Exactly two years to the day I started writing this post I published Map and Territory's most popular post of all time, "Doxa, Episteme, and Gnosis" (also here on LW). In that post I describe a distinction ancient Greek made between three kinds of knowledge we might translate as hearsay, justified belief, and direct experience, respectively, although if I'm being totally honest I'm nowhere close to being a classics scholar so I probably drew a distinction between the three askew to the one ancient Attic Greeks would have made. Historical accuracy aside, the distinction has proven useful over the past couple years to myself and others, so I thought it was worth revisiting in light of all I have learned in the intervening time.

Nuanced Distinctions

To start, I still draw the categories of doxa, episteme, and gnosis roughly the same as I did before. To quote myself:

Doxa is what in English we might call hearsay. It’s the stuff you know because someone told you about it. If you know the Earth is round because you read it in a book, that’s doxa.
Episteme is what we most often mean by “knowledge” in English. It’s the stuff you know because you thought about it and reasoned it out. If you know the Earth is round because you measured shadows at different locations and did the math to prove that the only logical conclusion is that the Earth is round, that’s episteme.
Gnosis has no good equivalent in English, but the closest we come is when people talk about personal experience because gnosis is the stuff you know because you experienced it. If you know the Earth is round because you traveled all the way around it or observed it from space, that’s gnosis.

There's more nuance to it than that, of course. Doxa, for example, also refers to thoughts, beliefs, ideas, propositions, statements, and words in addition to its connotations of hearsay, common belief, and popular opinion. Episteme, to Plato, was the combination of doxa and logos, contrary to my example above where I root episteme in observational evidence, although then again maybe not because "logos" can mean not only "reason", "account", "word", and "speech" but also "ground" or "ultimate cause". And gnosis, despite its connotations in English as a special kind of insightful knowledge about the true nature of existence as a result of its use by Christian mystics, shares the same root or is the root via borrowing of the word for "knowledge" in most European languages, English included.

Further, the boundaries between the three categories are not always clear. We've already seen one way this is so, where I described episteme in a way that it's grounded by gnosis via the direct experience of observation, but this is an empiricist perspective on what episteme is and there's an equally valid notion, in terms of category construction, of episteme as reasoning from first thought within a traditional rationalist perspective. Another is that all knowledge is in a certain sense gnosis because there must have been some experience by which you gained the knowledge (unless you really want to double down on rational idealism and go full Platonist), although this need not confuse us if we understand the difference between the experience of something and the something quoted/bracketed within the experience. And similarly, all knowledge we speak of must first become doxa in our own minds that we tell ourselves before it becomes doxa for others by being put into words that draw common distinctions, hence episteme and gnosis can only be generated and never directly transmitted.

Additional Categories

In addition to doxa, episteme, and gnosis, we can draw additional distinctions that are useful for thinking about knowledge.

One is metis, or practical wisdom. This is the knowledge that comes from hard won experience, possibly over many generations such that no one even knows where it came from. Metis is often implicit or exists via its application and may look nonsensical or unjustified if made explicit. To return to my original examples, this would be like knowing to take a great circle route on a long migration because it's the traditional route despite not knowing anything about the roundness of Earth that would let you know it's the shortest route.

Related to metis is techne, or procedural knowledge or the knowing that comes from doing. In English we might use a phrase like "muscle memory" to capture part of the idea. It's like the knowledge of how to walk or ride a bike or type on a keyboard or throw a clay pot, and also the kind of knowledge that produces things like mathematical intuition, the ability to detect code smell, and a gut sense of what is right. It's knowledge that co-arises with action.

I'm sure we could capture others. Both metis and techne draw out distinctions that would otherwise disappear within doxa and gnosis, respectively. We can probably make further distinctions for, say, episteme that is grounded in gnosis vs. episteme that is grounded in doxa, gnosis about other types of knowledge, and doxa derived by various means. We are perhaps only limited by our need to make these distinctions and sufficient Greek words with which to make them.


Rather than continuing down the path of differentiation, let's look instead at how our three basic ways of knowing come together and relate to one another. In the original post I had this to say about the way doxa, episteme, and gnosis interact:

Often we elide these distinctions. Doxa of episteme is frequently thought of as episteme because if you read enough about how others gained episteme you may feel as though you have episteme yourself. This would be like hearing lots of people tell you how they worked out that the Earth is round and thinking that this gives you episteme rather than doxa. The mistake is understandable: as long as you only hear others talk about their episteme it’s easy to pattern match and think you have it too, but as soon as you try to explain your supposed episteme to someone else you will quickly discover if you only have doxa instead. The effect is so strong that experts in fields often express that they never really knew their subject until they had to teach it.
In the same way episteme is often mistaken for gnosis. At least since the time of Ptolemy people have had episteme of the spherical nature of the Earth, and since the 1970s most people have seen pictures showing that the Earth is round, but astronauts continue to experience gnosis of Earth’s roundness the first time they fly in space. It seems no matter how much epistemic reckoning we do or how accurate and precise our epistemic predictions are, we are still sometimes surprised to experience what we previously only believed.
But none of this is to say that gnosis is better than episteme or that episteme is better than doxa because each has value in different ways. Doxa is the only kind of knowledge that can be reliably and quickly shared, so we use it extensively in lieu of episteme or gnosis because both impose large costs on the knower to figure things out for themselves or cultivate experiences. Episteme is the only kind of knowledge that we can prove correct, so we often seek to replace doxa and gnosis with it when we want to be sure of ourselves. And gnosis is the only kind of knowledge available to non-sentient processes, so unless we wish to spend our days in disembodied deliberation we must at least develop gnosis of doxastic and epistemic knowledge to give the larger parts of our brains information to work with. So all three kinds of knowledge must be used together in our pursuit of understanding.

That sounds pretty nice, like all three kinds of knowledge need to exist in harmony. In fact, I even said as much by concluding the original with an evocative metaphor:

It’s coincidental that ancient Greek chose to break knowledge into three kinds rather than two or four or five, but because it did we can think of doxa, episteme, and gnosis like the three legs of a stool. Each leg is necessary for the stool to stand, and if any one of them is too short or too long the stool will wobble. Pull one out and the stool will fall over. Only when all three are combined in equal measure do we get a study foundation to sit and think on.

Alas, I got some things wrong in the original with how I described the relationship between these three aspects of knowledge, specifically in the way things fall apart when the three aspects are not balanced. I won't reprint those words here to avoid spreading confusion, and will rather try to make amends by better describing what can happen when we privilege one kind of knowledge over the others.

To privilege doxa is to value words, thoughts, and ideas over reason and experience. This position is sometimes compelling: as the saying goes, if you can't explain something, you don't really understand it, and to explain it you must have and generate doxa. Further, doxa lets you engage with the world at a safe distance without getting your hands dirty, but this comes with the risk of becoming detached, unhinged, ungrounded, unroot, disconnected, and otherwise uncorrelated with reality because, on its own, doxa is nothing more than empty words. The people we pejoratively claim to put doxa first are sophists, pundits, ivory-tower intellectuals, certain breeds of bloggers, and, of course, gossips. The remedy for their condition is to spend more time thinking for oneself and experiencing life.

When we privilege episteme we believe our own reason over and above what wisdom and experience tell us. The appeal of favoring episteme lies in noticing that wisdom and experience can mislead us, such that if we just bothered to think for 5 minutes we would have noticed they were wrong. And, of course, sometimes they are, but if we continue down this path we run into infinite inferential regress, the uncomputable universal prior, the problem of the criterion, epistemic circularity, and more mundane problems like making commonly known mistakes, ignoring our experiences because we don't understand them, and otherwise failing because we didn't reckon we would. Putting episteme first is the failure mode of high modernists, logical positivists, traditional rationalists, and internet skeptics. If we fall victim to their mistakes, the solution lies with finding the humility to accept that sometimes other people know things even when we don't and to trust our lived experiences to be just as they are, nothing more and nothing less.

Finally, privileging gnosis is to rely on our experiences at the expense of reason and wisdom. There's a certain logic to the radical empiricism of this approach: what I can know for sure is what I experience with my eyes, ears, nose, tongue, body, and mind, and every other way of knowing is a secondary source. But this leaves out the important contributions of what we can know about the world that lies beyond our direct experience where we learn from others and from reasoning, effectively giving up the epistemic benefits that come with language. Solipsists, hippies, mystics, and occultists are among the folk who tend to value gnosis over episteme and doxa. For them we might advise listening more to others and spending more time at rigorous, precise, and careful thought to balance out their over-strong belief in what they experience.

Walking the middle way between these three attractors is not easy. If nothing else, there's a certain temptation that can arise to identify with the way of knowing you like best and the people who engage most with that way of knowing. I encourage you to resist it! You can hang out with and wear the attire of an intellectual, a rationalist, or a hippie without succumbing to their stereotypical epistemological failure modes of excess doxa, episteme, and gnosis. There is no special virtue in making wrong predictions about the world, regardless of how you came to make that wrong prediction. Instead, you can aspire to remain a sharp blade that cuts to the truth no matter the whetstone used to hone the blade or the stance from which the cut is made.

Beyond Distinctions

If it's the case that there's no special privileging of one kind of knowledge over another and the path to truth lies with combining them all, you might ask why make any distinctions at all? Certainly it feels at times useful to draw these distinctions, but as we've seen these distinctions are blurry, nuanced, and blend into each other. What about the alternative of unifying these kinds into a single concept that captures them all?

By itself the English word "knowledge" fails to do that adequately because it tends to point towards explicit knowledge and disregards that which is known implicitly and that which is inseparable from its embeddedness in the world, and we know this because it's noteworthy to point out ways that things like gnosis and metis and techne can count as knowing. So what is the thing that ties these notions all together?

I think it's worth considering what it means to know something. Knowing is an intentional act: a subject (you, me, them) knows an object (the something known). Thus it is a kind of relationship between subject and object where the subject experiences the object in a particular way we consider worth distinguishing as "knowing" from other forms of experience. In knowing the object seems to always be something mental, viz. the object is information not stuff, ontological not ontic. For example, you might say I can't know the cup on my desk directly, only the experience of it in my mind—the noumenon of the cup is not known, only the phenomenon of it. And from there we can notice that knowing is not a single experience, but composed of multiple motions: initial contact with a mental object, categorization of the object in terms of ontology, evaluation of it, possible recollection of related mental objects (memories), integration into a network of those related mental objects, and self-reflection on the experience of the mental object.

Given the complexity of the knowing act, I'm inclined to infer that even if the neurological processes that enable knowing can be thought of as a unified system, its complex enough that we should expect it to have many aspects that to us would look like different kinds of knowledge. When certain aspects of that process are more salient than the others, we might see a pattern and label that knowing experience as doxa, episteme, or gnosis. So knowledge is neither a single kind or multiple, but a holon both composed of distinct kinds and cut from a single kind, codependent and inseparable from one another. Thus there are different kinds of knowledge and there is just one kind of knowing, and holding both perspectives is necessary to understanding the depths of what it means to know.

More to say?

There's always more to say. For example, I chose to leave out a more detailed discussion on the etiology of knowledge, which confuses the matter of a bit since it can mean putting one kind of knowledge causally first which can be mistaken for thinking one kind is more important than the others. Maybe I'll return to this topic in another two years or more and have additional insights to share.

New Comment
2 comments, sorted by Click to highlight new comments since: Today at 9:37 AM

I think I bounced off this at first because I wasn't sure how much of this was new vs a bit more polished. Could you briefly summarize/table-of-contents the "diff" between this and last post?

Everything that is not a literal quote from the previous post is new.