An abbreviated version of the following paper was published under the same title in The Proceedings of the Twenty-First World Congress of Philosophy, vol. 9, ed. Stephen Voss (Ankara, Turkey), 2006, pp. 29-33.
According to Buddhist ontology, every (samsaric) being is impermanent, unsatisfactory, and devoid of self-nature. Anicca, dukkha, anatta: these are the famous three marks (tilakkhana) upon which the whole of Buddhism rests. I would like to consider a well-known Buddhist argument for the third of these marks, that of anatta, an argument one could call ‘The Chariot.’ The argument aims to show that no (samsaric) being is a self, or has self-nature, or is a substance. My thesis will be that, successful as this argument may be when applied to things other than ourselves, it fails when applied to ourselves.
The Chariot argument in the Milindapanha is presented in the form of a dialog between King Milinda (Menander, Menandros) and the monk Nagasena.[i] Rather than clutter my text with quotations from this dialog, I will simply report its main dialectical moves as I understand them. My interest is not historical or exegetical: I aim to present and evaluate a fascinating argument that is of enduring philosophical interest.
Nagasena begins their dialog by introducing himself to the King as ‘Nagasena,’ but immediately adds that ‘Nagasena,’ like any name, is a convenient designation that does not denote any ego or individual or person (puggala). Milinda questions this assertion, insisting that there must some individual who bears the name ‘Nagasena,’ for otherwise there would be no one who applies himself to meditation, keeps the precepts, earns merit, etc.
The issue dividing the interlocutors seems to be this. Although both agree that there is a reality independent of mind and language, they disagree about its nature. Milinda claims that it contains unitary and self-same individuals corresponding to such proper names as ‘Nagasena.’ It is this claim that Nagasena denies. For the latter, reality consists of radically impermanent and insubstantial momentary entities that we, wielding words and concepts, group together into unities for our purposes. Thus the issue is whether in reality there is an ontological unity corresponding to the linguistic unity of the name ‘Nagasena,’ or whether there is no such ontological unity but only disconnected momentary entities that we collect for conventional purposes under the name ‘Nagasena.’
It is no part of Milinda’s position as I shall reconstruct it that the individuals denoted by proper names be absolutely permanent entities: they could well be relatively permanent.[ii] Thus one is not forced to choose between saying that ‘Nagasena’ has no referent in reality and saying that it has an absolutely permanent referent. Charitably construed, Milinda’s position is that the unitary and self-same individuals corresponding to names like ‘Nagasena’ are relatively permanent entities possessing relative self-nature. If Milinda’s position so construed were correct, then of course Nagasena’s would collapse.
Milinda tries to prove to Nagasena that there must be a unitary and self-same individual by the following reasoning. He first obtains Nagasena’s agreement on three points: (i) Nagasena is not identical to any one of his proper parts, whether these be physical (brain, liver, ...) or mental (sensations, perceptions...); (ii) Nagasena is not identical to the mere sum of his parts; (iii) Nagasena is not identical to something wholly distinct from his parts. Having agreed to these three points, Nagasena would seem to be committed to denying that anything could serve as the referent of ‘Nagasena.’ Milinda concludes that in this case the name ‘Nagasena’ is an empty or meaningless sound and that Nagasena speaks falsely insofar as he implies, by his use of the sentence “I am called ‘Nagasena,’” that there is such a being as Nagasena. Milinda is clearly assuming that if a name is to be genuine, and not an empty sound, it must designate something real.
Milinda’s argument against Nagasena may be framed as the following reductio ad absurdum: (a) ‘Nagasena’ is a meaningful name. (b) A meaningful name must have a referent. (c) The referent of ‘Nagasena’ cannot be one of the monk’s parts, or the sum of his parts, or something wholly distinct from his parts. (d) If ‘Nagasena’ does not designate an individual, ‘Nagasena’ has no referent. Therefore, (e) if ‘Nagasena’ does not designate an individual, then ‘Nagasena’ is a meaningless name, which is absurd. Therefore, (f) ‘Nagasena’ does designate an individual.
In his reply to Milinda, Nagasena in effect argues against premise (b) by trying to show that the term ‘chariot,’ though it lacks a referent, is not for that reason a meaningless term. The king having arrived at their meeting in a chariot, the monk challenges the king to show him the chariot. Nagasena enumerates the parts of the chariot, and gets Milinda to agree that the chariot cannot be identified with any of its proper parts taken singly. He also gets the king to agree that the chariot cannot be identified with the parts taken collectively as a mere sum. Finally, the king admits that the chariot cannot be identified with anything wholly distinct from the parts. From this three-fold admission, Nagasena concludes that the king is bound to infer that the term ‘chariot’ is an empty sound and that his majesty speaks a falsehood insofar as he implies, by his use of the term ‘chariot,’ that there actually is something in the world to which the term applies. Nagasena in effect convicts Milinda of the incoherence of saying, on the one hand, “I came in a chariot,” but also, on the other, “There is no chariot.”
Milinda tries to extricate himself from the jaws of this contradiction by replying that the word ‘chariot’ is but a convenient designation for pole, axles, wheels, chariot-body, etc. With this concession, however, the king falls into Nagasena’s trap. The monk triumphantly announces, in effect, that a human being is analogous to a chariot. Just as the term ‘chariot’ is merely a convenient designation with no referent in reality, the same holds for the name ‘Nagasena.’ In reality, all there are are items which, when grouped by us, become parts of wholes, chariot-parts in the one case, human-parts in the other.[iii] There is no soul, ego, substantial self, individual, or person to be found. There is no underlying substratum that unifies and supports the parts, remaining numerically identical through changes of parts. There is nothing real to which I refer when I say ‘I’ and to which ‘Nagasena’ refers when he or someone else says ‘Nagasena.’ There is no self!
So has Nagasena won the debate? Has he established the doctrine of no-self? I can’t see that he has.
The underlying argument seems to be as follows.
P1. No concrete partite thing is identical to any one of its proper parts.[iv]
P2. No concrete partite thing is identical to the mere(ological) sum of its proper parts.
P3. No concrete partite thing is identical to something wholly distinct[v] from each of its parts.
C. Singular terms denoting concrete partite things, useful as they are for counting and classifying, do not refer to anything real.
The premises of this argument are exceedingly plausible. Thus it is surely obvious that the king’s chariot is not identical to its right wheel, or to any other proper part, or to any two proper parts, etc. It is also obvious that the chariot is not identical to the mere sum of its parts: the sum of the chariot’s parts can exist even if the chariot does not exist, as when the chariot is disassembled. It is the same sum whether the chariot is assembled or disassembled. As for the third premise, it also seems quite clear that there is not, in addition to the parts, some further physical or metaphysical entity that is the ‘real chariot’ or essence or substratum of the chariot which could subsist in splendid isolation from the parts. That is no more the case than that there is a little man – a homunculus – inside my head looking through my eyes, and hearing through my ears, etc.
The premises, then, seem to be true; but does the conclusion follow? One obvious response is that the argument is a non sequitur since it ignores a fourth possibility: that terms like ‘Nagasena’ and ‘this chariot’ refer to wholes of parts in a definite arrangement, where this arrangement is a feature of reality and is not introduced by our use of such terms as ‘Nagasena’ and ‘chariot.’ Thus a chariot is neither a sum of disconnected chariot-parts, nor something wholly distinct from the parts, but a sum of parts connected in the right way. One might argue that ‘this chariot’ does indeed refer to something in reality, namely, a whole of connected chariot-parts, where both the parts and the connectedness are features of mind and language-independent reality. Common sense will insist that we cannot effect the assembly of chariot-parts simply by applying the term ‘chariot’ to them. It is the other way around: ‘chariot’ applies to them because of their antecedent connectedness. If this is right, then things like chariots would have a sort of relative self-nature which would entail relative permanence and relative satisfactoriness. We would thereby avoid two extremes, that of saying that ‘Nagasena’ refers to nothing in reality, and that of saying that it refers to an absolutely permanent self wholly distinct from each of Nagasena’s parts.
This defense of Milinda against Nagasena, however, is workable only if there is indeed some distinction in reality between the mere sum of chariot-parts and the connected parts. It is open to the Buddhist, however, to argue that there cannot be any such distinction, and thus to argue against even relative self-nature. To simplify the discussion, suppose we have a whole W consisting of two parts, a and b. Then from (P1) we know that ~(W = a) and ~(W = b). From (P2), we know that ~(W = (a + b)). From (P3), we know that there is no c wholly distinct from both a and b such that W = c. The objector, however, wants us to consider a fourth possibility, namely, that W = aRb, where R is an entity in the world that connects a and b, and in connecting them, makes the difference in reality between the whole and the mere sum of its parts. What the objector is saying, then, is that W is not a, not b, not their sum, and not some c wholly distinct from a and b; W is the unity of a and b, where this unity has its ontological ground in the unifier, or connector, R. Thus it is not the use of a name in thought or speech that confers unity upon that which is named; there is an antecedent ontological unity in the thing named. To say that there is this antecedent ontological unity is just to say that the anatta doctrine is false.
One problem with this suggestion is that in many cases no entity R -- no unifier -- is given. Thus if W is a bolt on which has been screwed a nut, all that analysis reveals is the bolt and the nut: there is nothing in reality that connects the bolt and the nut. We need not assume that in the past someone screwed the nut onto the bolt; it is conceivable that the two were together for all past time. But even if someone or something did screw the nut onto the bolt, now all we have is a bolt and a nut with nothing that connects them. We may agree that the bolt and nut are connected – in that the nut is threaded onto the bolt – but this connectedness is not a further constituent of reality that can be found by analysis. Not finding any constituent of reality that either is, or is the ground of, connectedness, philosophers of an empiricist and analytic bent will deny that there is any such thing.[vi] If we introduce the term ‘nolt’ to refer to the bolt-cum-nut, then our Buddhist analyst will say that ‘nolt,’ like ‘chariot,’ refers to nothing in reality. It does not refer to the sum of the parts, because all or almost all will agree that a whole cannot be identified with the mere sum of its parts: ther parts can exist without the whole existing. And of course all will agree that ‘nolt’ does not refer to the bolt or the nut or to something wholly distinct from both. Thus our natural thought that there is a connectedness in reality that justifies our belief that ‘nolt’ refers to something, and that what it refers to has relative self-nature, turns out, on the Buddhist analysis, to be an illusion.
Indeed, a defender of Nagasena could take a further step. Beyond making the empiricist point that no ground of connectedness appears or is found in experience, he could make the ‘rationalist’ or dialectical point that any further constituent introduced to do a connecting job will be unavailing for essentially Bradleyan reasons. Suppose two boards are connected with a nail. The resulting whole has three parts: two boards and a nail. An ordinary person will say that one of the parts, the nail, connects the other two. The Buddhist philosopher, however, could point out that there is no difference in reality between the unconnected sum of two boards and a nail, and the same three things connected. For even given that someone in the past drove the nail through the two boards, what we have now before us is simply two boards and a nail. At the present moment there is nothing in reality -- nothing that analysis can uncover -- that is or accounts for the connectedness. There is nothing we can point to, nothing empirically discernible, that is the connectedness or the ground of the connectedness. If we call the two boards nailed together ‘noard,’ then our Buddhist will say that ‘noard’ does not refer to anything. For the noard is not identical to one of its proper parts, nor is it identical to the mere sum of its proper parts, nor is it identical to something wholly distinct from its parts.[vii]
It is furthermore clear that introducing a fourth entity to connect the nail to the boards will only generate a Bradley-style vicious infinite regress. For we can then reiterate the inquiry by asking what grounds the difference between the mere sum of these four items and the concrete whole having them as parts. The ground of connectedness continues to elude the analytic understanding.
Thus one can see that the Chariot is an intriguing argument that cannot be easily dismissed. We want to say, with King Milinda and with common sense, that a whole of parts is more than a mere sum of parts, and that this something more -- the unity of the parts -- is something real as opposed to something introduced by our conceptual or linguistic activities, or by our craving for permanence. But since we cannot find this ‘something more’ by analysis, the pressure is on to write it off as illusory.
But even if the Chariot succeeds in showing that nonpersons lack self-nature, does it also show that persons lack self-nature? It may be that to argue by analogy as Nagasena does, applying to persons what is true of nonpersons, is a mistaken procedure. Indeed, I will now argue that the analogy is mistaken, and that a person is a whole of parts in an importantly different sense than that in which a chariot is a whole of parts.
Suppose my mental state passes from one that is pleasurable to one that is painful. Observing a beautiful Arizona sunset, my reverie is suddenly broken by the piercing noise of a smoke detector. Not only is the painful state painful, the transition from the pleasurable state to the painful one is itself painful. The fact that the transition is painful shows that it is directly perceived. It is not as if there is merely a succession of consciousnesses (conscious states); there is in addition a consciousness of their succession. For there is a consciousness of the transition from the pleasant state to the painful state, a consciousness that embraces both of the states, and so cannot be reductively analyzed into them. But a consciousness of their succession is a consciousness of their succession in one subject, in one unity of consciousness. It is a consciousness of the numerical identity of the self through the transition from the pleasurable state to the painful one. Passing from a pleasurable state to a painful one, there is not only an awareness of a pleasant state followed by an awareness of a painful one, but also an awareness that the one who was in a pleasurable state is strictly and numerically the same as the one who is now in a painful state. This sameness is phenomenologically given, although our access to this phenomenon is easily blocked by inappropriate models taken from the physical world. Without the consciousness of sameness, there would be no consciousness of transition.
Another example is provided by the hearing of a melody. To hear the melody Do-Re-Mi, it does not suffice that there be a hearing of Do, followed by a hearing of Re, followed by a hearing of Mi. For those three acts of hearing could occur in that sequence in three distinct subjects, in which case they would not add up to the hearing of a melody. (Tom, Dick, and Harry can divide up the task of loading a truck, but not the ‘task’ of hearing a melody, or that of understanding a sentence.) But now suppose the acts of hearing occur in the same subject, but that this subject is not a unitary and self-same individual but just the bundle of these three acts, call them A1, A2, and A3. When A1 ceases, A2 begins, and when A2 ceases, A3 begins: they do not overlap. In which act is the hearing of the melody? A3 is the only likely candidate, but surely it cannot be a hearing of the melody. For the awareness of a melody involves the awareness of the (musical not temporal) intervals between the notes, and to apprehend these intervals there must be a retention (to use Husserl’s term) in the present act A3 of the past acts A2 and A1. Without this phenomenological presence of the past acts in the present act, there would be no awareness in the present of the melody. But this implies that the self cannot be a mere bundle of perceptions externally related to each other, but must be a peculiarly intimate unity of perceptions in which the present perception A3 includes the immediately past ones A2 and A1 as temporally past but also as phenomenologically present in the mode of retention. The fact that we hear melodies thus shows that there must be a self-same and unitary self through the period of time between the onset of the melody and its completion. This unitary self is neither identical to the sum or collection of A1, A2, and A3, nor is it identical to something wholly distinct from them. Nor of course is it identical to any one of them or any two of them. This unitary self is given whenever one hears a melody.
Now consider a synchronic example, the hearing of a chord. Hearing the major chord C-E-G, I hear that it is major, and hearing the minor chord C-E flat-G, I hear that it is minor. How is this possible? The hearing of the major chord cannot be analyzed without remainder into an act of hearing C, an act of hearing E, and an act of hearing G, even when all occur simultaneously. For to hear the three notes as a major chord, I must apprehend the 1-3-5 musical interval that they instantiate. But this is possible only because the whole of my present consciousness is more than the sum of its parts. This whole is no doubt made up of the part-consciousnesses, but it is not exhausted by them. For it is also a consciousness of the relatedness of the notes. But this consciousness of relatedness is not something in addition to the other acts of consciousness: it includes them and embraces them without being reducible to them. Once again the unitary self is given: it is given whenever we hear a chord.
There is also this consideration. Phenomenologically, mental change is not existential change, but alterational change, or in a word, alteration. Existential change, as when something comes into being or passes away, is not a change in something, or at least it is not a change in the thing that suffers the change: a thing that ceases to exist is no longer available to be that in which this change occurs, and a thing that comes to exist is not available prior to its coming to exist to be that in which this change occurs. We express this by saying that there is no substratum of existential change. Alteration, however, requires a substratum: alteration occurs when numerically one and the same individual is in different states at different times.
Pali Buddhism, with its interconnected doctrines of radical impermanence and universal selflessness, implies that that ultimately there is no alteration, that all change is existential change. For alteration requires substrata of alteration, and substrata are incompatible with anatta. But if mental change were existential change, there could be no consciousness of it. If the pleasant visual sensation simply passes out of existence to be replaced by the painful auditory sensation, then there is a change all right – a change in the way things are – but not a change of which there could be any consciousness in the one in whom the change occurs. Furthermore, there would be no awareness of dukkha – the starting point of Buddhist soteriology – because there would be no possibility of a perceived contrast of the dukkha-state with the earlier sukha-state (the pleasant awareness of the sunset).
To be aware of the change from the pleasurable state to the painful one, I must endure through the change. Therefore, since there is consciousness of mental change, mental change is alteration and thus requires a substratum that is numerically identical across the change. The point was appreciated by Kant, who wrote that “A coming to be or a ceasing to be . . . can never be a possible [object of] perception.” (CPR A 188 = B 231)
What this shows is that there is direct awareness of the self as that in which the two distinct states are united. The fact of experienced mental change refutes the anatta doctrine. There is not just an awareness of one state followed by an awareness of a second; I am aware of myself as the transtemporal unity of the two states. Unity, of course, is not identity: so talk of the unity of the pleasurable and painful states is consistent with their numerical distinctness. The self, therefore, is directly given in the experience of mental change; but it is of course not given as a separate object wholly distinct from its states. It is given in and through these states as their transtemporal unity. The self is not one of its states, nor the sum of all of them, nor something wholly distinct from all of them; the self is their self-unifying unity. Thus one must not think of the substratum of mental change as wholly distinct from its states. It is not like a pin cushion into which pins are stuck. A pin cushion without pins is conceivable; a self without conscious states is not. The self is not an unconscious something that supports consciousness; it itself has the nature of consciousness. Consciousness/self-consciousness is a sui generis reality that cannot be understood in terms of crude models from the physical world.
Note how the conscious self (the person) differs from an artifact like a chariot. We saw how it is natural to say that a chariot is the unity of its parts (and not their mere sum or something wholly distinct from each of them). But in the case of an artifact, the unity is not given; it cannot be isolated in the way an analyst would like to isolate it. And if it cannot be isolated, that provides some (though not a compelling) reason to deny it any mind-independent reality. This renders aid and comfort to the anatta theorist. Standing pat on his empiricist principles, he can say: “I just don’t find this unity or selfhood or atta you prate about; it is not empirically discernible, and to say that it is something hidden behind what is empirically discernible smacks of mystification or speculation.”
In the case of the conscious self, however, the unity is given: it is something I live through (er-leben) in the experience (Erlebnis) of mental change. Mention was made of the diachronic unity of the self in the consciousness of the transition from a pleasurable state to a painful one. But there is also the synchronic unity of the self in the consciousness of the unity of two distinct simultaneous states. Suppose that while enjoying the visual perception of the sunset, I am also enjoying the olfactory sensation of burning incense. Phenomenologically, what we have here is a whole of parts, which, though numerically distinct, are given as one, as unified. The unity of the chariot is not given either synchronically or diachronically; the unity of the self is given both synchronically and diachronically.
The upshot is that the conscious self is a whole of parts in an importantly different way than an artifact is a whole of parts. A conscious self is not merely an external collection of conscious states, but a self-unifying unity of them. Thus even if the Chariot shows that artifacts lack self-nature, it cannot show that the same holds for those conscious and self-conscious unities of experience we call persons.
This result can be further supported by the following argument. If Nagasena is right, then there is nothing in reality denoted by ‘this chariot.’ There is no ontological unity in the world corresponding to the linguistic unity of the expression. This implies the identitarian claim that there is the unity of the chariot, but it is linguistically/conceptually constituted: we project unity onto a given manifold of ‘sub-chariot’ entities.[viii] Identitarian claims are of the form, ‘There is an X but X is identical to Y,’ while eliminativist claims are of the form, ‘There is no X.’ In the present case, the identitarian claim alone is reasonable. For the eliminativist line that there is no unity of the chariot, whether ontologically grounded or constituted by our linguistic/conceptual operations, would imply the nihilist view that there is no chariot at all. Nihilism has little to recommend it, here as elsewhere. Everyone must grant that there is a difference between a chariot and the sum of its parts, for the simple reason that the sum can exist without the chariot existing, as when the chariot is disassembled. This is a difference that must be accounted for if we are to account for the existence of the chariot. If the difference is not accounted for by a principle[ix] of unity – a unifier – in the chariot, then it must be accounted for by something external to the chariot such as some mind’s unifying activity. The existence of the chariot is equivalent to the unity of its parts; so if this unity were just nothing – as it would be if it were neither grounded in the chariot nor grounded in something external to it – then the chariot would be just nothing, which contradicts our experience.[x] Note also that the existence/unity of the chariot is presupposed as a prior condition of the analytic project: to analyze the chariot into the sum of its parts is to presuppose that it is more than the sum of its parts. I shall return to this point.
Now if there is the unity of the chariot, but this unity derives from the unifying power of the mind, then minds must be self-unifying unities. In other words, if the unity of the chariot derives from the unity of a concept which subsumes a manifold of data, and this concept expresses the unity of a conceiving which is itself a synthesizing of a manifold of data, then the synthesizer or unifier must be a self-unifier: it must be the ground of its own unity. How then could minds lack self-nature? If, in reality, Nagasena’s mind -- call it M1 -- were just a bunch of disconnected momentary data, then its unity would have to derive from some other mind, call it M2. (Don’t forget: the difference between a complex entity and the sum of its constituents is real and must be accounted for on pain of nihilism; this principle applies to minds and non-minds alike.) The unity of M2's mind, in turn, would require for its unification M3, and so on into a regress both infinite and vicious. To avoid this regress, we must say that at least one mind possesses an intrinsic principle of unity. We must say that at least one mind is a self-unifying unity of consciousness and self-consciousness.
The upshot is that Nagasena’s position is self-limiting or self-restricting: if he is right that things like chariots lack intrinsic self-nature, then there must be other things, minds, that do possess intrinsic self-nature. For again, it would be nihilism to claim that the chariot has no unity whatsoever: that would be tantamount to claiming that the chariot does not exist. It exists, and its existence is equivalent to the unity of its parts. Given that the chariot possesses unity, and its unity is not intrinsic to it, then its unity must derive from an external unifier such as a mind. But we have just seen that any mind in a position to do the unifying job would have to possess self-nature. It follows that the anatta doctrine cannot be unrestrictedly valid. Indeed, it cannot be unrestrictedly valid even in the realm of samsara if we count minds as samsaric beings.
Let us conclude by considering how the Buddhist theory of anatta stands to Hume’s bundle theory of the self. Both theories deny the existence of substantial selves, and for the same reason, namely, that nothing can be singled out within experience as the self: the self is not an isolable or discriminable datum. Thus when Hume goes in quest of his self, he comes up with nothing: all he stumbles upon are particular perceptions. He concludes from the nongivenness of a self distinct from his perceptions that the self is “...nothing but a bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and are in a perpetual flux and movement.”[xi]
There are two mistakes here, one of which Hume shares with the Buddhists: they imagine that if there is a self, it must turn up as a separate object of experience. It must be something isolable, identifiable, discriminable. But this is an absurd demand to make, since the self is precisely the subject of experience. As such, it cannot be found as just another object of experience among objects of experience. In the diachronic example I gave, the self is experienced not as an object alongside objects, but as the conscious transtemporal unity of the pleasurable and painful states. In the synchronic example, the self is experienced not as an object among objects, but as the conscious simultaneous unity of the visual and olfactory states.
There is no substitute for doing a little phenomenology under bracketing of crude theories that try to assimilate mental phenomena to physical phenomena. I see a bird, and I hear it sing. The bird may or may not exist in reality. Either way, the seeing and the hearing are of the same object, where this ‘of’ expresses the objective genitive. But they cannot be of the same object unless they form a unity which is experienced as more than the sum of its parts, but as a peculiar conscious interfusing of these parts. The unity of consciousness is not hidden or inferred or merely postulated, but given: one experiences the unity of consciousness when one simultaneously sees and hears the same object, but one does not experience the unity of consciousness as the result of a collecting of two antecedently given data (the seeing and the hearing)
Hume compounds his first mistake -- that of expecting the self to turn up as a separate object -- with a second when he identifies the self with a bundle of perceptions. This is an exceedingly crude physical metaphor, suggesting as it does a bunch of sticks, and it is difficult to attach any philosophical sense to it. Does Hume mean to suggest that the self is a collection of perceptions? A collection in what sense? A set of perceptions? A mereological sum? Obviously the self can be neither, since a self must be more than a mere set or sum of data. It must be more, since a set or a sum consisting of my olfactory sensation and your visual sensation would not constitute a self. The conscious data need to be unified, not in some external way, but by an internal principle of unity.
Consider the books in my library. They are more than a set, and more than a sum, since the set and sum could exist even if the library does not exist, as will happen if the library is sold off to different bookdealers at my death. What makes my books a library is presumably their proximity, arrangement, and common ownership by me. Thus A is next to B is next to C...; the logic section is on the topmost shelf; each book stands in the ownership relation to me. But these are all external relations. The perceptions that constitute the hearing of a melody, however, are not externally related to each other. The perception of the last note of a melody must include (in the mode of retention) the earlier perceptions of the previous notes. Thus the very identity of my present consciousness as I hear the last note depends on the preceding perceptions. This implies that my present consciousness is internally related to the preceding perceptions. In what sense, then, can a self be a bundle of perceptions?
It is important to observe that the Buddha does not make Hume’s second mistake, since it is quite clear to the Buddha that whatever a self is, it cannot be a bundle of constituents. For a bundle is dependent for its existence and identity on its constituents. But as the Anattalakkhana Sutta makes clear, a self is an entity that is absolutely permanent and free of all conditioning factors.[xii] Thus it is quite wrong to think of Buddhists as giving a reductive theory of the self in terms of bundles of subpersonal constituents. The Buddhist theory is an eliminativist theory. It is not a theory that identifies the self with something such as a bundle of perceptions; it is a theory that eliminates the self.
I for one resist this elimination. It seems to me that I am relatively a self, that my being is relatively permanent, and that my life is relatively satisfactory. There’s plenty of dukkha in my life, but my life is not dukkha all the way down. I will be told that I am living in illusion. I will respond that something has gone wrong with Buddhist theorizing. The problem, as it seems to me, is an exclusive and unjustifiable reliance on analysis as the way to truth. The commonsensical suggestion that a chariot is a whole of connected parts, and not a mere sum of parts, was rejected on the ground that this connectedness cannot be found by analysis, as indeed it cannot. But this is to presuppose that only what can be found by analysis is real. We have, however, no good reason to accept this presupposition, and a very good reason to reject it. We ought to reject it because the project of analysis itself presupposes the existence of wholes which are more than mere sums. It is only a unitary whole which needs analysis, and yet no such whole can be exhaustively understood by analysis. Thus the very project of analysis presupposes that there are unitary wholes which cannot be exhaustively understood by analysis. Every analysis presupposes a prior synthesis.
What I want to suggest is that each of us, as a self, is just such a whole. Each one of us is a synthetic unity of consciousness and self-consciousness. Analysis may lay bare the contents of our mental lives, but it will never disclose their unifying principle: not because there is no such principle, but because it is not the sort of thing that can turn up under analysis.
[i]. Cf. A Sourcebook in Indian Philosophy, eds. Radhakrishnan and Moore, (Princeton, NJ: Princeton University Press, 1957), pp. 281-284.
[ii]. An absolutely permanent entity is one that exists at all times, while a relatively permanent entity is one that exists at some, but not all, times. An absolutely impermanent entity is one that exists in a radically momentary fashion.
[iii]. Exercise for the reader: explain how Nagasena’s doctrine implies that the whole/part relation has no ultimate ontological validity.
[iv]. A concrete thing is one that is causally active/passive. A partite thing is one that has at least two proper parts. A proper part of a whole W is a part of W that is not identical to W.
[v]. X is wholly distinct from y if and only if x and y have no parts in common.
[vi]. Empiricism and analysis are of course not the same; but Buddhists are proponents of both.
[vii]. Compare Hume’s regularity theory of causation. The theory states that event-token e1 causes event-token e2 if and only if (i) e1 temporally precedes e2; (ii) e1 and e2 are spatiotemporally contiguous; and (iii) whenever any event of e1's type occurs, an event of e2's type occurs. This theory implies that causation ‘in the world’ is nothing more than one event’s contiguously following another. It implies that there is nothing in reality at the level of individual events that connects the cause-event-to the effect-event. The “constant conjunction” expressed in (iii) is a mere empirical (hence contingent) regularity with no ontological ground. The regularity theory is no doubt a deep affront to common sense. But if causation as a bringing-about at the level of individual events is not empirically detectable, why should we countenance it?
[viii]. A wheel is an example of a ‘sub-chariot’ entity. But a wheel too is a whole of parts. One of these is a spoke, which in turn is a whole of parts. . . . Eventually we come to wholly selfless, wholly insubstantial, radically impermanent items which are the ultimate matter of conceptual formation. These true entities or dharmas, as the ultimate matter of conceptual formation, are themselves conceptually unformed.
[ix]. I am using ‘principle’ here in the old-fashioned sense of ontological ground.
[x]. The same goes for wheels, spokes, and so on. A spoke, as a complex entity, is distinct from the sum of its parts. Thus the existence of a spoke needs accounting. When we come to the ultimate selfless items which make up the bedrock of being, however, we come to simples. Since a simple lacks constituents, one cannot oppose a simple to the mere sum of its constituents.
[xi]. David Hume, A Treatise of Human Nature, ed. Selby-Bigge (Oxford, 1975), p. 252. See also p. 207.
[xii]. Cf. my companion piece, “No Self? A Look at a Buddhist Argument,” International Philosophical Quarterly, vol. 42, no. 4 (December 2002), pp. 453-466.