The first argument for Sarvastivada
by David Bastow
Vol. 5 No. 2 Oct.1995
Copyright by Asian Philosophy
THE FIRST ARGUMENT FOR SARVASTIVADA
ABSTRACT Philosophers belonging to the Buddhist school of Sarvastivada
believed in the real existence of past and future dharmas. This paper
explores the implications, soteriological and philosophical, of an
argument for this belief presented at the beginning of an early
abhidharma text. The argument is two-fold: that past states of mind
can be directly perceived; and that the temporal and causal context of
these states of mind, including their karmic future and the
possibility of an alternative saving future, can also be directly
perceived. The paper relates the Sarvastivadins' theory of time to
Buddhist concerns with self-knowledge and with conditional-ivy; and
suggests that the argument is an early example of their adherence to
the epistemological position of Direct Realism.
From the very beginnings of Buddhism, different interpretations of the
Buddha's message began to appear. Around the time of Asoka these
differences began to be formalised, and gave rise to different schools.
Sarvastivada was one of the major schools which arose about this time; as
distinct from the Mahasanghikas, the Sthaviras and the Pudgalavadins. On
the whole these different schools shared the material of the Sutra Pitakas;
but the abhidharma texts they severally produced were more sectarian, more
concerned with their different doctrinal theories.
One of the books of the Sarvastivada Abhidharma Canon, the Vijnanakaya
(probably composed about 200 BCE), has as its first section a chapter
called the Maugdalyayana-skandhaka.  This chapter consists of various
arguments designed to establish the Sarvastivadin belief in the real
existence of past and future dharmas, and to refute the opinion expressed
in its first sentence:
The sramana Maugdalyayana says: The past and the future do not exist; the
present and the unconditioned (asainskrta) exist.
Many arguments are given in the chapter; but most of them are variants of
that with which it begins. It is with this main argument of the chapter
that the present paper will be concerned. (This argument does not appear in
the summary of the Sarvastivadin position expounded and criticised by
Vasubandhu in the Abhidharmakosa, which of course was written about six
hundred years later; and so it is not part of the debate, which has
received some modem attention, between Vasubandhu and Sanghabhadra. )
In the paper's three sections I present the argument and give an analysis
of its structure; I discuss the doctrinal context of the argument; and I
consider matters related to its cogency as a philosophical proof. Two
themes run throughout these three sections: the possibility of
self-awareness, seeing one's own states of mind; and the philosophical
theory of conditionality, of the causal and therefore ontological
interdependence of dharmas, and its 'religious' expression in the doctrine
of the Four Noble Truths.
In this section I shall give a close paraphrase of the argument as it
appears in the text;  and then an account of its argumentative structure
as I see it. The argument begins with a reference to a pronouncement of the
Buddha taken from a Sutra text--probably Anguttara Nikaya III section 69
(i.e. A i 201-3). "There are three akusala-mulani (roots of ill, roots
leading to bad consequences). These are lobha, greed; dvesa, anger; and
moha, confusion." From this agreed premise the argument proceeds, first
taking the case of lobha.
There is no doubt then that there has been, is, will be a seeing that lobha
is akusala (otherwise translated: a seeing of akusala--presumably meaning
akusala-dharmas--in or through lobha). The lobha that is thus seen--is it
past, present or future? If it is past or future, then it must be admitted
that past or future exist. So could it be present? To allow this would
involve admitting that there are in one pudgala two simultaneous cittas,
states of consciousness; but this cannot be admitted. However, there must
be seeing of either past lobha, or future lobha, or present lobha;
otherwise it could not be that someone sees that lobha, akusalamula, is
akusala. And in that case it would not happen that someone becomes repelled
by lobha, detached, freed from lobha, obtains nirvana (or has obtained or
will obtain nirvana).
The same argument is then applied to other things that can be seen with
respect to lobha: it can be seen that lobha is a fetter, a bondage, an
anusaya; and further that lobha is to be rejected, to be left behind, to be
abandoned, to be fully known (prajna).
Later in the chapter other variants of what is essentially the same
argument appear. They all use the same crucial move about the
non-simultaneity of two cittas. One variant  refers directly to Sutra
texts on mindfulness:  when someone has, for example, eye-bondage--that
is the fetter of lustful visual perception--he truly knows that he has
eye-bondage. Another variant,  again based on a Sutra text,  relates
to the use of one's mind to discipline one's mind. "With teeth against
teeth, with tongue against the palate, by thought (cetasa) he tames,
disciplines, subdues thought (cittam)." In all the variants it is argued
that the two states of consciousness cannot be simultaneous; so if the
seeing, the true knowing, the taming, are happening now, at the present
time, then what is seen, truly known, tamed, must exist at some time other
than the present.
To return to the first form of the argument, relating to the seeing of
lobha; this is now taken up again in a slightly different way.
There has been, there is, there will be a seeing that lobha produces
suffering in a future life (or, a seeing of future-life suffering in
lobha). Is what is thus seen of the past, the present or the future? Again
the crucial move in the argument is a principle which rules out the
possibility of what is seen being present. Here what is seen is to be
understood as the future suffering; the 'non-simultaneity principle' this
time is that it cannot be allowed that the same person at the same time
performs an action and experiences its karmic consequences.  That is,
the lobha and its karmic effects must be separated in time. But what is
seen must exist in the past, the future or the present; otherwise it could
not be that someone sees that lobha akusalamula produces in the future a
painful retribution. And so on as before.
The same argument is then said to apply to the other akusala-mulani, dvesa
and moha; and to a long list of akusala actions and states of mind, mostly
motives related to attachment, asravas, hindrances, etc.
I shall now summarise the structure of this argument as I interpret it. In
fact there are two arguments, one deriving from the possibility of seeing
lobha, and the other deriving from the possibility of seeing various truths
[A 1] It is possible to see one's own state of mind (or in other
variants it is possible truly to know one's own state of mind, or it
is possible to grapple with, to subdue one's own state of mind).
[A 2] The seen state of mind cannot be simultaneous with the seeing of
it. [First principle of non-simultaneity.]
[A 3] However, whatever can be seen must exist, must be real. So if
the act of seeing is present, then the seen state of mind must be real
in some non-present mode.
[B 1] It is possible to see certain truths about this state of mind,
this lobha; to see certain things in or through the lobha.
Namely: (i) lobha, greed, is akusala;
(ii) greed is an anusaya, a binding proclivity or habitual
(iii) greed ought to be set aside;
(iv) greed leads to future suffering.
[B 2] All these four clauses in different ways involve seeing the
future. With respect to clause iv, an argument is given: the effects
of greed cannot be simultaneous with the act of greed itself. [Second
principle of non-simultaneity.]
[B 3] However, what can be seen must exist, must be real. So if the
lobha is present, these seen futures for lobha must be real in a
The 'second principle of non-simultaneity' is part of the general theory of
karma, and needs no special comment. But perhaps an explanation is needed
of what I have called the 'first principle of non-simultaneity'. As far as
I know this is not stated as a principle in the Sutras, but it seems to
have been generally accepted by the abhidharmikas; it follows from their
account of the structure of the mind, and it has its own plausibility. 
There were many variants of the Buddhist reductive analysis of the mind,
differing in detail; but its main principles were not controversial.
According to dharma theory, the person is to be analysed into a complex
series of short-lived psychophysical experience-events. This series changes
all the time, and so has diachronic complexity; but any momentary
cross-section will itself reveal another dimension of synchronic
complexity. Analyses of this momentary cross-section will reveal a group of
causally related mental states; but always at the centre of the group is
consciousness--in the sense of awareness and intentionality
(citta/vijnana). All mental activity involves a grasping of something other
than itself. But this consciousness-dharma is always accompanied by a
variety of simultaneous mental factors (cetasikas)--attention, volition,
thought of one kind or another, moods and motives which may be good, such
as faith, mindfulness, equanimity; or bad such as delusion, hate, conceit.
Some of these are always present, some vary from time to time, from moment
to moment. But consciousness is always there. The non-simultaneity
principle then is that an analysis of a state of mind at one time, one
moment, can contain only one consciousness element. If an analysis did
contain two consciousnesses, each would have its associated mental factors,
which might in fact be incompatible with each other, opposed to each other,
as would be hate and equanimity, passion and dispassion. So
consciousness-dharmas must succeed each other. 
The argument in the Vijnanakaya applies this general principle of
non-simultaneity of cittas to mental activity of a special kind of
complexity; when the person is aware of his or her own mental state. This
introspective serf-consciousness is of course of fundamental importance for
the Buddhist path. Only by applying oneself in intense meditation to a
detailed knowledge of the elements which are wrongly synthesised into a
person belief can the wayward mind be controlled, wisdom and detachment
substituted for ignorance and desire. So when one 'sees lobha', there are
two levels of consciousness: the greed-occurrence of which one is aware is
itself a first-level mental complex, with consciousness at its centre,
grasping the object of desire. But then the mindful seeing of the greed is
a second-level consciousness. If these cannot be simultaneous, the seen
lobha must be past or future with respect to the seeing of it. Presumably
it will typically be in the past, though the text does not say so. Since
whatever is seen must exist, the past lobha-dharma must exist. That is,
although it is past, there must be a sense in which it is real at the time
of the seeing.
In this section I wish to explore two themes which are implicit in the
Vijnanakaya arguments about time. They are (a) self-awareness in time; and
(b) the idea that conditioned dharmas necessarily exist in a context of
causal and therefore temporal relationships. Themes (a) and (b) are
prominent respectively in arguments [A] and [B] as described in section 1.
A question which naturally occurs to the Western student of Sarvastivada
is: why should a Buddhist school think it important to work out a position
on the metaphysics of time? As far as I know this question is not directly
addressed by any Sarvastivadin text, so perhaps even to ask it involves
some misunderstanding. What can be shown is that an understanding of the
interrelations between past, present and future is crucial for the
(ultimately reductive) self-awareness and self-understanding of the
Buddhist person; and also for this person's awareness of his soteriological
predicament. The Sarvastivadina were sufficiently impressed by the
importance of these interrelations to make the further ontological move:
that the reality of the present implies the reality of the context with
which it is intertwined. The ancient doctrine of impermanence (later
interpreted as momentariness) does not by itself constitute an adequate
theory of time for the Buddhist; it must be developed to take account of
the doctrine of conditionality.
(a) Self-awareness in time
The first move in the Vijnanakaya argument is the postulation of
'seeing'--which must mean some kind of direct knowledge (the perceptual
metaphor will be discussed in my final section)--of a state of mind. The
use of the non-simultaneity principle in the text makes it obvious that the
two states of consciousness, the seeing and what is seen, belong to the
same pudgala. So we have an example of what may be called self-awareness;
though of course in Buddhist thought reference to the self is no more than
a facon de parler. Self-awareness plays many different roles in Buddhist
theory. To understand the present example we need to place it in its
context, and to distinguish it from other types and levels of
serf-awareness important in Buddhist thought.
At the fundamental level is the self-awareness which is necessary for
self-construction. Our belief in our own existence as persons, as abiding
selves, rests ultimately on the grasping by one consciousness-dharma of a
series, a history, of other consciousness-dharmas. The basic Buddhist
vision, explicated in terms of dharma-theory, is reductive; instead of an
abiding self, to which from time to time various experiences are appended,
there is a stream of short-lived dharmas, psychophysical experiential
elements, with no abiding substantial core.
But a full account of what constitutes the ordinary person must obviously
include something that binds the elements together; what makes us people is
our awareness of ourselves as people. So there must be synthetic activity;
self-construction (one of the several meanings of samskara). The agent, the
subject of this constructive activity, must itself lie within the stream of
dharmas--for the stream of dharmas is all that there is. Out of the dharmas
a self is constructed, by 'running through and holding together' (to use
Kant's potent phrase) past experiences, and perhaps also images of possible
futures. That is, a present dharma--a consciousness dharma--has available
to it, in one way or another, a temporally ordered series of experiences.
Out of them it constructs a self, by claiming them as its own.  Thus I
have available to me certain experiences, as from the inside, of being a
child, a young man, etc.; my self-awareness, my belief in myself as David
Bastow, involves my adopting these experiences, thinking of them as part of
my own life, regarding my present awareness of them as my memories.
But persons are not just contemplators of the past, of what they count as
their own past; they are agents in the world. Deliberate action on my part
involves my looking into the future, seeing various possible futures,
acting with the intention that one of them will come about. This activity
of self-projection into the future is analysed in the ancient Buddhist
causal cycle (Pali paticca-samuppada, Sanskrit pratitya-samutpada) which
explains the causality of the round of rebirth, the perpetuation of the
self or rather of self-construction, the continued fabrication of an
artificial and insubstantial self. The fundamental motivating agent in this
round of self-fabrication is desire, thirst, greed, grasping after the
future. So the first level of self-awareness is that basic awareness of the
past and future which contributes to self-construction. The pudgala thus
constructed is an interweaving of the present with the past and the future;
a being which exists purely in the present moment cannot be said to be a
The Buddha's attitude to such self-construction by awareness of dharmas
across time is complex. Ultimately the constructed self is to be seen for
what it is; the misguided synthesis is to be brought to an end. So as
Dhammapada 348 says: "Let go the past. Let go the future. Let go the
present ..."  But an initial stage on the path of self-deconstruction
may be described as taking oneself seriously. Here a higher level of
self-awareness comes into play.
We may use a distinction from an entirely different perspective to explain
this. The Danish philosopher Soren Kierkegaard, in his first book
Either/Or,  distinguishes initially between two ways of life; what he
calls the aesthetic and the ethical. The aesthetic person lives a life
ruled by feeling, by immediate impulse.  So he lives from day to day,
takes no thought for the morrow, never considers the meaning of his life,
the overall pattern of his past and future actions. His actions arise in
the moment, . from his immediate desires and needs. But Kierkegaard gives
us reasons for rejecting this style of life. He says that the aesthetic
person never really becomes a self, a person, an individual; he never makes
any real choices, so cannot be said to be really free. By contrast, the key
idea of the ethical life is choosing oneself, taking responsibility for
one's past; and making commitments, actively projecting oneself into the
future. In general to set out on the ethical life is to think of oneself as
'having a life', which embraces past, present and future, and is to be
given a shape, a purpose, a meaning. This involves a heightened form of
self-awareness, more explicit than that involved in basic
self-construction. We may call this 'taking oneself seriously' the second
level of self-awareness.
There is no doubt that the Buddha's initial advice to those who wished to
follow him was to choose themselves, to take responsibility for themselves,
to realise their autonomy. He explicitly rejected the autonomy-denying
fatalisms and nihilisms of his day . Partly this was a matter of taking
one's actions and one's motivations seriously because of their long-term
karmic consequences; but his primary message was of the possibility of
realising true freedom by a transcendence of the karmic cycle of action and
How is this to be done? The peculiarly Buddhist answer to this question
crucially involves a further kind, a third level, of self-awareness; and
this is what is involved in the Vijnanakaya argument. This is a mindful
awareness of one's own mental states, which in general is aimed at
self-deconstruction, but in particular is aimed at changing one's mental
make-up so as to be free of greed, hatred, confusion, the three roots of
ill. This awareness must be analytical and objective.
As the Vijnanakaya text says, it is seeing lobha for what it is that leads
to rejection of it, to detachment and liberation from it. Though the greed
itself is akusala, a cause of bad consequences, the seeing of it is
typically kusala, leading towards freedom. The seeing, as opposed to the
seen, is dispassionate, realistic, unfettered; or rather it is potentially
such. To be aware of one's greed is for the moment to transcend the greed;
the awareness is not itself greedy. But of course to be aware of greed is
not in itself to be free of the fetter of greed. Action, the path of
self-mastery and wisdom, is necessary to bring this freedom to fulfilment.
text) of reactions to the seeing of lobha, that is rejection, detachment,
etc., can be interpreted in two ways: as a project of self-change in time,
and as a project aimed at the transcendence of time. The first undertaking
is that of Kierkegaard's ethical person, who becomes a fully realised
person only as he is committed to a day-to-day working out of
life-projects, enterprises which give life a meaning. Kierkegaard's
examples are marriage and a career which is a vocation. Similarly the
Buddhist path could be thought of as a life-project with the aim of
serf-fulfilment. In that case, awareness of one's own greed is knowledge
which is valuable for its pragmatic usefulness in the path-project, like a
racing driver becoming aware of a fault in the engine of his car. The
seeing of greed is a step towards the freedom to act without greed; this is
an example of what Sartre calls consciousness's ability to nihilate,
surpass, transcend, move beyond what it is conscious of.
But the notions of path, goal, self-realisation are paradoxical in the
context of the Buddhist rejection of ideals of selfhood. So we need to bear
in mind another way of interpreting the seeing of lobha and the project of
liberation which should arise from it. This is that, while the first order
act of greed is in time, in the moment, the seeing of it is already a
transcendence of the moment, and indeed of the constraints of the temporal
order. It is a distancing from the serf pursuing its projects in time, a
partial realisation of freedom from action, from conditionality, from time
itself; a preliminary glimpse of an end of projects. 
(b) Causal and temporal relationships
Argument [B] goes beyond self-awareness, the seeing of a state of mind. It
is concerned with the placing of the state of mind, for example lobha, in a
context--primarily the context of its possible futures. Lobha has these
futures in that it is tied to them by various types of causality. So what
is seen is not a future detached from the present, as would be a future
event seen in a crystal ball, but rather a causal web by which dharmas of
the three times are linked together.
In this subsection I wish to suggest that a motive for the Sarvastivadins'
concern with the ontological status of past and future dharmas was their
vivid appreciation of the importance of this causal web. Not that causality
was a new concern within Buddhism; the doctrine of paticca-samuppada was
always central in the Buddhist understanding of the world of samsara, and
paticca-samuppada itself was to be understood in the context of the Four
Noble Truths. That is, as the Second of the Truths it places present
suffering in the context of the past karmic activity which has brought it
about, and present desire in the context of the future suffering which will
follow from it. But also the Third and Fourth Noble Truths tell of a
different causality leading to freedom from ignorance, from desire and
therefore from suffering.
Why then should an appreciation of the Four Noble Truths and their
ontological implications be particularly associated with the
Sarvastivadins? In brief, because they made 'vision' of the Truths central
to their account of the Path. The Vijnanakaya text is an early
manifestation of Sarvastivadin doctrinal developments not merely in that it
argues to the real existence of past and future dharmas ('sarvasti'), but
also in that the details of the arguments reveal the background concerns
which made this 'real existence' important for the school.
These doctrinal developments have been surveyed and analysed by Erich
Frauwallner,  and especially in a recent comprehensive analysis by
Collett Cox.  The developments were concerned mainly with the Buddhist
path; with its aim and with its method. The Sarvastivadins gradually
produced a newly focused and systemafised account of the path, as the
abandonment, the setting aside of 'anusayas'. This term occurs relatively
rarely in the Sutras (though the sutra term 'asrava' is an enigmatic
equivalent); but it does occur in our Vijnanakaya text, associated with the
notions of lettering and bondage. The term means something like a long-term
habit of mind and of motivation, deep-seated and recalcitrant to change.
There are various more or less elaborate lists of these mental fetters. The
lists always contain the three akusala-mulani, greed or desire, hatred, and
ignorance or confusion, but as the tradition develops there comes to be
more and more emphasis on the cognitive element, the conquest of ignorance,
doubt, confusion. These constraining habits of mind are on the
Sarvastivadin account seen as the root cause of suffering; their removal is
the key to liberation.
How are they to be removed? The answer is a consolidation and
systematisation of the culmination of the path as described in the
. . . he applies and bends his mind to the knowledge pertaining to the
extinction of all asavas. He knows as it really is [the Four Noble
Truths as related to suffering]. He knows as they really are 'these
are asavas; this is the origin of asavas; this is the extinction of
asavas; this is the path leading to the extinction of asavas'. To him
thus realising, thus seeing, his mind is set free from the asavas of
sensuous desire, of existence, of ignorance. 
In the Sarvastivadin texts, the primary method for the abandonment of the
anusayas is vision (darsana) of the Four Noble Truths. On many accounts,
vision has to be supplemented by cultivation (bhavana); but as Cox makes
clear these two paths were complementary, and basically similar in mode of
operation. With respect to each anusaya the application to it of the Four
Noble Truths must first be understood, and then experientially realised.
Cox summarises the path-structure as follows:
Praxis consists of a gradual process through which one is sequentially
disconnected from specific defilements through the application of
vision and cultivation, culminating in the complete abandonment and
future non-arising of all such defilements. 
It seems to me that we have in this refocused understanding of the path two
linked motives for the Sarvastivadin concern for the reality of past and
future. First, the notion of an anusaya, a long-term habit of mind, itself
emphasises the temporal interconnections which constitute a person. This is
a powerful causality which has to be broken down, a new interpretation
perhaps of pratitya-samutpada; to break it down one has to understand it
and grasp its whole temporal extension into past and future.
Second, and more importantly, the renewed and systematic emphasis on the
Four Noble Truths is itself an emphasis on the importance of appreciating
the reality of past and future. As explained earlier, the Truths operate
precisely by placing suffering in its causal and temporal context. So the
reality of this context--the causal web in which our present existence is
situated--needs to be metaphysically safeguarded; and it also needs to be
epistemologically justified. These are precisely the concerns of the
The four characterisations of lobha in the Vijnanakaya argument [B] restate
and redescribe this causal and temporal web. These characterisations are
that lobha is:
(ii) a fetter, an anusaya
(iii) to be abandoned
(iv) a cause of future suffering.
The first summarises the other three, so need not be commented on
separately. Clause iv refers to a structurally simple type of causality,
the simple karmic link between action (in this case the mental act of
greed) and experience (in this case suffering). The 'seeing' here is a
straightforward confirmation of the second of the Noble Truths. The
causality is one-directional and not in itself serf-perpetuating. Action
leads karmically to experience; but experience, being resultant and
therefore not karmically potent, does not necessarily lead to renewed
Clause ii refers to a more complex kind of psychological causality. The
text says that as it can be seen that lobha is akusala, so it can be seen
that lobha is samyojana, bandhana, anusaya, upaklesa, paryavasthana. 
Cox says that this is "a stylised list used frequently to refer to all
varieties of defiling factors" p. 69.
Of the five terms anusaya came to be taken as the generic term for these
akusala habits of mind. The metaphor of bondage is obvious in the first two
elements of the list, samyojana and bandhana. The point is that an
occurrence of lobha is inadequately understood unless it is seen as one
phase and manifestation of a self-perpetuating system, reaching into the
past and the future. These mental characteristics are long-term,
persistent, difficult to get out of. They are felt by those striving for
freedom as a constraint, and are recalcitrant to their efforts towards
freedom. They are not to be removed by a simple act of will. To see, as in
the Vijnanakaya argument, that greed is a fetter, is to see the future, in
that the present instance of greed will unless drastic action is taken be
followed whether one likes it or not by future greedy acts, all karmically
potent. The complex psychological implications of the metaphor of bondage
are very well articulated by Vasubandhu at the beginning of his lengthy
chapter on anusayas in the Abhidharmakosa (v la). The picture is one of
positive feedback--an occurrence of greed itself results in a series of
such occurrences; it leads to a strength-erring of the bond, tying the
defilement more securely to the person; it leads to ignorance and error
about the true nature of the object of the desire; and it strengthens its
own cause, namely ayoniso-manaskara, misdirected thought or attention.
Clause iii in the Vijnanakaya text describes quite a different future which
may be seen in lobha. Lobha is to be set aside, discarded, fully and
ultimately known. This makes it clear that the positive feedback of the
anusayas is not inevitable; though it is typical, and for ordinary people
almost universal. The reflective person though sees in lobha not only its
typical akusala future, but also its proper future, that future which is in
accordance with true values. To see lobha as it is to see not only that it
leads to suffering, but also that it should be rejected, discarded.
So the Sarvastivadins' belief in the reality of past and future does not
commit them to an absolutely fixed future, to a deterministic universe. A
true vision and comprehension of the present dharmas sees them as nodes
with at least two types of future latent within them. (As was made explicit
in later texts, future dharmas exist which as a matter of fact never become
According to the developed Sarvastivadin doctrine of the Path, what makes
the difference between these two futures is the persistence or abandonment
of anusayas, mental dispositions to false and confused world views and to
the passions which manifest and nurture these world views. Abandonment is
achieved by a vision of the Four Noble Truths, that is of the complex
causal context of the dispositions. The Vijnanakaya argument [B] is
concerned precisely with the vision of the causal and temporal context of
lobha as one of these dispositions.
Finally I wish to consider the arguments in the Vijnanakaya as
philosophical arguments. My aim here is not to make a final judgement on
their validity, but rather to indicate the more general philosophical
issues which are raised by them. It seems to me that large areas of
philosophical debate are opened up by all the central aspects of these
As I said in my first section, there are really two distinct arguments: the
argument [A] from the seeing of lobha, which is presumably meant to show
the reality of a limited set of past dharmas; and the argument [B] from the
seeing of the temporal and causal context of the lobha dharma. It is worth
remarking how limited these claims are. The matter at issue is not, at
least as far as these arguments are concerned, the reality of all dharmas
in the past and future history of the universe. The Buddhist theoretician
is concerned only with making sense of soteriological understanding and
practice; the Sarvastivadins were working out a metaphysics of time which
they thought would underpin that soteriology.
The two arguments [A] and [B] have a similar form. Both begin (1) with an
'empirical' claim, that a certain kind of dharma can be seen. Both then
argue (2) that these seen dharmas cannot be present at the time of the
seeing; they must be past or future dharmas. And both conclude by invoking
the principle (3) that whatever can be seen must be real, must exist, at
the time of the perception. So past and future dharmas must be real.
[A 1] States of mind such as greed may be 'seen'. As I have argued in
section 2(a) above, this is a reference to the analytical 'self'-awareness
involved in the Buddhist meditative technique of mindfulness. As such it is
for the Buddhist non-controversial; and insofar as it makes an empirical
claim about what is possible for the Buddhist practitioner I have 'no
comment to make on it.
[A 2] What is seen (when greed, etc. is seen) cannot be simultaneous with
the seeing of it. I have explained (in section 1) the context of this in
abhidharmic psychological analysis; but there is also a context of
philosophical debate, which should be mentioned here. In Kathavatthu V.9
(this is a Theravadin (Sthaviravadin) abhidhamma text from about the same
time as the Vijnanakaya) there is a brief debate in which the following
questions are asked:
Is there knowledge of the present?
Is there knowledge of that knowledge by that same knowledge?
Is there knowledge that the knowledge is knowledge by that same
Is the knowledge the object (arammana) of that same knowledge?
The only arguments given in answer to these questions seem to support the
denial of the identification of the first- and second-order knowledges.
They present analogies, at first ambivalent--'Does one feel a feeling by
that same feeling?'; but finally more decisive:
Does one cut a sword with that same sword?
Does one touch a finger-tip with that same finger-tip?
The Commentary links the debate clearly with the Vijnanakaya argument:
'Because two knowledges cannot be simultaneous in one self-conscious
subject, knowledge of the present cannot be known by the same act of
knowledge.' Mrs Rhys Davids' comment (on the Commentary) is pertinent: 'In
other words, self-consciousness is really an act of retrospection; its
object is not present but past'.  But the Kathavatthu passage can also
be seen as prefiguring a much later debate, described in chapter 5 of Bimal
K. Matilal's Perception.  This is about whether an awareness-event
(vijnana) is self-revealing, so that to be aware of something is by virtue
of the activity of that same dharma to be aware that one is aware; or
whether a second-order awareness-event is necessary for such
self-consciousness. As Matilal describes the debate (pp. 148-149), the
Nyaya position is that for serf-awareness two knowledges are necessary;
Dinnaga though believes that mental events are by their nature
serf-luminous, sva-samvedana. As in other matters, the position taken in
the Vijnanakaya seems to be closer to Nyaya than to Dinnaga.
But in my view the matter is more complex than 'one vijnana or two
vijnanas'; self-awareness, knowledge of knowledge, is not a unitary
concept. The distinctions described in my section 2 (a) are relevant here.
Whether or not there is a sense in which conscious states are
self-luminous, what I described earlier as the three levels of
serf-awareness cannot be identified with their object, for the object
consciousness may well exist in their absence. A mental dharma can well
exist without being made part of a synthesised person-construct; without
being part of a meaningful 'life'; and without being the object of mindful
introspective analysis. In the seeing of lobha there are two distinct
states of consciousness, playing the distinct roles of object and subject
As for the pastness of the perceived lobha, this is guaranteed by the
'first principle of non-simultaneity'. But even without this principle it
can plausibly be argued that some memory-knowledge of our own states of
mind has at least the appearance of direct perception; for example my vivid
present recollection of my greedy desire for the last piece of cake at the
party I attended yesterday evening.
Two philosophical comments on the self-awareness theme remain to be made.
The first is about the possibility of introspection. For reasons which will
become apparent my discussion of this will follow my remarks about [B 1].
The second relates to the interpretation of 'seeing' as direct immediate
cognition; and the argument (3) that what is directly cognised must exist.
Comment on this will form the final part of this Section.
[B 1] To see a dharma as it is involves seeing at least something of its
past and its possible futures.
The 'empirical' 'claim here is that it is possible in seeing lobha to see
its karmic consequences in future-life suffering; and also its alternative
possible future leading towards liberation and its long-term status as an
anusaya, a fetter. Obviously there is need of some metaphysical theory to
explain the possibility of the seeing of dharmas from the three times. The
Sarvastivadins later developed a sophisticated theory of 'karitra',
according to which past, present and future modes of reality are
distinguished by their functions, their causal powers. But what is needed
for the Vijnanakaya argument is a theoretical account of the relation
between lobha and its past and futures which makes possible the seeing of
them through it. Just such a theory is to be found in the Maha-vibhasa.
 It lies behind the arguments in that text for the reality of past and
future dharmas. The theory rests firmly on the concept of conditionality
which pervades the early texts.
The dharmas we are considering here are conditioned, samskrta; that is,
they are by their nature interrelated with the complex set of dharmas
involved in their causation, and are themselves part of the causal
conditions for future dharmas. The claim is that to see them for what they
are is necessarily to see them within this system of interrelationships,
linking past, present and future dharmas. What they are includes their
interrelations with other dharmas, past and future; their reality implies
the reality of the whole causal web of which they are merely one node. This
relational ontology is implicit in the Sutta expositions of
paticca-samuppada, where the theory of conditionality is a middle way
between existence and non-existence, between eternalism and annihilationism
(Samyutta Nikaya, Sutta 12 (Midanasamyutta), for example section 15,
section 17; S ii 15-19). There certainly are links between past, present
and future, but they are not substantial (as in eternalism), and they are
not merely external (as in annihilationism). The remaining possibility is
that the momentary dharmas are in their very nature internally related to
the past which conditions them and the future (or futures) which they
condition. The idea is abstractly expressed in one version of the
Maha-vibhasa argument: the present could not be the present except in the
context of past and future.
So if past, present and future dharmas are (though in a sense momentary)
internally related to each other, then it makes sense to claim that someone
who sees a present dharma as it is will see its context of conditioning and
conditioned dharmas; for they are a part of what it is. So to see in the
fullest sense a present lobha-dharma (actually by argument [A] not present
but past), is to see not the past and future phases of this particular
dharma but other past and future dharmas which are internally related to
The position has clear analogies with the Sankhya theory of satkaryavada,
that the effect is present in the cause. In fact in Abhidharma-kosa v 26
Vasubandhu criticises one version of the Sarvastivada theory of time as
being indistinguishable from the Sankhya theory of parinama. But there are
also clear distinctions, as Braj Sinha argues in chapter 8 of Time and
Temporality in Samkhya-Yoga and Abhidharma Buddhism.  The Sankhya view
depends on a strong theory of substance; of abiding elements which are
merely rearranged in the process of change and evolution. The Sarvastivadin
theory operates without substantial continuity; the relations between
present and past and future, between cause and effect, are real but merely
I now turn back briefly to the 'self-awareness' theme prominent in argument
[A]. Is lobha, greed, the kind of thing that can be inwardly seen,
introspected? Modern philosophical psychologists have cast doubt on the
appropriateness of a perceptual model for what seems to be direct knowledge
of our own states of mind, such as my knowledge of my own greedy desires.
For example, Gilbert Ryle in The Concept of Mind:
Motives and moods are not the sorts of things which could be among the
direct intimations of consciousness, or among the objects of
introspection, as these factitious forms of 'Privileged Access' are
ordinarily described. They are not experiences, any more than habits
or maladies are experiences. 
Here Ryle moves from his own analysis of such statements as 'I greedily
desired that piece of cake' to a rejection of the possibility of this
greedy desire being an object of introspection, in the sense of direct
perceptual (though non-sensory) knowledge of the state of mind. For Ryle,
terms such as desire and greed serve as action explanations not because
they name a mental episode which precedes and precipitates the action; but
rather because they indicate the general context in which the action was
done. For example, 'the action was motivated by greed' would be analysed as
equivalent to something like 'the action was aimed at short term
self-gratification without much thought for its effects on other people.'
So knowledge that an action was done from greed would rest on an
understanding of this context. This would involve knowledge of mental
episodes, but none of these episodes would in itself constitute greed.
Similarly for motives of hatred and confusion; these cannot be
introspected, for one cannot introspect a context or a propensity.
What would be the Sarvastivadin response to such a counter-theory? My view,
perhaps surprisingly, is that the Sarvastivadin would find some aspects of
the modern theory congenial. This follows from principle [B 1]. As I argued
above, this implies that present reality can be understood only in the
context of past and future reality--in fact the very idea of the present
implies the existence of a past and a future. The 'realism' of
Sarvastivadin dharma-theory does not imply the existence of dharmas as
self-contained isolated quanta of existence; in fact this is the very
opposite of the truth.
So the Rylean and the Sarvastivadin positions meet in the assertion that
the under- standing of greed, etc. involves not just presently occurrent
phenomenological states, but a whole context of past and future thoughts
and actions. Both would agree for example that greed is only greed if it
rests on an acquisitive belief in the self, and leads to a certain type of
morally deficient action.
There is though a fundamental difference between the two accounts. For Ryle
the links between a greedy state of mind and the consequential actions
would be logical; without a particular pattern of thought and action it
would not count as greed. That is just what greed means.
In the Vijnanakaya argument also, the links between past, present and
future phenomena are necessary, but here the necessity is real rather than
merely nominal. So also the epistemology is different, being not a matter
of linguistic analysis but of direct knowledge of the structure of
consciousness, as in the true dispassionate phenomenological epoche in
which the essential structure of a state of consciousness is known.
So for Ryle mental states cannot be introspected, directly intuited,
because they logically involve a past and future context. The
Sarvastivadins agree that mental states do involve a past and future
context; but to see these states as they really are is, as in the
Vijnanakaya argument, precisely to intuit this complex causal and temporal
[A 3, B 3] My final discussion is of the epistemological principle (3)
which is crucial to both arguments, [A] and [B]. Put at its simplest, this
is the principle that 'if x can be seen, then x exists'. This seems
innocuous, almost tautological, but is here put to powerful use. In fact
the Vijnanakaya argument uses an extension of the principle, for the
knowledge of lobha is not literally by sight; mental rather than visual
consciousness is involved. But the analogy with visual cognition is
important; it is no doubt meant to imply that this 'knowledge by the mind'
is in some way direct (knowledge by acquaintance) rather than by
description. This is certainly a plausible account of mindfulness, and of
short term memory images.
But the application of the principle to past and future dharmas implies a
more specific meaning for 'direct knowledge'. That is, it must be assumed
that in this 'seeing' the perceptual consciousness grasps the very object
(the lobha-dharma) itself, not an intermediary representation. For only on
this interpretation is the argument at all plausible. If what is presently
grasped (in argument [A]) is not the past greedy desire itself but some
representation of it, perhaps causally derived from a memory trace, there
is no reason to conclude that the past desire is now real. Similarly
argument [B] has little force if what is now grasped is an imagined
representation of a predictable future. So what the argument implies is
that the cognitive consciousness grasps not representations of the past and
future dharmas (however fight might be the causal story linking
representations and dharmas), but these very dharmas themselves.
The plausibility of the argument as a whole depends therefore on the
possibility of theoretical support for this account of the relation between
the perceiving consciousness and its object. Such support may be found in
the epistemological theory known as Direct (or Naive) Realism. In modern
epistemology little attention has been paid to this theory, for it has been
thought to collapse rapidly into Representative Realism. For surely direct
perception implies true and complete perception, whereas it is easy to show
that perceptual knowledge is often partial, subjective, misleading. So what
is immediately perceived must be an internal object, at a remove from the
external reality which it aims to represent. If general perception involves
an internal representation, then it would seem that this could easily be
invoked as above, to defeat the Vijnanakaya argument of the reality of past
and future dharmas.
Indeed the Vijnanakaya arguments rely on an unusually extreme version of
Direct Realism. One could be a Direct Realist with respect to the
perception of present trees and chairs, but invoke mediating
representations to explain our knowledge of past and future experiences. On
the other hand, the Vijnanakaya's Direct Realism with respect to past and
future would seem a fortiori to imply Direct Realism with respect to the
It follows that a full defence of our argument would require a strong
defence of Direct Realism as an epistemological theory, with an attack on
the notion of representations as internal objects; or at least a
demonstration that the postulation of such internal objects is unnecessary.
The present paper is not the place for such a defence. I wish instead to
comment briefly on the historical context. There are good reasons for
thinking that Direct Realism was in Indian philosophical debate seen as a
viable and sophisticated theory; and that the Sarvastivadins themselves did
in their epistemological arguments see things from a Direct Realist point
The wider context of Indian philosophical epistemology is admirably
presented in Bimal K Matilal's Perception.  This book is largely
concerned with the debate in classical Indian philosophy between adherents
of Direct and Representative Realism (and also the theories of
Phenomenalism and Idealism, which standardly develop out of the two
Realisms). It is one of the many fascinations of the book for the student
of Sarvastivada that Matilal shows how Direct Realism, which has a strong
initial plausibility, can be developed in a far from naive way; and was so
developed by Nyaya and Mimamsa. Matilal's main theme is the contrast
between Nyaya realism and Buddhist phenomenalism; for him the main stream
of Buddhist epistemology is represented by Dinnaga and Dharmakirti, and he
does not link his extended discussions of Direct Realism to the
Sarvastivada tradition.  For Matilal Direct Realism naturally goes with
a belief in the real existence of external objects, trees and horses;
whereas the Buddhist ontology ultimately reduces these to briefly
manifested dharmas. Still, the Direct Realist view can survive in these
changed surroundings, emphasising the distinction between perceiving
dharmas and perceived dharmas, and insisting that no perceptual event can
take place unless there is an object, a collection of object dharmas, to be
This Direct Realist refusal to accept the possibility of perception with no
external object is evident, for example, in a central Sarvastivadin
(Vaibhasika) text, though one written many centuries later than the
Vijnanakaya. This is Sanghabhadra's defence of Sarvastivada against
Vasubandhu's attack in the Abhidharma-kosa.  In Vasubandhu's account of
the debate on the Sarvastivada theory of time he presents four arguments
supporting the Sarvastivada position. The second and third of these are
versions of the abstract argument already noted in the Vijnanakaya: 
vijnana arises only in the presence of sense organ and sense object;
manovijnana may have as its object past and future dharmas; so these
dharmas must exist. In his (Sautrantika) criticism of these arguments
Vasubandhu makes the standard Representative Realist move. He allows that
there is a sense in which past and future dharmas may be the object of
present vijnana; obviously we can now think about and even have knowledge
of the past and the future; but this has no implications with respect to
external reality. 'Object' (alambana) in this sense can for Vasubandhu mean
no more than internal representation. In the debate which follows, in the
Abhidharma-kosa and in Sanghabhadra's response, the issue is generalised.
Surely, says the Sautrantika, there are many cases of vijnana with a
non-existent object (asadvisayalambaka), i.e. with an internal object,
alambana, but with no external object, visaya. He cites for example dream
objects, the mistaken idea of the 'self', the knowledge of someone who
hears a sound that before the sound arose there was no sound ... In each
case the Sarvastivadin is at pains to show that the perception is a
grasping of some existing thing. The perceived object may not be correctly
known; there may be confusion about its temporal status (as in dreams where
something that is actually past is thought to be present). But the
principle is at all costs to be preserved, that vijnana must have an
object, and that not merely an internal object. In fact for the Direct
Realist the external existent takes the place of the internal object.
The Vijnanakaya argument is worthy of attention not just because of the
intrinsic philosophical interest of debates on the nature of time, but also
because the argument draws its conclusions from two central areas of
Buddhist concern: self-awareness and self-analysis, and conditionality.
Many different types of self-awareness are involved in the Buddhist
understanding of the person; the seeing of lobha which is invoked in our
argument is a direct unmediated grasping of a mental element, part of a
dispassionate analysis of the stream of consciousness which outside the
Path is falsely taken to constitute a self. But the mental element thus
Chosen to play the central role in the 'first argument' has a particular
potency. As a form of desire it is a basic cause of the suffering cycle of
samsara; as an anugaya it is a manifestation of a long term
self-perpetuating character trait. So its present reality is inextricable
from its relations with past and future dharmas. These complex causal
relations are the burden of pratitya-samutpada and of the Four Noble
Truths; it was in support of the reality of those causal relations, of that
causal web, that the Sarvastivadins argued to the reality of past and
 This chapter is in DE LA VALLEE POUSSIN, LOUIS (1925) Le controverse du
temps et du pudgala dans le Vijnanakaya, Etudes Asiatiques publiees a
l'occasion du 25me anniversaire de l'Ecole Francaise de l'Extreme-orient
(Paris), pp. 346-358, which is a translation of the first pan of Nanjio
1281, Tokyo, XXIII, 9, pp. 1-72. I have with the help of Mr Victor He,
University of Sunderland, compared de la Vallee Poussin's translation with
Taisho 26.1539, 531a-536a. This comparison revealed only one substantial
problem of translation, which is noted below (note 8).
 This first chapter of the Vijnanakaya does contain (DE LA VALLEE
POUSSIN, op. cit., note 1, pp. 352-353) a version of an argument about
vijnana: that vijnana relating to past and future cannot be analambana
vijnana; for the Buddha clearly defined vijnana as that which discerns or
grasps its object, the visible, sound, smell, taste, the tangible, dharmas.
This argument is reported and commented on by Vasubandhu in Abhidharma-kosa
v 25 a-b. It may be an abstract version of the argument discussed in the
present paper; but as it stands it is too schematic for its import to be
clear. The 'first argument' of the Vijnanakaya makes clear, as the vijnana
argument does not, what precisely is meant by awareness of past and future.
 DE LA VALLEE POUSSIN, op. cit., note 1, pp. 346-349.
 Ibid., p. 350.
 Samyutta Nikaya iv 89.
 DE LA VALLEE POUSSIN, op. cit., note 1, p. 352.
 Majjhima Nikaya i 242.
 At this point de la Vallee Poussin's translation is misleading. 'Ce qui
est l'object de cette vue, est-ce l'attachement passe? est-ce l'attachement
futur? est-ce l'attachement present?' But for the object of the seeing here
to be lobha would make nonsense of the argument; what is seen (in or
through the lobha) must be the suffering which derives from and is
temporally distinct from the lobha. And the Chinese in fact asks merely:
'Is what is seen of the past, the present or the future?'
 Compare COX, COLLETT (1988) On the possibility of a non-existent object
of consciousness: Sarvastivadin and Darstantika theories, Journal of the
International Association of Buddhist Studies, 11 (1), footnote 52:
'According to the Sarvastivadins and Darstantikas, two instances of thought
(citta) or perceptual consciousness (vijnana) cannot occur simultaneously.'
Our passage in the Vijnanakaya is the earliest text to which Collett Cox
refers to justify her statement. See also Abhidharmakosa ii 34d: the citta
and caittas which have the same object are simultaneous, but 'at a given
moment only one single citta can arise'.
 It is interesting that Gilbert Ryle uses as what he admits is a rather
feeble argument against the possibility of introspective self-awareness the
implausibility of 'attending twice at once': "... the occurrence of such an
act of inner perception would require that the observer could attend to two
things at the same time. He would for example be both resolving to get up
early and concomitantly observing his act of resolving;" RYLE, GILBERT
(1949) The Concept of Mind (London, Hutchinson) p. 164. Of course from the
Sarvastivadin point of view this objection simply begs the question.
 For a fuller analysis of this constructive synthetic activity, see
BASTOW, DAVID (1986) Self- construction in Buddhism, Ratio, XXVIII (2), pp.
 See BASTOW, DAVID (1995) Becoming a changed person, Philosophical
Investigations, 18 (1), pp. 49-64.
 It is significant, and in accord with Sarvastivadin thinking, that
when the past and the future are abandoned, so is the present. As will be
argued later in this paper, past, present and future are inextricably
intertwined. The realisation of impermanence and no-self cannot be
understood as a mode of living purely in the present.
 KIERKEGAARD, SOREN (1959) Either/Or (New York, Anchor Books). First
 This is not the only type of aesthetic life described in Either or,
but it is pertinent here, as a realisation of 'momentariness' which is
obviously far from the Buddhist ideal.
 BASTOW, DAVID (1994) Levels of self-awareness in Pali Buddhism,
Scottish Journal of Religious Studies, XV (1), especially pp. 5-9.
 This interpretation of Sarvastivada finds no direct support in my
text, so I shall not discuss it further. But in his important book: Time
and Temporality in Samkhya-Yoga and Abhidharma Buddhism (New Delhi,
Munshiram Manoharlal, 1983), and especially in Chapter 9, Brai M. Sinha
discusses "the significance of reflection as a special mode of being and
the central role that it plays in the transcendence of temporality" (p.
143). Sinha's notion of 'reflection' points to the ability of consciousness
to step outside its own subject-object structure; in the present case a
radical depersonalising of lobha.
 FRAUWALLNER, ERICH (1971) Abhidharma-Studien III. Der Abhisamayavadah,
Wiener Zeitschrift fur die Kunde Sudund Ostasiens, 15, pp. 69-102.
 COX, COLLETT (1993) Attainment through abandonment: the Sarvastivadin
path of removing defilements, in: R. E. BUSWELL & R. M. GIMELLO (Eds) Paths
to Liberation (Honolulu, HI, Kuroda Institute; University of Hawaii Press)
 Digha Nikaya i 84.
 COX, op. cit., note 19, p. 77. See also Abhidharma-kosa v 3, 4 and 5.
 DE LA VALLEE POUSSIN, op. cit., note 1, p. 347; see also COX, op.
cit., note 19, p. 69 and footnote 27. Upaklesa and paryavasthana refer
particularly to the occurrent manifestations of the dispositions.
 Kathavatthu (Pali Text Society 1979) p. 314. See BASTOW, DAVID (1995)
Debates about time in the Kathavatthu, Buddhist Studies Review,
forthcoming. For reasons explained in that paper, the PTS translation
(Points of Controversy) by S. Z. Aung & C. A. F. Rhys Davids (London, 1960)
Luzac for the Pali Text Society is radically misleading, being an
interpretation based more on the Commentary
(Kathavatthuppakarana-Atthakatha, Pali Text Society 1989) than on the text
oft he Kathavatthu.
 AUNG & RHYS DAVIDS, op. cit., note 23, p. 183, footnote 1.
 MATILAL, BIMAL K. (1986) Perception (Oxford, Clarendon Press) chapter
5: 'Knowing that one knows'.
 The Maha-vibhasa is an immense survey of Sarvastivadin doctrine dating
from several centuries after the Vijnanakaya (see for example HIRAKAWA, A.
A History of Indian Buddhism, Honolulu, HI, 1990; pp. 135-136). The
relevant arguments about time are translated by Louis DE LA VALLEE POUSSIN
(1937) Melanges Chinois et Bouddhiques V, pp. 8-25. For a full discussion
of these arguments, see BASTOW, DAVID (1994) The Maha-vibhasa arguments for
Sarvastivada, Philosophy East and West, 44 (3), pp. 489-499.
 SINHA, op. cit., note 17, p. 125 "While for Abhidharma Buddhism
temporal becoming exemplifies the constancy of the "continuum" (samtana),
the Samkhya-Yoga speaks of the constancy of the continuant (samtani).'
 RYLE, op. cit., note 10, p. 115.
 MATILAL, op. cit., note 25. The first sentence of the Introduction is
'Naive Realism is not really naive'.
 Except in the 'Chronological Table of Philosophers' at the beginning
of the book, where he says (p. xiii) 'Vaibhasika phenomenalistic realism
believes that percepts are as much real as the perception itself and the
external world is directly grasped in our conception-free perception'.
 The most complete account of early philosophical developments in this
area is the paper by Collett Cox, op. cit., note 9. Professor Cox does not
explicitly link her description of these debates to the choice between
Direct and Representative Realism, but there seems little doubt that this
is what is at stake here. That is, to take the title of Cox's paper, the
question 'Can there be a non-existent object of consciousness?' really
means: can vijnana take as its object an internal representation, for which
there may be no similar or simultaneous external objective grounding? For
the Darstantika, who believes in internal representations, there is no
problem here; for Sarvastivada there are no internal objects, so the only
possible objects of perception are external, so in the absence of such
external objects perception is impossible.
 Vasubandhu's discussion is in Abhidharma-kosa V 25-26. Sanghabhadra's
response in the Nydyanusara is translated in DE LA VALLEE POUSSIN, op.
cit., note 26, pp. 25-128.
 See note 2 above.