he is concerned with is not an evidential or
heuristic principle, but a metaphysical-cum-semantic one.
The provision of a criterion of diachronic identity for things of a
kind would not merely be a statement of what would count as
evidence for the identity over time of a thing of that kind, but a
statement of what the identity over time, and hence the
persistence of such a thing, would consistin.
Lowe distinguishesthree approachesto the provisionof such a
criterion and argues that two are unsatisfactory, whilst the
third, which he favours, implies the existence of persisting
entities for which no criterion of diachronic identity can be
given. These three approaches he labels 'the property instantiation approach', 'the temporal parts approach' and 'the
substantial constituents approach', the last being the one he
favours.
However, I wish to begin by motivating and then explaining
an alternative point of view from which these issues can be
approached. I shall then introduce two senses in which it might
be held that the identity over time of some persisting things is
ungrounded, but argue that neither Lowe, nor anyone else, and
in particular not Kripke (with whose views Lowe indicates that
he thinks his own have affinities)hasproventhat there must be, or
are, things whose identity over time is ungrounded in either of
these senses.
But my main concern will not be to argue against the
possibility of ungrounded diachronic identities (forwhich there
may well be good arguments) but to get clearer about what

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

80

II-HAROLD W. NOONAN

precisely we are seeking when we enquire after the criterion of


diachronic identity, or the persistence conditions, of things of a
certain kind.
II
At first sight there seems to be no difficulty. For any kind of
persisting thing K, in addition to the identification problem for
Ks, i.e. the problem of specifying logically necessary and
sufficient conditions for being a K, there is also the problem of
specifying the logically necessary and sufficient conditions for
the identity of a K existing at one time and a K existing at
another. Thus the problem of personal identity over time is the
problem of saying 'what are the logically necessary and
sufficient conditions for a person P2 at a time t2 to be the same
person as a person P, at an earlier time tl' (Swinburne
1976:223). The (less interesting) problem of ship identity over
time is precisely analogous:what are the logically necessaryand
sufficient conditions for a ship S2 at a time tz to be the same ship
as a ship S, at an earlier time tj? And mutatismutandisfor the rest
of the problems of identity over time discussedby philosophers.
In addition to these problemsof diachronic identity, there are
also problems of synchronic identity, which can be stated
similarly. Thus there is the problem of synchronic identity for
persons:'what are the necessaryand sufficientconditions for two
persons [i.e. persons identified by distinct descriptions] at a
given time to be the same person?'(Swinburne 1976:228);there
is the less interesting problem of synchronic identity for ships,
and so on.
That these problems exist, that they make sense, and that we
can fruitfully discuss them, seems evident. For what else are we
doing when we debate such puzzle cases as Locke's Prince and
Cobbler, Shoemaker's Brown and Brownson, the Ship of
Theseus, or the (non-fictional) cases of 'split-brain' patients?
Either these debates are a lot of nonsense, it seems, or these
problems about identity are genuine ones. And so most
philosophers are content to hold that they are indeed genuine
ones (cf. Kripke's unpublished lectures on identity over time for
similar remarks).
But when one looks closely at the formulations of these
apparently genuine problems it is easy to become puzzled. They

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

81

seem to be requests for the specification of the satisfaction


conditions of certain relations--personal identity over time, or
ship identity over time, for example. But how can this be?There
are not different kinds of identity to be differently analyzed.
There is just the onerelation of identity, and there is nothing in
any way puzzling about it. As David Lewis puts the point:
'Identity is an utterly unproblematic notion. Everything is
identical to itself. Nothing is identical to anything else. There is
never any problem about what makes something identical to
itself; nothing can fail to be. And there is never any problem
about what makes two things identical; two things never can be
identical' (1986:192-3).
How then can it make senseto ask, for example, what makesa
person existing at one time identical with a person existing at
another? If the person existing at the earlier time is identical
with the person existing at the later time, the question is a
request for an account of what makes a thing identical with
itself. Whilst if the earlier person is distinct from the later one,
the question is a request for an account of what makestwothings
identical. In either case it is unanswerable. The same problem
confronts a request for an account of what makes two persons
(personsspecified by distinct descriptions)at one time identical.
Either they are, or they aren't. If they aren't nothing makes
them identical. If they are, then their identity is the identity of a
thing with itself, and so again makes 'them' identical. Of course,
statements of identity can be informative, and so the possibility
is still left open of evidential or heuristic principles stating what
evidence would count in favour of claims of personal identity or
ship identity, or whatever. But the 'semantic-cum-metaphysical'
problem of what constitutesidentity (whether diachronic or
synchronic) for things of a kind now begins to look like a
nonsense.
III

The basic thought underlying this argument is twofold: first,


identity is not, as Locke said, 'suited to the idea', there isjust the
one relation of identity, the relation everything has to itself and
nothing else, and secondly, this notion is unanalyzable in any
'For this line of thought see, apart from Lewis, B. Brody 1980.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

82

II-HAROLD W. NOONAN

more fundamental terms. How then can there be botha problem


about personal identity and a distinct problem about ship
identity? How can there be botha problem about diachronic
personal identity and a distinct problem about synchronic
personal identity? And, more fundamentally, how can any of
these problems make sense?
One response to this difficulty would be to side with Locke
and to deny the uniquenessof the identity relation. But there is a
better way, which is to deny that the genuineproblems which
philosophersare concerned with when they debate topics under
the title of 'problems of (synchronic and diachronic) identity'
are problems about identityat all. Rather, what they are
problems about is kind-membership.
This suggestion is, of course, a wholly unoriginal one, the
clearest expression of which is in Quine (e.g. 1976).
According to Quine 'any collection of particle stages,
however spatio-temporallygerrymanderedor dispersed'counts
as a physical object. The world's water is a physical object.
There is a physical object part of which is a momentary stage of a
silver dollar sometime in 1976 and the restof which is a temporal
segment of the Eiffel Tower through its third decade. Any two
momentary objects, taken at different moments, are time slices
of one physical object-time slices indeed of many such.
However, most such physical objects are irrelevant to our
concerns, and go unnamed in our language. But some do not;
though ontologically on a par with the rest, these occupy a
favoured place in our language and conceptual scheme. For any
such favoured kind K of physical object there is the problem of
specifying the conditions a physical object has to satisfy to be a
K. Thus there is the problem of specifying the conditions a
physical object has to satisfyto be a ship, or a person, a river,or a
body of water.
Now according to Quine the temporal parts of a physical
object need be related in no way that is of interest to us. But
when we consider, say, what conditions a physical object has to
satisfy to be a river, the situation is different. It is not enough for
a physical object to be a river that its momentary stages have a
certain character; in addition they must be inter-related in a
certain way-they must be river-kindred.
In Quine's view it is the
specification of this relation which philosophersare concerned

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

83

with when they debate the 'criterionof diachronic identity' for


rivers.But in specifyingthe conditions of riverkinshipwe are not
stating conditions for identity,but merely conditions for beinga
river.
It is, in Quine's view, the same with the problem of synchronic
identity for rivers. Insofar as it makes sense it is not a problem
about identity at all. Though it is not sufficient for a physical
object to be a river that its momentary stages have a certain
character, still, it is necessary. A momentary stage of a river
differs in intrinsic character from, say, a momentary stage of a
cow. Thus there is the problem of saying what this character is.
But nothing can count as a river-at-a-momentunless its partsat
that moment stand to one another in certain relations. It is these
relations that are discussed under the misleading title 'the
criterion of synchronic identity for rivers'.
This Quinean conception of problems of synchronic and
diachronic identity as reducible to problems about kind
membership seems to me very plausible, indeed wholly
compelling. But it is bound up in Quine, as in the exposition
above, with an idea that many philosophersfind a good deal less
compelling-the idea that everyday things like ships and people
are 'four-dimensionalworms', with temporal as well as spatial
parts.
But I want to suggest that we can take on board the more
attractive of these Quinean ideas without committing ourselves
to the less attractive one. However, in order to make good the
claim that the so-called problem of identity over time for a
particular kind of thing K is not a problem about identity at all,
but solely a problem about kind-membership, we need a
formulation of the problem in which the notion of identity does
not occur. In the absence of such a formulation the claim is a
fraud.
So what might such a formulationbe if we do not presuppose
the four-dimensional ontology?
We can approach an answer to this question if we begin by
asking what informationa solution to the problem of K-identity
over time would provide. The answer is that it would provide an
account of the distinction between those changes a K can
survive, and those it cannot, that is, an account of the sort of
history that is a possible one for a K, an account of the variations

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

84

II-HAROLD W. NOONAN

and constanciessuch a historymustdisplay,and those it may


display (notice that the only necessityin questionhere is de
dicto).
In askingwhat K-identityover time consistsin, then, what one
is asking for, in part, is a specification of certain necessary
conditionsof being a K, namely those identifiableby specification
of the relation R satisfying the following schema:
(1) Necessarily for any (thing of kind) K, x, for any times t
and t', if x exists at t and x exists at t' then Rxtt'.
The hope is that such conditions can be informatively
specified, i.e. can be specified without use of the very concept K
which is being analysed (whether the concept of identityneeds to
be employed is neither here nor there).
But, of course, in asking what constitutesK-identity over time
one is asking for more than the specification of certain necessary
conditions of K-hood.
The four-dimensional theorist can explain this 'more' very
simply: what one is asking for, he can say, is a specificationof a
relationR such that it is a sufficientcondition of a physical object
being a K that all its temporal parts are pairwise related by R
(sometimes, as in Perry 1972, such a relation is called the 'unity
relation' for Ks). But if we do not presuppose the fourdimensional ontology, we must express the request differently,
namely, as the requestfor a specificationof a relationR such that
for any x, it suffices for x's being a K that R relates all ordered
triples <x,t,t'> where t and t' lie within the period of x's

existence. That is, the request is fora specificationof a relationR


satisfying the condition:
(2) Necessarily, for any x, if for every t and t' ifx exists at t
and x exists at t' then Rxtt', then x is a K.
However, the specification of such a relation may be wholly
uninformative (naturally the same is true of the specificationof a
four-dimensionalunity relation). To ensure that this is not so (to
ensure, in other words, that the specification contributes to the
analysisof the concept of a K) we need to appeal once more to
schema (1) and require (at least) that the relation satisfying (2)
be specified as the relation whose satisfaction by any ordered
triple <x,t,t'> is entailed by (and entails) thejoint satisfactionby

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

85

that triple of some set of relations R', each of which satisfies


schema (1), but is specified without the use of the concept ofa K.
(In other words, the specification of the relation satisfying (2)
must be of the form: 'the relation R such that necessarily, an
ordered triple <x,t,t'> satisfies R iff it satisfies all of R, ...
R,'-where R1 ... Rn are relations satisfying schema (2) and
specified in the description given without the use of the concept
of a K.)
This, then, I suggest, is what the request for 'a criterion of
diachronic identity for Ks', or an account of what constitutes Kidentity over time, comes down to when properly expressed. It
would perhaps be better described as a request for 'the
diachronic criterion of K-hood'.
What of the request for a criterion of synchronic identity for
Ks, and its customarily assumed distinctness from the request for
a criterion of diachronic identity? Familiar examples (e.g. in
Perry 1972) make it evident that it is at least logically possible to
be in a state in which one's grasp of the concept of a K is partial in
such a way as to make it tempting to say that whilst one grasps
the criterion of synchronic identity for Ks one does not grasp
their criterion of diachronic identity. But how is the demand for
a criterion of synchronic identity to be expressed if the notion of
identity is not to be used?
Once again, if the four-dimensional ontology is presupposed
the answer is simple, as we have seen. But what if it is not? Then,
I suggest, the only intelligible question to be asked is: What are
the necessary conditions for a K's existence at a time? That is,
what conditions C satisfy the following schema:
(3) Necessarily, for any K,x, for any time t, if x exists at t,
then Cx at t.
Anything sensible that can be said in answer to the request for
a criterion of synchronic identity for Ks must therefore be
comprised in the answer to this question.
Of course, one such necessary condition is that x be a K. So
there is no hope, unless we presuppose the four-dimensional
ontology, of treating the request for a criterion of synchronic
identity for Ks as wholly distinct from the request for a criterion
of diachronic identity. But it may nonetheless be possible, by
suitable choice of'C' in (3) to give an informative specification of

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

86

II-HAROLD

W. NOONAN

somenecessary conditions of K-hood without thereby specifying


or presupposingthe diachronic criterion of K-hood. Many such
necessary conditions will have nothing in common with what
philosophers typically have in mind when they talk of the
criterion for the synchronic identity of Ks, but this need not be
true of all of them. A subset of such necessaryconditions, in the
case of spatially extended objects, for example, will concern the
interrelations of their (spatial) parts, and when we ask for a
criterion of synchronic identity for, say, ships or tables, it is
largely information about this that we are seeking (see once
again Perry 1972). In the case of persons our interest in a
criterion of synchronic identity is rather an interest in the
relationshipswhich must obtain between simultaneouslyoccurring, co-personal, mental states. But this can similarly be
understood as an interest in the truth-yielding specifications of
(3), with 'K' read as 'person'.
But what now of what Lowe calls the exclusion
principlefor Ks,
the principle that two Ks cannot occupy the same place at the
same time, i.e. that Ks occupying the same place at the same
time must be the same? Such a principle cannot be regarded as
specifying necessary conditions of K-hood in the manner of an
instance of schema (3), yet it seems, at first sight, a perfectly
intelligible thesis about K-identity at a time. How, then, can this
be reconciled with the Quinean thesis that all questions about
criteria of identity, whether diachronic or synchronic, for things
of a particular kind must reduce to questions about the criteria
for membership in that kind?
The easiest way to understand the role of such exclusion
principles, I think, is to revert yet again to the point of view of
the four-dimensionaltheorist. From this point of view there is, of
course, no difficulty whatsoever in the idea of two physicalobjects
being in the same place at the same time. So what, from this
point of view, can we be doing when we say (using the concept of
a familiar kind of physical object): two Ks cannot occupy the
same place at the same time? The answer is that even though we
are not specifying a necessary condition of K-hood (a Fregean
'mark' of the concept) we are specifying a constraint on the
concept of a K: a condition any concept must satisfy if it is to
qualify as the concept of a K (or equivalently, a condition the
unity relation for Ks must satisfy). And, of course, the role of the

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

87

principle remains the same even if the four-dimensionalpoint of


view is rejected.
But, if this is the way in which exclusion principles are to be
understood,we can reaffirmthe Quinean thesisof the reducibility
of questions about identity criteria to questions about kindmembership. For, if this is the case, for any kind K, whether the
exclusion principle for Ks is true will be determined once the
necessary and sufficient conditions for membership in the kind
(the marksof the concept K) have been fixed: if true its statuswill
thus be that of a merely derivative truth which does not have to
be mentioned in a full account of the concept (cf. the way in
which the specificationof the marksof K-hood, together with the
facts, will determine, without the aid, or the possibility, of any
further delimitation of the concept, the truth-value of'Ks exist').
Whether, once this is accepted, one continues to speak of the
exclusion principle forKs (forthose kinds for which the exclusion
principles are true) as an aspect of the conditions of K-identity at
a time is, of course, a matter of no importance. But it is
important, if we choose to do so, to note the difference between
this aspect of the conditions of K-identity at a time and those
which can be subsumed under a specificationof the conditions C
satisfying schema (3); and to be aware also that it will be
impossible to specify the marks of the concept of a K which
determine the truth of the exclusion principle without specifying
the diachroniccriterion of K-hood. (Here, I think, I am in
complete agreement with Lowe.)
I now wish to note two additional points.
The first of these is simply that one who rejects the fourdimensional ontology but accepts the frameworkfor discussion
outlined above can nonetheless still mimic four-dimensional
terminology. He can, for example, speak of an ordered pair
<x,t>, where x is a K and t a time at which x exists, as a 'K-stage'.
Lowe refers to, and criticises, philosophers who use the 'thingstage' terminology more generally than he proposes,but it may
be that these philosophershave something like this frameworkin
mind when they do so, and if so their terminology is harmless.
My second point is more important.I have alreadyemphasized
that one of the featuresof the four-dimensionalscheme is that it
entails that the familiarcontinuantsof oureverydayacquaintance
are a mere subset of the totality of physical objects, ontologically

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

88

II-HAROLD

W. NOONAN

on a par with the rest,and assigneda favouredpositionin our


language and conceptual scheme only because of interestrelative considerations.But this is certainlyone reasonwhy
somephilosophersare reluctantto acceptthe four-dimensional
scheme.2My pointnowisthatifsomeonedoesobjectto thefourdimensionalschemefor this reason,he oughtto thinkno better
of the alternativeframeworkjust outlined. For like the fourdimensionalschemeit entailsthe (possible)existenceoffarmore
physicalobjectsthanarerecognisedin oureverydayconceptual
scheme, and like the four-dimensionalscheme it entails that
theseadditionalentitiesareontologicallyon a parwiththemore
familiarinhabitantsof ourconceptualscheme-or at leastwith
thosefor which it is possibleto specifyinformativediachronic
criteriaof kind-membership.
The reasonforthisis verysimple.To say thata conditionCis
a necessarybut insufficientconditionforbeinga K is to saythat
forsomethingto satisfyC withoutbeinga K.
it is at leastpossible
So if thereareseveral
necessaryconditionsofK-hoodidentifiable
by appropriatespecificationsof R in schema(1) (as theremust
be if aninformative
diachroniccriterionofK-hoodisspecifiableat
all) then foreach suchconditionit mustbe possibleforthereto
be objectsotherthanKssatisfyingit. Butsuchobjects,insofaras
they can satisfythe conditionin question,and assumingthe
conditionis not one whichcan be satisfiedby bothabstractand
concrete objects, must be concrete objects if Ks are concrete
objects, and thus ontologically on a par with Ks.
For example, suppose it is held, to revert to the familiar
terminology for a moment, that the criterion of diachronic
withpsychological
personalidentity is physical continuitytogether
continuity-psychological continuity and physical continuity
by themselves only being necessary conditions. Then as I
understand it, that is to say that physical continuity is only a
necessarycondition of personhood, which is to say, in turn, that
it is at least possible for there to be something which is not a
in mind S. Shoemaker. In fact it was from Shoemaker's
2 Here I particularly have
1984, in which he expresseshis objection to the four-dimensionalscheme, that I got the
expression 'ontologically on a par'. Shoemaker does not define it, but I am taking it that
it is a sufficientcondition forKs and A*s being ontologically on a par that they are both
kinds of concrete object, and it is a sufficientcondition forKshaving ontological priority
over A*s that A*s are set-theoretical constructions out of Ks.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

89

person (because not psychologically continuous) which is


physically continuous throughout its existence. Equally, it is to
continuity is only a necessary condition of
say that psychological
personhood, which is to say, in turn, that it is at least possiblefor
there to be something which is not a person (because not
physically continuous) which is psychologically continuous
throughout its existence. But only a subject of mental states can
exhibit psychological continuity, and only a concrete object can
be a subject of mental states.
The proponent of physical-cum-psychological continuity as
the criterion of diachronic personal identity might not have
thought himself committed to the possibility of non-personal
physically discontinuous subjects of mental states, but on my
interpretation of his proposal, just as much as on the fourdimensional interpretation (on which he will be regarded as
proposing an account of the 'unity relation' for persons), he is.
To put the point somewhat differently: the proponent of
physical-cum-psychological continuity as the criterion of diachronic personal identity will of course hold that in such cases of
'mind-swapping' (information transfer) as are described in
Williams 1970, personal identity will not obtain between the Abody person before the mind-swap and the B-body person
afterwards.But if he interpretspsychological continuity in such
to the
a way that he claims that these cases are counterexamples
of
condition
is
sufficient
a
that
proposal
psychologicalcontinuity
I
as
is
he
then
over
time
committed,
personal identity
understand his position, to the claim that there is present in the
case, in addition to various physically (and psychologically)
continuous persons, two other physically discontinuous but
psychologically continuous subjects of mental states. The
proponent of physical-cum-psychological continuity as the
criterion of diachronic personalidentity might not have thought
that he was so committed, but on my interpretation of his
proposal, just as on the four-dimensional interpretation, he is.
The same point holds, mutatismutandis,whatever the concept
for which a criterion of diachronic identity is being proposed
and whatever the specific content of the proposal. To hold that
informative criteria of diachronic identity can be given for the
kinds of persisting things which we talk and think about in our
everyday lives, is to be committed, on my interpretationof what

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

90

II-HAROLD W. NOONAN

the request for such a criterion comes to, just as much as on the
four-dimensionaltheorist'sinterpretationof that request, to the
possibility (and, therefore, if the facts are so disposed, to the
actuality) of a host of entities not acknowledged in our everyday
thought and talk.3
Having explained my favoured frameworkfor the discussion
of so-called problemsof diachronic and synchronicidentity I can
now introduce the first sense which I can give to the notion of
ungrounded identity over time. I shall refer to this as
ungroundednessI.
The identity over time of persistingthings of a kind K will be
ungrounded1, then, if no informative diachronic criterion of
K-hood can be given.
Now, of course, it is not excluded that the specification of an
informative diachronic criterion of K-hood might require the
use of the concept of some other kind of persistingthing, K*. But
on pain of vicious regress, this cannot always be so. So either
there are kinds of persisting things whose identity over time is
ungrounded1,or else there are some kindsof persistingthings for
which diachronic criteria of kind-membershipcan be specified
without the use of the concept of any (other) kind of persisting
thing. But if this is so, and if informative diachronic criteria of
kind-membership can be specified for every kind of persisting
thing, there will be a clear sense in which both the concept of a
persisting thing, and concepts of particular kinds of persisting
thing, are redundant.All facts about persistingthingswill supervene on facts specifiable without the use of any such concept.
Now in Kripke's unpublished, but much discussed, lectures
on identity, he introduces the notion of a 'holographic state' of
the universe at a moment, a sort of three-dimensionalpicture of
the universe as it is at a moment.4 The important thing about
this notion is that the holographic state of the universe at any
instant is to give us the complete state of the universe at that
instant withoutprejudiceto whethersuccessiveholographsare showing
3Though I must add that as far as I can see not everyQuinean physicalobject will have
to be acknowledged as a real (and concrete) entity; for example, this will not be so for
that Quinean physical object composed of a momentary stage of a silver dollar in 1976
and the Eiffel Tower through its third decade.
4For those who do not know Kripke's lectures the best source of information about
them is Shoemaker 1984.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

91

thesameordifferent
objects.The holographic state of the universeat
an instant is to be thought of as providing all the information
there is about the state of the universe, and the propertiesof its
inhabitants, at that instant, but not information about their
numerical identity with, or numerical distinctness from, items
depicted in earlier or later holographs.
Now it would be a sufficient condition of the redundancy of
any concept of a persisting thing, and hence a sufficient
condition for all identity over time to be groundedl, if there were
no facts about the universe 'over and above' facts about its
holographic states, i.e. if the totality of facts about the universe
supervenedon the totality of facts given by its holographicstates.
Let us call this 'the strong holographic supervenience thesis'.
But even if the strong holographic supervenience thesis is
false, identity over time may still always be grounded,. For all
that this requires is a weaker supervenience thesis, namely that
the totality of facts about the universe supervene on the totality
of facts given by its holographic states togetherwith thoseof their
whichcanbe
(whichmayincludee.g. causaldependencies)
relationships
specifiedwithoutemployingtheconceptof a persistingthing.This is a
weaker thesis because there may be relationships between
holographic states specifiable,without the use of the concept of a
persisting thing, which do not supervene on the totality of facts
given by those holographic states.
I shall call this second supervenience thesis 'the weak
holographic supervenience thesis'. Identity over time is ungrounded1 if, but only if, it is false. I return to this question
in section V.
IV
There is another line of solution to the puzzle I outlined in
section II, however, which has also had an influence on the
discussionof problems of identity over time. We can call this, for
brevity, but somewhat misleadingly, the Fregean solution.
According to Frege's familiar proposal we can introduce the
functor 'the direction of' by the stipulation that:
the direction of a = the direction of b iff a is parallel to b.
We thereby fix the criterion of identity for directions as the
relation of parallelism between lines.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

92

II-HAROLD

W. NOONAN

With this fixed we can go on to explain 'x is a direction' as


meaning 'for some line a, x is the direction of a',sand we can go
on to explain the predicates of directions in terms of those
satisfiable by lines, subject to the constraint that the truthconditions of each statement of the form 'the direction of a is F'
be given by a statement of the form 'line a is PF', where 'F*'
denotes a property of lines for which parallelismis a congruence
relation. Thus, starting from a specification of the criterion of
identity for directions we can go on to explain the whole
'language game' in which we speak of directions.
This approach is susceptible to generalization. Foranykind of
object K we can ask (although we cannot always receive an
answer): (i) what entities play the role for Ks that straight lines
play for directions, and (ii) what relation plays the role for Ks
that parallelism plays for directions?
We can apply this Fregean approach to problems of
diachronic identity and we can do so without presupposingthe
Quinean four-dimensionalscheme.' The entities which stand to
Ks as straight lines stand to directions might be ordered pairs of
persisting things (of a distinct kind K*) and times, and the
relation which serves as the criterion of diachronic identity for
Ks might be a relation between pairs of such ordered pairs.
Where a and b are A*s the criterion of diachronic identity for
Ks could then be given in the form:
the Kof which K*a is a manifestation at t = the Kof which
K'*bis a manifestation at t' iff <a,t>R<b,t'>.
5At this point, of course, Frege himself proceeded differently, explicitly defining
directionsas classesof parallel lines (or rather,as the extensionsof certain concepts). This
is why I warned that it was somewhat misleading to describe the approach outlined in
the text as 'Fregean'. However (apart from the fact that the problem which causes Frege
to take this line (The 'Caesar'problem) is not solved by it, but only pushed furtherback)
if we were to be faithful to Frege here it would appear completely implausible to suppose
that the Fregean approach could have application to any problems of identity apart
from those for types of abstractobject. For a good discussion of Frege's views on these
matters, to which this section is heavily indebted see Wright 1983.
6But what is the relation of this approach to problems of diachronic identity to the
approach outlined in the previous section? The answer, I think, is that the Fregean
approach is just another way of developing the Quinean insight that questions about
identity criteria reduce to questions about kind-membership;in specifying the Fregean
criterion of identity for directions one is specifying exactly what it is to be a direction,
namely something for which questions of identity and distinctness are reducible to
questions of parallelism between lines. That is all there is to being a direction (see again
Wright 1983).

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

93

Being a K could then be explained as being the K of which


some <a,t> is the manifestation,and the satisfactionconditionsof
predicates satisfiable by Ks could be explained in terms of the
properties possessed by K*s at times (equivalently, relations
between K*s and times, or propertiesof orderedpairs off*s and
times).
Thus, if one is prepared to discount the possibility of
disembodiedexistence, one might take the problemof diachronic
personal identity to be the problem of specifying the relation
which body a existing at time t must bear to body b existing at
time t', iff the person occupying a at t is to be identical with the
person occupying b at t'. Here the K*s are bodies.
Or, if one finds the idea of a Lockean thinking substance
intelligible, one might take it to be the problem of specifyingthe
relation which thinking substancea, existing at time t, must bear
to thinking substance b, existing at time t' iffthe person in which
a thinks at t is to be identical with the personin which b thinksat
t'. Here the K*s are thinking substances.
But although the Fregean approach can be applied to
problems of diachronic identity without presupposingthe fourdimensional scheme, it is not, of course, incompatible with it; so
one might also take the problem to be that of specifying the
relation which must hold between person-stagea, existing at t,
and person-stage b, existing at t' iff the person of which a is a
stage at t is to be identical with the person of which b is a stage at
t'. Here the K*s are person-stages, and if they are momentary
the reference to times is redundant.
The Fregean approach, then, provides a second way of
making unproblematicsense of requestsforcriteriaof diachronic
identity. But it should be noted that it, too, requires the
recognition, as ontologically on a par with the familiar things of
our everyday acquaintance, of a multitude of extra entities
which go unremarked in our everyday thought and talk. For if
the criterion of diachronic identity for a familiar kind of
continuant K is given as an equivalence relation R, between
ordered pairs of K*s and times there will be equivalence
relations entailed by R, or entailing R, equally capable of
serving as criteria of diachronic identity. And if the availability
of a criterion of diachronic identity for Ks in terms of K*s does
not oblige us to regardKs as set-theoretical constructionsout of

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

94

II-HAROLD W. NOONAN

K*s (in which case, as noted in footnote 5, the Fregean approach


to problemsof diachronic identity becomes completely implausible) then we are no more obliged to regard the unfamiliar
entities for which these latter relations can serve as criteria of
diachronic identity as set-theoretic constructions out of K*s.
There will thus be no reason not to regard them as ontologically
entirely on a par with Ks, albeit of less interest to nonphilosophers.
Now I can introduce a second sense I can give to the notion of
ungrounded identity over time-ungroundedness2.
The identity of Ks over time will be ungrounded2 iff no
Fregean criterion of diachronic identity for Ks can be given.
Now I said above that the applicability of the Fregean
approach to problems of diachronic identity did not requirethe
four-dimensional framework, but, of course, on pain of vicious
regress, if Fregean criteria of diachronic identity are to be
available for all kinds of persistingthing, there must be a classof
non-persisting things to relations between which the identity
over time of at least some of these things reduces.
Thus it will be a sufficient condition of identity over time
being ungrounded2 that such a class of non-persisting things
does not exist.
But even if this is so, it is to be noted, it will not immediately
follow that identity over time is ever ungroundedl. This could
only be so if the weak holographic supervenience thesis entailed
the existence of such a class of non-persisting things, and it is
unclear that it does so.
V
In his unpublished lectures Kripke argues explicitly against the
strong holographic supervenience thesis, and perhaps against
the weak holographic supervenience thesis also. Thus his
arguments present a challenge to the thesis that identity over
time is never ungrounded1, and Kripke does explicitly take
himself to be arguing for a sense in which identity over time is to
be regarded as a primitive or ungrounded fact.
In this section I wish to explain why Kripke's argument does
not convince me that identity is ever ungrounded1.I should add,
however, that my rejection of Kripke's argument is very
tentative, for, as will emerge, there is a substantial price to be

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

95

paid for its rejection, and whether that price is worth paying is
debatable.
Kripke's argument (for which those who do not know
Kripke's lectures should consult Shoemaker 1984) concerns a
homogeneous disc, stationary in one world, rotating in a second.
The identity of the disc is not in question, but consider the
portion of matter which makes up its northern half at some time
t: if the disc is stationarythis will be the sameportion as makes up
its northern half at any later time t'. Not so if the disc is rotating.
Thus the two worlds will differ with respect to what identities
hold over time, and hence with respect to facts statable in the
form 'Some portion of matter is in P at t and P' at t". If then they
do not differ in their holographic states the two worlds are a
counter-example to the strong holographic supervenience
thesis, and if they differ neitherin their holographic states norin
the relations between those states which are specifiable without
the use of the concept of a persisting thing they are a counterexample to the weak holographic supervenience thesis-and
hence a proof that identity over time is sometimesungrounded1.
Kripke certainly takes his disc example to play the first of
these roles, and he may take it to play the second. However, I am
unconvinced.
First, it has to be emphasized that for Kripke's example to
work the two worlds must differ not at all in their momentary
holographic states; it is not enough that they exhibit no such
difference locally, i.e. no such difference where the disc is
located. Nor is it enough that they exhibit no such difference
whilst the disc is in existence. For either holographic supervenience thesis to be refuted by the example the two worlds must
differ not at all, anywhere, anytime, in respect of their
momentary holographic states. For to refute a supervenience
thesis one must describe two possibilities differing not at all in
respect of the proposed supervenience base, but differing in
supposedly supervening facts.
This point, I think, puts Kripke'sexample in a ratherdifferent
light from that in which it initially appears. At first sight the
example seems perfectly commonplace-one simply imagines
the disc and notes that it will look exactly the same in each
momentary holograph whether it is spinning or not. But this is to
ignore the point of the previous paragraph, and once that is

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

96

II-HAROLD W. NOONAN

taken into account it becomes clear that the example is a good


deal less ordinary than it might at first sight have appeared. It is
no good for instance to think of the disc as set in motion in world
1 (the world in which it rotates) by someone's giving it a twist.
For if this happens in world 1 but not world 2 the two worldswill
differwith respect to their holographicstates beforethe disc is set
spinning. Either the disc is always spinning in world 1, then, or it
came into existence spinning there, or its spinning was startedby
a cause which was not present in world 2 but whose absence
from that world makes its holographic states no different from
those in world 1.
I do not say that these points prove the unintelligibility of
Kripke's example, but only that they make it clear that it should
not be treated as wholly uncontentious.7
If the example is accepted as an intelligible one, however, it
does refute the strong holographic supervenience thesis, and
since, despite the words of cautionjust given, I do not feel able to
deny its intelligibility, I accept that that thesis is refuted. But
what matters from the point of view of this paper is whether
Kripke's disc also refutes the weak holographic supervenience
thesis, and that I am not prepared to admit.
For even if the existence of two worlds related as Kripke
requires if the strong holographic supervenience thesis is to be
refuted has to be admitted, these two worlds must still differ in
respect of the counterfactualstrue of the relationswhich obtain
between their holographicstates. For example in world 2, where
the disc is stationary, it will be true that if its northern half had
been marked in a certain way at a time t then its northern half
would have had that mark on it at a subsequenttime t', whereas
in world 1, where the disc is spinning this will not be so
(assumingt' is sufficientlylater than t); but other counterfactuals
will be true, e.g. that if the northern half of the disc had been
marked in a certain way at t its southern half would have
exhibited that mark at t' (cf. Shoemaker 1984:224).
If, then, these counterfactuals can be stated without the
'7If, that is, its purposeis to refuteone of the holographicsuperveniencetheses,but it can
be treated as wholly uncontentious if it is taken (as Kripke also takes it) as a counterexample to the thesis that spatio-temporal continuity (under a sortal) suffices for
transtemporal identity, for then it does not require that worlds I and 2 be globally
indistinguishable in respect of their holographic states.

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE, IDENTITY AND TIME

97

employment of the concept of a persistingthing the two worlds


will not be related in such a way as to refute the weak
holographic supervenience thesis, or to establishthe existence of
identities which are ungrounded,. But it does seem that these
counterfactualscan be stated in this way. (Of course, as actually
stated they presuppose the identity over time of the whole disc,
but that is irrelevant.)
There is a natural counter to this argument, however, which
is to point out that if the weak holographic supervenience thesis
is to be maintained in this way in the face of Kripke's disc
example (while the strong holographic supervenience thesis is
acknowledged to be refuted by that example) then the
counterfactualsin respect of which the two worlds differmust be
accepted as barelytrue. For they cannot be true in virtue of
differences in the two worlds' holographic states-there are
none. Nor can they be true in virtue of the facts about identity
over time in respect of which the two worlds differ-for these
latter differencesare supposed to obtain, according to the weak
holographic supervenience thesis, becauseof the difference in
counterfactuals, not the other way round. (The same point can
be approached by recalling the point of insistence on the weak
holographic superveniencethesis,namely, to enable retention of
the thesis that identity over time is always grounded1,i.e. that
there is no kind of persisting thing for which a diachronic
criterion of kind membership can not be given. But if reference
to counterfactual relations between momentary holographic
states has to enter into an account of what it is to be a K, the
obtaining of such counterfactual relations can hardly be
explained as grounded in the existence of a K.)
It was this counter-argument that I had in mind when I said
earlier that a high price had to be paid for resistanceto Kripke's
argument-the price is the acceptance of the possibility of
barely true counterfactuals.
I am not, however, convinced that the price is too high to pay.
For the alternative is to postulate as the grounds of the
counterfactuals facts about identity over time which ex hypothesi
cannot be grounded in anything else. But our understanding of
what it is for these identities to obtain and to differ between the
two worlds is mediated only by our understanding of the
counterfactuals which obtain there. So it is unclear that

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

98

II-HAROLD W. NOONAN

postulation of the identity facts can be anything other than a


way of preserving the letter of the thesis that counterfactuals
cannot be barely true.
I conclude, albeit tentatively, that Kripke'sdisc example does
not prove that identity over time is ever ungrounded,--though
it does prove (assuming that it refutes the strong holographic
supervenience thesis) that eitherthis is so or counterfactualscan
be barely true.
There is a final point to be made about Kripke's argument,
however: even if it should be accepted without reservations it
does not entail that the mere asking of questions about
diachronic identity criteria (as such questions are interpretedin
section III) is illegitimate. For even if it does establish that
identity over time may sometimesbe ungrounded1,it establishes
this only for entities of a very special kind-homogeneous masses
of matter which (at some time in their existence) are proper
parts of larger masses of the same kind of homogeneous matter.
And it is unclear that reference to entities of these kinds would
ever be required in an account of the criteria of diachronic
identity (diachronic criteria of kind membership) of the kindsof
entity that our more familiar sortal terms signify. But if not then
even if Kripke'sargument is wholly correct it leaves a very large
area within which queries about criteria of diachronic identity
are entirely legitimate.
VI
Jonathan Lowe's own argument for ungrounded identity is
differentfrom Kripke's and it is unclear how, even if it is valid, it
can refute the (weak) holographic supervenience thesis. (It does
not present us, as Kripke'sargument does, with a pair of possible
worlds identical in the supposed supervenience base but
differing in the allegedly supervening facts.) Consequently, I
cannot see that it establishesthe existence of identitieswhich are
ungrounded1.But that, I think, is not its aim. Rather its aim is to
refute the idea that a class of non-persisting entitiesexists, to
relations between which the existence of persisting things
reduces. That this is so is, of course, part and parcel of the fourdimensional picture, and is also entailed by the thesis that
identity over time is never ungrounded2, i.e. that Fregean

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

SUBSTANCE,IDENTITYAND TIME

99

criteria of diachronic identity can be given for all persisting


things.
If the idea he objects to were correct, Lowe argues,
everything's persistence would ultimately be accounted for by
reference to change. But the very notion of a change presupposes
that of the persistence of at least somethingthrough that change
and so cannot be appealed to in giving a perfectlygeneralaccount
of what the persistenceof objectsconsistsin. The persistenceof at
least somesort of thing must thus be primitive or ungrounded in
that it consists neither in relationships between non-persisting
things, nor in the persistence of other sorts of things.
I agree with part of this. It does seem right to say, and Lowe
seems to argue effectively, that the notion of change is bound up
with the notion of persistence.If there is to be change there must
be that which persists through the change. For if nothing
persisted through the change, then, as Lowe says, with what
right could we regard the events occurring before the change as
belonging to the same possible world as the events subsequentto
it? And how then could we regard the change as a change?For if
nothing persists it is not a change in anything, but at best a
change in how things are in (what features are present at) a
certain spatial location. But in the absence of any frameworkof
persisting objects how can we make sense of the same spatial
location first exhibiting one feature and then another? I agree
with Lowe then that the hypothesis of a global existence-change
is incoherent: if a possibleworld is one in which there is change it
is also one in which there are persistingthings (not necessarily,as
Lowe points out (footnote 12) to be picked out by count nouns)
which provide the background to change.
But to say this is only to say that the applicability of the notion
of change to a possible world entails the applicability of the
notion of persistence, and that, as far as I can see, is quite
compatible with the four-dimensionalscheme, or with the thesis
that Fregean criteria of diachronic identity are available for all
persistingthings. What does follow, if either of these is correct, is
that whenever anything persists,something
elseceasesto exist, and
elsecomes into existence. But that is not to say that really
something
nothing persists (cf. the well-known objection to Lewisian
Counterpart Theory that it entails that reallyall properties are
essential-it doesn't).

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions

100

II-HAROLD W. NOONAN

Thus I fail to see that this argument of Lowe's establishes its


conclusion.s
To sum up: I began by presenting a puzzle which seemed to
cast doubt on the intelligibility of requests for criteria of
diachronic identity. I then presented two responses to this
puzzle and in the light of them defined two senseswhich could be
given to the notion of ungrounded identity over time. I argued,
however, albeit tentatively, that no proof of ungrounded
identity over time in either sense had been given, either by
Kripke in his famousunpublishedlecture or by Lowe. I have not
claimed, however, that no such proof can be given.

REFERENCES
Brody, B., 1980, Identityand Essence,Princeton, Princeton University Press.
Lewis, D., 1986, On the Pluralityof Worlds,Oxford, Basil Blackwell.
Perry, J., 1972, 'Can The Self Divide?', TheJournalof Philosophy,69.
Quine, W.V.O., 1976, 'Worlds Away', The Journalof Philosophy,73.
Shoemaker, S., 1984, 'Identity, Properties and Causality', in Identity,CauseandMind,
Cambridge, Cambridge University Press.
BritishPhilosophy,
Swinburne, R., 1976, 'Personsand Personal Identity', in Contemporary
ed. H. D. Lewis, London, George Allen and Unwin.
Williams, B., 1970, 'The Self and the Future', Philosophical
Review,79.
Wright, C., 1983, Frege'sConception
of Numbersas Objects,Aberdeen University Press.

8Independently of this argument Lowe also directs some criticism at the fourdimensional notion of a 'thing-stage', arguing that even if the existence of such entities is
admitted they are incapable of being identified independently of the things of which
they are the stages. I lack the space to go into this; but note that this will not be so if
Lewis's account of thing-stages,more particularly,person-stages(mentioned by Lowe in
his footnote 6), is accepted. In fact, the argument of Lewis's that Lowe refersto seems to
me very powerful (but one of its premissesis the weak holographic supervenience thesis
so it is not wholly uncontentious). Lowe objects to its first premiss ('First: it is possible
that a person-stage might exist. Suppose it to appear out of thin air then vanish again')
that all it does is to introduce us to the fanciful notion of a very short-lived person. But
this is merely to draw attention to the fact, which Lewis explicitly acknowledges, that
person-stages as he characterizes them are entities it would be perfectly correct to
describeas personsif they were not proper partsof similarlycharacterizablewholes. This
does not entail that it is correct to describe them as personswhen they areproper partsof
such wholes (though it might be, as the philosophersreferredto by Lowe in his footnote 5
claim; but this would not affect the cogency of Lewis's argument).

This content downloaded from 200.3.149.179 on Tue, 28 Oct 2014 17:31:43 PM


All use subject to JSTOR Terms and Conditions