Two Versions of the An-Atta Teaching

This essay concerns two different versions of the early Buddhist an-atta teaching:
One is the “identity” version, a warning against Attachment to any identity or self-image dependent on any Impermanent condition in the world. The essay “I and my identity” explained this version and gave many examples of its relevance to problems in everyday life.
The other version is what I will call “the ontological no-self doctrine,” a philosophical doctrine denying the existence of a “self.” Theravada Buddhists regard belief in this doctrine as a touchstone of Buddhist orthodoxy. Especially among those not engaged in Buddhist practice, this doctrine has become the most famous doctrine attributed to early Buddhism.

The question as to what the suttas teach and do not teach about the existence/non-existence of a self is a complicated question. In this essay I will confine myself mainly to (1) describing key differences between these two versions of the An-Atta teaching, and presenting arguments showing that the “identity” version can be put into practice in a way that contributes to the goal of achieving “the cessation of dukkha,” liberating practice, but nothing similar can be said of the philosophical no-self doctrine.

Two sutta passages presenting the identity version of the An-Atta teaching.
The first part of the present essay focuses mainly on two parallel and closely adjacent sutta passages which show clearly that the identity version of the an-atta teaching is certainly among the things taught in the Pali suttas, and which give a rather concrete illustration of the basic ideas.
This will be followed by another part which will focus on one of the most important of what I call “ambiguous” sutta passages, the latter half of the Anattalakkhana Sutta. This passage is often assumed to be stating an argument central to the ontological no-self doctrine. I show how succinct wording in this sutta makes it ambiguous enough to be able to be understood in this way. But I also argue that more careful attention to wording and context favors resolving these ambiguities in favor of the view that the primary original intention of this sutta was to teach instead the identity-version of the an-atta teaching.

The Five Khandhas.
All of the passages discussed in this chapter employ the list of five khandhas (commonly translated as “the five aggregates) as a framework for making their points, so I need to preface a discussion of the passages themselves with some comments on the khandhas.
Khandha in Pali refers to a “heap” of things. I think usage and context makes it clear that, in the standard list of five khandhas, the term refers to a logical grouping, a set of general categories under which can be included many different phenomena. For example, when it appears as the second item on the list of khandhas, the term vedanā/”feeling” functions as a general category under which can be included many different feelings and different kinds of feelings. So for example when the first passage quoted below speaks of attachment to vedanā /”feelings,” this does not refer to attachment to feelings in general (attachment to having feelings rather than having no feelings). Attachment to vedanā /feelings refers rather attachment to some particular kinds of feelings, such as calm feelings as opposed to agitated feelings. To make this point clear I will often speak below of “khandha-phenomena,” referring to individual phenomena which can be included under one of the more khandha-categories.
As to the individual items on this list, one thing to note is that we find in the suttas no clear explanation of exactly how they should be understood. Thus Sue Hamilton rightly remarks (p. 70),
[the khandhas] are hardly explained in any coherent way in the Sutta Pitaka: there is no text which gives a full and clear account of what is being referred to by the term khandha.

Everyone probably agrees that, at least when taken collectively, this five-khandha list means to be an exhaustive list of something, what Noa Ronkin (Early Buddhist Metaphysics p. 44) calls a “totality formula.” But what is it an exhaustive list of? Consensus today seems often to take it for granted that this list should be understood ontologically, as an exhaustive list of all components of the being of human beings. I will argue in this chapter for a different understanding, first citing one sutta passage in which the khandhas clearly represent an exhaustive list of all possible objects of Attachment. This is actually one among very many sutta passages which do not mention atta or an-atta at all, but which obviously treat the khandhas an an exhaustive list of all objects of Attachment, the reason they are often called pancupadanakkhandha, “the five Attachment khandhas.”
After this first passage in which the khandhas clearly figure as objects of attachment, I will cite a parallel passage in which it seems clear that the khandhas function as an exhaustive list of all conditions in the world in terms of which a person might define an “identity” to which they might become attached. This is the meaning Bikkhu Bodhi has in mind in another quote below when he call the khandhas, “building blocks which we typically use to construct our sense of personal identity.”

Here then is a list of the five khandhas with translations I will use in this book:
rupa “body”
vedanā “feeling”
sañña “perception”
sankhara “construction”
viññana “state of mind”
Translations of the first three terms here are relatively non-controversial. Translations given here of the final two items sankhara and viññana, depart somewhat from common translations, for reasons explained in Appendix Two below.

Two parallel passages.
I pass on now to a discussion of the two parallel sutta passages occurring in close proximity to each other in the Samyutta Nikaya. Each of these passages goes through each of the five khandha-categories, repeating verbatim the same formulas applied to each. For the sake of brevity, I will cite only those parts of these passages involving the second khandha-category vedanā “feeling(s)”.
The first passage runs as follows.
If one is not devoid of lust, desire, affection, thirst, passion and craving in regard to vedanā/Feeling, then with the change and alteration of this [particular] Feeling [sa vedanā] there arise in him sorrow, lamentation, pain, displeasure, and despair. (SN 22:2)
This passage is easily recognizable as a somewhat more elaborated paraphrase of a basic idea-complex upadana-anicca-dukkha, Attachment-Impermanence-Distress, that occurs frequently in the suttas: Attachment to any Impermanent X makes a person susceptible to Distress when X changes.
Further:

First note that the mention of this feeling (sa vedanā) in the first part of the passage means that what is being referred to is not change in feelings in general, but a change in some particular feeling-state.
– “Lust, desire, affection, thirst, passion and craving” for vedanā/Feelings, serves here as an elaboration on the idea of Attachment to some particular feeling-state.
– “The change and alteration” taking place in some particular vedanā/Feeling is a concrete manifestation of the “Impermanence” of all individual phenomena belonging to the general category “Feelings” (and by extension, all phenomena belonging to any of the khandha-categories).
– “Sorrow, lamentation, pain, displeasure, and despair” makes more vivid the dukkha/Distress that afflicts a person when something changes that she has been Attached to. Note also that what is described here is clearly not physical “suffering,” but emotional “distress,” which is why I think that “Distress” is a better translation of the Pali term dukkha when it occurs in the context of unpleasant experiences due to changes in what a person has become attached to.
Note especially the practical significance of Impermanence here. The meaning of this passage does not depend on a general point that all feelings are constantly changing. For example, pleasant calm feelings might actually last for some time. What this passage envisions is just a case in which, for example, at some future point in time the pleasant calm feelings that a person might at present be inflexibly Attached to might actually disappear to be replaced by unpleasant anxious feelings.
Following the passage quoted above, there follows a description of a contrasting case in which a person “is devoid of lust, desire…for a particular feeling,” in which case “there do not arise in him sorrow, lamentation…” when this feeling changes. That is, what is described here is a way of achieving “the cessation of dukkha.”
Next I want to cite another sutta passage occurring in the section (SN 22:1) just prior to the one commented on above. This passage uses verbal formulas identical to the ones used in the passage quoted above. Except that, where the previous passage speaks of dukkha due to “Lust, desire, affection, thirst, passion and craving” for some Impermanent feeling, this passage speaks of dukkha due to “regarding [some Impermanent feeling] as atta” (vedanām attato samanupassati.) I offer this as support for one of my main arguments in this chapter: that in this passage at least, “regarding some Impermanent X as atta,” is closely connected in meaning to “being Attached to this Impermanent X.”
Here then is the passage in question:
The uninstructed worldling… regards [a particular] Feeling as atta (vedanām attato samanupassati)
. He lives obsessed by the notions, “I am [this particular] Feeling, [this] Feeling is mine (aham vedanā, mamam vedanā-ti)” As he lives obsessed by these notions, that [particular] Feeling [sā vedanā] changes and alters. With the change and alteration of [this] Feeling, there arises in him sorrow, lamentation, pain, displeasure, and despair.
This same passage later asks us to imagine the opposite case, in which
The instructed noble disciple… does not regard [some particular] Feeling as atta… He does not live obsessed by the notions, “I am [this] Feeling, [this] Feeling is mine.” As he lives not obsessed by these notions, that [particular] Feeling (sā vedanā) of his changes and alters. With the change and alteration of [this] Feeling, there do not arise in him sorrow, lamentation, pain, displeasure, and despair.“
Most of the phrasing in this passage is identical to the phrasing of the passage cited and commented on above, with one exception:
Whereas the previous passage says:
the instructed noble disciple is devoid of lust, desire, affection, thirst, passion and craving in regard to vedanā
The present passage says:
The noble disciple… does not regard [a particular] Feeling as atta… He lives not obsessed by the notions, “I am [this] Feeling, [this] Feeling is mine.”
Otherwise, both passages go on to say in identical words:
With the change and alterations of [a particular] Feeling, there do not arise in him sorrow, lamentation, pain, displeasure, and despair.

So these parallels suggest again an equivalence between an attitude described by saying
“[he does not experience] lust, desire, affection… craving for [a particular] Feeling”,
and virtually the same attitude equivalently described as
“[he does not] regard [a particular] Feeling as atta… He lives not obsessed by the notions “I am [this] Feeling, [this] Feeling is mine.”
Thus this second passage shows fairly clearly that the identity version of the an-atta teaching is at least among the teachings found in the Pali suttas. It also shows that this version of the an-atta teaching is closely related to another cluster of ideas found very often in the suttas, the idea that Attachment to some Impermanent X makes a person susceptible to dukkha when X changes. This also then gives us a clear explanation of how the identity version of the an-atta teaching can serve as a guide to Buddhist practice aiming to achieve “the cessation of dukkha.”

Contrasting suppositions defining differences between two versions of the an-atta teaching.
First then I want to spell out more clearly two clear differences between the two versions of the an-atta teaching involved in my discussions.
First, the “identity” version of the an-atta teaching necessarily assumes friction between two elements: (1) some Impermanent and changeable X (e.g. wealth, or calm feelings) which I might be attached to as essential to my identity, and (2), “me” who might at one time feel this changeable X to be an essential and inseparable part of my being, but who at some future time be forced to exist without this X.
Thus this identity version of the an-atta teaching is not a philosophical theory about the makeup of the being of human beings, not about the existence/non-existence of anything. If I am attached an identity as “an always calm person,” this involves no more than simple commonsense assumptions that “calmness” exists and that “I” exist who might relate to calmness in two different ways.
The presupposition of friction between two elements explained above, marks an essential difference between this identity version of the an-atta teaching, and what I call here “the philosophical no-self” version. This philosophical doctrine denies the existence of a single entity, a “self,” as part of the makeup of the being of human beings.
Secondly, “Impermanence” plays an important part in both versions of the an-atta teaching, but a very different one in each case.
In the “identity” version, the “Impermanence” and changeability of all khandha-phenomena is what makes them dukkha-causing for anyone who is Attached to any particular khandha-phenomena, or who “regards some X Impermanent khandha-phenomena as atta,” essential-to-my-identity.
By contrast, in the philosophical no-self doctrine, (1) the khandhas themselves are understood philosophically as an exhaustive list of all components of the being of human beings, so that (2) the assertion that all khandha-phenomena are Impermanent is construed as an indirect way of asserting that no “permanent” self also exists as part of the being of human beings.

A Key Question
But all of this raises a key question: Once we remove Impermanence from its association with conditions in the world that might cause me dukkha if I regard them as essential to my identity, and make Impermanence instead part of a doctrine denying the existence of a “permanently existing” atta/self, this raises the issue: How exactly would belief in a “permanent” self cause me dukkha? And how would rejecting this belief contribute to achieving “the cessation of dukkha”?
Nothing in the suttas even attempts to explain such a connection. In an appendix below I consider the views of three recent authors attempting to explain how belief in a self might cause dukkha, and explain problems with each.

The Anattalakkhana Sutta.
Next I want to comment on a sutta passage that can be described as ambiguous because, as I will show below, succinct wording in this passage makes it able to be understood either (1) as an expression of the “identity” version of the an-atta teaching, as explained clearly in the second passage above, but also (2) as an expression of the ontological no-self teaching. The passage I want to comment on here is the last part of the Anattalakkhana Sutta (SN 22:59). Steven Collins (Selfless Persons p. 98) rightly remarks that the basic arguments spelled out here (including much of the wording), is repeated in a very large number of other sutta passages.
As is true in very many of these passages, this passage is again very repetitious, going through each of the five khandhas and repeating the same formulas of each. Again for convenience’s sake, I cite here again only the sections of this sutta dealing with the second khandha, vedanā “feelings.” One thing that must be kept in mind is a point made above: Although, in accordance with the abstract and succinct style of expression in the suttas, this passage uses the single term vedanā, this term should be understood as a general category under which can be included many kinds of feelings and many instances of feelings. It also must be kept in mind that feeling is only one of five general categories of phenomena dealt with in this passage.
Here is the passage in question:
Buddha: Is Feeling permanent or Impermanent (vedanām niccam va aniccam va)?
Impermanent, sir (aniccam, bhante).
Is what is Impermanent distress or happiness (yam pan-āniccam dukkham va tam sukham va)?
Distress, sir.
Something that is impermanent, distress, changing-by-nature, is this proper to be regarded [as]: “This is mine, this I am, this is my self.” (Yam pan-āniccam, dukkham, viparināma-dhammam, kallam nu samanupassitum: “etam mama, eso aham asmi, eso me attā-ti”)
No sir.
Therefore, bikkhus, any kind of feeling whatsoever, whether past, future, or present, internal or external, gross or subtle, inferior or superior, far or near –
All feelings should be regarded as they really are with correct wisdom thus: “This [is] not mine, this I am not, this [is] not my self” (sabbā vedanā “na etam mama, na eso aham asmi, na me eso atta-ti,” evam-etam yatha-bhutam sammappaññāya datthabbam).
Seeing this the instructed noble disciple experiences revulsion against all feelings (evam passam… sutavā ariya-sāvako vedanāyapi nibbindati)
Experiencing revulsion, he becomes dispassionate. (Nibbindam virajjati)
Being dispassionate he is liberated; (virāgā vimuccati.)

So we can first consider how this passage could be understood as an expression of the philosophical no-self teaching.
First, we would have to come to the passage already assuming
(1) that the question which this passage means to address is an philosophical question concerning the existence/non-existence of a “permanent” atta /“self,” and
(2) that the five khandhas should themselves be understood philosophically, as an exhaustive list of all components of the being of human beings.
Neither of these assumptions can be derived from anything said in this sutta itself.
But if we take these assumptions as a context, this means that when this sutta first says that all khandha-phenomena are Impermanent, and then says that therefore all khandha phenomena should be regarded as an-atta, “not self,” this should be taken as an indirect way of asserting that no “permanent” self also exists as part of the being of human beings. In other words, this gives a negative answer to a question hypothetically being asked about whether any “permanent” self exists as part of the being of human beings.
But, although succinct wording in this passage renders this a possible reading, I want to note two difficulties.
First, this reading leaves no obvious place for the connection between Impermanence and dukkha clearly suggested in the first part of the passage.
Buddha: Is Feeling permanent or Impermanent (vedanām niccam va aniccam va)?
Impermanent, sir (aniccam, bhante).
Is what is Impermanent distress or happiness (yam pan-āniccam dukkham va tam sukham va)?
Distress, sir.
Something that is impermanent, distress, changing-by-nature, is this proper to be regarded [as]: “This is mine, this I am, this is my self.” (Yam pan-āniccam, dukkham, viparināma-dhammam, kallam nu samanupassitum: “etam mama, eso aham asmi, eso me attā-ti”)
Next, note the way the passage ends. If we take what is said in the first part of the passage about the Impermanence of all khandha-phenomena as the premise of a logical argument leading to the conclusion that no “permanent” atta /”self” exists as part of the being of human being, then why doesn’t the passage end by stating this conclusion? It could easily have done so in the words of Vachagotta in another passage (SN 44:8), na atthi atta, “there is no self.”
But this is not the way the passage actually ends. Instead it ends by describing, not a change from believing to not believing in a “permanent” self, but a change in the noble disciple’s attitude toward the khandhas: He “experiences revulsion” (nibbindati) against all khandha-phenomena, which brings him to a state of “dispassion” (virāgā), which in turn constitutes being “liberated” (vimuccati), the goal of Buddhist practice. On this reading, what we have here is an implied contrast with what is said earlier in the passage: To “experience revulsion” against all khandha-phenomena is the opposite of “regarding some khandha-phenomenon as atta.”
Another problem also appears here: suppose that the being of “the noble disciple” consists only and entirely of khandha-phenomena. In this case, what could it be in the being of the noble disciple that “experiences revulsion” (nibbindati) against all khandha-phenomena, and in doing so is “liberated” (vimuccati)?
Following out all the implications of this, first we would have to suppose, quite implausibly, that a collection of khandha-phenomena (making up the entire being of the noble disciple), “experiences revulsion” against this same collection of khandha-phenomena. Rather, the normal implication would be that something exists in the being of the noble disciple different from the khandhas, which is able to experience revulsion against the all Impermanent khandha-phenomena.
Secondly, what about “liberation,” mentioned in the last sentence quoted above. If, according to the khandhas-only doctrine, the entire being of a human being consists only and entirely of khandha-phenomena, then the change which is “liberation” could only consist in a change taking place from one state of this collection of Impermanent khandha-phenomena, to another state of this same collection of Impermanent khandha-phenomena. And if all khandha-phenomena are Impermanent, how then could “liberation” be the decisive and enduring change which the suttas always assume it to be?
These problems just described point us instead to understanding this passage as an expression of the same teaching clearly presented in the second sutta passage discussed above in the first part of this chapter.
In this reading, we would assume, first, not that the passage means to address the theoretical philosophical question about the existence non-existence of a “permanently” existing self. We should assume rather that, like the vast majority of suttas, it is giving advice for liberating practice, explaining how one might achieve “the cessation of dukkha.”
And, secondly, we should not assume a philosophical understanding of the khandhas, as an exhaustive list of all components of the being of human beings. We should assume rather an understanding implied in the two sutta passages quoted at the beginning of this chapter: They are an exhaustive list of all possible objects of Attachment, and all possible conditions in the world on which a person might come to depend for maintaining a particular “identity” or self image. This involves then an easily understood connection between Impermanence and dukkha suggested in the Anattalakkhana Sutta being commented on here: Attachment to any X Impermanent condition in the world as essential-to-my-identity, makes me susceptible to dukkha /Distress when X changes. Conversely, if I cease to be Attached to any Impermanent khandha-phenomena as essential to my identity, but instead realize that being Attached to any such identity makes me susceptible to Distress, then I will “experience revulsion” against all khandha-phenomena, become “dispassionate” toward all khandha-phenomena, and achieve “liberation,” the goal of Buddhist transformative practice.
What is it in a person that experiences revulsion against all khandha-phenomena, becomes dispassionate, and thus achieves the change which is “liberation?” As noted above, it is implausible that what this passage has in mind is khandha-phenomena experiencing revulsion against khandha-phenomena, and a state of liberation consisting in a change in khandha-phenomena, the only things that could be intended if we assume that the entire being of human beings consists only and entirely of Impermanent khandha-phenomena.
Rather, the natural assumption which would make sense of all this, is the assumption that, when “revulsion,” or “dispassion” of a human being, we usually take for granted that these words imply the existence of an experiencing human subject which experiences this revulsion and becomes dispassionate.
(It is important to note the difference between what is being asserted here, and the views of a minority of scholars who think that the suttas teach something like Hindu belief in an eternally existing Atman. This is a specific, explicitly and consciously held religiously-charged belief in an eternally existing Atman.
What I am speaking of here is nothing like this. I am only referring here to minimum assumptions about the existence of a human subject implicitly taken for granted all the time by all normal people. What is taken for granted here is (1) that the very idea of human “feeling” for example, includes the idea of a subject-who-feels. (No one imagines human feelings as free-floating phenomena belonging to no one.) Another minimal assumption taken for granted by everyone all the time is (2) that each individual human subject exists continuously through time as one and the same single subject in this life. This is very different from a religious belief in an “eternally” existing self able to survive death.
Probably for most people these minimal taken for granted assumptions seldom rise to the level of conscious explicit reflection. They are “default” assumptions, implicitly and unconsciously taken for granted in the way all normal people perceive the world. It remains in place so long as no one makes it a matter of conscious reflection, and so long as some people do not adopt a conscious belief rejecting these assumptions. What I claim to be doing here is no more than trying to make explicit neither more nor less than what is implicitly taken for granted all the time by all normal people, so long as no one tries to explicitly reject this taken for granted assumption, and tries to consciously adopt the explicit belief that no human subject really exists.
One other distinction I think important to make for the sake of clarity is the distinction between (1) minimal assumptions that are implicitly taken for granted about a human subject in the way all normal people experience the world, and (2) conscious beliefs explicitly affirming the existence of a human subject, or (3) states of mind described as being “self-conscious” or having a “sense of self.” I argue that in the case of the vast majority of people, what I call “the minimal assumptions implicitly taken for granted” by everyone, seldom rise to the level of conscious thoughts and beliefs as in (2) above. As to (3), in my view it is in the nature of the human subject that this subject as perceiving subject can never appear at the same time as one more object among other objects-perceived. So insofar as being “self-conscious” or having a “sense of self,” describe cases in which the “self” is an object of awareness, they are different from what I am calling a human perceiving subject. For example, if a particular “sense of self” is something that might vary from time to time, then we are talking about an object-of-awareness that a person might become Attached to as part of this person’s “identity,” or self-image. All this must be kept in mind when I say above that the ending of the Anattalakkhana Sutta takes for granted the existence of an experiencing human subject who experiences revulsion against the khandhas, becomes dispassionate, and achieves liberation.

A Parallel passage in the Salayatana Samyutta section of the Samyutta Nikaya.
A further argument in favor of reading the Anattalakkhana Sutta as an expression of the identity-version of the an-atta teaching can be drawn from the fact that a passage (SN 35:1) using much of the same wording found in the Anattalakkhana Sutta, except that this passage uses a list of six ayatanas instead of the five-khandha list.
As described in the suttas, the six ayatanas (sal-ayatana) consist in a list of six senses and their objects: the usual five external senses (sight, hearing, smell, taste, and touch) by which a person can become aware of anything in the external material/social world, plus another “inner sense” called mano by which a person might become introspectively aware of anything happening in her own mental/emotional life.
“The six sense-bases” is the usual translation of sal-ayatana, but the term ayatana is better translated as “region,” so I will translate sal-ayatana as “the six sense-regions,” or simply “the six regions.” These six regions divide up the entirety of the world-perceived according to the six senses which are our sole means of perceiving anything. Just as in the case of the five khandhas, the suttas often use the six-sense-region schema as a framework for presenting ideas concerning Impermanence-Attachment-dukkha: Everything included in any of the the six sense regions is Impermanent, and therefore dukkha causing for anyone Attached to anything included in any of these six regions.
All this is relevant in the present context because, while it is at least plausible that the five khandhas can be understood philosophically, representing an exhaustive list of all components of the being of human beings, this is completely implausible in the case of “the six regions.”
As explained in the suttas, salayatana, “the six regions,” does include the six senses themselves. These senses are certainly among the components normally making up the being of human beings, but it is completely implausible that they can be understood to include every component of the being of human beings. Also “the six regions” in the suttas also include the objects of the six senses. Since the objects of the five external senses make up the entire external material world made up of objects of these five senses, it is even more implausible to take the six sense-bases as an exhaustive list of all components internal to the being of human beings.
Consider in this light the sutta passage mentioned above, in which the six regions take the place of the five khandhas, but which otherwise uses mostly the same wording found in the Anattalakkhana Sutta. Remarks above make it completely implausible to read this as an expression of the philosophical no-self doctrine. Rather, when this passage says that all objects of the five external senses should be regarded as an-atta, this clearly suggests the “identity” version of the an-atta teaching, not the philosophical no-self version. It also suggests that what is urged upon aspiring Buddhists is to refrain from Attachment to any identity or self image dependent, not only on transitory conditions within one’s own being, but also on transitory conditions in the external world, of the kind Rupert Gethin describes in a quote below, such as “student, teacher, banker, lawyer, politician, craftsmen, Buddhist monk” and so on.

Two helpful metaphors.
Finally, I want to cite two sutta passages using metaphors which help give us a clearer understanding of what sutta authors had in mind when they urged aspiring Buddhists to develop the ability to regard everything belonging to the khandhas as an-atta “not-self.”
In the first passage (SN 35:101), the Buddha asks his disciples to imagine a case in which someone carries off a pile of “grass, sticks, branches, and leaves” and “burns them, or does what they want with them.” Then he asks,
Would you think, ‘This person is carrying us off, burning us, or doing what they want with us?’ (amhe jano harati vā dahati vā yathāpaccayam vā karoti)?
The answer is “no”
Why?
Because that is neither self nor belonging to self (Na hi no etam…atta vā attaniyam vā).
In the same way, the eye is not yours, give it up. Giving it up will be for your welfare and happiness. (Evameva kho… cakkhu na tumhākam. Tam pajahatha. Tam yo pahinam hitāya sukhāya bhavissati).
Cakkhu/“eye” here stands for the entire eye ”region”–the faculty of vision as well as the entire world of visual objects. The text first elaborates on this to include all feelings (vedanā) that are stimulated by contact with visual objects. Then it goes through each of the remaining five of the six sense-regions, repeating identical formulas of each.
The point: in the normal case, each of us tends to construct an identity dependent on some particular condition in the world, be it an identity as “a wealthy person” dependent on wealth, or “an always calm person” dependent on the ability to remain always calm. In this case, whatever happens to my wealth or my calm feelings, I feel as “happening to me.” I would feel deeply threatened by losing my wealth or being unable to remain calm. This is what it would mean to regard wealth or calm feelings as atta/”myself.”
The opposite Buddhist ideal is to develop the difficult ability to be able to regard all possible conditions in the world, external or internal, as an-atta, “not-me.” If I were able to reach this ideal, there would be no conditions or events in the world that would feel emotionally threatening to me. Losing my wealth or being unable to remain calm would just be just more individual events among many events happening in the world, of no more concern to me than a pile of brush my neighbor is burning in her backyard.

A second sutta passage (SN 22.85) I want to comment on here begins with the story of an individual who comes to the home of a wealthy householder offering to be his servant, but actually intending to first gain his trust by serving him faithfully and then kill him. Sure enough, this is what happens.
Buddha then comments that the apparent “servant” was actually always a killer (vadakho), from the time he first appeared, even though the householder did not realize this. Then he applies this as a metaphor for the way that an unlearned ordinary person (assutavā puthujjano) is attracted to something belonging to the rupa/”body”-khandha, grasps at it and commits to the view “this is ‘my self’ (rūpam upeti upādiyati addhitthāti atta me’ti).
This is because he
does not understand body(-phenomena) which is Impermanent, as Impermanent; does not understand body, which is dukkha, as dukkha; does not understand body, which is an-atta/”not-self” as not-self; does not understand body, which is conditioned (sankhatam) as conditioned; does not understand body which is a murderer, as it really is, a murderer (vadhakam rūpam vadhakam rūpan’ti yathābhūtam nappajānā ti)
They approach rupa/”body”, are attached to it, and commit to the view “this is ‘my self’ (rūpam upeti upādiyati addhitthāti atta me’ti)
In usual fashion, what is said of the first “body”-khandha, is repeated verbatim of each of the other four khandhas in turn.
The earlier metaphor of the burning brush-pile is about the attitude a good Buddhist should take toward all Impermanent conditions and events in the world. Whereas people normally feel some conditions and events as closely tied to with “me”–regarding them as atta–a good Buddhist would learn to put distance between himself and all these conditions and events in the world, so than nothing happening in the world (e.g. losing my job) would be felt as “happening to me.”
This latter passage about the deceptive and treacherous murderer changes perspective. The point of this passage can perhaps best be put in terms of a self-image or “sense of self.” In the case of the first “body”-khandha, for example, a person might normally feel an attraction to a self-image as “a good-looking person,” “a strong, healthy person,” and so on. But Buddhist teaching draws attention to something normally overlooked when this happens: Allowing Impermanent conditions like good-looks or bodily health to become part of one’s intimate sense of self is like inviting a murderer into one’s house, because this puts one in deep emotional jeopardy due to the changeability of these conditions.

Appendix One. Three modern authors explaining the identity version of the an-atta teaching.
Although some of my arguments above run contrary a common consensus which emphasizes the alleged presence of the philosophical no-self doctrine in the earliest stages of Buddhist tradition, I am not at all alone in claiming that the identity version of the an-atta teaching suttas is certainly among the things taught in the suttas. I only think that, apart from Rupert Gethin, writings by other scholars generally pay insufficient attention to a very fundamental distinction between this and the philosophical no-self doctrine, and especially the importance of this distinction when it comes to interpreting passages like the Anattalakkhana Sutta.
First, then, here is Peter Harvey’s explanation of the identity version of the an-atta teaching (Selfless Minds, p. 44), in which he also, as I do, connects “identification” with “Attachment” or “clinging”:
Identification, whether conscious or unconscious, with something as ‘what I truly and permanently am’ is a source of attachment; such attachment leads to frustration and a sense of loss [dukkha] when what one identifies with changes and becomes other than one desires… Contemplation of phenomena as Impermanent, dukkha, and not-self is a way of undermining craving for [tanha] and clinging [upādāna] to such phenomena.

Bikkhu Bodhi
Next, here is a quote from Bikkhu Bodhi (In the Buddha’s Words, p. 22), a very learned Theravada Buddhist and translator of some of the most recent translations of the Pali suttas. Here Bodhi gives virtually the same explanation Harvey gives, in terms of “identity” and “identification,” but now also brings the five-khandha list into his explanation. In this excerpt Bodhi refers to the five khandhas as “the five aggregates.”

These five aggregates [the five khandhas] are the building blocks that we typically use to construct our sense of personal identity… Whatever we identify with, whatever we take to be a self or the possessions of a self, can all be classified among these five aggregates. The five aggregates are thus the ultimate grounds of “identification” and “appropriation,” the two basic activities by which we establish a sense of selfhood. Since we invest our notions of selfhood and personal identity with an intense emotional concern, when the objects to which they are fastened—the five aggregates—undergo change, we naturally experience anxiety and distress. In our perception, it is not mere impersonal phenomena that are undergoing change, but our very identities, our cherished selves, and this is what we fear most of all.
However… a noble disciple has clearly seen with wisdom the delusive nature of all notions of permanent selfhood and thus no longer identifies with the five aggregates. Therefore the noble disciple can confront their change without anxious concern, unperturbed in the face of their alteration, decay, and destruction.

Bodhi speaks here of “identification” and “appropriation,” as “the two basic activities by which we establish a sense of selfhood.” These seem to fit well with the two metaphors used in the two passages quoted immediately above: The passage using the metaphor of burning brush urges aspiring Buddhists to cease “identifying with” conditions that should be regarded instead as foreign, “not-me.” The passage using the murderer metaphor urges aspiring Buddhists to cease “appropriating” or “regarding as atta” conditions in the world which will be harmful and dukkha-causing to them if they are introjected and made central to one’s sense of self.

Rupert Gethin
Finally I want to quote and comment on the views of another contemporary scholar, Rupert Gethin. Gethin is one of the few scholars who shows awareness of differences between what I call the “personal identity” version of the an-atta teaching and what I call the “philosophical no-self” version. He describes these two teachings under two headings.
What I call the philosophical no-self doctrine, he describes as a “delusion,” “a failure to perceive the world as it actually is.”

The Buddhist critique of the notion of a self rests on the claim that we never in fact experience an unchanging self, and that there is therefore no reason to posit an unchanging self underlying experience. In other words, the idea that one exists as a permanent, unchanging self is born of faulty reasoning based on the failure to perceive the world as it actually is. This notion of self is born of delusion (moha) or ignorance (avidyā/avijjā). (Foundations of Buddhism, p. 146)
In another passage, Gethin makes clear that he understands the specific denial that a “permanent” self exists in a very radical way, as a denial of the existence of a continuously existing self in this life, rejecting the commonsense assumption that, while my experiences might change from day to day, I who have some experiences today am the same “I” who had different experiences yesterday.
Here he includes as a “delusion” the assumption that
an “I”..exists…. who has various experiences… [that] although my experiences may vary there is something–me–that remains constant. (Ibid. p. 134)
But Gethin also explicitly recognizes the difference I argue for, between this philosophical no-self doctrine, on the one hand, and the identity version of the an-atta teaching. He describes this latter idea as “another strand” of the an-atta teaching, which he describes again using the concept of “identifying-with” some part of the world, which he also, along with Harvey and Bodhi, connects with tanha/Craving and upādāna/Attachment.
There is another strand to the Buddhist critique of the notion of self which sees it as intimately bound up with craving [tanha…] and attachment [upādāna].
Thus for Buddhist thought, to understand the world in terms of self is not only to see it wrongly [the “deluded” belief that a human subject exists], but to be led by greed, desire, and attachment.
One’s sense of ‘self’ springs … from the desire to identify and claim some part or parts of the universe as one’s own, as one’s possession, and say of them ‘this is mine, I am this, this is my self’…. We continually crave to be particular kinds of person.
In so far as they are entangled with craving, the notions of self and of personal identity can, from the Buddhist point of view, only lead to suffering…. The appropriating of some part or parts of the universe as mine… the desire to construct my ‘self’ or personal identity…. drives me to accumulate ‘possessions’—both physical and psychological—that define and reinforce my sense of my own selfhood as student, teacher, banker, lawyer, politician, craftsmen, Buddhist monk – as some kind of person as opposed to some other kind of person. And when I feel that what I regard as my self, that what I regard as by rights mine, is in danger of being taken from me, I become angry, frustrated, fearful. (Ibid. p. 146-147)
Gethin says here that “we continually crave to be particular kinds of person,” and gives as examples, “student, teacher, banker, lawyer, politician, craftsmen, Buddhist monk.” So he is here clearly describing the identity version of the an-atta teaching rather than the philosophical no-self doctrine. Attachment to an identity as “teacher” or “banker” for example, is attachment to an identity dependent on changeable conditions in the world, and so can cause me dukkha if conditions change to make it impossible to maintain one of these particular identities. This insistence on the Impermanence affecting all these particular identities (particular “kinds of person”) that I might become Attached to, is clearly different from asserting that I myself do not exist continuously through time as the same single person, so that I cannot be the same person I remember being yesterday.

Appendix Two: Some further notes on the five-khandha list.
Terms representing the five khandhas (rupa, vedanā, sañña, sankhara, viññana) have been translated and understood in a number of different ways. As Sue Hamilton rightly remarks:
[the khandhas] are hardly explained in any coherent way in the Sutta Pitaka: there is no text which gives a full and clear account of what is being referred to by the term khandha.
And on the other hand, I think the most important point to be kept in mind is that, taken collectively the five khandhas occur in the suttas as a universalizing device (what Noa Ronkin calls a “totality formula”), representing absolutely all possible objects of Attachment, and (to use Bodhi’s words) absolutely everything one might take as “building blocks” which a person might use to “construct a sense of personal identity.” For present purposes at least, once one grasps this general point about the khandhas taken collectively, it matters little how one understands individual items on this list. My central theses in this book are probably compatible with any number of understandings of individual items.
Nonetheless, a few remarks will be helpful here about how I think the fourth and fifth items on this list should be understood and translated.
Viññana, when listed as the fifth khandha, is sometimes translated as “consciousness.” But, since the khandhas taken collectively represent impermanent and changeable phenomena that might cause dukkha when they change, this is possibly misleading if it refers merely to the fact of being conscious. The only change in “being conscious” is a change to not being conscious, most likely not what is suggested when viññana is used to refer to something changing and dukkha-causing. It seems also true that I can only suffer dukkha from changes that I am aware of, so when Viññana occurs as the fifth khandha, it must refer not to conscious awareness itself, but to some changeable object of conscious awareness.
For these reasons I think viññana as the fifth khandha is best understood as a particular state of mind, or “mental state.” A closely related possibility is that viññana as the fifth khandha should be understood in the meaning it has in sutta descriptions of the six sense-regions. These descriptions sometimes use viññana in reference to the six senses and their respective objects. For example, visual perceptions, changing moment by moment, are commonly referred to by the term cakkhu-viññana, “eye-consciousness”. Along these lines, “consciousness-moments” is another possible translation of viññana when used as the fifth khandha.
I also advocate a slightly different translation and understanding of the fourth khandha, sankhara. Taken simply in itself, the term sankhara refers to anything “thrown together,” or to the activity of throwing something together. Taken in the context of early Buddhist thought, I think sankhara probably has a very general meaning referring to the “conditioned” character of all phenomena in the world, the fact that all phenomena “arise dependently” on other phenomena. In one of the quotes cited above, as often in the suttas, the participial form sankhata is in fact best translated as “conditioned.”
Understood in this general way, as Walpola Rahula remarks (What the Buddha Taught, p.57), sankhara describes something characteristic of everything belonging to any of the khandhas. “Conditions” or “constructions” are some of the better English words that might evoke these ideas, and I will generally opt for the translation “constructions.

Appendix Three. Self-belief as a cause of dukkha?
One of my arguments above is that the identity version of the an-atta teaching has a very clear and easy to understand relation to core teachings in the suttas regarding Impermanence, Attachment, and dukkha, and thus a clear connection to transformative and liberating practice aimed at achieving “the cessation of dukkha.” By contrast, it is not easy to understand how maintaining commonsense assumptions about the existence of a self as human subject could be a cause of dukkha for anyone, or how rejecting such assumptions could contribute to “the cessation of dukkha.” It is hard to prove a negative, but I doubt if there is any possible causal connection.
Given the importance of liberating practice in Buddhism, one would expect that if there were such a connection, it would be well-known to anyone familiar with Buddhist tradition. But in fact, I have been able to find only three modern authors who even attempt to explain such a connection. Each is somewhat different from the others, and as I explain below, there are problems with each.
But one reason that their proposals are worth considering is because this helps make more clear the problems involved in trying to actually try to put the philosophical no-self doctrine into practice at all.

Eduard Conze.
My first author is Eduard Conze, an older but very learned scholar in the field of Buddhist studies. He thinks that “The specific contribution of Buddhism to religious thought lies in its insistence on the doctrine of ‘not self’ (an-atta).” And he goes on to say that “The belief in a ‘self’ is considered by all Buddhists as an indispensable condition for the arising of suffering.” (Buddhism. Its Essence and Development, p. 18)
But how exactly is “belief in a self” a “condition for the arising of suffering”? Conze attempts to explain this, using suffering from a toothache as an example:
If there is a tooth, and there is decay in that tooth, this is a process in the tooth, and in the nerve attached to it. If now my ‘I’ reaches out to the tooth, convinces itself that this is ‘my’ tooth… and believes that what happens to the tooth is bound to affect me, a certain disturbance of thought is likely to result.
The Buddhist sees it like this: Here is the idea of ‘I’ a mere figment of the imagination, with nothing real to correspond to it. There are all sorts of processes going on in the world. Now I conjure up in my imagination, the idea of ‘belonging,’ and come up with the conclusion that some not particularly well-defined portion of the world ‘belongs’ to that ‘I’ or to ‘me.” (Ibid.)
What seems valid here is that being an experiencing human subject, and experiencing all parts of my body as belonging to “my” body, is a necessary condition for suffering pain from a toothache. If I did not exist as an experiencing human subject, and did not experience my body as “my” body, then I would not suffer from toothache. To this extent, one could say that being such a human experiencing subject, and experiencing my teeth and the rest of my body as “my tooth” and “my body,” is a “cause” of my being able to suffer toothache pain.
But one thing misleading about this quote from Conze comes when, instead of just describing the factors that are necessary presuppositions making our experience of toothache the experience that it is, he claims to be explaining the genesis of this particular state of affairs, and describes this as a more or less conscious process which has allegedly brought it about: He asks us to imagine an “I” (which is itself an imagined reality, not really real), reaches out to the tooth, and convinces itself that it is my tooth. He describes tooth decay as one among “all sorts of processes going on in the world” which, as they exist in themselves belong to no one. Tooth decay is thus one such phenomenon which, considered in itself belongs to no one. But then
I conjure up in my imagination, the idea of ‘belonging,’ and come up with the conclusion that some not particularly well-defined portion of the world ‘belongs’ to that ‘I’ or to ‘me.’
No such genetic account of “the cause of suffering” is ever given in the Pali suttas. Nor does Conze refer to any other Buddhist writing from which he has drawn this account. Its purpose seems to be to suggest that, since my experiencing toothaches the way that I do is the result of conscious processes brought about by my own conscious and voluntary activity, I can alter everything by just refraining from creating an imagined “I,” and refrain from “convincing myself that this is my tooth”–and this is one way that it might be possible to bring about “the cessation of suffering,” at least the cessation of toothache pain.
But of course if we take the next step and imagine how this can actually done, we will recognize that this is not a real possibility. It is worth reflecting on the reason: I do not bring myself into existence as a conscious experiencing subject or “I.” Nor can I engage in conscious activity that would bring about the kind of fundamental change in the way I experience things, that would cause me to cease experiencing toothaches the way that I do. Human beings have very limited ability to alter a number of fundamental aspects of the way they experience the world, including the way each of us experiences our own physical bodies as “my” body.
Modern neuroscience has actually offered us a good model for understanding the physical causes that cause human beings to experience the world the way that they do: Our neuron-based perceptual and cognitive apparatus causes us to experience the world the way we do. And in principle this quite likely applies to the way that we experience toothaches due to tooth decay. All the factors necessarily presupposed in the phenomenon of “suffering from a toothache” are indeed most likely in this sense “constructed” phenomena. But they are not constructed by “me,” and I cannot un-construct them at will, to make them fundamentally different from the way they presently are. No one thinks that they can free themselves from toothaches by just choosing not to be an “I” or choosing not to experience a tooth in their mouth as “my” tooth.
So there is some sense in which it does seem plausible to claim, on a very theoretical level, that being the “I” that I am and experiencing the world the way that I do, is the “cause” of my being able to suffer from a toothache. But this can play no part in Buddhist practice able to be practically effective in helping an actual Buddhist practitioner make progress in reaching “the cessation of dukkha.” This is because the fact of being an I of this kind, and experiencing my teeth the way that I do is not something that I have brought about, and not something I can change at will.

Andrew Olendzki.
My next author is Andrew Olendski, a current meditation teacher in Theravada tradition, who in a recent book also makes an attempt to explain some connection between being a self and suffering dukkha. The relevant passage occurs in a book-chapter entitled “Self is a verb.” Here he asks us to imagine being a self as the product of an activity which he calls “selfing”. He then says that “selfing is optional,” implying that engaging in this “selfing” activity is optional: Individuals could avoid engendering a “self” as part of their being, if they would just choose not to engage in this “optional” selfing activity.
Olendzki also makes it clear that by “self” here, he is referring to a human subject-who-desires. This has specific reference then to what he takes to be Buddhist teaching: that “desires” are the cause of suffering. There could be no desires, and no suffering, unless there existed a human subject-who-desires.
Desires can only manifest if the person who desires is created. The self (as a noun) is created as an (imaginary) subject of desire. (Unlimiting Mind P. 132)
The upshot: A self-who-desires only comes into existence as the result of an action of “selfing,” which is “optional” and need not happen. If this selfing ceased, then there would be no self-who-desires. And if there were no self-who-desires, there would be no desires, and so there would be no suffering.
But of course we can see here the same difficulty that we saw in the case of Conze above. Even though Olendzki tends to avoid first-person expressions, using passive and impersonal expressions instead, what he says could only have a practical effect on my life if “I” as a reader could actually manage to “cease selfing,” that is, cease bringing myself as a subject-who-desires into being. It is only then that the ideal Olendzki suggests could be practically realized: suffering would cease because suffering is caused by desiring, and desiring would cease because I would have caused myself to no longer be a subject capable of desiring.
This runs into the same difficulties noted above in the case of Conze’s explanation. And in fact, shortly after the passage quoted above, Olendzki seems to recognize the implausibility of what he is saying. He continues to want to use “selfing” as a verb, and speaks of “manag[ing] not to self.” But he entirely shifts the meaning of this phrase. “Not selfing” no longer has anything to do with ceasing to bring into being my self as human subject-who-desires. “Managing not to self” is equated purely and simply with “ceasing to grasp,” or “ceasing to cling.”
Try locating the grasping reflex in your own experience, the subtle attitude of holding onto or pushing away what suits or vexes “me,” and see what happens when it is replaced at any given moment by equanimity. Rest assured that if you manage not to self for a moment or two, you will not cease to exist. You will however cease to cling, and for a moment at least there will be no one who suffers.
What he says in this last quote is quite plausible, and a quite plausible account of sutta teaching about how “the cessation of suffering” can be brought about by the cessation of Clinging. But it achieves this plausibility only by entirely changing the subject. He is no longer talking an “I” capable of desiring, ceasing to exist or to be brought into being. He is just talking about a (clearly existing) “I” of the usual kind for whom it is possible to actually cease to have certain kinds of desires.

Mark Siderits
I treat finally some proposals of Mark Siderits (Buddhism as Philosophy p. 76-77) as to how belief/disbelief in a self might be connected to liberating practice, bringing about “the cessation of suffering.”
First, Siderits has in mind a rather unique idea of what the “suffering” in question consists in: It is existential angst or depression over the question of whether my life is a meaningful life or not. Seen in this context, “the cessation of suffering” equates to the cessation of existential angst or depression.
How does Siderits think could we achieve this? By realizing that a necessary condition for a life being meaningful or meaningless is the existence of a human subject for whom events in life can have a meaning. If no such human subject exists, neither can there be such a thing as a meaningful or meaningless life.
In his words:
For there to be depression over the lack of ultimate meaning, there must be a subject for whom meaninglessness is a source of despair. When the Buddhist denies that our lives have meaning, it is not because they hold that our lives are inherently meaningless. It is rather because they hold that meaning requires something that does not ultimately exist, the subject for whom events in a life can have meaning. If there is no such subject–if there is no self–then there is equally no subject whose life can lack all meaning. There is no one whose life either has or lacks meaning, there is just the life.
A first problem with this, as an interpretation of early Buddhist teaching in the suttas, is that nothing in the suttas describes the dukkha /’suffering’ that Buddhist practice means to bring an end to, as the problem of angst over the question of whether one’s life is meaningful or meaningless. The kind of dukkha which the suttas claim to be able to put an end to, is suffering caused by changes in conditions in the world which have become objects of desire/attachment. I think many Buddhists would probably be surprised by the claim that Buddhism “denies that our lives have meaning.”
Siderits does not appeal to any ideas present in the suttas or in later Buddhist tradition, as the source of what he says here; I have not seen anyone else who has found anything like these ideas in any part of Buddhist tradition.
Thus the ideas quoted above are probably best seen as an example of what Siderits elsewhere calls “fusion philosophy.” This is his attempt, not just to interpret teachings found in Buddhist writings, but to fuse together some aspects of “Buddhist philosophy” found in these writings, with ideas from other philosophical traditions he has learned from and espouses. The main such tradition in his case is Anglo-American “analytic philosophy,” and the influence of this latter tradition is not difficult to trace in this excerpt. Analytic philosophers have traditionally tried to adhere to a model of thinking that makes philosophy as similar as possible to the impersonal, objective and theoretical orientation of the physical sciences. This has given them some antipathy to the more “subjective” and emotional elements that have been characteristic of “existentialist” philosophers concerned with questions about the meaning of life.
Coming to Siderits actual proposal, however, we can consider two possibilities:
(1) It seems at least plausible that through my own efforts I could cease to be concerned about the meaning of life, cease to be the kind of “I” that is so concerned, and so cease to “suffer” from existential angst. This would be unlike the cases of Conze and Olendzki described above, where it is not really possible that I could by my own efforts entirely cease to be the kind of self as human subject capable of suffering from toothaches or is capable of desiring. But in this case, we are not talking about ceasing to be a self or “I” of any kind. We are only talking about exercising a capability that “I” might already have, and that is assumed here to be a capacity that all normal people have, to cease caring about life’s meaning.
(2) But it’s possible that Siderits has something more in mind–not a case in which I would just exercise a choice that all human subjects have, to cease caring about the meaning of life. It’s possible that what he means is something more radical, that I would cease to care about the meaning of life because I had ceased to be a self of any kind, and for this reason had lost all the capabilities normally inherent in the being of human beings. In this case, the same objections made about Conze and Olendzki above apply in his case also. This is not something that anyone could achieve by their own efforts.