By James Giles
Philosophy East and West
Volume 43, Number 2
April 1993
(C) by University of Hawaii Press


The problem of personal identity is often said to be
one of accounting for what it is that gives persons
their identity over time. However, once the problem
has been construed in these terms, it is plain that
too much has already been assumed. For what has been
assumed is just that persons do have an identity. To
the philosophers who approach the problem with this
supposition already accepted, the possibility that
there may be no such thing as personal identity is
scarcely conceived. As a result, the more
fundamental question--whether or not personal
identity exists in the first place--remains unasked.
Consequently, the no-self theory, that is, the
rejection of the notion of personal identity
altogether, is never fully considered. One of the
reasons for the ignoring of the no-self theory seems
to be the failure of many philosophers to
distinguish between reductionism and the no-self
view. The reasons for this error are perhaps
understandable. For there is a sense in which the
two theories are in agreement. Both theories, for
example, reject the notion of a substantive self
which somehow exists beyond the bounds of
experience. The difference, however, is that while
the reductionist accounts then go on to resurrect
the self and, consequently, its identity, in terms
of putative psychological relations or various
theories of the body, the no-self theory lets the
self lie where it has fallen. This is because the
no-self theory is not a theory about the self at
all. It is rather a rejection of all such theories
as inherently untenable. And since reductionism is
just one more theory about the self, it, too, must
be untenable. In explaining the distinction between
these two theories it is instructive to turn to the
philosophy of mind, where we find a similar
distinction being employed. Here a distinction is
often drawn between reductive materialism and
eliminative materialism. Both these theories are in
agreement so far as they reject the existence of
mental phenomena. But while the former attempts to
do so by showing how the notion of the mental can be
reduced to the physical--that is, that those things
called minds are really just brains or states of the
brain--the latter rejects the notion of the mental
as fundamentally confused, and so hopes to eliminate
the idea of the mental altogether. The eliminative
materialist would argue that although the reductive
materialist is right to reject the idea of the
mental, he is wrong to think he can reconstruct the
mental in terms of the physical. This is because, it
is argued, discourse about things like
intentionality and awareness is simply not reducible
to discourse about things like neurological states.
Importing this distinction into the discussion of
personal identity, we could then say that the
no-self theory is an eliminative rather than a
reductive theory of personal identity. The no-self
critique of the reductive theory of


identity would then be that the concept of self and
personal identity cannot be reduced to our ideas
about psychology or bodies and, therefore, that such
concepts have to be eliminated.(1) We must not,
however, push the analogy too far. For although some
eliminative materialists see the elimination of the
theory of the mental as implying an elimination of
mental language, an analogous implication does not
hold for the no-self theory. That is, within the
no-self theory we can, as I shall soon argue,
eliminate the notion of the self and its identity,
and yet on pragmatic grounds continue to permit the
use of the language of personal identity.

A further but related point that separates
reductionism from the no-self theory is that since
reductionism seeks to give an account of personal
identity--a notion which has its roots firmly
embedded in the soil of the strict or nonreductive
theory (that is, the view that personal identity is
something simple and unanalyzable)--then it has
already accepted a certain view into which it must
now force the structure of human existence. And
this, it would seem, cannot but lead to distorted
accounts of how we undergo our experience. The
no-self theory, on the other hand, has no such prior
commitments. And in this sense it is more
phenomenologically based than are the reductionist
theories. To borrow Husserl's phrase, it goes back
to the things themselves. That is, it starts with an
examination of experience rather than with an
attachment to the project of how to account for
personal identity. This does not mean, of course,
that the no-self theory need not face the issue of
why someone might come to believe in his own
identity. For if there is no such thing as personal
identity, then it is essential that we can offer
some other account of why someone might be led to
think there is.

It is just this sort of approach to the problem
of personal identity that is found in the writings
of the most important no-self theorist, David Hume.
Hume was the first Western philosopher to unmask the
confusions attending our idea of personal identity
and subsequently to reject the idea as a fiction. It
will be worth our while, therefore, to start by
conducting a detailed examination of his position.
The problem, however, is that there is much
disagreement about how we are to interpret Hume on
this point. On the one hand, there are some
commentators who, while agreeing that Hume does
reject the idea of personal identity, go on to
proclaim that his account is deeply confused and
inconsistent with what he says elsewhere. On the
other hand, there are those who argue that his
account is quite consistent, but only because he did
not really reject the notion of personal identity
but only a particular version of that notion. It
shall be my contention in what follows that both
these accounts are wrong. The latter is wrong
because it fails to accept what Hume clearly says,
namely, that identity of the self is a fiction. The
former is wrong because it finds confusion and
inconsistency where, if we are fair to Hume, there
is none. This is not to say that Hume is completely


clear and exhaustive in his treatment of personal
identity, but only that within his writings we have
the makings of a solution to the problem of personal

What, then, is Hume's position? If we attend to
the section "Of personal identity" in Book I or A
Treatise of Human Nature, an answser is immediately
forthcoming. Hume starts by pointing out that
although some philosophers believe we are
continuously aware of something we call the self,
when We look to our experience there is nothing to
substantiate this belief. We are never, says Hume,
aware of any constant invariable impression that
could answer to the name of self. What we
experience, rather, is a continuous flow of
perceptions that replace one another in rapid
succession. "When I enter most intimately into what
I call myself," says Hume, "I always stumble on some
particular perception or other, of heat or cold,
light or shade, love or hatred, pain or pleasure. I
never catch myself at any time without a perception,
and never can observe any thing but the
perception."(2) Within the mind, he continues, these
perceptions "successively make their appearance;
pass, re-pass, glide away, and mingle in an infinite
variety of postures and situations" (p. 253). And
there is nothing to the mind but these perceptions.
There is consequently never any simplicity within
the mind at one time nor identity at two different
times. Nor, says, Hume, do we have any idea of a
self; for every real idea must be derived from some
one impression: "but self or person is not any one
impression, but that to which our several
impressions and ideas are supposed to have
reference" (p. 251).

With this much said against the notion of
personal identity, Hume turns to the question of why
we have such a proclivity to ascribe identity to our
successive perceptions. Prefacing his reply to this
question, Hume points out that a distinction must
first be made between "personal identity as it
regards our thought or imagination, and as it
regards our passions or the concern we take in
ourselves" (p. 253). His concern here, he says, is
with the former. He then starts his answer by
distinguishing between the ideas of identity and
diversity. In the former case we have the idea of an
object that persists, invariable and uninterrupted,
through a particular span of time. It is this that
comprises our idea of identity. In the case of
diversity we have the "idea of several different
objects existing in succession, and connected
together by a close relation" (p. 253). Now although
these two ideas are plainly distinct, it is certain,
says Hume, that in our "common way of thinking" we
generally confound them. That is, we often claim
that an object at one time is identical with an
object at another time, when in fact the two are
little more than a succession of different objects
connected by a close relation. To justify these
absurd ascriptions of identity we either come up
with the notion of a substantive self by feigning
the continued existence of our perceptions, or we
imagine the existence of something mysterious which
binds our many


perceptions together. Even where we do not do this
we at least have a propensity to do so. We can see,
then, says Hume, that because we often assert the
existence of such fictions, the problem of personal
identity is not merely a verbal dispute. It is
natural to ask, therefore, what it is that induces
us mistakenly to attribute identity to something
while, being a succession of objects, it is really
an instance of diversity. Hume feels that the
answer must lie in the workings of the imagination.
The reason why we might make such an attribution,
says Hume, is that "the passage of thought from the
object before the change to the object after it, is
so smooth and easy, that we scarce perceive the
transition, and are apt to imagine, that 'tis
nothing but a continu'd survey of the same object"
(p. 256) . This is especially true when the
alteration is either relatively small or when it
occurs gradually and insensibly.

From here Hume goes on to consider the various
other ways in which we ascribe identity to objects
which are variable and interrupted. This lays the
ground for his account of personal identity. For the
identity of the mind, like that of such changing
things as plants, animals, repaired ships, rebuilt
churches, and republics, says Hume, is only a
fictitious identity. It must therefore be similarly
explicable in terms of the workings of the
imagination. Hume concludes his account with the
important remark that all "nice and subtile"
questions concerning personal identity are best
considered as grammatical rather than philosophical
difficulties. Thus, except where the notion of a
fictional entity or principle is involved, all
disputes about personal identity are merely verbal
disputes and can never possibly be decided.

A hurried reading of this section of the
Treatise might well lead one to conclude that Hume
is confused; for there are some apparent
inconsistencies. These difficulties, however, are
resolved by keeping in mind the scope of Hume's
project and not allowing his claims to be taken out
of context. Let us start our appraisal of Hume by
dealing with a criticism raised by one of his
commentators. In a paper called "Hume on Personal
Identity," Penelhum argues that Hume's account is
an immense blunder, "an excellent example of how
complex and far-reaching the consequences of a
mistake in linguistic analysis or conceptual
investigation can be."(3) According to Penelhum,
Hume's fatal mistake is to think that one object
cannot have many parts. This is false, claims
Penelhum, because whether or not an object has parts
will depend upon what sort of object it is. A
melody, for example, is still one melody even though
it consists of a succession of notes. A related
mistake, we are told, is Hume's muddling of the
distinction between the numerical and specific
senses of identity. To remain the same through a
period of time is to remain the same in a specific
sense: to be exactly the same as one was at an
earlier time. For an object to change, however, it
must remain numerically the same; for it must be one
and the same object which is doing the changing.


Unfortunately for Penelhum, Hume makes it quite
plain that he does allow for one object to have many
parts, and he even allows that the parts may undergo
certain changes without affecting the identity of
the whole:

Suppose any mass of matter, of which the parts
are contiguous and connected, to be plac'd
before us; 'tis plain we must attribute a
perfect identity to this mass, provided all the
parts continue uninterruptedly and invariably
the same, whatever the motion or change of place
we may observe either in the whole or in any of
the parts. (P. 255)

What Hume does not allow is that an object can
maintain its identity with a coming and going of its
parts. It might seem, however, that this is all
Penelhum needs to have his point go through. For
what is a melody but one thing which is a succession
of other things? And if this is so, then Hume would
appear to be wrong in holding that a succession of
objects cannot be one object.

This, however, will not sustain Penelhum's
objection. For if a melody is an object, it is a
different sort of object than are the objects with
which Hume is concerned. A melody, by definition, is
a temporal sequence of musical notes: it is
something whose existence is necessarily spread out
over time. It is therefore logically incapable of
existing instantaneously in the specious present.
But there are no such logical restrictions on the
existence of trees, ships, churches, or persons. We
can easily imagine a tree, for instance, which
suddenly pops into existence only to disappear
immediately. And we can also imagine, it would seem,
a person who exists but for a second. It is true
that trees and persons as a matter of fact tend to
exist for varying durations of time, but I see
nothing logically incoherent about their existing
merely in the specious present. Indeed, it is just
because our idea of a person does allow for the
momentary existence of a person that we are able to
ask questions about personal identity in the first
place. For only when we have allowed this can we
raise the question of whether a person existing at
this particular instant is the same as an earlier
person existing at an earlier instant. We cannot,
however, ask the same question about a melody, for
melodies do not exist at particular instants; only
their notes do.

So Penelhum's alleged counterexample of one
melody that contains several parts is quite
acceptable to Hume. It is acceptable, first, because
Hume never denied that one object cannot consist of
several parts and, secondly, because even though a
melody is a succession of parts, this is a necessary
feature of a melody in a way that it is not a
necessary feature of a person.

What about Hume's purported muddling of the
numerical and specific senses of identity? This
would be a strange error for Hume to make since, as
Penelhum is aware, Hume himself draws this very
distinction in the course of his argument. But,
according to Penelhum, Hume has made


this mistake just so far as he thinks that for
something to remain the same it must not change. For
an object to remain unchanged is for it to remain
the same in the specific sense. But for an object to
change through time it must remain the same in the
numerical sense. Of course, says Penelhum, an object
can lose its identity by changing, but only if the
object is by definition an unchanging thing.

It should not be too difficult to see what is
wrong here. Consider Hume's example of a ship that
has its parts gradually replaced. It is possible
that eventually none of the parts of the original
ship remain. And yet the ship of today might be
exactly similar to the earlier ship. We can express
this relationship by saying the two ships have a
specific identity; that is, they exactly resemble
each other. It would be false, however, to say that
the two ships have a numerical identity; for there
is nothing about them that is numerically the same.
In contrast to this, Penelhum argues that the only
reason for saying that something has lost its
identity (that is, has become another thing) is that
it is by definition an unchanging thing. Since
nothing in the definition of a ship seems to rule
out the possibility of the ship having its parts
repaired, then even though the ship of today shares
not one plank or bolt in common with the earlier
ship, the two are nonetheless numerically identical,
that is, one and the same ship. But were we to
accept this scenario we would immediately be faced
with a difficulty. For how could we distinguish
between the situation where a ship has persisted
without changing any of its parts, and the very
different situation where it has changed all of its
parts? The natural way for us to mark this
distinction would be to say that in the former case
the ship of today is exactly one and the same as the
earlier ship, while in the latter case it only
resembles the earlier ship. That is, in the first
case the two ships are numerically identical, but in
the second case they are only specifically
identical. But on Penelhum's account we cannot draw
this distinction, because there is no distinction to
be drawn. Both cases have equal claim to being
instances of numerical identity. And yet it is
obvious that there is an important distinction to be
made here. In the case of the unaltered ship, what
we have is a ship that is identical in the strongest
sense with the earlier ship. In the case where the
ship suffers a total change of its parts, it can be
the same as the earlier ship only in a weaker sense
And thin distinction, which is natural to mate, is
merely the distinction between numerical and
specific identity. Hume's account of identity
allows us to make this distinction; Penelhum's does
not. It seems, therefore, that it is Penelhum who
has muddled things.

We should not, however, be led to conclude that
the distinction between numerical and specific
identity is central to Hume's discussion. For Hume
is not really interested in specific identity and
only introduces the distinction by way of showing
how we might come to ascribe (numerical) identity
where there is none. One such way--and there are


other ways--is to confound the ideas of numerical
and specific identity and to claim that two
successive objects are numerically identical when
they are only specifically identical.

This, it seems, is the mistake made by James
Noxon.(4) Noxon tries to argue that this distinction
lies at the heart of Hume's thesis and that Hume's
real purpose is to show how identity terms can be
meaningfully applied to persons once the
numerical/specific distinction is made clear.

But this is an unfounded reading of Hume. For
not only does Hume refer to the notion of specific
identity only in passing, but also nowhere does he
mention or even imply that the real identity of
these variable and interrupted things is a specific
identity. If we turn to the quotations cited as
support for this view, we can see the error
immediately. When Hume says that a seedling which
becomes a large tree is still the same oak, and that
an infant which becomes a man is still the same
individual, he cannot be referring to specific
identity; for it is evidently false to say that a
seedling is exactly similar (that is, specifically
identical) to a one-hundred-foot tree, and just as
false to say that an infant is exactly similar to an
adult. There may be some vague similarities in
either case, but not enough to justify a claim to
specific identity. Noxon seems aware of this problem
and tries to avoid it by asserting that specific
identifications are based on "points of resemblance
evaluated in light of general knowledge of the
changes which things of a certain sort undergo
during a certain period of time."(5) But I am not
even sure if I know what this means. How is the
knowledge that a seedling will change into a large
oak supposed to aid us in evaluating points of
resemblance between the two? For, regardless of our
knowledge of arboreal ontogeny, the fact remains
that a sprouting acorn bears little or no
resemblance to a massive oak.

Further, it seems clear that Hume, in his
examples, is intent on denying similarity rather
than affirming it. Not only does the tree have no
particles of matter in common with the seedling, but
even the "figure of its parts is different.
Likewise, the infant not only grows into an adult,
but is "sometimes fat, sometimes lean." Noxon could
always protest here that the tree remains exactly
similar to the seedling so far as it is still an
oak. But Hume says more than that the tree is still
an oak; he says it is still the same oak. And this
strongly suggests that he is not here using "same"
in its specific sense. If he only meant that the
seedling is as much an oak as the tree (a pointless
thing to say), there would have been no need for the
insertion of the words "the same" before "oak."

But does this not leave us with a problem? For
if Hume is not here referring to the specific
identity of the oak, then, we might think, he must
be referring to its numerical identity; but as we
have just noted, the tree is said to have not one
particle in common with the seedling, which means,
for Hume, that the two cannot be numerically
identical. The solution to this problem is to see
that Hume is not referring to the tree's


identity in any sense: he is referring to our
attribution of identity to the tree (that is, our
attribution of numerical identity). This is evident
from the line which immediately precedes his example
of the tree. Hume remarks here "tho' everyone must
allow, that in a very few years both vegetables and
animals endure a total change, yet we still
attribute identity to them, while their form, size
and substance are entirely alter'd" (p. 257). The
reference to the growing oak and the growing man are
merely illustrative examples of a changing plant and
a changing animal to which we attirbute identity.
Since, however, there can be no identity where the
form, size, and substance of a thing has entirely
altered, the identity which we attribute in such
cases can only be a fictitious one; that is, it is
the work of the imagination, not a property
belonging to the object or, better, to the
succession of objects to which we attribute it. And
this is true not just for plants and animals but for
all things which are variable and interrupted: for
example, repaired ships, rebuilt churches, rivers,
republics, and persons. So when Hume discusses "the
identity" of such things, he is only discussing how
we come to attribute identity to them, not their
actual identity.

The only reason we might think that Hume is
making positive statements about the actual identity
of things is that we may fail to notice the
structure of his overall argument and locus only on
specific remarks. If we view the section "Of
personal identity" as a whole, wee will see that in
the first four paragraphs Hume discusses the actual
identity of the self and categorically rejects the
notion as untenable. He does, however, feel that we
have a natural propensity to ascribe identity to
ourselves, and it is to an explanation of this
propensity that Hume devotes the rest of his
discussion. This division in the test is also
attended, for the most part, by a division in the
type of language that Hume uses to discuss identity.
In the first part, Hume speaks in a categorical way.
Thus it is claimed, for example, that "there is no
inmpression constant and invariable" (p. 251) .
"there is no such idea (of the sell)" (p. 252), "I
always stumble on some particular perception or
other" (p. 252), and, finally, "there is properly no
simplicity in (the mind) at one time, nor identity
in different" (p. 253). However, when we come to the
second part of the discussion, the language becomes
more psychological than categorical. That is, Hume's
concern here is more with how we imagine, suppose,
ascribe, or attribute identity rather than with the
actual identity of things. And so the second part
begins by asking "What then gives us so great a
propension to ascribe an identity to these
successive perceptions, and to suppose oursleves
possest of an invariable and uninterrupted existence
thro' the whole course of our lives?" (p. 253; my
emphasis). This psychological language continues,
for the most part, throughout this latter half: the
repaired ship "is still consider'd as the same; nor
does the difference of the materials hinder us from
ascribing an identity to it" (p. 257); although


plants and animals undergo a total change, "yet we
still attribute identity to them" (p. 257); a man
who hears an intermittent noise "says it is still
the same noise" (p. 258); and because an earlier
church is demolished before its successor appears we
do not think of them as being different, "and for
that reason are less scrupulous in calling them the
same" (p. 258; my emphasis throughout the last four
quotations). Likewise, when Hume says we can extend
our identity beyond our memories, his discussion
makes it plain that he is only talking about "the
most established notions of personal identity" (p.
262; my emphasis) , that is, what is commonly
believed to be true about personal identity, and not
about personal identity itself.

This use of psychological language to discuss
the supposed identity of interrupted and variable
objects should keep us alert to the fact that Hume
is here discussing only the origin of our belief in
such identity, and not the actual identity of what
is really an instance of diversity. Once we see this
we should not be worried by Hume's occasional use of
categorical statements in what is overtly a
discussion of psychology. Thus, when, in the second
part of the text, Hume states that a seedling which
becomes a tree "is still the same oak," all that is
before this makes it plain that he can only mean
that it is still called the same oak, or it is still
supposed to be the same oak, or some such thing, not
that it is actually still the same oak. Likewise,
when he says that the infant becomes a man "without
any change in his identity," all that he can mean is
"without our attributing any change to his

There will be those, no doubt, who will charge
that I am riding roughshod over what Hume says and
am twisting the text to suit my own purposes. But if
this is true, then my critics must explain why Hume
says that although the addition or removal of even
an inconsiderable amount of a mass of matter
"absolutely destroys the identity of the whole,
strictly speaking; yet as we seldom think so
accurately, we scruple not to pronounce a mass of
matter the same, where we find so trivial an
alteration" (p. 256). For in saying this Hume has
not only given the reasons why growing plants and
humans cannot maintain their identity over time
(they are continually losing and gaining amounts of
matter), but he also explains why we scruple not to
pronounce them the same (we find their
monment-to-moment alterations too trivial to note).
Of course, over many years the alterations that a
seedling has undergone will no longer be trivial,
"but where the change is produc'd gradually and
insensibly we are less apt to ascribe to it the same
effect (that is, the loss of its identity)" (p.

All of this should make it clear that, contrary
to what many of Hume's interpreters would have us
believe, for Hume it is an error to attribute
identity to variable and interrupted things. He says
in several places that the attribution of identity
to instants of diversity is both a


mistake and an absurdity (see especially pp.
254-255). But we must be careful here; for its being
a mistake does not imply what Penelhum thinks it

After noting Hume's claim that we are making a
mistake in referring to a person over time as the
same person, Penelhum remarks that "a little effort
of imagination is enough to indicate just how much
chaos would result from adopting Hume's diagnosis as
the source of a prescription and using a different
proper name whenever we noticed the slightest
change, even in ourselves (or rather in the separate
people that we would be from minute to minute)." And
if this is a mistake, continues Penelhum, it is one
whose correction "would require a complete overhaul
of the concepts and syntax of our language."(6)

Yet there is no reason to think that the correct
prescription for Hume's diagnosis need be using a
different proper name every time we notice a change
in someone or something. Indeed, Hume even says that
a rebuilt church, which we only imagine to be the
same as an earlier one, can still be called the same
as its predecessor "without breach of the propriety
of language" (p. 258). He does not, unfortunately,
elaborate on this. However, at the end of the
section on personal identity something is said that
suggests how it might be permissible to call two
things the same which in fact are only imagined to
be the same. Here Hume states that all disputes
about the identity of successive objects "are merely
verbal, except so far as the relation of parts give
rise to some fiction or imaginary (i.e. mysterious)
principle of union" (p. 262). Thus if two persons
are arguing about whether or not an earlier church
is the same as its rebuilt predecessor, and neither
of them is asserting the existence of a fictional
entity or principle of union which somehow unites
the two churches, then their dispute will be merely
about how the word "same" is to be used in these
circumstances. That is, they will not be disputing
an actual identity but only the linguistic
conventions surrounding our use of identity terms.
If we imagine that our disputants finally agree that
our linguistic conventions permit us to call the two
churches the same (a conclusion to which Hume would
give his assent), then we can see why it is
permissible to call two things the same which are in
fact different This is because there are two levels
at which the notion of identity can be employed: one
which deals with questions about identity at the
metaphysical or ultimate level, and one which deals
with them at the verbal or conventional level.

It is disappointing that Hume does not have more
to say about the two levels of "disputes." For it is
with such an account that we are able to explain why
it is acceptable for us to continue to talk in terms
of selves and personal identity despite the fact
that there are no such things. We can, however,
arrive at a fuller understanding of what the
two-level account involves by turning to another
version of the no-self theory. This


is the no-self theory as propounded by the Buddha
and various of his followers. The Buddhist theory
can offer some insights. For at the very heart of
this theory lies the doctrine of the two levels of
truth. Although the different schools of Buddhist
thought disagree on the exact nature of the
distinction to be drawn between the two truths,
there are enough similarities-at least in the early
Hinayana schools--for us to give a general account.
This will be useful to our project because it will
allow us both to see what Hume might have been
getting at and to acquire more munitions with which
to fend off this attack on the no-self theory.
I hasten to add, however, that I am not here
attempting a scholarly exposition of Buddhist
thought; for my present interests are confined to an
exposition of the no-self theory of personal
identity. It is just that Buddhism has some valuable
contributions to make here. There are, of course,
problems involved in the cross-cultural discussion
of ideas; Hume and the Buddha, after all, lived
their lives in very different social and historical
contexts. And yet I do not think that these
difficulties need detain us; for when we go to the
texts where Buddhist thinkers are grappling with the
problem of personal identity, we find their concerns
are essentially the same as Hume's.

In the earliest texts of Buddhism, the Pali
Canon (about 500 B.C.), we come across a distinction
drawn between two types of discourse: that of direct
meaning and that of indirect meaning. The former
type of discourse is said to be one whose meaning is
plain while the latter type needs to have its
meaning inferred with reference to the former. In
the discourses of indirect meaning, words are used
which apparently refer to persisting entities such
as a self or an I which, according to the Buddha,
are merely "expressions, turns of speech,
designations in common use in the world which the
Tathagata (i.e, the Buddha) makes use of without
being led astray by them."(7) That is, although we
may use words like "self" and "I," we should not be
led into thinking that they actually refer to
something, for they are but grammatical devices.
This nondenoting aspect of these expressions is
something which must be inferred in light of the
discourses of direct meaning. In this latter type of
discourse, the nonexistence of anything permanent or
enduring, such as the self or I, is asserted, and
the misleading features of language--those features
which lead us astray into the belief in an I--are
made explicit. Here there is no need for inference,
since the meaning of such discourse is plain.

As it happens, however, we are apt to confuse
the two types of discourse: "there are these two who
misrepresent the Tathagata. Which two? He who
represents a Sutta (i.e., discourse) of indirect
meaning as a Sutta of direct meaning and he who
represents a Sutta of direct meaning as a Sutta of
indirect meaning."(8) Although the Pali Canon does
not elaborate here, we can easily see what sorts of
errors are being referred to. On the one hand we
might think that someone who is using the


words "self," "I,"or "Buddha" (which are mere turns
of speech) is in fact denoting a particular entity.
Or, on the other hand, we might think that someone
who is denying the existence of the self cannot
really mean what he or she is saying and so we might
be tempted to infer a further meaning which would
still allow the existence of the self. We might, for
instance, think that the person making this claim is
only denying the existence of a certain type of

The discussion of the two types of discourse is
continued in the various Buddhist commentaries on
the Pali Canon, and here we are introduced to the
related ideas of two levels of truth. In one
commentary it is stated that all "Buddhas (i.e.,
enlightened beings) have two types of speech;
conventional and ultimate. Thus 'being', 'man',
'person', (the proper names) `Tissa', 'Naga' are
used as conventional speech. `Cagegories',
'elements', 'sense-bases' are used as ultimate
speech." Because of this division in speech, we are
told that the Buddha "declared two truths: the
conventional and ultimate, there is no third. Words
(used by) mutual agreement are true because of
Worldly convention; words of ultimate meaning are
true because of the existence of elements."(9)
Although the various elements are said to be the
constituents of which everything else, including
what we call the self, is made, it is not because
the elements are more basic than the self that the
self is said ultimately not to exist. It is simply
because there is nothing in the world, not even an
assemblage of the elements, that can be identified
with the self. Although the Buddha cites various
characteristics that something must have if it is to
be considered a self, the most important is that of
permanence or identity over time. But when we look
to our experience, there is nothing but impermance:
our bodies, feelings, and thoughts are forever
coming and going. In this sense the Buddha is in
complete agreement with Hume: where there is
diversity there can be no identity. None of this,
however, implies that statements which make use of
words like "self," "I," "you." " "Tissa, " or
"Buddha.' are false or nonsensical at every level of
discourse. For they can be true at the conventional
level, which means that they can be true because of
their being used in accordance with mutual
agreement, that is, linguistic convention.

A good illustration of how this distinction is
to be drawn is given in a well-known passage from
the The Questions of King Menander (about A.D.
100).(10) In this dialogue, the Indo-Greek king
Menander puts various questions about the nature of
the self to the Buddhist monk Nagasena. At the
opening of the dialogue, Menander asks "How is your
Reverence known, and what is your name? " The
somewhat provocative answer given to the king is
"I'm known as Nagasena, your Majesty, that's what my
fellow monks call me. But though my parents may have
given me such a name... it's only a generally
understood term, a practical designation. There is
no question of a permanent individual implied in the


of the word." Menander is quite astonished by this
reply and eventually asks "If your fellow monks call
you Nagasena, what then is Nagasena?" He asks
whether Nagasena is any part of the body or the
mind, or whether he is all of these things taken
together, or whether he is anything apart from them.
To all of this Nagasena replies merely "No, your
Majesty." Menander then exclaims triumphantly "Then
for all my asking find no Nagasena. Nagasena is a
mere sound! Surely what your Reverence has said is
false!" But Nagasena is not to be dealt with so
swiftly and, in good Socratic fashion, replies by
himself asking a question: "Your Majesty, how did
you come here--on foot, or in a vehicle?" "In a
chariot," says Menander. Nagasena then asks what the
chariot is, whether it is the pole, axle, wheels,
frame, reins, or yoke, or whether it is all these
taken together, or again whether it is something
other than the separate parts. Menander replies in
the negative. With this Nagasena fires back at the
king his own reasoning: "Then for all my asking,
your Majesty, I can find no chariot. The chariot is
merely a sound. What then is the chariot? Surely
what your Majesty has said is false! There is no
chariot!" Menander protests that he has not said
anything false: "It's on account of all these
various components, the pole, axle, wheels, and so
on, that the vehicle is called a chariot. It's just
a generally understood term, a practical
designation." Nagasena's rejoinder is to praise
Menander for this remark and to point out that the
same holds true of himself. For it is because of his
various components that he is known by the practical
designation "Nagasena." However, he adds, in the
ultimate sense there is no person to whom the name

It is easy to mistake this passage, as some have
done, for a statement of reductionism.(11) But the text
clearly disallows this interpretation. The
reductionist view would be that the person of
Nagasena can be reduced without remainder to his
various impersonal constituents. But when the king
asks if Nagasena is all of the parts of his body and
mind taken together, Nagasena answers "No." This is
because Nagasena rejects any notion of a person that
exists in the ultimate sense: a person is not
ultimately something other than his parts (the
strict theory), nor is a person ultimately the sum
of his parts (the reductionist theory). This does
not mean, however, that the word "Nagasena" is a
mere sound; for it is more than that: it is a
generally understood term whose proper use is
determined by mutual agreement concerning how, when,
and where it is to be used. Or, as Nagasena says, it
is because of his various components that he is
known as "Nagasena," even though "Nagasena" does not
refer to anything.

It is crucial to see, however, that in
relegating such terms to the realms of conventional
discourse, Buddhism is nor proposing a
conventionalist theory of personal identity. That
is, for Buddhism, a person's identity is not
something ultimately to be decided by convention.
For to


hold this position would be to commit the very error
against which the Pali Canon warns us, that is, to
represent a discourse of indirect meaning as a
discourse of direct meaning. It would be to conflate
the levels of truth and think that questions at the
ultimate level can be answered by agreements at the
conventional level.

We are now able to see why the no-self theory
does not imply that our language is in need of an
overhaul. For it is quite consistent with the
nonexistence of the self or I that we continue to
employ the words "sell" and "I" in their practical
everyday usage, provided we do not mistake them for
denoting some particular entity at the ultimate
level, or, as Hume would say, feign the existence of
a fiction. This is why, contrary to what many of
Hume's critics think, Hume's own use of the
first-person pronoun does not undermine his theory.
In Hume's statement "when I enter most intimately
into what I call myself, I always stumble on some
particular perception" (p. 252), the word "I" is
being used at the convectional level: it is merely a
generally understood term whose proper use is
determined by mutual agreement. We should not,
therefore, think that in using the first-person
pronoun Hume has committed himself to the existence
of a self at the ultimate level.

Some will no doubt find it paradoxical that we
can use personal language correctly when there is
nothing to which these terms ultimately refer. It
was reasoning akin to this, it seems, that led
Descartes to his famous proclamation "I think,
therefore I am." I must exist, reasoned Descartes,
because even when I doubt that I exist there is
still an I that is doing the doubting. But Descartes
has become led astray by his own language, for there
is no need for the "I" in "I think" or "I doubt" to
refer to anything. What Descartes was aware of, as
both Hume and the Buddha would agree, was just
thinking, not an I that was doing the thinking.
Consequently Descartes might just as well have said
(and should have said if his concern was with
ultimate rather than conventional truth) "there is
thinking, therefore there are thoughts." And such a
deduction, if we may call it that, does not suffice
to prove the existence of an I.

A possible response here would be to say that
although there need be no reference to an I when we
use the nounal sense of "thinking" or "thoughts,"
when the verbal sense "I think" is employed, then
plainly there must be some reference to a subject;
for what is it that thinks? To this it can be
replied that although the term "think" does require
a subject, this is little more than a grammatical
requirement. And so we might just as well employ a
nonreferring grammatical subject rather than the
misleading term "I." This is a point that is
recognized by George Christoph Lichtenberg, who
says, about Descartes' dictum, "We should say,'It
thinks', just as we say 'It thunders'. Even to say
cogito is too much if we translate it with 'I
think'. To assume the 'I' to postulate it is a
practical need."(12) Thus, since the use of the
verbal sense "thunders" also


requires the introduction of a subject, we bring in
the word "it" and say "it thunders." But this does
not mean that the grammatical subject "it" here
refers to anything. All we are saying when we say
"it thunders" is "there is thunder." Consequently,
since the requirement that the word "thinks" have a
subject is also a convention of grammar, or, as we
might say with Lichtenberg, a practical need, we
could likewise employ "it" to serve this purpose. We
could say "it thinks, therefore there are thoughts,"
and the appearance of "it" here would no more imply
a reference to an actual subject than would "it" in
"it thunders."(13)

There remains, however, a further problem which
needs our attention. It was mentioned earlier that
one of the criticisms leveled at Hume is that his
account of personal identity is inconsistent with
what he says elsewhere. The passages which are
supposed to contradict the view of the section on
personal identity come from Book II of the Treatise,
where in Hume discusses the nature of the passions
or emotions.(14) Because Hume refers here to the
importance of the role played by the self as the
object of such passions as pride and humility, some
writers have been quick to accuse him of
contradicting his earlier claims about the self.
Thus Norman Kemp Smith cites the following passages
from Book II as being incompatible with Hume's
claims from Book I:

'Tis evident, that the idea, or rather
impression of ourselves is always intimately
present with us, and that our consciousness
gives us so lively a conception of our own
person, that 'tis not possible to imagine, that
any thing can in this particular go beyond it.
(P. 317)

The stronger the relation is betwixt ourselves
and any object, the more easily does the
imagination make the transition, and convey to
the related idea, the vivacity of conception,
with which we always form the idea of our own
person. (P. 318)(15)

Kemp Smith suggests that the reason why Hume so
freely makes use of the idea of the self in Book II
when he has just dismissed the notion in Book I is
that Book II was written before Book I, and,
consequently, by the time Hume had got around to
writing Book I he had forgotten what he had said in
Book II.(16) Before we accept this picture of a
strangely distracted Hume, it seems we should look
at what is actually being said about the self in
Book II. And what we find in the opening pages of
Book II is this: "`Tis evident, that pride and
humility, tho' directly contrary, have yet the same
object. This object is the self, or that succession
of related ideas and impressions, of which we have
an intimate memory and consciousness." And a few
lines later we are told about "that connected
succession of perceptions, which we call self" (p.
277). These hardly seem like the words of someone
who is totally incognizant of the view expressed in
the section on personal identity. Indeed, by taking
the time to insert these descriptions of the self in
Book II, it seems that Hume is


anxious to remind us of the conclusions reached in
Book I. And so, when Hume says in Book II that an
awareness of ourselves is always intimately present
to us, we should understand this in terms of what
was said in Book I, namely, that to enter intimately
into what is called the self is just to encounter
various impressions. That is, what is always
intimately present to us is just those particular
perceptions whose succession we call the self (the
fact that Hume uses the word "intimately" in both
the passages from Book I and Book II also suggests
that he is intent on discussing the same notion of
the self in both places).

Still there is an apparent problem here. And
this is highlighted by the fact that Hume often
speaks in Book II of our idea or impression of the
self as though it were something which occurs
instantaneously within our awareness, a singular
perception which we experience as the object of
pride and humility. But, as Hume has told us, there
is no impression or idea of the self; there is only
the smooth and uninterrupted progress of thought
within our imagination. But how is it that a smooth
and uninterrupted progress of thought could appear
instantaneously within awareness? The notion of an
uninterrupted progress is a temporal notion; it is
something that occurs over an extended period of
time. Accordingly, it is difficult to see how such a
train of perceptions could be an object to which we
could, in an instant, direct our emotions.

I doubt, however, that this is a major
difficulty. Hume himself is aware that a different
account is needed to deal with those occasions, such
as in certain emotional states, where we suddenly
come across the object of what we call
self-awareness. He says, we have noted, that in
order to discover what is responsible for the
tendency to believe in personal identity, a
distinction must be drawn between "personal
identity, as it regards our thought or imagination,
and as it regards our passions or the concern we
take in ourselves" (p. 253). Personal identity of
the first sort is his professed concern in Book I,
while in Book II it is the latter sort. The problem
is that Hume never explains just how we are to
understand the relation between the two sorts of
personal identity. He obviously wants the personal
identity that concerns the passions to be basically
the same as that which the imagination constructs.
This is evident from the descriptions of the self
given in his discussions of the passions. And yet
his theory seems to require that the object of the
passions be capable of appearing to awareness in a
way that does not render it a succession.

The way for us to deal with this problem, and so
fill out our picture of the no-self theory, is to
see that when we enter a psychological or emotional
state that seems to depend on an instantaneous
awareness of self, what we are doing is not
reflecting on a succession of related ideas (which
could not be done at any one instant); rather we are
latching on to a particular collection of some of
these ideas which, by virtue of their


being related, can instantaneously present
themselves in a condensed form to our awareness. It
is with this discovery of a constructed or condensed
self-image, as we shall call it, that we are brought
to a central point within the no-self theory,
namely, that although we may on various occasions
have experiences of something that we take to be
ourself, on closer examination this object of our
awareness turns out to be nothing more than a
collation of related images.

Even some nonreductionists, such as John
McTaggart, for example, Seem to be alive to the
possibility that the establishment of what we call
self-perception would not be enough to prove that a
self exists, but only that something is perceived as
being a self.(17) This suspicion that we might perceive
something which we mistakenly believe to be ourself
is borne out by an examination of our states of
self-awareness. We can start by noting that although
we do experience occasions of self-awareness, there
are numerous instances in which it is plain that we
have no awareness of anything that can be considered
a self. Take, for example, my awareness in the
activity of listening to music. It is often the case
in listening to music that one can become immersed
in the music to the point that there is no room in
consciousness for the awareness of an I. Here there
is just the experience of the music: the gradual
unfolding of the melody or theme. And listening to
music is hardly an exceptional case. Other
activities, such as reading or writing, playing
chess, daydreaming, and making love, all provide
instances in which one can go on being thoroughly
unaware of oneself. It is, of course, possible for
me to shift my awareness to a point where I am aware
that I am listening to music, but then I am no
longer immersed in the music. It is important to
see, however, that when I am thus not immersed, then
my experience of the activity is drastically
altered. This is because, in entering a state of
self-awareness, I undergo a "giving-up" or a
"forsaking" of the activity in which I was
previously engaged. My listening gets left behind as
I conjure up the I to which I now direct my
attention. Here the act of listening loses its sense
of spontaneity; it becomes something I must struggle
with, something impeded by the new awareness that it
is I who am doing the listening. The act of
self-awareness is thus a reflective and complicated
act which involves both a stepping back from the
flow of experience and the introduction of a further
element (or rather collation of elements) into that
experience. It is in this sense, then, that
self-awareness can be called a secondary phenomenon,
for the object of self-awareness is not part of the
basic fabric of experience; rather it is something
which experience itself fabricates and then takes as
its object. It is, consequently, understandable that
self-awareness, with its complicated and reflective
qualities, is but an infrequent visitor to

This view of the secondariness of self-awareness
is supported, I think, by what we know about the
ontogeny of consciousness. The develop-


mental psychologist Jerome Kagan, for example,
delineates a sequence of four psychological
functions which lead up to but do not imply the
emergence of self-awareness. According to Kagan, the
first function to appear which comes close to
implying the existence of consciousness is
recognition memory. This makes its appearance in the
first eight months of life and is displayed in the
ability to discriminate between familiar and novel
stimuli. A few months later there appear the
functions of retrieval and inference, by which Kagan
means "the retrieval of schemata without any cues in
the immediate field and the generation of inferences
following the relating of those schemata to present
experience."(18) These functions are demonstrated by
a child who looks persistently for a toy it just saw
hidden a moment earlier. Here the child maintains a
schemata of the hiding of the toy and infers that
the toy must still exist somewhere. The fourth
function, which emerges at about sixteen months,
constitutes the child's ability to monitor, select,
and control the other functions. This function is
the awareness of one's potentiality for action and,
for Kagan, comes closest to our notion of

The difficulty, however, appears with the
emergence of the fifth function, namely, the
awareness of self as an entity with characteristics.
This is a difficulty because, although the function
of self-awareness seems to depend on the appearance
of the earlier functions, nothing in the earlier
functions implies that self-awareness need emerge:
"To explain why a 30-month-old declares, laughingly,
'I'm a baby,' as she crawls on the floor and sucks
on a bottle, requires more than an awareness of
one's ability to act. This behaviour presupposes not
only a schema for one's actions but also a schema
for the self as an object with variations in
attribute."(19) Consequently, says Kagan, some new
processes must be introduced into the developmental
sequence if the self-awareness function is to make
its entrance. But just what this must involve Kagan
cannot say; it remains, we are told, part of an
ancient enigma.

However, it is not too difficult to see that an
answer to this "enigma" is already contained within
the account. We need only observe that Kagan's
schema for the self as an object with variations is
just what we have called the constructed self-image.
That is, it is a collation of earlier experiences
which presents itself in a condensed form to our
awareness. Once we spell out the schema for the self
in this way we can see that the process needed to
generate it is not altogether dissimilar to the
process which generates the fourth function, that
is, the awareness of the potentiality for action.
The child arrives at its action schema by directing
its awareness toward its own abilities, that is, by
monitoring, selecting, and controlling the more
primitive functions of recognition memory,
retrieval, and inference. Once the child has
acquired the capacity thus to direct its awareness,
it is but a small step then to turn this awareness
toward some of the perceptions which constitute the
succession of related ideas


and so arrive at a schema for the self. However,
although the self schema will contain components not
to be found in other schemata, the components of
other schemata will nevertheless be discoverable
within the self schema. This is what kagan means, I
take it, by saying that the function of
self-awareness depends on the earlier functions.
Thus the infant's awareness of being a baby will
require more than an awareness of the ability to
act, but will still involve such an awareness. This
is because to have the awareness "I'm a baby" is,
among other things, to be aware that one acts like a
baby; and even a thirty-month old will know that
crawling on the floor and sucking on a bottle are
things that are (normally) only done by babies. The
reason, however, why self-awareness will include
more than the child's awareness of its ability to
act is that, in addition to acting, the child also
is acted upon by others, has an idea of its own
appearance, experiences emotions, and so on--all of
which are condensed into the object of
self-awareness. Thus, self-awareness is a secondary
phenomenon on the ontogenetic theory because it does
not involve the acquisition of a new ability: it is
merely the deployment of an earlier function upon a
wider range of objects.

Having established that self-awareness is both a
complicated and rare occurrence, we can now move on
to the phenomenology of the constructed self-image.
For in laying bare its structure we will see that
although it is the object of what we call
self-awareness, it is constituted by nothing more
than a collection of transient images.

One of the central features of the constructed
self-image is that it is a condensation of related
experiences. It is this feature that helps it to
masquerade as being a self; for, being a
condensation of our experience, it appears both as a
singular thing (although, as we shall see, this need
not be so) and, at the same time, as something which
contains our experiences. Since the experiences of
which it is composed may be taken from the different
times of our life, we can also see how the
constructed self-image might, in an instant, convey
something of the notion of identity over time; that
is, present itself as something that has persisted
throughout our lives.

The psychological process of condensing several
experiences into a single image or idea is not
unique to the structure of self-awareness. On the
contrary, it is a process commonplace in much of our
psychology. Thus, when I call up an image of my
friend Mary, I am not presented with a single
portrait, as it were, of Mary. What I find is
several different images of Mary which have been
collated and superimposed upon one another. I may,
for example, have an image of her hair in the wind,
which at the same time incorporates the movement of
her dress. This composition may be further blended
with an image of her face which is at once a frontal
view and a profile, and this in turn may allow the
blueness of her eyes to merge with the turquoise of
her earrings. In the same instant this


will all be bathed in emotional qualities that will
reveal the feelings I have for Mary. Such images
are, of course, transient, with certain of their
parts being introduced, expelled, and replaced in an
instant. However, as long as I dwell on the image of
Mary, there will continue to be an indistinct
network of constituent images, the older of which
are forever dying away while new relations take
their place. A similar process of condensation also
seems to take place in the construction of dream
images, and here, because of what Freud has called
primary process or the unrestrained character of
hypnagogic cognition, the resultant complex may
incorporate numerous diverse elements that are held
together by only the remotest of connections.(20) We
may, for example, encounter someone in our dreams
who displays the characteristics of several
different people at once: in one way he is our
childhood playmate, in another way he is our old
schoolteacher, while in yet a third way he is
someone whom we saw only yesterday. It is as if we
were looking at one of those anatomy textbooks which
depicts the different organ systems on a series of
overlapping transparent pages. On the bottom page is
the skeletal system, on the next page is the nervous
system, and then comes the circulatory system, and
so on. The end result is a single image of the human
body in which the different organ systems are
nevertheless discernable.

The constructed self-image is put together in
much the same way as other condensed images. When I
enter a state of self-awareness, the I that is
summoned before my consciousness is not a simple
entity that infixes itself changelessly in my mind.
It is rather a composite of various fading images
which will have some reference to how I see and feel
about myself. I may, for example, have an image of
my face as it appeared to me in the mirror this
morning which is nevertheless infused with features
of previous images of my face. Thus, although my
eyes and lips might appear to me as they did today,
my cheeks and the shape of my face might seem more
like those of myself of twelve years ago. Or again
it might include features of how I would imagine
myself to look in twenty years. This composite image
of my face might itself be superimposed on some
familiar scene, say, the beach where I often go for
walks. Here the sand dunes might be incorporated
into the cheeks and the rising of the waves into the
forming of a smile. And all of this will be
presented in a suffusion of affective tones which
will exhibit the emotional evaluations I have of
myself. Like other of our condensed images, the
self-image will have but an ephemeral existence; the
constituent images, continuously dissolving as new
associations, make their way into the complex. Just
what the constituents of the constructed self-image
are will naturally be different for each person,
since each person will see herself in a different
way. While for one person it may consist mainly of
idealized images of her physical appearance, for
another it may be a mixture of certain sensations or
emotions, while for a third it may be


images of how others respond to her. To verify that
this is so one need only ask different persons to
describe what it is they are aware of when they are
aware of themselves. It will then be seen that such
images vary quite markedly. Further, it also seems
to be that although some people have a relatively
consistent image of themselves over time (they have
always seen themselves as a whirling complex of
emotions), others do not (they now see themselves as
more of a physical entity than they did five years

I said above that the constructed self-image
need not exist as a singular thing; that is, at any
one moment it is possible for a person to have more
than one constructed self-image. And this much seems
to foflow both from what has been said about the
possible inconsistencies in the self-image over
time, and from the nature of the psychological
mechanism of condensation. For, in the first place,
if one can construct two or more different images of
oneself at different times, then the stage is set
for the possibility of constructing as many
different self-images at the same time. It might be
objected that consciousness could not operate in
this divided way, but various other phenomena, such
as the findings of research on brain bisection, give
us sufficient grounds to question such an

Further, because the constructed self-image is a
condensation of related ideas, it seems likely that
certain of our ideas which normally appear when we
think of ourselves are such that they bear little
relation to other ideas that also appear in the same
instant. Consequently, any attempt at condensing
these ideas into a composite idea will tend to fail.
Now in most cases the solution to this dilemma will
be somehow to repress one of the inconsistent
experiences or at least deny it entry into the
constructed self-image, that is, not allow it to be
an element in the state of self-awareness. The
problem, however, is that some of these offending
experiences may present themselves as too salient or
meaningful to accept banishment from the self-image.
What we have, then, is an experience or group of
experiences that is at once pulled into and yet
expelled from the self-image. Under these conditions
it is quite conceivable that the exiled experiences
might set up their own camp in which they, too,
could fly the flag of the self-image. There would
then exist two or more groupings of experience, each
of which would present itself as the appropriate
object of self-awareness. To enter a state of
self-awareness in these circumstances would be to
have one's awareness divide itself between two
objects competing for the same phenomenological
status. In understanding how this could occur in
consciousness we must not think of the constructed
self-image as an object to which I direct my
attention in the same way that I might direct a beam
of light upon a plane in the sky. For if we use this
analogy we will be tempted to see the constructed
self-image and my consciousness of it as if they

two separate entities, the latter of which somehow
engulfs the former. We will then imagine that to have
two or more simultaneous self-images will involve
little more than having one instant of consciousness
which simultaneously engulfs more than one thing,
much as one beam of light might illuminate two
planes. But, in reality, the constructed self-image
just is an instant of consciousness. For when I
enter into a state of self-awareness, what happens
is this: my consciousness throws itself into a
certain configuration that is structured in the form
of a condensed self-image which permeates or
diffuses over my world at that instant. In this
situation we can begin to see how parts of my
experience, which are neither assimilable nor
repressible, might fissure themselves off to become
a separate sphere of consciousness.

A Person in this state might then begin to
experience himself as two distinct persons.
Something like this could well play a role in the
genesis of the type of dissociative condition known
as multiple personality. Here the problem of having
more than one self-image would be dealt with by
producing a schism in awareness which would
subsequently be attended by the appearance of two or
more personalities, each of which would be matched
to the appropriate self-image. I am not saying, of
course, that the constructed sell-image is to be
identified with the personality, but only that
whatever sort of image we have of ourselves will
tend to relate both to the way we behave and to the
way we think, and that, as a consequence, having two
or more radically distinct self-images will tend to
correlate with having two or more radically distinct

We now have to ask what it is that leads someone
into perceiving the constructed self-image as being
a self. And here again the work of Buddhist
philosophers is most helpful. According to Buddhist
theory, what we call a person is really just an
aggregation of the five khandhas or elements. These
are: physical form, perceptions, feelings, motives,
and consciousness. But none of these elements,
whether considered separately or in combination, can
rightly be identified with the self, for they lack
the various qualities which we attribute to the
self.(22) This, however, does not stop one from
mistakenly identifying oneself with one or another
of the elements, and indeed this is a ubiquitous
confusion from which Buddhism hopes to set us free.
But what is it that leads a person to this mistaken
identification? To answer this we need to refer back
to our previous discussion of the conventional and
ultimate levels of truth. There we saw that although
personal names and personal pronouns do not at the
ultimate level refer to anything, at the
conventional level it is quite acceptable to use
such expressions for pragmatic reasons. Thus the
Buddha uses the language of the self as convenient
designations without being led astray by them. The
problem is that, unlike the Buddha, many of us do
get led astray by the expressions we use; that is,
in failing to notice that we are using language at
the level of convention, we end up


thinking that there must be something to which the
words "I" or "self" refer. And so we turn our gaze
inward (because this is where the self is supposed
to exist) and, coming upon one or another of the
elements, or a collection of the elements, hasten to
identify it with our self. Buddhism underlines the
importance that language plays here by making a
didactic use of the Pali word ahamkara, which can
mean both "the utterance of 'I'" and "I-make."(23)
That the same word has both meanings helps to
suggest a connection between the two meanings: not
only that the language of the self leads to the
fabrication of a self but also that a fabricated
self leads to a misconstrual of the language of the
self. Here, then, we see the cyclical nature of the
trap in which the straying language user is caught.
In uttering "I" one is led to misidentify an element
in one's experiences as the self. Having affected
this delusory identification, one then goes on to
make similar utterances firm in the belief that
these utterances ultimately refer to oneself. For
Buddhism, it is thus that we come to weave the first
threads of an ever-expanding veil of delusion. For
in believing that we have a sell, we are easily led
to other experiences which depend on this illusory
self. Pride and humility, for example, can only get
their foothold if there is something which we
perceive to be the self and to which we can relate
the causes of our pride or humility. I cannot be proud
of my successes or humiliated by my failures unless
I believe there is an I to whom these successes and
failures belong. This is why, on Buddhist theory,
the giving up of the belief in the self--which we
must do if we are to be free from delusion--is also
attended by the cessation of pride, humility,
embarrassment, envy, and other self-oriented (and
hence delusive) emotions.

The Buddhist method for overcoming the delusion
of self is to engage in meditation, a practice which
may be described as pure internal analysis or
sustained inward gazing. Through this technique, the
supposed internal self comes to be seen for what it
is: a mere collection of transient elements. This
realization loosens the grip of the belief in sell
and so dissolves the constructed self-image back
into the elements from which it came. In one Pali
text, the Visuddhimagga Sutta or The Path of
Purificalion, we are told that through the
contemplation of voidness or, what is the same
thing, not-self, "the misinterpreting (insisting)
that 'a self exists' is abandoned." This abandoning
occurs "because it has been clearly seen that there
is no core of permanence and no core of self."(24)
The insight thus acquired, the Sutta tells us, is
"the death of formations"; the formation that we
took for a permanent self dies before our very eyes.

Here at last we are in a position to see how
Hume's idea of personal identity as it regards the
imagination is related to the notion of the
constructed self-image, or personal identity as it
regards the passions. When I come to believe that I
am the same person I was twelve years ago, it is
because of the smooth and uninterrupted progress of


that is produced in my imagination when I reflect on
the succession of related ideas. This is the
imaginary self which is contemplated over time and
whose identity is fictitious. When, on the other
hand, I believe that I am perceiving my self in an
instant of self-awareness, such as when I experience
a self-oriented emotion, what is happening is that
my awareness is being directed to an object which is
merely a condensed version of extracts from the
succession of related ideas. This is the constructed
self-image, which is a rare and secondary aspect of
consciousness. In either case, what I come upon is
merely a collection of experiences: in either case,
there is no sell to be found.


This essay is a version of part of a doctoral thesis
which was accepted by the University of Edinburgh in
1989. I should like, therefore, to thank the
University of Edinburgh and the Committee of
Vice-Chancellors and Principals of the Universities
and Colleges of the United Kingdom, from both of
whom I received the scholarships which helped to
make this research possible.

1 - For a discussion of how the bodily version of
reductionism misconstrues the nature of the
body, see my "Bodily Theory and Theory of the
Body, " Philosophy 66, no. 257 (July 1991):
2 - David Hume, A Treatise of Human Nature, ed. J.
A. Selby-Bigge (Oxford: Clarendon Press, 1978),
p. 525. Further references are to this edition
and are indicated in the text. Unless stated
otherwise, all italics within this and
subsequent quotes are Hume's.

3 - Terence Penelhum, "Hume on Personal Identity, "
The Philosophical Review 64 (1955): 571.

4 - James Noxon, "Senses of Identity in Hume's
Treatise, " Dialogue: Canadian Philosophical
Review 8 (1969): 367-384.

5 - lbid., pp. 372-373.

6 - Penelhum, "Hume," p. 578.

7 - Trans. K. N. Jayatilleke in Early Buddhist
Theory of Knowledge (London: Alien & Unwin,
1963), p. 319.

8 - Ibid., p. 361.

9 - Trans. Steven Collins in Selfless Persons:
Imagery and Thought in Theravada Buddhism
(Cambridge: Cambridge University Press, 1982),


10 - In Sources of Indian Tradition, ed. Wm.
Theodore de Bary (New York: Columbia University
Press, 1958) , vol. 1, pp. 103-105. The
following translation comes mainly from this
edition. However, the editor omits the
important final remark made by Nagasena. This
line, which I have included, can be found in
The Sacred Books of the Buddhists, vol. 23,
Milinda's Questions, trans. I. B. Horner
(London. Luzac, 1963), bk. 1, pp. 37-38.

11 - For example, Derek Parfit, in Reasons and
Persons (Oxford: Claredon Press, 1984), p. 273.
Parfit is one of those who fail to distinguish
between the eliminative no-self theory and
reductionism. Consequently, he mistakenly
thinks that the Buddhist position supports his

12 - Trans. J. P. Stern in Lichtenberg: A Doctrine
of Scattered Occasions (Bloomington: Indiana
University Press, 1959), p. 270.

13 - Stern does not seem to me to be aware of this
point in his discussion of Lichtenberg. See
ibid., p. 270.

14 - There are also the so-called "second thoughts"
in the Appendix to the Treatise (pp. 633-636),
where Hume seems to express doubts about the
truth of his account of personal identity.
Here, however, I follow Norman Kemp Smith in
thinking that the passage in question is in
fact a reaffirmation of his principles and that
his only doubts are with his theory's ability
to account for states of self-awareness. This,
as will be seen, is not a difficulty for a more
expanded version of Hume's account. For Kemp
Smith's remarks on the Appendix, see his The
Philosophy of David Hume: A Critical Study of
its Origins and Central Doctrines (London:
Macmillan, 1931), pp. 553-560.

15 - Kemp Smith, The Philosophy of David Hume, p.

16 - Ibid., p. vi.

17 - John McTaggart, The Nature of Existence
(Cambridge: Cambridge University Press), vol.
2, chap. 36.

18 - Jerome Kagan, The Second Year: The Emergence of
Self-Awareness (Cambridge, Massachussetts:
Harvard University Press, 1981), pp. 148-149.

19 - Ibid., p. 149.

20 - For a discussion of primary process, see
Sigmund Freud, The Standard Edition of the
Complete Psychological Works vol. 5, The
Interpretation of Dreams, ed. J. Strachey
(London: Hogarth Press, 1964), pp. 599-609.

21 - See Thomas Nagel's "Brain Bisection and the
Unity of Consciousness," in Personal identity,
ed. John Ferry (Berkeley: University of
California Press, 1975).


22 - For the Dialogue in which this argument is
given, see The Collection of the Middle-Length
Savings (Majjhima Nikaya), trans. I. B. Horner
(London: Pali Text Society, 1957), vol. 1, sec.

23 - See J.A.B. van Buitenen, "Studies in Samkhaya
II," Journal of the American Oriental Society
77 (1957): 15-25.

24 - Buddhaghosa, The Path of Purification
(Visuddhimagga) , 2d ed., trans. Bhikkhu
Nyanamoli (Columbo: A. Semage, 1964), p. 815.