In contrast to the dominant role that mettā (lovingkindness) and the other Brahmavihāras (compassion, sympathetic joy, and equanimity) play in contemporary Buddhist practice, they seem to have played a relatively minor role in the earliest tradition. One looks in vain for much elaboration on mettā’s dhammic role; largely it seems to have been seen as a skillful means to counteract hatred or ill will. (E.g., MN 114.7, DN 184.108.40.206, SN 46.51.iii). Secondarily it seems to have been seen as a means of physical and mental benefit (AN 11.16, Iti. 1.27) and protection (AN 4.67, SN 20.5).
In practice it seems to have been used originally as a kind of ad hoc skillful method to help dispel hindrances and focus the mind, and to protect oneself from snakes, bad dreams, and other troublesome or dangerous pests. It does not however seem to have been a central part of practice. For example, none of the Brahmavihāras are explicitly listed as parts of the Eightfold Path. And while practices of lovingkindness, compassion, sympathetic joy, and equanimity can be seen as integral to Right Effort, the effort to cultivate the skillful and de-nourish the unskillful, nevertheless they do not explicitly constitute that aspect of the path as, say, jhāna practice explicitly constitutes Right Concentration.
Brahmavihāras in the Early Tradition
The particular formulas for Brahmavihāra practice changed as well quite significantly over the centuries since the earliest material was compiled. The Nikāyas actually contain a very compressed set of practices, most particularly that of boundless radiation:
He abides pervading one quarter with a mind imbued with loving-kindness (compassion, altruistic joy, equanimity), likewise the second, likewise the third, likewise the fourth; so above, below, around, and everywhere, and to all [as to himself], he abides pervading the all-encompassing world with a mind imbued with loving-kindness (compassion, altruistic joy, equanimity), abundant, exalted, immeasurable, without hostility and without ill will. (MN 7.13-16).
I have put the phrase “as to himself” in square brackets; this is a phrase we will return to below.
The famous Mettā Sutta of the Sutta Nipāta (Sn. 143-152) outlines a framework of achievements and efforts that surround any practice of mettā, suggesting that it cultivates the path factors of Right View, Speech, Action, and Livelihood. As to one skilled in that practice:
He should be capable, straight, and very upright, easy to speak to, gentle and not proud, contented and easy to support, having few duties and of a frugal way of life … And he should not do any mean thing, on account of which other wise men would criticize him. (Sn. 143-145; Norman, p. 19).
As to formal meditation however, the Mettā Sutta contains only one additional set of practices to that of unbounded radiation in all directions. It is given in the barest sketch:
Let all creatures indeed be happy [and] secure; let them be happy-minded. Whatever living creatures there are, moving or still without exception, whichever are long or large, or middle-sized or short, small or great, whichever are seen or unseen, whichever live far or near, whether they already exist or are going to be, let all creatures be happy-minded. (Sn. 145-147).
Here instead of carving up the world directionally, and radiating good will in various and all directions, one is instead to carve up the world in terms of various gross aspects of the creatures within it, by size, by visibility, by distance, by time.
Buddhist scholar Andy Olendzki has noted* that in none of these earliest practices does one find any reference towards a practice of directing mettā towards oneself. It is true that in Ñāṇamoli’s (2009) translation above we find the phrase “to all as to himself”, thus apparently including oneself in the practice of boundless mettā radiation, however the Pāli word at issue is something of a problem. This is the word “sabbattatāya”, presumably a compound of “sabba + atta”; but this should mean something like “all selves” or “all beings”, e.g., “and to all” in the above quotation. There is no additional phrase “as to himself” in the passage, and it may stem from the 5th c. CE commentator Buddhaghosa. (Vsm. IX.47).
Olendzki, Anālayo (2015a: 21n46; 2015b), and Rune Johansson (1981: 90) put forward the possibility that the word may actually be “sabbatthatāya” or “in every way”. Anālayo notes that several editions of the texts do indeed preserve this form. (2015b: 17). Further, Anālayo notes that “none of the relevant Chinese or Tibetan texts have a reference to oneself. Instead, wherever equivalents can be found, these all correspond to the basic notion “everywhere”.” (2015b: 19). Maurice Walshe translates like passages (DN 13.76-78, 220.127.116.11) as “everywhere, always”, not “to all as to himself”.
To direct mettā everywhere, or to all beings, is of course by implication to direct it to oneself. As we have seen, the Buddha does say to direct it towards other kinds of beings in the Mettā Sutta, and in a sutta protecting one from snakes. (AN 4.67). Hence one might argue one was directing the attitude towards oneself as well. The knock on this kind of analysis is that the Buddha did not believe that there was a self accessible to meditative insight, so telling one in formal meditation to direct an attitude of mettā towards such a self is odd; this I believe is why Olendzki and Anālayo do not believe the Buddha intended us to practice in that way.
But then mettā is not intended in the early tradition as an insight practice.
If indeed Richard Gombrich (1996, 2009) is right that at least at times in his early teaching the Buddha intended the Brahmavihāras as direct routes to nibbāna, then he left the practices for such a direct path severely underdeveloped, as he did not for example with the “direct path” outlined for mindfulness practice. (Anālayo 2003). A Brahmavihāra-based path also appears philosophically underdeveloped. There are several places in the Canon where such practice is claimed not to be sufficient for liberation (e.g., MN 7.14-17, MN 52.8, AN 3.63). For example in the Aṭṭhakanāgara Sutta, Ānanda states that the Brahmavihāras are “conditioned and volitionally produced”, and so any awakening stemming from them must stem from insight into their being impermanent, dukkha, and non-self.
That is to say, Brahmavihāra practice is not in itself a form of insight practice, and it is only insight practice which leads to awakening in the early texts, because it is only insight practice that leads to complete non-clinging, through awareness of the impermanent, unsatisfactory, and selfless nature of reality. This is no doubt the reason why Anālayo in his recent book calls the Brahmavihāras routes towards “temporary” rather than complete liberation. (2015: 21).
Only equanimity (upekkhā) might, if understood as complete non-clinging, be understood as true awakening: it is after all the one Brahmavihāra which is also explicitly an enlightenment factor (bojjhaṅga). So if indeed the Buddha intended the Brahmavihāras to be a direct path towards liberation, it would probably have been through practice of equanimity in particular.
That said, of course the Brahmavihāras are entirely compatible with whatever we might consider the awakened state. Perhaps indeed they are necessary for it, or necessary accompaniments of it. It is only that they were most likely not intended as direct paths to liberation in the early texts, as for example mindfulness practice was.
So if, pace Gombrich, we are to see at least the first three Brahmavihāras as non-insight related practices, as skillful means for ridding the mind of ill will, and for protecting oneself from various forms of nastiness, then perhaps it matters less how we hold the self when in formal mettā meditation, for example. After all, the Buddha did hold to a kind of self as locus of transformative effort. Many of us will be familiar with passages from the Canon such as, “Oneself, indeed, is one’s own protector. What other protector could there be?” (Dhp. 160). Or in one of the last statements before the Buddha died, “[Y]ou should live as islands unto yourselves, being your own refuge, with no one else as your refuge …” (DN 16.2.26). And so on.
Indeed, as Anālayo (2015a) points out following passages in the Canon (e.g., AN 4.95), for the Buddha “benefiting oneself is superior to benefiting others.” (Though of course it is most superior to benefit both). This is because one can only teach properly from a position of understanding:
In this way, one’s verbal teaching will be grounded in one’s own practice and will be complemented by teaching undertaken by way of example. (2015a: 19).
One may have good intentions but if one acts upon them without understanding one will not in fact work for the true benefit of others. Or in other words, put on your own oxygen mask before helping your neighbor to do so. Insofar as we make transformative effort, our locus of action begins in the self as “protector” and “refuge”, however that is understood.
Brahmavihāras in the Later Tradition
One does not find practices of mettā, or to my knowledge any of the other Brahmavihāras, directed at oneself in the early texts. This was important enough for Buddhaghosa to comment upon it in his own formulation of the practice of mettā in the Visuddhimagga, from where we receive the basis of much contemporary practice:
First of all [lovingkindness] should be developed only towards oneself, doing it repeatedly thus: ‘May I be happy and free from suffering’ or ‘May I keep myself free from enmity, affliction and anxiety and live happily’.
If that is so, does it not conflict with what is said in the texts? For there is no mention of any development of [lovingkindness] towards oneself in what is said in the [Canon]. (Vsm. IX.8-9).
Buddhaghosa’s desire to answer this question is, I think, enough of an indication that the role of the self in mettā meditation was a live issue in his day, even though it seems he did understand the word as “sabbattatāya”, and as possibly indicating self-reference. He says though that if it did include reference to the self along with other beings, that was “without making the distinction ‘This is another being’” (IX.47), a claim which is somewhat difficult to follow.
Nevertheless his answer is that developing mettā towards oneself, particularly as the first step in a practice of the Brahmavihāras, is apt because one naturally holds oneself dear:
… if he develops it in this way: ‘I am happy. Just as I want to be happy and dread pain, as I want to live and not to die, so do other beings, too’, making himself the example, then desire for other beings’ welfare and happiness arises in him. (Vsm. IX.10).
And then he refers to the famous verse from the Mallikā Sutta:
Having traversed all quarters with the mind,
One finds none anywhere dearer than oneself.
Likewise, each person holds himself most dear;
Hence one who loves himself should not harm others. (SN 3.8).
Buddhaghosa’s approach to practicing the Brahmavihāras is personal, involving an attitude towards oneself, and recollections of particular token individuals from one’s past. This is quite a radically different approach than that found in the early texts, which are exclusively generalist in character, referring to beings as types. In the early texts mettā is radiated outwards like “a mighty trumpeter were with little difficulty to make a proclamation to the four quarters” (DN 13.77), or it is applied to beings based on size, distance, time, species, and so on.
Buddhaghosa on the other hand asks us to begin with ourselves, towards whom we presumably find the production of lovingkindness the easiest. Then we move to a revered teacher, then towards a dear friend, then towards a neutral person, then towards a hostile person. (Vsm. IX.11-12).
Interestingly, and I think this is little noticed, he does not advocate our trying to view a hostile person with lovingkindness. Instead he advocates we look “towards a hostile person as neutral.” Presumably this is to rid ourselves of ill will towards the hostile person. Then if we can rid ourselves of ill will towards them, perhaps we can return to the practice at a later time and find them in the neutral category, at which point we can view them with lovingkindness.
Indeed the majority of Buddhaghosa’s elaboration of mettā practice deals with how we should approach our feelings towards hostile people, and how we should work to counter such negative feelings in the case that we cannot easily come to view them as neutral. (IX.14-39).
Buddhaghosa does not use the same formula of practice for the other three Brahmahvihāras however. For compassion he recommends starting with an unfortunate person, or if none is known, an evil person (who we are to assume will suffer for their deeds). After one has aroused compassion through them, one turns to a dear person, a neutral person, and then a hostile person. (IX.78-80). For sympathetic joy one starts by recalling the happiness of a good friend, then a neutral person, then a hostile person. (IX.85-86). For equanimity one begins with the neutral person, since the neutral person is the one to whom one is most naturally equanimous. Then one uses that sense of equanimity to infuse one’s attitude towards a person dear to one, a person who is a good friend, a hostile person, and even oneself. (IX.88-89).
Buddhaghosa’s advice for practicing the Brahmavihāras is quite a bit more complex and nuanced than one finds in the Canonical texts. That said, the practices he suggests have the apparent disadvantage of
tak[ing] living beings as their object. The problem implied here is that [lovingkindness], for example, supposedly comes close to holding the view of a self, since it takes living beings as its object. This helps to appreciate an advantage of the description of boundless radiation in the early discourses, which does not explicitly mention any individual person as the object. (Anālayo 2015b: 21).
So we can see that the practice of mettā and the other Brahmavihāras changed pretty radically over the first millennia or so of Buddhist history. They began as visualizations of generalized emotional radiation towards particular directions or types of beings. This radiation did not explicitly include oneself, although it might be argued that oneself was included implicitly.
Buddhaghosa’s practice on the other hand essentially involved recollections of token individuals, to whom one directed emotional radiation. This included oneself explicitly, in particular as that whom one holds most dear.
One knock on the early tradition is that its Brahmavihāra practices were underdeveloped. In the Vatthūpama Sutta (MN 7) for example, there is no distinction made between the practices of each Brahmavihāra; all use precisely the same simple formula of directional radiation. Only the emotion is changed. This is at the very least a highly compressed template for practice, and may indeed suggest a certain measure of superficiality.
It also should be said that while the early tradition did not explicitly include self-directed practice, the early teachings do contain references to the self as a locus of effort, intention, concern, refuge, and practice. In that context it is therefore perhaps not so unusual to include an explicit reference to self-directed mettā practice as in Buddhaghosa.
The knock on the later, Buddhaghosan tradition is that it appears to introduce oneself and other token selves as objects of meditative focus. Insofar as such practices as these are intended to lead to insight, such introductions may not be particularly helpful. They may lead to the subtle psychological reification of beings as permanent, or as worthy of attachment.
However, insofar as we do not see the Brahmavihāras as themselves insight practices, but instead as forms of skillful means, perhaps we can relax a bit around the Buddhaghosan approach: what it lacks in insight it may make up for in detail and effectiveness. Each practitioner will have to be the judge.
It may also be that we will decide to use one form of practice at one time and another at another. Perhaps for example we come to believe that Buddhaghosa’s approach is good for the novice, but that the radiative approach is better for the advanced practitioner. Or the other way around. It may even be that one or another practice will prove more effective under controlled scientific testing.
Nowadays the practice of the Brahmavihāras in general, and mettā in particular, seem to dominate the contemporary Buddhist scene, only slightly behind the mindfulness movement in public recognition. This is thanks to teachers such as Sharon Salzberg and others who have made it their particular aim. One might even say that Right Effort has been transformed in contemporary parlance into something more like Right Lovingkindness, and the Brahmavihāras have become an explicit part of the Eightfold Path.
I don’t think this is a bad thing, but it does appear to be a change from what went before. Another change is that since these practices are taken as useful for overcoming self hatred, we will need an approach to mettā which diverges from Buddhaghosa’s own. After all, he advocated beginning with the self since the self was the paradigm source of mettā, none being dearer to us than ourselves. Insofar as we find self-hatred arising instead of self-love, mettā practice cannot begin with the self. I have heard that contemporary teachers are aware of this issue and teaching with this in mind. Although I do not know specifics, this does not surprise me.
It may be that in the Buddha’s world, sparse populations and tight local communities meant that Brahmavihāra practice was not as important for them. Perhaps the everyday shared miseries of those days brought kindness and compassion to the surface more often than it is brought nowadays in our hectic, crowded, and often comfortable lives. This is only speculation, but anyhow a shift in emphasis seems clear.
* Most recently in a series of lectures on January 16 at New York Insight.
Bhikkhu Anālayo (2003), Satipaṭṭhāna: The Direct Path to Realization (Cambridge: Windhorse, 2003).
Bhikkhu Anālayo (2015a), Compassion and Emptiness in Early Buddhist Meditation (Cambridge: Windhorse 2015).
Bhikkhu Bodhi, Various Nikāyas (Boston: Wisdom).
Gil Fronsdal (trans.) (2005), The Dhammapada (Boston: Shambhala).
Richard Gombrich (1996), How Buddhism Began (New Delhi: Munshiram Manoharlal).
Richard Gombrich (2009), What the Buddha Thought (London: Equinox).
Rune EA Johansson (1981), Pali Buddhist Texts, 3rd Ed. (London: Curzon Press, Scandinavian Institute of Asian Studies).
Bhikkhu Ñāṇamoli and Bhikkhu Bodhi (2009), The Middle Length Discourses of the Buddha (Boston: Wisdom).
Bhikkhu Ñāṇamoli (1999), The Path of Purification (Onalaska WA: BPS Pariyatti).
K.R. Norman (2001), The Group of Discourses (Sutta-Nipāta), 2nd Ed. (Oxford, Pali Text Society).
Maurice Walshe (trans.) (1995), The Long Discourses of the Buddha (Boston: Wisdom).