Issues in the Buddhist Doctrine of Non-Self and Its Later Development: The Absence of the Self, or the Overcoming of Self-Centeredness?

I. Introduction

Does the Buddhist doctrine of non-self (wuwo shuo 無我說) refer to the “absence of the self,” or does it teach the “overcoming of self-centeredness”? The former amounts to an ontological denial of the “I” or the self. By contrast, the latter does not concern whether the “I” or the self exists; rather, it calls for abandoning of self-centered thinking. In early Buddhist texts, the term non-self (anattā; wuwo 無我) is most often used as a predicative nominal expression, to say that the five aggregates (pañcakkhandha/pañcaskandha; wuyun 五蘊), which constitute one’s empirical existence, are said not to be the “I” or the self. A typical example is the scriptural passage, “material form (rūpa; se 色) is non-self” (rūpaṃanattā) (SN III 21, etc.). Here, the Pāli term anattā may be translated either as non-self (wuwo 無我), meaning “without self,” or as not-self (feiwo 非我), meaning “not the self.”¹ Yet when its predicative use is taken into account, the two renderings ultimately convey the same meaning. Whether material form is understood as “lacking” the “I” or the self, or as “not being” the “I” or the self, both interpretations reject the identification of material form with the “I.” The same applies to the other aggregates, such as feeling (vedanā; shou 受), perception (saññā/saṃjñā; xiang 想), and volitional formations (saṅkhāra/saṃskāra; xing 行). In short, all empirical phenomena known as the five aggregates are “other than” the “I,” and the “I” is absent from them. Therefore, the statement “material form is non-self” can be understood as an instruction to recognize the distinction between material form and the “I,” and not to project the idea of the “I” or the self onto material form. This may be understood as a teaching that awakens one to the need to overcome self-centeredness by freeing oneself from “the state of being bound to the five aggregates,” that is, the five clinging aggregates (pañcupādānakkhandha; wuquyun 五取蘊).

However, the denial of the identity between the five aggregates and the “I” may also encourage the idea that it presupposes a transcendent “I” beyond the aggregates. In fact, the claim that “the five aggregates are not the ‘I’” is not a denial of the existence of the “I”; it only means that the five aggregates and the “I” itself are distinct. One may even say that it implicitly affirms the existence of the “I,” since it refers to an “I” distinct from the five aggregates. Of course, Buddhism does not acknowledge any entity within the empirical world other than the five aggregates. It is therefore also possible to argue that saying “the five aggregates are not the ‘I’” is, in effect, equivalent to saying “there is no ‘I.’” Yet not all phenomena or dharmas (fa 法) are explained solely through the five aggregates, and beyond the five aggregates there exists the unconditioned (asaṅkhata; wuwei 無爲) state of nirvana (nibbāna/nirvāṇa; niepan 涅槃). The Nikāyas also contain statements that seem to suggest the existence of “someone” who abides in this transcendent and unconditioned state. One such passage says: “For one who has been ‘extinguished’ [with respect to the defilements], there is no measure [by which one may determine existence or nonexistence]. There is nothing by which one may speak of such a one.” Here, we can detect a nuance suggesting that there exists a state to which the empirical categories of “existence” and “nonexistence” do not apply, and that a transcendent being abiding in such a state is, in some sense, acknowledged.

In this context, scholars such as Hajime Nakamura (中村元), Rhys Davids, and Radhakrishnan have argued that the non-self taught by the Buddha was ultimately intended to reveal a true being comparable to the ātman of Brahmanism.² Even the more recent scholar Johannes Bronkhorst argues that the Buddhist conception of the self clearly corresponds to the Upaniṣadic conception of the self as eternal, blissful, and unchanging (Bronkhorst 2024, 63–64). Although these views may at first appear to conflict with the doctrine of non-self, they are rooted in the denial of the identification of the five aggregates with the “I.” If empirical phenomena are repeatedly denied as “not the ‘I,’” the subject who performs this act of denial may, paradoxically, become more distinct. Transcendent entities such as ātman or puruṣa have historically served precisely this role as the subject. What is noteworthy, however, is that the Nikāyas also contain references to a being who carries out such a denial of identification. One passage states: “Whatever aggregate (khandha/skandha; yun 蘊), element (dhātu; jie 界), or sense base (āyatana; chu 處) there may be . . . one does not conceive it, does not conceive within it, does not conceive from it, and does not conceive it as ‘mine.’ One who does not conceive in this way (na maññamāno) does not cling to anything in the world.” Not conceiving any phenomenon as “mine” is a typical practice of non-self. By referring to “someone” who does not engage in such conceiving, this passage appears to suggest the existence of a transcendent being beyond the five aggregates and other empirical phenomena.

The problem is that once such a transcendent being is posited in any form, one is likely to become caught in craving and attachment to it. This, precisely, would be a return to self-centeredness. Concepts such as ātman and puruṣa are typical examples. They have a transcendental character, while also functioning as the goal of religious practice. Yet from the moment their existence is posited, they tend to operate by excluding factors that do not contribute to, or that obstruct, their realization. This is exactly the oppressive and violent aspect of self-centered thinking. Large-scale ideological conflicts and religious persecutions witnessed throughout history have often had this kind of self-centeredness lurking behind them. At their center stands an unshakable “truth” shared only by their followers. The Buddha’s disciples, who were familiar with the doctrine of non-self, could hardly have failed to recognize this problem. In order to prevent such a situation, they shifted toward firmly establishing the logic of “the absence of the self.” By advancing the claim that “there is, from the beginning, nothing that can be called the ‘I,’” they sought to uproot the very seed of self-centeredness. In this way, after the rise of the Abhidhamma schools, interpretations of non-self tended to settle into a discourse aimed at clarifying “the absence of the self.” In order to ensure the overcoming of self-centeredness, Buddhist thought moved toward the position of “the absence of the self.”

As a result, there are sufficient grounds and reasons for discussing Buddhist non-self both as the “overcoming of self-centeredness” and as the “absence of the self.” However, whether one emphasizes the overcoming of self-centeredness or the absence of the self, it is essential to understand the background and context in which each view was taught. Only then can the significance and legitimacy of each line of reasoning be properly clarified, and the problems that arise from absolutizing either one be avoided. In particular, a discourse on non-self that ignores this doctrinal-historical background risks mistaking the variously transmitted teachings on non-self for nothing more than a disorganized collection of inconsistent doctrines. In order to eliminate suffering, one may first need to overcome self-centeredness. In other words, one must realize that the five aggregates, which are subject to suffering, are not oneself. Furthermore, in order to bring the cessation of suffering to its final completion, it must become clear that there is no self that undergoes suffering. Yet in the course of Buddhist doctrinal history, attempts to overcome self-centeredness came to face the criticism that they might generate a new form of self-centeredness, while the view of “the absence of the self” also revealed a tendency toward self-centeredness when asserted in a dogmatic manner. This is especially true of the latter view, which, despite being criticized as “nihilism” by other religions, has held an authoritative status in traditions such as Theravāda (shangzuobu 上座部) as a non-negotiable orthodox teaching. This paper therefore seeks to examine the forms of the doctrine of non-self and its interpretive traditions from these two perspectives through a comparative analysis. In doing so, it aims to clarify both the strengths and limitations revealed by each perspective and to move toward a fuller understanding of the doctrine of non-self as a whole.

II. Classification of the Doctrine of Non-Self by Type

It is no longer academically acceptable to regard the vast body of Nikāya literature simply as a collection of the Buddha’s direct words. The Nikāyas also include many teachings delivered by the Buddha’s disciples, and in many places they reveal traces of tension and opposition arising from differences in the dispositions and modes of practice among those disciples. Moreover, it has become clear that the final editing of the Nikāyas was not completed until around the sixth century (see Lamotte 2006, 258). This makes it reasonable to suppose that perspectives from later sectarian Buddhism may have found their way into the Nikāyas. Reflecting this textual-historical process, the doctrine of non-self (anattā; wuwo 無我) in the Nikāyas does not appear in a single uniform form. Rather, it can be divided into various aspects and types. These include heterogeneous elements, ranging from what appears to be the original form taught during the Buddha’s lifetime to types that appear to have been refined during the period of sectarian Buddhism. In previous research, I argued that these can be broadly classified into three groups and, more specifically, into twelve types (Lim 2021, 1–22; 2022, 1–24). They may be organized in the following chart.

Table 1.

Classification of the Doctrine of Non-Self in the Nikāyas

Of course, it remains possible that the Nikāyas contain additional types of non-self that are not included in the classification chart above. Therefore, the twelve types presented here may be revised in the future as further evidence comes to light. At the same time, objections may be raised to this typological classification. For example, the Anattalakkhaṇa Sutta (SN III 66–68) presents a form that combines Type 3-1 and Type 1-2. From this, one might argue that distinguishing these two types is unnecessary. However, Type 1-2 also appears independently in many other suttas. In light of this, the distinction between the two types can be regarded as justified. Although the classification into twelve types proposed here requires further discussion, it nevertheless has significance insofar as it provides at least a general picture of the various forms in which the doctrine of non-self appears in the Nikāyas. In particular, the three types in the first group and the three types in the second group deserve special attention because they occur frequently and are closely related to one another in content. The author therefore identifies these as the mainstream forms of non-self discourse in early Buddhism (Lim 2021, 17). By contrast, the six types in the third group contain ideas that differ markedly from those of the first two groups and even display mutually contradictory characteristics. Consequently, if these third-group types were to be regarded as the orthodox doctrine of non-self, the first two groups would have to be rejected or treated as heterodox. Yet, in the Nikāyas, these third-group types actually constitute a minority (Lim 2022, 16).

For the development of the argument in this paper, the types classified in my previous research may be summarized as follows. Type 1-1 in the first group is the typical case in which anattā, the original term for non-self, is used as a predicative noun. It is found in about seventy-five passages (Lim 2021, 7–8). A representative example is the scriptural passage quoted earlier: “Material form (rūpa; se 色) is non-self [or not-self]. Feeling (vedanā; shou 受) . . . perception (saññā/saṃjñā; xiang 想) . . . formations (saṅkhāra/saṃskāra; xing 行) . . . and consciousness (viññāṇa/vijñāna; shi 識) are also [the same].” Variations of this scriptural passage, such as “[material form] should be observed as non-self (anattato) [or as not-self],” appear in about 117 passages (SN II 111, etc.; Lim 2021, 8). Type 1-2 refers to cases in which the term anattā, that is, non-self, does not appear. It is found in about seventy-nine passages (Lim 2021, 8–9). For example, this type states only that the five aggregates and related phenomena are not “mine,” that such an “I” does not exist, and that they are not “my self.” This type may be said to reveal only the aspect of not-self (feiwo 非我), namely that the empirical phenomena known as the five aggregates are not the “I.” Type 1-3, the final form in this group, is significant because it explains the meaning of anattā from within the texts themselves. It is found in about twenty-one passages (Lim 2021, 10–11). A representative passage states: “Bhikkhus, material form is impermanent; what is impermanent is suffering; what is suffering is non-self. What is non-self (yad anattā taṃ) is not ‘mine,’ such an ‘I’ does not exist, and it is not ‘my self.’” This type is especially noteworthy because, while explaining the meaning of anattā, it connects the term to the not-self formulation found in Type 1-2.

The three types of non-self in the second group problematize the “state of being bound by the fetter of views” (diṭṭhisaṃyojanasaṃyutto) and teach that one must go beyond it. Type 2-1, the first case in this group, treats both the “view of self” and the “view of non-self” as objects to be overcome. It is found in three passages (Lim 2021, 12–13). One passage states: “In him, the view ‘I have a self’ arises as true and firm. . . . In him, the view ‘I have no self’ . . . In him, the view ‘I perceive non-self by means of self’ . . . In him, the view ‘I perceive self by means of non-self’ arises as true and firm. Bhikkhus, . . . this is called ‘being bound by the fetter of views’ (diṭṭhisaṃyojanasaṃyutto)” (MN I 8, etc.). This type makes clear that even the “view” asserting non-self can be regarded as a state of bondage and may therefore become an object of criticism. Type 2-2 refers to the state after the defilements have been extinguished. It is found in about nine passages (Lim 2021, 14–15). One example is the passage quoted earlier: “For one who has extinguished [the defilements], there is no measure [by which existence or nonexistence may be determined]. There is nothing by which one may speak of such a one. All dharmas have been cut off, and all paths of language have also been completely cut off.” This passage is distinctive in that it evokes a transcendent state after linguistic discrimination has disappeared. Type 2-3, the last case in this group, describes the subject who negates identification between the “I” and empirical phenomena such as the five aggregates. It is found in about seventeen passages (Lim 2021, 15–16). The passage quoted earlier belongs to this type: “Whatever aggregate (khandha/skandha; yun 蘊), element (dhātu; jie 界), or sense base (āyatana; chu 處) there may be . . . one does not conceive it, does not conceive within it, does not conceive from it, and does not conceive it as ‘mine.’ One who does not conceive in this way does not cling to anything in the world.” The three types in this second group show the view that the ultimate state in which non-self is realized exists, in a transcendent manner, beyond bondage to “views.”

The six types of non-self in the third group, by contrast, are characterized by an ontological clarification of the “absence of the self.” Type 3-1 indicates the absence of a sovereign controller (zhuzhaizhe 主宰者) who exercises control over the five aggregates and related phenomena. It is found in about twenty passages (Lim 2022, 4–5). A representative passage states: “If this material form were the self . . . one could say of material form, ‘Let this material form of mine be thus,’ or ‘Let this material form of mine not be thus.’ But, bhikkhus, material form is non-self. Therefore . . . [even if one speaks in this way] . . . it does not become so” (MN I 230–231, etc.). Type 3-2 refers to cases in which the existence of a transcendent subject or observer apart from the field of experience is said to be meaningless. It is found in one passage (Lim 2022, 5–6). For example, when someone asserts the existence of a self that transcends feeling, the text asks in response whether one could still say “I exist” if no feeling were felt at all. Type 3-3 refers to cases in which a “sentient being” or self is analyzed into empirical elements such as the five aggregates, and its substantial existence is then denied. It is found in two passages (Lim 2022, 8–10). One passage states: “Why do you imagine a ‘sentient being’? Māra, you have fallen into a wrong view. There is only a heap of formations (saṅkhāra/saṃskāra; xing 行), and no ‘sentient being’ can be found here. Just as the name ‘chariot’ exists when parts have been assembled, so too, when the aggregates are present, there is only the conventional expression (sammuti), ‘sentient being.’” These three types share a common feature: they reject the absolute beings or objective realities asserted by non-Buddhist religions and philosophies.

Within the third group, Type 3-4 acknowledges the process of birth (jāti; sheng 生) through dependently arisen conditions, while denying a substantial self that exists through dependent origination. It is found in one passage (Lim 2022, 10–11). For example, in relation to consciousness (viññāṇ/vijñāna; shi 識), the third link of the twelvefold dependent origination, the text says that if one asks, “Who consumes the food called consciousness?” the question is not valid. In this connection, the Buddha says that the question should instead be changed to, “For whom is the food called consciousness?” There is no fixed subject, no “someone,” who enjoys the phenomenon called consciousness. Rather, future birth arises with consciousness as its condition. This is expressed in the statement: “The food called consciousness is the condition for the arising of rebirth in the next life.”³ The meaning is that there is no “someone” corresponding to an original self; there is only new birth through dependently arisen conditions. This content is close to later ontological interpretations of non-self that find the basis of non-self in the doctrine of dependent origination. In the Nikāyas, however, it is mentioned only once, and its focus is not so much to explain non-self itself as to clarify the possibility of new birth.

Type 3-5 consists of a collective critique of four kinds of views of self that may be held with regard to the five aggregates. It is found in fifty-one passages (Lim 2022, 12–14). One representative passage states: “The uninstructed ordinary person . . . [1] observes material form as the self, [2] observes the self as possessing material form, [3] observes material form as being in the self, or [4] observes the self as being in material form, and is caught in [such views as] ‘I am material form, and material form is mine.’”⁴ These four kinds of claims about the self are repeated for each of the five aggregates and are therefore collectively referred to as the twenty forms of identity view (sakkāyadiṭṭhi; youshenjian 有身見), in the form 4 × 5 = 20. These overlap in certain respects with Type 1-1 and Type 1-2, which deny identification with the five aggregates, and they may also be said to overlap with Type 3-2, which refers to the absence of a transcendent self. Nevertheless, they can be treated as a distinct type because they represent an attempt to comprehensively formalize the doctrine of non-self. Finally, Type 3-6, the last case in this group, shows a tendency toward the absolutization of non-self. It is found in about forty-eight passages (Lim 2022, 14–15). A representative scriptural statement is: “All formations (saṅkhāra/saṃskāra; zhuxing 諸行) are impermanent (anicca/anitya; wuchang 無常), and all dharmas (dhamma/dharma; zhufa 諸法) are non-self” (MN I 228, etc.). Another passage, found in one instance, may also be assigned to this type: “Bhikkhus, the statement ‘all dharmas are non-self’ is an established element (dhātu; jie 界), established as Dharma, and determined as Dharma, whether before or after the appearance of the Tathāgata or before his appearance.”⁵ This type, in particular, may be regarded as one of the bases on which some schools, including Theravāda (shangzuobu 上座部), came to present the doctrine of non-self itself in a dogmatic manner. The six types in the third group discussed above share the feature of offering an objectivist and absolutist interpretation of non-self.

III. The Perspective of Disidentification and Its Later Development

As Tilmann Vetter points out, the earliest form of the doctrine of non-self (anattā; wuwo 無我) is closely related to the five aggregates (pañcakkhandha/pañcaskandha; wuyun 五蘊) as the constituent factors of personhood (Vetter 1988, 36–37). In other words, it was an admonition not to project a fixed notion of “I” or “self” onto the physical and mental phenomena that constitute one’s existence. The six types in the first two groups among the twelve types of non-self classified into three groups in the previous chapter are representative examples. Although described in various ways, they share the common feature that their essential content is the “disidentification” from the five aggregates. The term “disidentification” is used primarily in contemporary psychotherapy (Baer 2009, 48; see also 156). The present author adopts this expression because it is corresponds well to the early form of the doctrine of non-self, which teaches that the five aggregates are “not I” and “not my self” (Lim 2016, 270). The objects of “disidentification” include not only ordinary notions of “I” or “self,” but also one’s personal dispositions, habits, perceptions, and judgments. Even the “view that advocates non-self” may itself become an object of disidentification. In this sense, the teaching means that one should not identify with anything that constitutes one’s personal identity, and it may therefore be understood as another way of expressing the idea of “overcoming self-centeredness.”

In order to understand clearly what “disidentification from the five aggregates” means, it is necessary to reflect on what each of the five aggregates signifies. First, it should be noted that the five aggregates refer to existential categories that constitute what is “one’s own.” They must be distinguished from “the elements of the world” as understood in natural science. As existential categories, the five aggregates constitute the “I” from the perspective of the first-person speaker. They include the way one experiences oneself and the world, as well as the content that appears as a result of that experience. Let us consider, for example, material form (rūpa; se 色), the first of the five aggregates, which is often translated as “body” or “material phenomenon.” It is explained in terms of the four great elements (mahābhūta; sida 四大): the earth element (paṭhavī-dhātu; dida 地大), the water element (āpodhātu; shuida 水大), the fire element (tejo-dhātu; huoda 火大), and the wind element (vāyo-dhātu; fengda 風大), along with phenomena derived from these four elements (upādā-rūpa; sida suozao se 四大所造色) (MN I 185). These are explained, respectively, as what is hard or soft, what flows or moistens, what is hot or cold, and what moves or vibrates (MN I 185–188). Yet no phenomenon can be soft or hot independently of the knowing subject who experiences it. A phenomenon experienced as soft or hot must already include a subjective response to it. The earth element or fire element disclosed in this way is therefore nothing other than one’s own experience. The same applies to the remaining aggregates: feeling (vedanā; shou 受), perception (saññā/saṃjñā; xiang 想), formations (saṅkhāra/saṃskāra; xing 行), and consciousness (viññāṇa/vijñāna; shi 識). In particular, these are also called “mental phenomena” (nāma; ming 名), in contrast to the body or material phenomena, and correspond to the gradual process by which one’s own cognition and thought become fixed (MN I 111–112).

The Buddha teaches that one should let go of desire and passion (chandarāga; yutan 欲貪) for each of the five aggregates (SN III 27). If one does so, he teaches, they will never grow again, like “a palm stump whose root has been cut off” (ucchinnamūlaṃ tālāvatthukataṃ). This shows that the five aggregates themselves can change according to one’s attitude toward them. Each aggregate is described as becoming more clearly defined through mutual influence with the others, and through their interaction one’s own existence unfolds in a particular way. Insight into the five aggregates therefore has the significance, above all, of self-reflection. To observe each of the five aggregates as “not oneself” (feiwo 非我), or as “lacking oneself” (wuwo 無我), means distancing oneself from the factors that constitute one’s identity. The reason such reflective insight is necessary is that, when one identifies the five aggregates with oneself, one is inevitably drawn into suffering when their impermanence and unreliability become apparent. This is expressed in the following statement: “For one who is caught in [the view], ‘material form is myself, and material form is mine,’ material form changes and becomes otherwise. Because material form changes and becomes otherwise, sorrow, lamentation, pain, displeasure, and despair arise” (SN III 3, 16, and 18, etc.).

This form of the doctrine of non-self reveals a practical purpose: to overcome existential suffering such as sorrow and lamentation by seeing the five aggregates as they really are, namely as “not oneself” or as “lacking oneself.” The following is a typical passage describing the attainment of liberation through this process:

“Bhikkhus, material form is non-self. Feeling is . . . perception is . . . formations are . . . consciousness is non-self. Bhikkhus, seeing in this way, the noble disciple who has heard the teaching becomes disenchanted with material form. He becomes disenchanted with feeling . . . with perception . . . with formations . . . with consciousness. Becoming disenchanted, he becomes free from passion. Through the fading away of passion, he is liberated. When liberated, there arises the knowledge that he is liberated. He understands: birth is exhausted, the holy life has been completed, what had to be done has been done, and there is no more ‘going toward such a state.’”⁶

This passage is found in exactly the same form at least seventeen times, and if similar forms are included, it appears more than 126 times. Here we can identify a sequence of practice: insight into non-self (anattaṃ passaṃ; wuwo guan 無我觀) → disenchantment (nibbidā; yanli 厭離) → fading away of passion (virāga; litan 離貪) → liberation (vimutti; jietuo 解脫) → knowledge of liberation (vimuttamiti ñāṇa; jietuo zhijian 解脫知見). According to this sequence, the doctrine of non-self is a practical form of self-reflection that culminates in the fading away of passion and liberation. It is not a theoretical “view” arising from a dogmatic concern. If the five aggregates were “objective elements of the world,” insight into them would have to take the form of natural scientific inquiry. The knowledge obtained through such inquiry would also take the form of an objective and theoretical “view.” In that case, however, the meaning of self-reflection found in the passage above would be weakened, and the possibility of personal transformation arising from such reflection would also be reduced.

In this connection, Steven Collins points out that in Buddhism all “views,” including the view of non-self, tend to be judged according to the emotional criterion of “attachment” (1982, 117). This means that what mattered more was the inner attitude with which one asserted a view, whether that view affirmed the self or denied it. From this standpoint, a particular view is not judged simply through the binary distinction between true and false. Instead, it is placed on a single continuum according to whether one is attached to it. This is because each view, differing only in degree, is an expression of inner desire. At the lowest point of this continuum stand the views or speculative reflections of ignorant ordinary people, competing with one another. At the highest point of the continuum, or in a realm beyond it, lies the silent sage’s “unconditioned freedom from views,” free from desire. According to this position, the proper way to respond to wrong views must be found in the transformation of one’s inner emotions and character; philosophical refutation of those views does not carry much significance (Collins 1982, 119). All kinds of views and attachment to them can only be results arising from the interaction of the five aggregates. Moreover, since the five aggregates are interwoven and mutually affect one another, it is not easy to isolate any one of them or treat it as separate from the others. Therefore, the appropriate response to the problem of views is to consider a comprehensive “distancing” from all five aggregates. This is precisely what it means to observe each of the five aggregates uniformly as “not oneself” or as “lacking oneself.”

The three types of non-self in the first group discussed in the previous section take as their content “disidentification” from the five aggregates. The three types in the second group, by contrast, concern bondage through views and the state beyond that bondage. These two groups are naturally connected. Reflective insight into the factors that constitute one’s own existence awakens the need to distance oneself even from one’s own views and claims. Yet such reflective insight, or “distancing,” is not a despairing renunciation of all views. The statement that material form, feeling, and so on are not “mine,” not “I,” and not “my self” means only that material form, feeling, and so on cannot be identified with the “I.” It does not mean that there is no “I.” Rather, this kind of disidentification requires a practical subject who distances himself by denying that his body, feelings, and related phenomena are himself (Lim 2020, 71–78). This situation is particularly evident in Type 2-2, which refers to the state after the extinction of the defilements, and Type 2-3, which describes the subject who denies the identification of the “I” with empirical phenomena. These types may be said to affirm the practical subject who denies that empirical phenomena can be the “I.” On the basis of these types, it cannot be concluded that the existence of the “I” is denied.

There is an even more important reason why the doctrine of non-self should not be understood simply as the negation of the self. The absence of the “I” or the self can give rise to serious ethical problems. An ontological denial of the self may lead to the evasion of responsibility for the results of one’s actions. It may also make it difficult to explain the process of rebirth and transmigration, which most Buddhist schools accepted. In order to resolve this problem, certain Buddhist schools developed distinctive theories that acknowledged a subject of action. A representative example is the Sāmmitīya (zhengliangbu 正量部), known as the Pudgalavādins, or advocates of the person (Pudgalavādin; gewo lun zhe 個我論者). This school posited the existence of the person (pudgala; gewo 個我) in order to clarify responsibility for action and, at the same time, explain the necessity of religious practice throughout the process of rebirth. According to Priestley, the Pudgalavādins understood the Buddha’s doctrine of non-self only as the teaching that the five aggregates and the self cannot be identified (2024, 10). The person was posited as the subject of action and was discussed as something that “cannot be said to be real in the same way as the [five] aggregates (yin 陰), nor to be different from [the five aggregates].” The person in this sense is similar to the six types of non-self in the first two groups discussed above, and it may be said especially to inherit the content of Type 2-2 and Type 2-3. The person proposed by the Pudgalavādins has an extended meaning that includes even the being who remains after attaining nirvana without remainder (nirupadhiśeṣa-nirvṇa; wuyu niepan 無餘涅槃). It is characterized as a practical and ethical subject who experiences both the process of rebirth and the state of nirvana. However, the Pudgalavādin theory provoked strong criticism from schools that understood the doctrine of non-self as an ontological denial of the self. According to Xuanzang (玄奘), the Chinese monk who traveled in India in the seventh century CE, the Sāmmitīya was at that time the largest school within the Śrāvakayāna (shengwensheng 聲聞乘), and its size was comparable to that of all other Buddhist schools, Mahāyāna and Hīnayāna combined (Châu 1999, 13–15). This suggests that, until that time, the perspective of disidentification may have held a dominant position. However, the schools that advocated the person did not expand their influence beyond India, and they left behind almost no texts that clearly present their thought. In the end, their position survived mainly as a target of criticism by later schools that advocated an ontological interpretation of non-self.

As Paul Williams points out (2022, 188–93), criticisms of the Pudgalavādins by other schools focused on rejecting the person as a substance. Yet the Pudgalavādins themselves appear to have been concerned less with ontology than with the problem of moral responsibility. In support of this point, the Treatise of the Sāmmitīya School (Sāmmitīyanikāyaśāstra; sanmidibulun 三彌底部論), which transmits their position, explains that the person should not be regarded as a true and ultimate being, but only as a hypothetical designation (prajñapti; jiashe 假設, also jiaming 假名) that exists conceptually (Lim 2024, 106–112). The person as a hypothetical designation is also described as a rejection of the polar logic of “existence” and “nonexistence”:

To say that [the person] is real or not real, eternal or not eternal, is a wrong view, and should therefore be rejected. For this reason, the person cannot be said to be real in the same way as the [five] aggregates, nor to be different from [the five aggregates]. Therefore, one also cannot say, “This person goes there and is born.”

This passage points out that asserting a “view” of either the “existence” or the “nonexistence” of the person may itself be a wrong view rooted in attachment and self-clinging. This content is consistent with the description of non-self in Type 2-1, which criticizes the “state of being bound by the fetter of views.” In particular, the statement that “one cannot say that this person goes there and is born” makes clear that the person proposed by the Pudgalavādins was not a substantial entity of the kind feared by their critics. Walser argues that this Pudgalavādin perspective recalls the Madhyamaka school (Zhongguan xuepai 中觀學派), which rejected all conceptual thinking while still acknowledging its hypothetical aspect. Further discussion is needed concerning the influence of Pudgalavāda on the formation of Madhyamaka logic. Nevertheless, it is highly possible that the mainstream early Buddhist mode of describing non-self, centered on disidentification, was inherited by the Pudgalavādins. It also seems worthy of further study whether this perspective may, in turn, have directly or indirectly influenced later discussions, including those of the Madhyamaka school.

IV. Ontological Interpretations of Non-Self and Their Later Development

An interpretation of non-self (anattā; wuwo 無我) centered on “disidentification” from the five aggregates (pañcakkhandha/pañcaskandha; wuyun 五蘊) leads one away from the state of being absorbed in the five aggregates. To be absorbed in the five aggregates means to be bound by one’s own feeling (vedanā; shou 受), perception (saññā/saṃjñā; xiang 想), or volitional formations (saṅkhāra/saṃskāra; xing 行). This is precisely the state of falling into self-centeredness. Disidentification from the five aggregates guides one out of such a state of self-centeredness. It may be said to open the possibility of freedom beyond the bondage of the five aggregates. However, this way of understanding non-self may encourage the misunderstanding that there exists a higher-level “I” beyond the empirical dimension. It may also lead to the idea that one must abandon a lower-level self-centeredness in order to realize such a transcendent “I.” In that case, however, the doctrine of non-self would become identical with the Upaniṣadic doctrine of ātman or the Sām. khya doctrine of puruṣa, both of which acknowledge a being beyond linguistic discrimination. Ātman and puruṣa are transcendent beings free from the bondage of the empirical world. In fact, scholars such as Radhakrishnan and Rhys Davids argue that “there is no statement that the self cannot be found outside the five aggregates,” and on this basis arrive at the conclusion that the Buddha’s doctrine of non-self is not different from the doctrine of ātman or the theory of puruṣa.⁷

Scholars who emphasize the distinction between Buddhism and Brahmanism strongly oppose this interpretation of non-self. A representative figure is Walpola Rahula, who presents two reasons. First, in the Buddha’s teaching, all beings are composed only of the five aggregates, and nothing else exists beyond them. Second, the Buddha completely denied the existence of ātman, soul, self, ego, and the like. Rahula also points to the statement “all dharmas are non-self” as an example of the Type 3-6 description of non-self discussed earlier. “All dharmas” (sabbe dhammā; yiqie fa 一切法) include not only conditioned dharmas (saṅkhata-dhamma; youwei fa 有爲法), that is, empirical phenomena, but also unconditioned dharmas (asaṅkhata-dhamma; wuwei fa 無爲法) beyond them, namely nirvana (nibbāna/nirvāṇa; niepan 涅槃). On the basis of this passage, Rahula argues that even if the discussion is extended beyond the five aggregates, as well as within them, there is no room for anything resembling a self (Rahula 1978, 57–58). He strongly insists that, even if the Buddha’s doctrine of non-self were entirely wrong, it would still be mistaken to assert ātman as long as these textual grounds exist (1978, 56). Rahula’s position is firmly grounded in the logic of the ontological denial of the self: “there is, from the beginning, nothing that can be called the ‘I.’” Yet he adds that his position does not deny the “I” when it is used as a reflexive pronoun or an indefinite pronoun (1978, 59–60). This claim makes it clear that one should not become attached to the notion of the self in any form, while at the same time indicating that the function and role of the “I” are not thereby denied. The self, or “I,” is acknowledged as the subject of ordinary experience, but transcendent entities such as ātman or puruṣa must be rejected. This view may be understood as being grounded in a reflection on the historical fact that even religious ideals can become objects of attachment and craving.

The six types of non-self in the third group discussed earlier indicate the absence of the “I” or the self in this context. Although they differ in expression, they share the aim of clarifying that the “I” or the self does not exist as an objective fact. This kind of explanation of non-self may be defined as the construction of a theory of non-self that is independent of one’s emotions or affective states. The emergence of this theory of non-self also seems to be deeply related to a change in the way the five aggregates are understood. When non-self is clarified in an objective manner, the five aggregates come to take on the character of objective phenomena unrelated to one’s own existence. For example, material form (rūpa; se 色), one of the five aggregates, is described in the following way: “Whether past, future, or present; internal or external; gross or subtle; inferior or superior; far or near, all material form . . . .” Such an expression reveals a perspective that regards the five aggregates as objective realities. Material form or material phenomena understood in this way become objective and universal phenomena free from the limits of experience and time. The discussion therefore shifts into a metaphysical context different from that of the types of non-self centered on disidentification discussed in the previous section. In this case, the five aggregates take on the character of objective entities that have nothing to do with reflective insight into oneself. Here, even if one gives up desire for the five aggregates, the earlier scriptural passage that they will never grow again “like a palm stump whose root has been cut off” can no longer apply. Understood in this way, the five aggregates become unrelated to the “I” and separate from one’s cognition. Therefore, no matter how one relates to them, the five aggregates themselves remain as they are.

A representative example of a description of non-self that reflects an objectivist perspective is the passage classified as Type 3-3: “. . . no ‘sentient being’ can be found here. Just as the name ‘chariot’ exists when parts have been assembled, so too, when the aggregates (khandha/skandha; zhuyun 諸蘊) are present, there is only the conventional expression (sammuti), ‘sentient being’” (SN I 135, etc.). Here, the expression for the “I” or the self appears as the common noun “sentient being,” and this “sentient being” is compared to a mere combination of “parts,” rather than something that truly exists. The “chariot” or “sentient being” is analyzed into individual “parts” or the five aggregates, and non-self is explained by saying that what actually exists is only the “parts” or the aggregates. This deconstructive mode of explanation becomes even more prominent in later texts. For example, the Milindapañha cites this passage directly and then asks which among the various “parts” of the chariot can truly be called the chariot. The Abhidharmakośabhāṣya (apidamo jushelun 阿毘達磨俱舍論) also states that the self does not exist and that only the worldly expression (saṃvṛti) “sentient being” (sattva; youqing 有情) is used. It then cites the passage again, presenting it as the statement of the nun Śaila. According to these texts, what truly exists is only the empirical elements known as the five aggregates; “sentient being” or self is nothing more than a conventional expression (sammuti) or a designation (prajñapti). The empirical elements called the five aggregates merely function in mutual connection, and nowhere can an independent person called a “sentient being” be found.

This explanation of non-self may be defined as an objective deconstruction of the self. This logic, too, has its own significance, because the removal of the idea of the “I” or the self is central to overcoming suffering. If the “I” or the self that suffers does not exist as an objective fact, then there is no ground on which suffering can stand. Clarifying this point therefore becomes, in itself, a soteriological message that removes suffering. Moreover, this kind of theory of non-self can be said to be compatible with the method of natural scientific inquiry, since it does not take issue with the inner attitude of the person who engages with it. In particular, the Type 3-6 description of non-self mentioned earlier is a typical example of this objective explanation of non-self. It appears in the statement: “All formations (saṅkhāra/saṃskāra; zhuxing 諸行) are impermanent (anicca/anitya; wuchang 無常), and all dharmas (dhamma/dharma; zhufa 諸法) are non-self” (MN I 228, etc.). All formations belonging to the realm of conditioned phenomena are said to be impermanent, but all dharmas, including both the conditioned (saṅkhata; youwei 有爲) and the unconditioned (asaṅkhata; wuwei 無爲), are defined as non-self. This means that the discourse on non-self goes beyond the conditioned empirical dimension and, for this reason, turns into a kind of metaphysical “view” (diṭṭhi). Indeed, the Nikāyas contain passages that refer to non-self as a “view.” One such passage states: “Bhikkhus, to [see] non-self as non-self is an unperverted perception (saññā/saṃjñā; xiang 想), an unperverted mind (citta; xin 心), and an unperverted view (diṭṭhi; jian 見).” This passage, found twice in the Nikāyas, is noteworthy because it openly presents a “view of non-self.” It clearly conflicts with Type 2-1, which criticizes the assertion of non-self as a “view,” as well as with Type 2-2 and Type 2-3, which describe a dimension beyond “views.”

Although ontological discourse on non-self was only a minority position in the Nikāyas of early Buddhism, in later periods it emerged as a mainstream mode of explanation in the texts of various Buddhist schools. A typical example is the Abhidharmakośabhāṣya, which interprets the meaning of the doctrine of non-self as the claim that there is no independently existing self. Furthermore, it clarifies, as an objective fact, that all conditioned phenomena are dependent existents that cannot arise by their own power alone:

Body and mind each arise on the basis of their own causes and conditions, and these causes and conditions, in turn, arise in dependence on their own causes and conditions . . . . In view of this fact, one who can act by one’s own power alone is called an “agent,” but no self of that kind is perceived.

In this passage, it is explained that there is no “agent” independent of causal dependence. The passage clarifies the absence of the self through causal dependence. In this respect, it is connected with the Type 3-4 description of non-self, which denies a substantial self that passes through the process of dependent origination (paṭiccasamuppāda/pratītyasamutpāda; yuanqi 緣起), while still acknowledging future birth arising through dependently arisen conditions. The Abhidharmakośabhāṣya further extends this logic by applying the principle of non-self to the process of rebirth and thereby explaining the issue of responsibility for action. One example is the following statement: “Although the fire that burns a field (liao yuan 燎原) ceases in each moment, it is said to ‘undergo transmigration’ (saṃsarati; lunhui 輪迴) because it continues in succession (santatyā; xiangxu 相續). In this way, the five aggregates [that continue through rebirth] are collected together and hypothetically designated for a time as a ‘sentient being.’” The meaning is that there is no subject that passes through rebirth, but the causal continuity of the five aggregates, moving from a previous life to a later life, may be hypothetically designated as a particular “sentient being.” In this way, the influence of a previous life on a later life is explained, and at the same time the issue of responsibility for action is also clarified. This forms the core logic of the “Refutation of the Self” chapter of the Abhidharmakośabhāṣya, which is devoted entirely to the interpretation of non-self. It may be said to show a clear tendency in the discourse on non-self that developed in sectarian Buddhism.

Ontological discourse on non-self also appears clearly in the Visuddhimagga (qingjing daolun 淸淨道論) of Southern Theravāda (shangzuobu 上座部). The following passage is an example:

He does not see another “maker” (kāraka) behind the causes. When the arising of the result occurs through the arising of the result, there is no separate “experiencer” (paṭisaṃvedako). He sees rightly with correct insight that the wise speak conventionally using names accepted in ordinary usage. (See Daerim 2004, 208; Vism 602)

In this passage, terms such as “maker” and “experiencer” are explained as conventional expressions only; they do not actually exist. A being corresponding to the “I” or the self is denied at its root. The Visuddhimagga also explains this objective fact of non-self in connection with the process of rebirth, in a way similar to the Abhidharmakośabhāṣya. For example, it adds the following statement: “There is no ‘maker’ of karma and no ‘experiencer’ of its result; only pure dharmas arise. This is right seeing. Thus, when karma and its result proceed together with their causes, their beginning cannot be known, as in the case of seeds and trees. [Therefore,] one does not hold that [karma and its result] will not arise in future rebirth . . .” (see Daerim 2004, 208–209; Vism 602). This passage also agrees with the hypothetical designation of a “sentient being” through the continuity of the five aggregates in the Abhidharmakośabhāṣya, which denies a subject of rebirth while still acknowledging future birth. Explaining non-self together with the teachings of rebirth and dependent origination in this way shares the same context as the Type 3-4 description of non-self discussed above. The ontological interpretation of non-self in the Visuddhimagga appears most clearly in the following verse:

There is only suffering, but no “one who suffers”;
there is only action, but no “agent”;
there is only cessation, but no “one who has ceased”;
there is only the path (magga/mārga; dao 道), but no “one who goes.”
(See Daerim 2004, 576; Vism 513)

Through this verse, we can see that the nonexistence, from the very beginning, of any being corresponding to the “I,” the self, or the agent is declared as a fundamental fact. The Abhidharmakośabhāṣya and the Visuddhimagga have traditionally held central positions in the scholastic systems of Northern and Southern Buddhism. As seen above, their interpretations and explanations of non-self are strikingly similar. If contemporary interpretations of non-self tend to lean toward the ontological denial of the self, this is probably due to the influence of these two texts.

V. Conclusion

There are various types of the doctrine of non-self (anattā; wuwo 無我), and each type has its own lineage and transmission. These types may be understood through two broad categories: the “overcoming of self-centeredness” and the “absence of the self.” The types that belong to the former category take as their content “disidentification” from the five aggregates (pañcakkhandha/pañcaskandha; wuyun 五蘊), which constitute the “I” or one’s own existence. By awakening one to the fact that the five aggregates, such as feeling (vedanā; shou 受) and perception (saññā/saṃjñā; xiang 想), are not oneself, they guide one away from the self-centeredness that arises from being bound to the five aggregates. The latter category, by contrast, focuses on demonstrating that there is, from the beginning, nothing that can be called the “I” or the self. by deconstructing the idea of the “I” or the self, these types fundamentally block any room for attachment and self-clinging to arise. Yet the former category tends to highlight the subject who carries out disidentification from the five aggregates, and this raises the problem that it may be confused with a transcendent self of the Upani adic type. The latter category, meanwhile, gives rise to controversy because it takes the form of a metaphysical “view” about the “absence” of the “I” or the self. The Buddha was critical of metaphysical views and regarded immersion in a particular “view” as an unhealthy condition requiring treatment. In conclusion, these two opposing currents may be said to have their own reasons for criticizing one another.

In relation to this issue, scholars such as Hajime Nakamura (中村元) and Tilmann Vetter have paid attention to the historical transformation of the doctrine of non-self.⁸ According to them, the earliest form of the doctrine of non-self was closer to not-self (feiwo 非我), meaning disidentification from the five aggregates, whereas in later periods the tendency toward an ontological denial of the self became more prominent. The claim that the content of the doctrine of non-self changed over time can offer a persuasive explanation for the diversity of non-self types found in the Nikāyas. According to philological discussions, the final editing of the Nikāyas took place at least about one thousand years after the Buddha’s passing into nirvana, and it is said that each Buddhist school at that time possessed its own collection of discourses (sūtra-piṭaka; jingzang 經藏). It is reasonable to assume that the discourse collections belonging to each school were colored by that school’s own doctrinal tendencies. It is also highly possible that the doctrine of non-self preserved in the Nikāyas includes perspectives specific to Theravāda (shangzuobu 上座部). Nevertheless, in the Nikāyas, the perspective of not-self, meaning disidentification from the five aggregates, appears as the mainstream tendency, whereas the ontological denial of the self appears only as a minority position. This suggests that, although the tendency toward the ontological denial of the self became dominant in sectarian Buddhism, it was not strong enough to alter the original character of the Nikāyas.

The situation changes, however, in the commentarial texts that emerged after the period of sectarian Buddhism. In later texts that exerted outstanding influence on Buddhist history, such as the Abhidharmakośabhāṣya (apidamo jushelun 阿毘達磨俱舍論) and the Visuddhimagga (qingjing daolun 淸淨道論), the tendency toward an ontological interpretation of non-self becomes clearly dominant. In particular, these texts connect the doctrine of non-self with the doctrines of dependent origination (paṭiccasamuppāda/pratītyasamutpāda; yuanqi 緣起) and rebirth, thereby moving toward the construction of a metaphysical system grounded in non-self. This system denies a substantial self that passes through the process of dependent origination, while still acknowledging new birth that arises through dependently arisen conditions. In this way, an interpretation of non-self became established in which rebirth continues through the continuity of karma, even though no self exists.⁹ By contrast, the Treatise of the Sāmmitīya School (Sāmmitīyanikāyaśāstra; sanmidibulun 三彌底部論), which received relatively little attention in later Buddhist history, met the opposite fate. This text contains ideas that are closely related to the original mainstream type of non-self in early Buddhism. It states that the self is acknowledged only as a hypothetical designation (prajñapti; jiashe 假設) and stands apart from the polar logic of “existence” and “nonexistence.” Yet this perspective continuously became the target of criticism from texts that advocated ontological non-self, because it seemed to recall an Upaniṣadic kind of substance. Moreover, although the Sāmmitīya (zhengliangbu 正量部), which possessed this text, once formed one of the most vigorous schools in India, it failed to expand beyond India. In the end, its position disappeared into the background of history with the decline of Indian Buddhism and came to be pushed aside as heterodox.

In relation to the transformation of the doctrine of non-self, it is also necessary to pay attention to entirely new discourses that emerged in Mahāyāna Buddhism. These are the doctrines of tathāgatagarbha (rulaizang 如來藏) and consciousness-only (vijñāptimātra; weishi 唯識). The former chiefly teaches that all sentient beings originally possess the potential to become buddhas, that is, buddha-nature (buddha-dhātu; foxing 佛性). In particular, tathāgatagarbha was regarded as an absolute reality and therefore drew criticism for conflicting with the doctrine of non-self. It was also sometimes considered similar to the self of the Upaniṣadic, which asserts a transcendent self. However, the core of tathāgatagarbha thought is often discussed as the claim that all beings are empty (śūnya; kong 空), and that emptiness (śūnyatā; kongxing 空性) is precisely the nature of nirvana.¹⁰ In this respect, all beings, regardless of whether they have attained awakening, in principle possess the same nature as nirvana. Even so, tathāgatagarbha thought does acknowledge an original ground of awakening, and for this reason it differs in orientation from positions that refuse to allow any kind of essential self. The doctrine of storehouse consciousness (ālaya-vijñāna; alaiyeshi 阿賴耶識) in the consciousness-only school shows a similar pattern. Storehouse consciousness refers to a consciousness (vijñāna; shi 識) whose identity continues even amid rebirth. It is described as originally pure, beginningless and endless, and constant. The relationship between storehouse consciousness and the phenomenal dharmas (zhufa 諸法) that it projects is discussed as indeterminate and impossible to define linguistically (Priestley 1999, 220). These concepts of tathāgatagarbha and ālaya-vijñāna may be seen as a kind of reaction that emerged against the historical atmosphere in which the logic of the ontological denial of the self was gaining dominance. Together with the person (pudgala; gewo 個我) of the Treatise of the Sāmmitīya School, these two concepts appear to have a genealogical connection with the descriptions of non-self in Type 2-2 and Type 2-3, which refer to the state after linguistic discrimination has disappeared. This issue will require more detailed discussion in future research.

References