Image courtesy of Aleksa D at

Where do the instructions for mettā practice come from? The suttas give relatively little instruction on what actually to do, beyond generating universal good will that we find in the Mettā Sutta (Sn. 143-152), and the practice of focusing on each direction, in an oft-cited pericope:

[W]ith his heart filled with lovingkindness, he dwells suffusing one quarter, the second, the third, the fourth. Thus he dwells suffusing the whole world, upwards, downwards, across, everywhere, always with a heart filled with lovingkindness, abundant, unbounded, without hate or ill-will.

Just as if a mighty trumpeter were with little difficulty to make a proclamation to the four quarters, so by this meditation, … by this liberation of the heart through lovingkindness he leaves nothing untouched, nothing unaffected in the sensuous sphere. (Tevijja Sutta, DN 13.76-77).

In one sutta (AN 4.67), the Buddha suggests directing mettā towards various species of snake, and from there to various living creatures. Perhaps this is the inspiration for the more general instructions one finds attributed to Sāriputta from the Paṭisambhidāmagga, which slices the categories somewhat differently:

May all women be freed from enmity, distress and anxiety, and may they guide themselves to bliss. May all men… all Noble Ones… all who are not Noble Ones… all deities… all human beings… may all those in the states of deprivation be freed from enmity, distress and anxiety, and may they guide themselves to bliss. …

May all breathing things… May all creatures… May all persons… May all who are embodied… May all women… May all men… May all Noble Ones… May all who are not Noble Ones… May all deities… May all human beings… May all those in the states of deprivation in [each] direction be freed from enmity, distress and anxiety, and may they guide themselves to bliss… (XIV.5-6).

I don’t know about you, but I find it hard to generate the appropriate feeling tone towards abstracted categories like “all in the eastern direction”, or “all creatures”. At least at the start of my practice, I find mettā is best directed towards particular people. Then when the felt sensation arises, the practice can move towards general categories.

Contemporary mettā practice such as that found in Sharon Salzberg’s wonderful book Lovingkindness instructs us to pick out particular people in our lives and direct good will to them, in a particular order. We start by wishing good will towards ourselves, then towards a benefactor to whom feelings of good will come naturally, then to a good friend, then to a neutral person, and finally to someone we find difficult. Andy Olendzki has described this practice as a form of fixed concentration on an emotional tone, where we hold onto the tone and work by shifting its focus from an easy person to a more difficult person. By practicing mettā bhāvana over a period of time, we hope to get ourselves to a place where we can maintain the same attitude of good will towards the difficult people in our lives that we do to our close friends.

I doubt I’m alone in finding it difficult to get myself to the same place with antagonists that I do with close friends, though.

The mettā practice of successive focus on particular people probably does not come to us from the Buddha, but rather from Buddhaghosa’s fifth century CE text, the Visuddhimagga. Or rather, this practice is most developed in that later text. In the sutta of the Simile of the Saw (MN 21) we do find that the Buddha counsels us to hold those particular people who would do harm to us with a mind of loving kindness, even if they “were to sever you savagely limb by limb with a two-handled saw”!* This is a very advanced practice, for sure.

It does not appear to me that the practice suggested in the Visuddhimagga is quite as demanding as the Buddha’s, at least in certain respects. First, the Visuddhimagga does not counsel us to try to hold the same attitude towards an enemy as towards a friend. Instead, its instructions are more relaxed: the practitioner should work to develop mettā “towards a very dearly loved friend, then towards a neutral person as a very dearly loved friend, then towards a hostile person as neutral.” (IX, 12).

In other words, we are to generate feelings of mettā towards a dear friend, then retain that same attitude towards a neutral person. But when switching to a hostile person, Buddhaghosa doesn’t counsel us to retain the same attitude. Rather, we are to hold the hostile person as we would someone to whom we have no particular feelings one way or the other. We do not need to hold them as we do a dear friend.

Basically, I think Buddhaghosa tells us to raise the level of friendliness in our thoughts so as to erase the hindrance of ill-will. Over time, this process may help to establish equanimity towards friend and foe alike. Then from such a place, once the distinction between “enemy” and “friend” has receded, one can manage mettā towards all alike.

A related issue comes in our attitude towards a “very dearly loved friend”. The first person who comes to mind is my wife. But such a person isn’t at all an apt subject for mettā practice. The Visuddhimagga tells us why, with a little story:

An elder supported by a family was asked, it seems, by a friend’s son, ‘Venerable sir, towards whom should lovingkindness be developed?’ The elder told him, ‘Towards a person one loves’. He loved his own wife. Through developing lovingkindness towards her he was fighting against the wall all the night. That is why it should not be developed specifically towards the opposite sex. (IX, 6).

The next time someone tells us the Visuddhimagga is a dry text, we should remember this poor young man, climbing the walls at night after performing mettā towards his wife!


* Thanks to Erik O’Donnell for pointing this to me over at suttacentral.

An earlier version of this essay appeared at the NYIMC blog.

No Comments

  1. Kevin K. on May 22, 2017 at 7:27 am

    Hi Doug,

    While I don’t have more than a small fraction of your exposure to the great teachers at BCBS, I was fortunate to participate in their “Entering the Path” course two years ago, and one of the most striking segments was Sharon Salzberg and Bhikkhu Analayo co-teaching metta practice. It clearly came as quite a revelation to Sharon to find out that the complex (with many categories of people to direct metta to) and highly discursive phrase-based practice she learned and teaches doesn’t come from the suttas. It’s also not taught in the Visuddhimagga (which I respectfully maintain is every bit as dry as it is reputed to be, your one emotionally resonant quote notwithstanding!) though it is certainly based upon it and its commentaries. Instead it seems quite clearly to have originated with Mahasi Sayadaw’s comprehensive (and dry!) instructions (see link below) – meaning this “ancient Buddhist meditation technique” was invented in late 19th century Burma along with noting practice, using the rising and fall of the abdomen as an object, the stages of insight, “vipassana jhanas” and pretty much everything else in the Mahasi and Goenka traditions.

    I have nothing against the complex, phrase-based practice for those for whom it works, but as Analayo pointed out in the course that practice doesn’t work at all for a lot of people (including himself). Many struggle for years trying to do a practice that doesn’t fit because they’re told it is the only authentic one rather than being offered more options.

    Developing the feeling in the heart in a simple, non-discursive way and then radiating it out as light and warmth in all directions – a sutta-based approach to metta.- has certainly been a revelation (and healing balm) for those relatively few of us who’ve been exposed to it. Interestingly Ajahn Brahm (in his excellent meditation manual “Mindfulness, Bliss and Beyond”) teaches a parallel approach using the lovely device of cultivating the feeling of metta towards an imaginary lost kitten that shows up at one’s door.

    At the moment though I think that students are about as likely to encounter instruction in sutta-based metta practice as they are to learn that samatha leading to jhana in the service of insight predates “vipassana” meditation techniques from Burma that are misrepresented as being what the Buddha practiced by about 2300 years. I’m not saying that we ought to prioritize older vs. newer practices but we do need to stop conflating the two, and that’s the message we’re getting loud and clear from the work Analayo in particular, along with Richard Gombrich, Bhikku Sujato and a handful of others.
    Here’s the link to the Mahasi tome:

    • Doug Smith on May 22, 2017 at 8:14 am

      Hi Kevin, and thanks for your thoughts and the link to Mahāsi Sayadaw’s text on the Brahmavihāras.

      I definitely agree with you and the folks you cite that it’s important not to conflate earlier with later teachings. but I’d want to stress that this is mostly for academic or scholarly purposes: we want to know where the ideas came from and how they developed. We also don’t want to conflate practices that may not work well together as they are intended for different ends. But many earlier and later practices do work well together, and certainly some later practices may be advances over their earlier alternatives. It is even conceivable that had the Buddha known of these later practices he would have preferred them.

      Of course we can’t know such things ourselves. My point is simply that scholarly interest in history of ideas aside, we should hold these distinctions of earlier and later pragmatically. For example, it may not matter to Sharon Salzberg’s updated teachings of the Brahmavihāras exactly how they developed. What matters is how effective they are now, with a given audience. They may not work for Anālayo but I find them helpful, at least in particular contexts. This is not of course to denigrate the earlier teachings of radiating mettā towards all beings. It is simply to expand practice in a way that I do not feel conflicts in any meaningful way from the early teachings.

      As to the Visuddhimagga’s teaching itself, it does direct mettā towards different categories of particular people, as I noted above. There are also phrases used therein, such as “May this good man be happy and free from suffering.” (IX.11) Other phrases used by those in anger, such as “Let him have no good fortune!” (IX.15) immediately suggest alternative phraseology with mettā.

      In fact, the Mettā Sutta of the Sutta Nipāta also suggests phrases such as, “Let all creatures indeed be happy and secure; let them be happy-minded.” (Sn. 145 on Norman’s translation). So I don’t in general think that using phrases is a later development, although perhaps which phrases were to be used became more established as time went on.

  2. Carl H on May 23, 2017 at 5:41 am

    It seems to me that one of the problems with metta practice is the unfortunate translation as “loving kindness”. John Peacock, as well as Christina Feldman and Akincano Weber, points out that the term is much closer to “boundless friendliness”. Christina teaches that metta practice is the practice of “befriending”, the lovely as well as the distressing. This is quite different than generating “loving kindness”, and for me much more accessable. Akincano believes that the brahmaviharas likely predated the Buddha. And also, that while the Theravadin tradition has seen them as objects of meditation, they are inherent human qualities to be cultivated and developed, not meditative states.

  3. Kevin K. on May 23, 2017 at 8:05 am

    Thanks very much Doug and Carl for your thoughtful comments.

    It seems likely that we’re stuck with “loving kindness” as the default translation for metta, but Analayo (“benevolence”) and Bhante Gunaratana (“loving friendliness”) can be added to the chorus of those you cited Carl who point out that “loving” has the feel of sticky romantic attachment while “kindness” could inappropriate evoke a sense of doing someone a favor.

    Sharon Salzberg points out that what we are training in in metta practice is actually attention – to whom and to what we direct it – and that metta is a byproduct of that. Analayo in turn points out that soaking it metta gives us the incredible chance to view the world through the lens of a mind that for that moment is without anger, and that that is an incredible basis for insight to arise.

    To your more scholarship-related comment Doug of course the phrases used in the Mahasi metta practice taught at IMS and Spirit Rock is based on the Visuddhimagga. Analayo addresses this in some detail in his most recent book, “Early Buddhist Meditation Studies” and my point on the practice side is that surely the much simpler, non-discursive way of practicing the Brahmaviharas described in the suttas deserves to be at least as widely known and practiced as the Visuddhimagga-based one.

    As Bhante Dhammika points out in his iconoclastic insider’s history of Theravada “Broken Buddha,” the fact that there were no meditation manuals in Theravada from the time of the Visuddhimagga (5th century AD) until the 20th century speaks volumes about how thoroughly dead meditation practice was in the tradition throughout most of its history. For that reason I find it essential to look at how samatha and vipassana as well as the Brahmaviharas have been practiced in the broader Indo-Tibetan tradition – within which there’s a great deal more consistency of instruction than one finds in modern Theravada/vipassana.

    For those interested in an overview of the origin various strands of often-contradictory meditation instructions this article by David Chapman remains the best thing I know of:

  4. Doug Smith on May 23, 2017 at 10:00 am

    Thanks for your great comments, Carl and Kevin. Yes, I agree that “loving kindness” isn’t really the best translation of “mettā” (I rather prefer Peacock’s “boundless friendliness” myself), but for the purposes of discussion I think it’s good enough. For better or worse it’s become relatively standardized in contemporary Western discussions of the practice.

    Indeed Carl, the Brahmavihāras likely did predate the Buddha. To take one example, at AN 7.66 the Buddha recounts the story of Sunetta, a past teacher who practiced the Brahmavihāras.

    I would say that they are inherent human qualities that can become objects of meditative practice, and have so become in many Buddhist traditions, including the earliest ones. (Not only Theravāda).

    As to your point Kevin that non-discursive practice deserves to be widely known, I absolutely agree. Anālayo’s discussion of it is great. I also think (?) that Bhante Gunaratana uses a similar practice as a route to the jhānas.

    The stuff about the development of Vipassana mediation is fascinating, I wrote a piece on it awhile back. I don’t find contradictions so much as different emphases as time went by. The development of Vipassana was essentially bound up in an increasing laicization of practice, due to the post-colonial impact of Christian proselytization. But it was conservative at its core; AFAIK no part of Vipassana strictly contradicts what one finds in the Nikāyas. (Where for example one does find the practice of “dry insight” or insight without jhāna).

Leave a Comment

You must be logged in to post a comment.

This site uses Akismet to reduce spam. Learn how your comment data is processed.