What This Concept Is Actually Saying
Most contemporary readers, encountering the word "self-harm," immediately picture something narrow: a person cutting themselves, an eating disorder, a suicide attempt. The Buddhist technical term 自害 (ātma-vihiṃsā) is, on the surface, exactly that — *by one's own power, harming one's own lifespan, body-mind, or wisdom-life*. But the classical elaborations in the two great Indian scholastic traditions stretch the word across three registers that almost never get bundled together in modern talk about self-harm: the body-life (壽命, lifespan), the physical form (色身, the body as object), and the wisdom-life (慧命, the lifespan of understanding). To grasp the concept you have to walk all three floors of this building, because most of us only ever look at the ground floor.
The Two Schools, Two Angles
The Abhidharma tradition (毘曇部), especially the *Abhidharmakośa* (《阿毘達磨俱舍論》), treats 自害 as a question about lifespan and the body it depends on. Its question is fundamentally metaphysical: under what conditions can a being's own lifespan be cut short by that being itself? This becomes a way of *classifying sentient beings* by their relation to death — who can die prematurely by their own hand, who can only be killed by others, who can do either, and who can do neither.
The Yogācāra tradition (瑜伽部), in the *Yogācārabhūmi* (《瑜伽師地論》), shifts the question from metaphysics to ethics and contemplative practice. Here 自害 is unpacked across three channels: as the first of seven faults arising from killing, as one of the ten ways precepts are damaged through self-mortification, and as the most subtle form of self-harm — clinging to a self-identity that quietly erodes one's capacity to understand.
These are not competing definitions. They are two lenses on the same object, and a contemporary reader needs both.
The Three Dimensions of Self-Harm
When the *Yogācārabhūmi* speaks of 自害, it is naming harm across three distinct layers:
1. Body-life (身命) — lifespan itself. This is the most obvious: a being that, by its own volition, ends its own life or shortens it. The *Abhidharmakośa*'s way into this is to ask, *whose lifespan can be so ended?* Not everyone's — and this is one of the more startling moves in the text.
2. Physical form (色身) — the body as object. This is the layer where self-harm shows up as *self-mortification*: the body not killed, but tortured, starved, burned, exhausted. The text lists specific practices — sitting among thorns (依棘刺), covering oneself in ash (灰坌), pounding oneself with wooden pestles (木杵), crouching like a fox (狐蹲住坐), tending sacred fires (事火), climbing into water (昇水), standing on one foot and rotating with the sun (一足住隨日而轉). These are not metaphors for the Yogācāra author; they are catalogued real practices of certain Indian ascetic traditions, and the text is naming them as a kind of self-harm, sitting opposite to the opposite extreme of indulgence (耽著受用之極樂行邊).
3. Wisdom-life (慧命) — the lifespan of understanding. This is the most subtle layer and the most important one for serious practitioners. The *Yogācārabhūmi* uses a striking analogy: a person too weak to harm an enemy fears being harmed first, so they approach the enemy with flattering deference — and *the very act of deference becomes the means by which the enemy destroys them*. So it is with 薩迦耶見 (satkāya-dṛṣṭi), the view of self: we cling to the identity-story we tell about ourselves, attach to its comforts, and in doing so we ourselves erode the root of our capacity for genuine understanding — what the text calls the 聖慧命根, the root-faculty of sacred wisdom. *We do not need an external enemy; the self we love is the enemy that undoes us.*
This third layer is what makes 自害 in Buddhism genuinely different from the contemporary psychological term. In modern usage, self-harm is what a person does when something has already gone wrong in them. In the Yogācāra analysis, the most dangerous self-harm is what an apparently healthy, functional person does constantly, in the ordinary course of loving themselves.
The Fourfold Classification from the Abhidharma
The *Abhidharmakośa* builds a four-part classification of sentient beings around the question of who can be self-harmed or other-harmed. This is unusual in Buddhist literature — most schemas are threefold or fivefold. The four categories are:
1. Only self-harmable, not other-harmable (唯可自害非可他害). This includes two strange categories of devas — the 戲忘念天 (heaven beings lost in forgetful play) and the 意憤恚天 (heaven beings of mental rage) — who die from their own excessive states: the first from an overwhelming surge of joy, the second from an overwhelming surge of anger. The text also includes the Buddha's own parinirvāṇa (諸佛自般涅槃) under this category — by entering final nirvāṇa, the Buddha's lifespan ended through his own act, not by another's hand. This is a famously delicate point in the Abhidharma; the Buddha's death is classified here as a *technical* form of self-harm, but the text is making a metaphysical point about agency, not suggesting anything pathological.
2. Only other-harmable, not self-harmable (唯可他害非自害). Beings still in the egg or womb (處胎卵) — their lifespan can be cut only by external action against the body that houses them; they cannot choose to end it.
3. Both self- and other-harmable (俱可害). Most sentient beings in the desire realm — ordinary humans, animals, most devas.
4. Neither self- nor other-harmable (俱非害). A surprisingly long list: beings in intermediate existence (中有), beings in the form and formless realms, hell beings, the inhabitants of Northern Kuru (北俱盧洲), those presently established in the path of seeing (正住見道), those absorbed in loving-kindness samādhi, cessation samādhi, or nothingness samādhi, certain sages and Buddha's disciples named in scripture (達弭羅, 嗢怛羅, 殑耆羅, 長者子耶舍, 鳩磨羅時婆, and so on), the last-life bodhisattva, the mother carrying a bodhisattva's embryo, all wheel-turning kings and their mothers during the gestation period.
The conceptual point: lifespan is not a uniform thing. Some lives are structurally protected by the state of the being inhabiting them; others are not. The two modes of lifespan described in the text — 隨相續轉 (continuing transformation, in which lifespan tracks the body's vulnerability) and 一起便住 (arising-and-then-abiding, in which lifespan is fixed once begun) — give the metaphysical machinery behind this classification. The first mode is "with obstacles" (有障), the second "without obstacles" (無障). From this the Abhidharma derives its proof that untimely death (非時死) is genuinely possible: a lifespan whose body is vulnerable *can* be cut short, and cutting it short is what we mean by self-harm.
Life Walk-Through: Mapping the Classical Schema onto a Contemporary Day
Let me walk the concept through a plausible contemporary life and see where each layer actually shows up.
Saturday morning. You're at a friend's wedding. The speeches land, the room is full of joy, and in a surge of feeling you laugh so hard or feel so overwhelmed that something in your chest gives way. This is the structure of the 戲忘念天 the *Abhidharmakośa* names: an excessive surge of joy that the body cannot absorb. The classical text is not using metaphor; it is identifying a class of death that proceeds from the inside out. Modern cardiology has analogues — takotsubo cardiomyopathy ("broken heart syndrome") and certain stress-induced arrhythmias — though we should be cautious about mapping a Buddhist metaphysical category onto a clinical one. The structure, however, is the same: *the body is destroyed by its own state, not by external injury.*
The same week, on the highway. Someone cuts you off. The rage spikes. Your face burns, your hands clench, and for a stretch of minutes your blood pressure reads as if you were fighting for your life. Most days this is just unpleasant. Some days, in people with the wrong constellation of factors, the spike is the trigger for a stroke or cardiac event. This is the structure of the 意憤恚天 — being destroyed by one's own anger. The text does not call these people weak or foolish; it calls them members of a class of beings whose lifespan is structured to be vulnerable to this kind of surge.
Monday morning, at the gym. You read about a "wellness influencer" who trains twice a day, fasts for sixteen hours, and posts their shredded physique online. Or you are yourself doing a punishing program, partly to compete, partly to punish yourself for a weekend of eating. The classical category here is the self-mortification side of the ten precept violations — thorns and ash and pestles rebranded as discipline and optimization. The *Yogācārabhūmi* does not say this is evil; it says it is one edge (邊) of a range of error, with indulgence (極樂行邊) as the opposite edge. Both extremes damage the body-as-object; the work of practice is to find the middle that does neither.
Wednesday, at work. Your team has a retrospective. You catch yourself composing a careful, strategic story about who you are in this team — the senior person, the competent one, the person who can be trusted with this kind of project. You notice the story has cost you something: you have softened a piece of honest feedback because it would disturb the story. Or harder: you notice that you have been quietly avoiding a piece of work because doing it would mean admitting you do not yet know how to do it. Here we have entered the third layer — wisdom-life. The story you tell about yourself, the identity you have grown attached to, has just been the thing that prevented you from learning something. You did not kill your body; you shortened, slightly, the lifespan of your understanding. The 薩迦耶見 (view of self) as 似怨家 (apparent enemy) is not a doctrine to believe or disbelieve; it is a structure you can watch operating in real time, in the small betrayals of curiosity by the need to be a certain kind of person.
Friday evening. A friend asks how you are. The honest answer is "tired in a way I cannot localize." You suspect, if you sit with it, that the tiredness is not primarily physical. It is the cumulative effect of a year of building a self-story that has gradually stopped being able to absorb new experience. You are not, in the clinical sense, depressed. You are, in the Yogācāra sense, slowly self-harming at the level of wisdom-life. The view you have of who you are is using up the energy that would otherwise go into seeing what is happening.
Now go back to the fourfold classification and ask: which category are most people you know in? Most ordinary adults in the desire realm of work, family, and ambition are in class three (俱可害) — vulnerable to self-harm and other-harm. But the Abhidharma point is that this is a *structural* fact about the kind of being we are, not a personal failing. And it is precisely because we are in class three that all three layers of self-harm are available to us, in different combinations, on different days.
Why Contemporary People Need This Concept
Three reasons, in increasing order of importance.
First, the contemporary vocabulary of "self-harm" is narrower than the reality. We have a clinical term that captures one slice of the phenomenon — usually the psychological-behavioral one — and we have very little language for the larger structure. When we hear *self-harm*, we think of a person in acute distress. But most self-harm, in the Buddhist sense, is *not* acute. It is the slow kind. It is the kind that looks like career-building, identity-maintenance, or righteous anger. The Buddhist concept enlarges the vocabulary so we can see what is happening.
Second, modern life amplifies each of the three layers separately, and amplifies them *together* in ways that compound.
The body-life layer is amplified by a culture that confuses intensity with meaning. People take substances to deepen experience, push their bodies to extremes to feel real, rage-scroll through feeds designed to manufacture surges of indignation. The infrastructure of contemporary attention is partly an infrastructure for manufacturing the kind of surge that can kill a 戲忘念天-style being.
The physical-form layer is amplified by wellness culture and hustle culture simultaneously — the same social-media timeline that sells ice baths and intermittent fasting sells optimization narratives that turn the body into a project to be improved, punished, and displayed. The classical categories of thorns-and-ash and indulgent-feasting have both been commodified and made aspirational.
The wisdom-life layer is amplified by identity performance. Contemporary life offers an unprecedented array of selves one can be — political, professional, aesthetic, spiritual, parental — and offers an unprecedented array of audiences for each. The cost is that the self one is performing becomes load-bearing in a way it did not have to be in earlier centuries. A person whose identity is online is structurally closer to the Yogācāra description: someone whose view of self is the very thing that quietly erodes the capacity to keep learning.
Third, and most importantly, the third layer — wisdom-life self-harm — is the one that most contemporary people will not recognize as self-harm at all. The Yogācāra analysis is doing something subtle here. It is saying that the most dangerous harm a person does to themselves is not visible as harm. The view of self does not feel like an enemy; it feels like a home. The 薩迦耶見 analogy is precise: the person who flatters the dangerous enemy does not experience themselves as flattering an enemy. They experience themselves as being reasonable, prudent, kind to themselves. This is exactly why the analogy works — and exactly why contemporary people, surrounded by an economy that monetizes the flattering of every kind of identity, are particularly vulnerable to this layer. The contemporary brain-science and AI-culture vocabulary here is suggestive but not conclusive: research on confirmation bias and on identity-protective cognition describes mechanisms that look structurally similar to what the Yogācāra text is naming, but the Buddhist claim is stronger than the empirical finding. It claims that *all* self-identity, however comfortable, is operating this way until it is seen through. That is a metaphysical claim, not a scientific one.
Common Misreadings and Clarifications
Misreading 1: "Self-harm" means physical self-injury. This is the most common reduction. The classical concept includes physical self-injury but is far wider. It includes being killed by one's own rage, by one's own unexamined identity-story, and by a regime of self-torture dressed up as discipline. If you only see the first layer you will miss the second and third entirely, and you will miss most of the cases.
Misreading 2: Self-care is the opposite of self-harm. It is not, in the classical framework, the *direct* opposite. The opposite of self-mortification is *indulgence* (耽著受用之極樂行邊), and the *Yogācārabhūmi* explicitly says both extremes are errors. Self-care, in a contemporary sense, can be a sophisticated form of indulgence; it can also be a sophisticated form of self-mortification (the punishing wellness routine dressed up as kindness). What the text is actually pointing toward is something more like maintenance without identification — taking care of the body without the body becoming either a project or a permission slip.
Misreading 3: Accepting oneself is the same as the 薩迦耶見 the text criticizes. This is the most subtle confusion and probably the most important to clarify. The Yogācāra text is *not* saying that self-acceptance is harmful. It is saying that clinging to a fixed view of who one is is harmful — because that view will, by its very rigidity, prevent the openness that genuine understanding requires. Self-acceptance that is fluid, that can absorb new information about oneself, that does not require the self to be a particular kind of object, is not what the text is targeting. The target is the *reification* of self — turning the ongoing process of being a person into a noun, and then defending the noun. This distinction is hard to hold in language but easy to feel in practice. The person who can say "I am wrong about myself" with some grace is not in the danger zone. The person who cannot, who must always be a certain kind of person in order to know who they are, is.
Misreading 4: The Buddha's parinirvāṇa being called "self-harm" means the Buddha harmed himself. This is a famous stumbling block. The *Abhidharmakośa* classifies the Buddha's final nirvāṇa as 唯可自害非可他害 — a being whose lifespan can only end through his own act, not through another's. This is a *technical* classification within the metaphysics of lifespan. It says something about the structure of the Buddha's life (it could only end through his own will, not through being killed), not a moral judgment about the act. Reading it as "the Buddha committed suicide" or "the Buddha harmed himself" misses the entire point of the category, which is to show that the Buddha's lifespan is not the kind of thing that can be externally destroyed.
Misreading 5: The classification is about who is more or less blameworthy. It is not. The fourfold classification from the Abhidharma is a metaphysical taxonomy of beings according to the structure of their lifespan. Beings in class one (only self-harmable) are not morally inferior; the 戲忘念天 are not worse people than you. Beings in class four (neither self- nor other-harmable) are not morally superior; the text groups together hell beings, form-realm devas, and the Buddha's most advanced disciples as all sharing this structural feature. The category is about the *kind of being*, not about how good or bad that being is.
Misreading 6: If wisdom-life self-harm is the most subtle, the other two layers don't really matter. They matter very much. The *Yogācāra* text itself treats all three as genuine and important. A person who has grasped the third layer intellectually but continues to be destroyed by rage, or who uses self-mortification as a spiritual practice, has not understood the concept at all. The three layers reinforce each other: a person with a fixed identity-story is more likely to need the surges of rage and indulgence that damage the body; a person whose body is depleted by self-mortification is more likely to grasp at identity for stability. The work of practice is to attend to all three, not to choose between them.
A Final Note on What the Concept Is Doing
What 自害 is ultimately doing, in its classical form, is *expanding the moral and contemplative vocabulary around self-harm until it becomes coextensive with most of what a person does to themselves across a life*. It is not pathologizing ordinary life. It is pointing out that the ordinary ways we relate to our own lifespan, our own body, and our own self-understanding contain — quietly, structurally, and most of the time without our noticing — operations that shorten us. The point is not to feel bad about this. The point is to see it, because once it is seen, the seeing itself begins to do something to the operation. The Yogācāra text does not end with a condemnation; it ends with a metaphor about an enemy that can only destroy you as long as you keep flattering it. The implication is that the deepest form of self-harm is also the most reversibly placed: it depends entirely on the relation you have to it, and the relation can change.