You are on page 1of 11

Origin and Identity

Author(s): Graeme Forbes


Source: Philosophical Studies: An International Journal for Philosophy in the Analytic
Tradition , May, 1980, Vol. 37, No. 4 (May, 1980), pp. 353-362
Published by: Springer

Stable URL: https://www.jstor.org/stable/4319382

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.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical
Studies: An International Journal for Philosophy in the Analytic Tradition

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
GRAEME FORBES

ORIGIN AND IDENTITY

(Received 15 June, in revised form 12 November, 1979)

0. In this paper I shall attempt to explore some of the connections between


our conception of the identity of objects and the thesis that living things
are essentially related to the entities from which they develop, the thesis of
the necessity of origin. Kripke has queried whether we can make sense of the
supposition that the Queen might have had parents different from her actual
ones.1 Behind this query there lies a general principle. To formulate it
precisely, let us borrow the term 'propagule' from biology; x is a propagule
of y iff x is one of the entities which grew or developed into y. Then the
thesis of the necessity of origin may be written as

(K) LI(Vx)EJ(Vy)L(Prop (x, y) -+ EI(Exists (y) -+ Prop (x, y))),

where 'Prop (x, y)' abbreviates 'x is a propagule of y'. My strategy in


exploring the connections between origin and identity will be to investigate
the consequences of scepticism about (K).

1. I shall begin by making two asumptions, the first of which I shall not
argue for. This is that the thesis of the necessity of identity is true, which
means that the extension of the identity relation is the same at every possible
world. The second assumption is that de re truths about metaphysical
possibility, and in particular the transworld identity claims they entail,
cannot simply be barely true; for any such truth there must be something of a
relatively less problematic nature in virtue of which it is true, or, equivalently,
if the required transworld identity relation obtains, there must be facts in
which its obtaining consists. 2 Pinning these notions down is a tricky business,
but it is easy to illustrate their intent with examples. Consider, for instance,
the case of identical (monovular) twinhood in humans. Identical twins are
produced when the normal processes consequent upon the formation of the
zygote (the entity which is the immediate result of the fusion of the father's
sperm with the mother's egg) break down in a certain way: after the first

Philosophical Studies 37 (1980) 353-362. 0031-8116/80/0374-0353$01.00


Copyright ? 1980 by D. Reidel Publishing Co., Dordrecht, Holland, and Boston, U.S.A

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
354 GRAEME FORBES

mitotic division of the zygote, the resulting daughter cells separate and
develop independently into distinct embryos. From our point of view, the
important fact is that mitotic division is physically symmetric, for each
daughter cell receives a copy of the chromosomes of the parent cell, and each
copy is, so to speak semi-original. A chromosome consists of a pair of
intertwined strings of DNA, and each string is itself a sequence of molecules
called nucleotides. Replication is effected by the intertwined strings
unravelling from each other, each string then acting as a template for the con-
struction, from materials present in the cell, of a sequence of nucleotides
exactly like the one from which it has just unravelled. Then each original
DNA string resumes the double helix structure by intertwining itself with the
new string whose construction it has just directed. Thus two chromosomes
are obtained from one, each new chromosome contains half the matter of the
original chromosome, and the new matter in such a pair of daughter
chromosomes has the same source. The two chromosomes now proceed to
opposite ends of the cell nucleus; since replication has been occurring with all
the chromosomes in the cell, division of the cell 'down the middle' produces
a pair of cells identical in genetic content, each containing half the genetic
material of the original cell.
Suppose, then, that A is an actual but untwinned human being. Could A
have been an identical twin? Granted that the intension of 'identical twin'
is fixed by the actual mechanics of mitosis, an affirmative answer to this
question amounts to the claim that there is a possible world in which A's
zygote undergoes fission as described, followed by non-standard separation,
and A is identical to one or other of the two resulting individuals B and C.
But we ought to be very reluctant to say that there is such a world. Of course,
A's zygote could divide in the required way; but from the account of mitosis
just sketched, it is evident that there is nothing in such a situation to
determine which of B and C is identical to A. Nothing makes A one of th
rather than the other, that is, there is nothing in virtue of which such an
identity might obtain. Perhaps the example is rather underdescribed. There
will be some worlds in which B turns out to be rather more like A as he is
the actual world than C does; but also vice-versa. And there will be some
worlds in which both B and C are equally dissimilar to A as he actually is. At
least with respect to these worlds, an identity claim would be completely
unmotivated. Moreover, the necessity of identity prevents us from saying that
A is identical to B in some worlds, and to C in others, depending upon how B

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
ORIGIN AND IDENTITY 355

and C in those worlds compare to A


who insists that A could have been
one of B or C rather than the oth
undetectable by us since it is enti
verification-transcendent as you pl
conception unintelligible is a meas
about metaphysical possibility whic
What can we say to someone who
simply does not see why there sh
required kind? Perhaps no argume
such a person should notice that t
with is a familiar one. Its present
discussed by writers on personal identity. To hold that A could be an
identical twin is just like holding that one or other of the products of a split
brain operation simply is identical with the original person, although there ar
no relatively less problematic features of that product's relationship to the
original person which differentiate him in any way from the other which
could reasonably be regarded as grounding the identity. This position has bee
held, but it does not recommend itself.3

2. Let us apply these considerations to the hypothesis that some organism,


say the oak tree presently growing outside my study window, could have had
a propagule distinct from the one it actually had. Suppose that this oak tree
has actually grown from an acorn dropped in the then naked soil some time ag
by a certain bird. Since we can best justify assertions that a given state of
affairs is possible by citing a non-vacuously true counterfactual whose
consequent says that that state obtains, a sceptic about (K) might argue as
follows. If that oak tree had been growing where it actually is through just
the stretch of time it actually has been there, but if a strong wind had been
blowing in the opposite direction from the one in which the wind was
blowing when the bird did plant the acom, then if that tree had grown from
an acorn dropped by the same bird with just the powers it actually had, it
would have had to collect the acorn from a closer tree than the one from
which it actually collected it. So the actual oak tree would have come from
a different acomr.4 In general, if we take a world w at a given time t' and an
actual object in a genuinely possible state at t' in w, and then specify how
things were in w at an earlier time t and perhaps also a little of the history

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
356 GRAEME FORBES

of w from t to t', then if we ask how w has to be if it is to be more similar


to the actual world than any other world meeting the specifications so far,
we find that in some cases, attributing a propagule to an object which it does
not have in other worlds in which it exists satisfies that requirement. In the
present case, the sceptic's counterfactual can be judged false only if we
assume the necessity of origin.
The argument our sceptic gave employed a backtracking counterfactual,
but I do not hold that against it. Nor do I object (for it would be a mistake to
do so) that it ignores the constraint on metaphysical possibility, the require-
ment of relatively less problematic grounding for transworld identity claims,
for which I have argued. The defect of the sceptic's example is that the
features it includes are insufficient to make the possible tree he described
identical to the actual one. It looks like the actual tree, perhaps, and has been
growing at the actual tree's location during just the period of time the actual
tree has been growing there; but this is not enough. Let us refer to the proces-
ses by which an organism comes into being as its generation system (G.S.); in
the case of an oak tree, these will be the interactions of the acom with the
soil around it and the consequent physical changes in the acom which cause it
to take root. Now in the world the sceptic described, there was only one
generation system, call it S', to consider; in other words, there was no rival
claimant for being the actual tree. But the system in question is clearly com-
possible with the actual G.S. of the actual oak tree, which we shall refer to as
S, so long as it is agreed that the location of S is not one of its essential
properties. So let us imagine a world where both S and S' occur. Which
output is the actual tree? We do not have to assume that the output of S
is the stronger claimant, as is sometimes suggested.5 The sceptic has to choose
between saying that (i) the output of S is the actual tree in such a world,
or (ii) the output of S' is, or (iii) neither is, and whichever answer he gives,
he will be faced with a certain powerful objection, deriving from our demand
for facts in virtue of which transworld identity claims hold. The difficulty
for the sceptic is that all three possible answers have the consequence that a
generation system which has a given output in one world w has a distinct
output in another world w', where the salient features of w' can be described
by simply extending the description of the salient features of w, so that w'
contains all the relevant features of w. For example, if the sceptic gives
answers (i) or (iii) to the question, then w is the world he described where
only S' occurred, and w' is a world just like w so far as S' is concemed, but

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
ORIGIN AND IDENTITY 357

in addition S also occurs in w'. And if the sceptic returns answer (ii), that in
S & S'-worlds it is still the output of S' which is the actual tree, then we can
invite him to consider a world w which only contains S, but S exactly as it is
in some S & S'-world w', such that it is agreed that the output of S in w is
the actual tree.6 Then w and w' are related as above. But in general, if w and
w are thus related with respect to a system which occurs in both, then if the
output of that system in w' is non-identical to a certain object, its output
in w cannot be identical to it, for w and w' have so been chosen that all the
features in virtue of which the output of the system in w has a certain
identity reappear in w'. In particular, for choices of w and w' meeting the
stated conditions for answers (i) and (iii), if the output of S' in w' is not the
actual tree, then the output of S' in w cannot be the actual tree either; and
for w and w' as described for answer (ii), if the output of S in w' is not the
actual tree, then its output in w cannot be either, contradicting the
descriptive condition by which w was introduced.

3. There is a general line the sceptic can take, on which he can produce a
number of variations, in order to resist this objection: the idea is to acquiesce
in the employment of the notion of that in virtue of which a certain
possibility is realized, but to deny that any case with both generation
systems contains all the relevant features of his case with S' alone, since, he
could say, the non-occurrence of S is a relevant feature of the latter, and is
one of the features in virtue of which the output of S' is the actual tree whe
other circumstances are as specified in the antecedent of the backtracking
counterfactual. But this position is surely untenable. How can what goes on
concerning another acom affect the identity of this tree? How can it be that
if certain things had happened which actually didn't then a given object
would have had a certain identity, but would have lost that identity without
any change in internal state whatsoever simply if some other causally isolated
process had also occurred?
Someone who thinks that this is possible believes that transworld identity
is extrinsically determinable, or as I shall put it for short, extrinsic. We can
get some grip on the content of this view by comparing the extrinsicness
thesis (hereafter 'E-thesis') for transworld identity with the E-thesis for
identity through time. Imagine a sequence of spatio-temporally continuous
thing-stages uninterrupted observation of which gives every appearance of the
presence of a single persisting thing, and let t and t' be times in the releva

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
358 GRAEME FORBES

period and s and s' the observed thing-stages at these times. Then the E-
thesis for identity through time says that whether or not s and s' are stages
of a single continuant is determined not just by intrinsic features of the
relationships amongst the various stages in the sequence that links s and s',
but also by the properties of causally unrelated thing-stages at t'. Reverting
again to the example of the split brain operation, the E-thesis would allow us
to hold that if the operation is performed on A and both half brains
transplanted to give 'new' individuals B and C, then neither B nor C is
identical to A, but if only one half is transplanted, say the B-half, while
the other is destroyed, then the resulting individual is identical to A. On
such a view, if the second case obtained, it would be appropriate for the
resulting individual to say 'Thank goodness the other half brain was
destroyed, otherwise I wouldn't have existed'. That is, his identity is fixed
by events which involve the other half brain and do not involve him.
However, we can imagine that the situation is set up in such a way that this
difference in identity would be completely unreflected by any other difference
in the states of that individual.
By analogy, the E-thesis for transworld identity says that whether or not
an object x in a world w is on the same 'transworld heirline' (Kaplan's phrase)
as an object y in another world w' tums upon whether certain events in which
y is not involved do or do not occur in w'. To see how false this is to any
conception of identity, compare transworld identity with the crossworld
relation which holds between x in w andy in w' iffy is the most similar thing
in w' to x as it is in w. Clearly, in order to determine the extension of this
relation for a pair of worlds it is necessary to look not just at how similar x is
in w to y in w', but also at how similar other things are in w' to the way x is
in w. That is, the E-thesis for crossworld similarity as defined is true. But
transworld identity is a different relation from crossworld similarity; when
the two notions are clearly distinguished, the E-thesis for the former loses any
vestige of plausibility.7
Are there any temporal phenomena which would support an E-thesis in
the modal case? In our example, the identity of a tree in a world is tuming
upon what other trees there are in that world. This conception has the follow-
ing temporal analogue: keeping one's eye on a particular tree-like arrange-
ment of wood throughout a period of time could not yield even prima facie
evidence that one had been observing the same tree throughout that period;
for that conclusion, one would also need information that no other trees

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
ORIGIN AND IDENTITY 359

meeting certain specifications had come into or gone out of existence


during that time anywhere else in the universe. If the modal version of
this is the price we have to pay for scepticism about (K), we are well advised
not to buy.8

4. These arguments against the sceptic all exploit the fact that he proffered
alternative generation systems for actual objects which were compossible
with their actual generation systems, forcing ad hoc variations in the identity
of certain objects from world to world according to whether or not the
actual generation-system also occurs in the world. But this defect can be
remedied by a modified scepticism, according to which any alternative origin
for an object must involve a generation system non-compossible with its
actual generation system. The case of organisms such as human beings, whose
generation systems involve propagule-fusion, is one to which such a scepticism
would fruitfully apply. Suppose A is a human being whose zygote is actually
formned from the fusion of a sperm b with an egg c, and let us call A's genera-
tion system Sl. Then S, is non-compossible with S2, a generation system
involving the fusion of the spenn b with an egg c' non-identical to c, since
b cannot survive fusion as an entity with an identity of its own. So a back-
tracking counterfactual can be constructed as before such that, barring the
necessity of origin, we would be strongly inclined to hold that some
antecedent world in which A is the output of S2 is more similar to the actual
world than any in which A is not. The counterfactual would therefore be
true, its antecedent would describe genuinely possible circumstances, and in
them A would have S2 ; so A could have had S2 .
However, modified scepticism succumbs to essentially the same objection
as was made against its unqualified predecessor. If A could have had S2, then
A could have had S3, a fusion of a sperm b' non-identical to b with the egg c
of SI. So let us consider S4, a fusion of the sperm b' from S3 with c' from
S2. This process is compossible with SI , and we can use this fact to show that
the sceptic is now committed to an indeterminacy thesis about transworld
identity just as objectionable as the E-thesis. Here are the four processes, each
with its output:

b c b cI b' c b' c'


s I \// s2 \\// S3 \\/ S4 \/
A 02 03 04

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
360 GRAEME FORBES

The problem with modified scepticism, essentially, is that if there are facts
in virtue of which A could have had S3, it is hard to see why there should not
be facts of the same kind in virtue of which it could have had S4, so that the
modification effected by this form of scepticism would simply rule out a
genuine possibility by ad hoc stipulation. One of the ways in which this fault
manifests itself is in the consequence of the position that the identities
of objects in a set of possible worlds will vary according as different choices
of an actual world in the set are made. That is, the identities of objects in a
structure are indeterminate, and the indetenninacy is differently resolved
(the transworld heirlines differently traced) as different models are defined
upon the structure. With respect to our example, for each Si, let wi, be a
world in which Si occurs. Then if w1 is actual, A, 02 and 03 are the same
thing, and 04 is something different, whereas if W4 is actual, 02, 03 and 04
are the same and A is distinct from all of these.9 So indeterminacy is induced,
not as extrinsicness was by surveying different parts of the modal universe
from the same place, but by altering one's perspective upon the same part of
the modal universe: the identities of 02 and 03 change according as w1 or
W4 is supposed to be actual. In terms of ordinary modal locutions, if C is
a set of conditions realized in W4 and not realized in any other world at least
as similar to the actual world as W4, then we cannot make a counterfactual
supposition beginning 'if C had actually been the case, ' without
disturbing the identities of actual and possible objects none of whose
properties are effected by the supposition. The temporal analogue of this
view would be that the identities of objects which exist at given times
sometimes alter simply in virtue of time's passing. One does not have to ho
a priori that time can have no causal powers to find this conclusion as un
ceptable as the E-thesis about transworld identity.

5. In fact, it is not difficult to see that modified scepticism must embrace


the E-thesis as well; for S2 and S3 are compossible and hence cannot have the
same output unless the identities of their outputs alter as we move from
worlds where they occur together to worlds where only one occurs. But a
formal manoeuvre can circumvent this difficulty: the sceptic need only
strengthen his necessary condition for systems S which involve propagules
distinct from those in SI to the condition that such a system is an alternative
generation system to SI for A only if it is a member of the set X of
generation systems non-compossible with SI and is non-compossible with

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
ORIGIN AND IDENTITY 361

every other member of X. Let us c


tion T does avoid commitment to the E-thesis about transworld identity,
it does not avoid indeterminacy. Speaking precisely, Condition T entails the
claim C that the identities of objects in a modal structure (W, A, v) change
simply in virtue of different choices of a world w from W to define a model
upon the structure.
Let us grant that for some object A there is an S which meets Condition
T and ascribes propagules other than those of SI to A. What might such an S
be like? Thinking of generation systems as mereological sums of events, the
model we have for one system being non-compossible with another is that
it have a part in common with the other. Hence the natural conception of a
system meeting Condition T is that of one which has a part in common with
every system that has a part in common with S1, and in addition, a part
in common with SI itself. So S is to be conceived of as an artificial hybrid,
constructed in thought out of the various physically possible generation
systems noncompossible with SI. No doubt such a thing is metaphysically
possible. But only one? It would be quite unreasonable to deny that if there
is one system satisfying Condition T which ascribes new propagules to the
output of SI then there is more than one, since such systems can differ from
one another with respect to which parts they have in common with some of
the natural systems which have a part in common with SI, though not with
respect to all, so that they are themselves mutually non-compossible. So let
Sk and Si be two such alternatives to Sl. Then it is easy to arrange that S,
is compossible with something non-compossible with Sk, for example, if Sk
has a part in common with Sm and S does not, where Sm is a process com-
possible with SI. Suppose also that Wk is a world in which Sk occurs. Then
we have the following situation: when w1 is actual, the output of Sk is
identical with the output of Sj by hypothesis. But when Wk is actual, the
output of Sk is distinct from the output of S, since Si is compossible with Sm,
which was stipulated to be non-compossible with Sk. Hence, by Condition
T, it is not, in this model, an alternative G.S. for the output of SI. This
verifies C, and since C is false, scepticism based upon Condition T must be
rejected.

6. The arguments of 4 and 5 traded upon the fact that non-compossibility is


not transitive. However, it is easy to manufacture a transitive relation out of
it, the relation of non-compossibility*, the ancestral of non-compossibility. If

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms
362 GRAEME FORBES

the sceptic about (K) is to hav


that an organism can have an alternative G.S. which ascribes it new
propagules only if that system is non-compossible* with its actual G.S. But
this is a self-defeating move if made by someone who wishes to maintain that
a person might have had on origin other than his actual one; for in the case of
creatures that reproduce sexually, or indeed for any case in which the shift to
qualified scepticism does make a difference, any two members of the species
will be such that their generation systems are non-compossible*, as we will
discover if we trace back far enough.10 On the basis of these results, there-
fore, I claim that there is no plausible way of combining a denial of (K) with
a correct account of the identities of actual and possible objects. And this, of
course, is exactly what we would expect if true attributions of origins to
organisms are necessary truths de re.

New College, Oxford

NOTES AND REFERENCES

1 S, Kripke, 'Naming and Necessity', in D. Davidson and G


of Natural Language (Dordrecht, 1972), pp. 312-314.
2 For one account of this distinction, see Michael Dumm
meaning? (II)', in: Truth and Meaning, ed. by Evans and McDowell (Oxford 1976),
p. 89.
3 See D. Parfit, 'Lewis, Perry and What Matters', in: The Identities of Persons, ed. by
A. 0. Rorty (California, 1976), pp. 91-107, esp. footnote 18.
4 This argument was suggested to me by J. L. Mackie.
See. J. L. Mackie, 'De what re is de re modality?', in: Journal of Philosophy 71
(1974), p. 560.
6 An anti-essentialist cannot consistently deny that there is such a world as w, on pain
of making the pair of the propagule and the location of the tree's origination essential to
it. But this argument would not trouble someone who wished to propose, as an alternative
to (K), that the place and time of an organism's origination is essential to it. In this
paper, I will have nothing to say about such a person.
7 Obviously, the claim I make here would be disputed by a counterpart theorist. Since I
do not have the space to argue against this theory here, it will have to be another premiss
of my treatment that objects exist in more than one possible world.
8 I am grateful to David Charles for comments on one earlier version of this section
and especially to Gareth Evans for spotting a flaw in another.
9 Suppose that all the outputs have the same career in their worlds. Then identities
compatible with the non-compossibility requirement will hold in both models or neither.
10 This sentence and its predecessor are due to Christopher Peacocke, for whose general
encouragement in the preparation of this paper I am grateful.

This content downloaded from


132.174.250.143 on Tue, 19 Oct 2021 17:52:28 UTC
All use subject to https://about.jstor.org/terms

You might also like