Phenomenology is the study of the structures of experience and consciousness. Literally, it is the study of "that which appears". The first time you look at a twig sticking up out of the water, you might be curious and ask, "What forces cause things to bend when placed in water?" If you're a curious phenomenologist, though, you'll ask things like, "Why does that twig in water appear as though bent? Do other things appear to bend when placed in water? Do all things placed in water appear to bend to the same degree? Are there things that do not appear to bend when placed in water? Does my perception of the bending depend on the angle or direction from which I observe the twig?"

Pehenomenology means breaking experience down to its more basic components, and being precise in our descriptions of what we actually observe, free of further speculation and assumption. A phenomenologist recognizes the difference between observing "a six-sided cube", and observing the three faces, at most, from which we extrapolate the rest.

I consider phenomenology to be a central skill of rationality. The most obvious example: You're unlikely to generate alternative hypotheses when the confirming observation and the favored hypothesis are one and the same in your experience of experience. The importance of phenomenology to rationality goes deeper than that, though. Phenomenology trains especially fine grained introspection. The more tiny and subtle are the thoughts you're aware of, the more precise can be the control you gain over the workings of your mind, and the faster can be your cognitive reflexes.

(I do not at all mean to say that you should go read Husserl and Heidegger. Despite their apparent potential for unprecedented clarity, the phenomenologists, without exception, seem to revel in obfuscation. It's probably not worth your time to wade through all of that nonsense. I've mostly read about phenomenology myself for this very reason.)

I've been doing some experimental phenomenology of late.


I've noticed that rationality, in practice, depends on noticing. Some people have told me this is basically tautological, and therefore uninteresting. But if I'm right, I think it's likely very important to know, and to train deliberately.

The difference between seeing the twig as bent and seeing the twig as seeming bent may seem inane. It is not news that things that are bent tend to seem bent. Without that level of granularity in your observations, though, you may not notice that it could be possible for things to merely seem bent without being bent. When we're talking about something that may be ubiquitous to all applications of rationality, like noticing, it's worth taking a closer look at the contents of our experiences.

Many people talk about "noticing confusion", because Eliezer's written about it. Really, though, every successful application of a rationality skill begins with noticing. In particular, applied rationality is founded on noticing opportunities and obstacles. (To be clear, I'm making this up right this moment, so as far as I know it's not a generally agreed-upon thing. That goes for nearly everything in this post. I still think it's true.) You can be the most technically skilled batter in the world, and it won't help a bit if you consistently fail to notice when the ball whizzes by you--if you miss the opportunities to swing. And you're not going to run very many bases if you launch the ball straight at an opposing catcher--if you're oblivious to the obstacles.

It doesn't matter how many techniques you've learned if you miss all the opportunities to apply them, and fail to notice the obstacles when they get in your way. Opportunities and obstacles are everywhere. We can only be as strong as our ability to notice the ones that will make a difference.

Inspired by Whales' self-experiment in noticing confusion, I've been practicing noticing things. Not difficult or complicated things, like noticing confusion, or noticing biases. I've just been trying to get a handle on noticing, full stop. And it's been interesting.

Noticing Noticing

What does it mean to notice something, and what does it feel like?

I started by checking to see what I expected it to feel like to notice that it's raining, just going from memory. I tried for a split-second prediction, to find what my brain automatically stored under "noticing rain". When I thought about noticing rain, I got this sort of vague impression of rainyness, which included few sensory details and was more of an overall rainy feeling. My brain tried to tell me that "noticing rain" meant "being directly acquainted with rainyness", in much the same way that it tries to tell me it's experiencing a cube when it's actually only experiencing a pattern of light and shadows I interpret as three faces.

Then, I waited for rain. It didn't take long, because I'm in North Carolina for the month. (This didn't happen last time I was in North Carolina, so perhaps I just happened to choose The One Valley of Eternal Rain.)

The real "noticing rain" turned out to be a response to the physical sensations concurrent with the first raindrop falling on my skin. I did eventually have an "abstract rainyness feeling", but that happened a full two seconds later. My actual experience went like this.

It was cloudy and humid. This was not at the forefront of my attention, but it slowly moved in that direction as the temperature dropped. I was fairly focused on reading a book.

(I'm a little baffled by the apparent gradient between "not at all conscious of x" and "fully aware of x". I don't know how that works, but I experience the difference between being a little aware of the sky being cloudy and being focused on the patterns of light in the clouds, as analogous to the difference between being very-slightly-but-not-uncomfortably warm and burning my hand on the stove.)

My awareness of something like an "abstract rainyness feeling" moved further toward consciousness as the wind picked up. Suddenly--and the suddenness was an important part of the experience--I felt something like a cool, dull pin-prick on my arm. I looked at it, saw the water, and recognized it as a raindrop. Over the course of about half a second, several sensations leapt forward into full awareness: the darkness of my surroundings, the humidity in the air, the dark grey-blueness of the sky, the sound of rain on leaves like television static, the scent of ozone and damp earth, the feeling of cool humid wind on my face, and the word "rain" in my internal monologue.

I think it is that sudden leaping forward of many associated sensations that I would call "noticing rain".

After that, I felt a sort of mental step backward--though it was more like a zooming out or sliding away than a discrete step--from the sensations, and then a feeling of viewing them from the outside. There was a sensation of the potential to access other memories of times when it's rained.

(Sensations of potential are fascinating to me. I noticed a few weeks ago that after memorizing a list of names and faces, I could predict in the first half second of seeing the face whether or not I'd be able to retrieve the name in the next five seconds. Before I actually retrieved the name. What??? I don't know either.)

Only then did all of it resolve into the more distant and abstract "feeling of rainyness" that I'd predicted before. The resolution took four times as long as the simultaneous-leaping-into-consciousness-of-related-sensations that I now prefer to call "noticing", and ten times as long as the first-raindrop-pin-prick, which I think I'll call the "noticing trigger" if it turns out to be a general class of pre-noticing experiences.

("Can you really distinguish between 200 and 500 milliseconds?" Yes, but it's an acquired skill. I spent a block of a few minutes every day for a month, then several blocks a day for about a week, doing this Psychomotor Vigiliance Task when I was gathering data for the polyphasic sleep experiment. (No, I'm sorry, to the best of my knowledge Leverage has not yet published anything on the results of this. Long story short: Everyone who wasn't already polyphasic is still not polyphasic today.) It gives you fast feedback on simple response time. I'm not sure if it's useful for anything else, but it comes in handy when taking notes on experiences that pass very quickly.)

Noticing Environmental Cues

My second experiment was in repeated noticing. This is more closely related to rationality as habit cultivation.

Can I get better at noticing something just by practicing?

I was trying to zoom in on the experience of noticing itself, so I wanted something as simple as possible. Nothing subtle, nothing psychological, and certainly nothing I might be motivated to ignore. I wanted a straightforward element of my physical environment. I'm out in the country and driving around for errands and such about once a day, so I went with "red barn roofs".

I had an intuition that I should give myself some outward sign of having noticed, lest I not notice that I noticed, and decided to snap my fingers every time I noticed a red barn roof.

On the first drive, I noticed one red barn roof. That happened when I was almost at my destination and I thought, "Oh right, I'm supposed to be noticing red barn roofs, oops" then started actively searching for them.

Noticing a red barn roof while searching for it feels very different from noticing rain while reading a book. With the rain, it felt sort of like waking up, or like catching my name in an overheard conversation. There was a complete shift in what my brain was doing. With the barn roof, it was like I had a box with a red-barn-roof-shaped hole, and it felt like completion when a I grabbed a roof and dropped it through the hole. I was prepared for the roof, and it was a smaller change in the contents of consciousness.

I noticed two on the way back, also while actively searching for them, before I started thinking about something else and became oblivious.

I thought that maybe there weren't enough red barn roofs, and decided to try noticing red roofs of all sorts of buildings the next day. This, it turns out, was the correct move.

On day two of red-roof-noticing, I got lots of practice. I noticed around fifteen roofs on the way to the store, and around seven on the way back. By the end, I was not searching for the roofs as intently as I had been the day before, but I was still explicitly thinking about the project. I was still aware of directing my eyes to spend extra time at the right level in my field of vision to pick up roofs. It was like waving the box around and waiting for something to fall in, while thinking about how to build boxes.

I went out briefly again on day two, and on the way back, I noticed a red roof while thinking about something else entirely. Specifically, I was thinking about the possibility of moving to Uruguay, and whether I knew enough Spanish to survive. In the middle of one of those unrelated thoughts, my eyes moved over a barn roof and stayed there briefly while I had the leaping-into-consciousness experience with respect to the sensations of redness, recognizing something as shaped like a building, and feeling the impulse to snap my fingers. It was like I'd been wearing the box as a hat to free up my hands, and I'd forgotten about it. And then, with a heavy ker-thunk, the roof became my new center of attention.

And oh my gosh, it was so exciting! It sounds so absurd in retrospect to have been excited about noticing a roof. But I was! It meant I'd successfully installed a new cognitive habit to run in the background. On purpose. "Woo hoo! Yeah!" (I literally said that.)

On the third day, I noticed TOO MANY red roofs. I followed the same path to the store as before, but I noticed somewhere between twenty and thirty red roofs. I got about the same number going back, so I think I was catching nearly all the opportunities to notice red roofs. (I'd have to do it for a few days to be sure.) There was a pattern to noticing, where I'd notice-in-the-background, while thinking about something else, the first roof, and then I'd be more specifically on the lookout for a minute or two after that, before my mind wandered back to something other than roofs. I got faster over time at returning to my previous thoughts after snapping my fingers, but there were still enough noticed roofs to intrude uncomfortably upon my thoughts. It was getting annoying.

So I decided to switch back to only noticing the red roofs of barns in particular.

Extinction of the more general habit didn't take very long. It was over by the end of my next fifteen minute drive. For the first three times I saw a roof, I rose my hand a little to snap my fingers before reminding myself that I don't care about non-barns anymore. The next couple times I didn't raise my hand, but still forcefully reminded myself of my disinterest in my non-barns. The promotion of red roofs into consciousness got weaker with each roof, until the difference between seeing a non-red non-barn roof and a red non-barn roof was barely perceptible. That was my drive to town today.

On the drive back, I noticed about ten red barn roofs. Three I noticed while thinking about how to install habits, four while thinking about the differences between designing exercises for in-person workshops and designing exercises to put in books, and three soon enough after the previous barn to probably count as "searching for barns".

So yes, for at least some things, it seems I can get better at noticing them my  by practicing.

What These Silly Little Experiments Are Really About

My plan is to try noticing an internal psychological phenomenon next, but still something straightforward that I wouldn't be motivated not to notice. I probably need to try a couple things to find something that works well. I might go with "thinking the word 'tomorrow' in my internal monologue", for example, or possibly "wondering what my boyfriend is thinking about". I'll probably go with something more like the first, because it is clearer, and zooms in on "noticing things inside my head" without the extra noise of "noticing things that are relatively temporally indiscrete", but the second is actually a useful thing to notice.

Most of the useful things to notice are a lot less obvious than "thinking the word 'tomorrow' in my internal monologue". From what I've learned so far, I think that for "wondering what my boyfriend is thinking about", I'll need to pick out a couple of very specific, instantaneous sensations that happen when I'm curious what my boyfriend is thinking about. I expect that to be a repetition of the rain experiment, where I predict what it will feel like, then wait 'til I can gather data in real time. Once I have a specific trigger, I can repeat the red roof experiment to catch the tiny moments when I wonder what he's thinking. I might need to start with a broader category, like "notice when I'm thinking about my boyfriend", get used to noticing those sensations, and then reduce the set of sensations I'm watching out for to things that happen only when I'm curious what my boyfriend is thinking.

After that, I imagine I'll want to practice with different kinds of actions I can take when I notice a trigger. (If you've never heard of Implementation Intentions, I suggest trying them out.) So far, I've used the physical action of snapping my fingers. That was originally for clarity in recognizing the noticing, but it's also a behavioral response to a trigger. I could respond with a psychological behavior instead of a physical one, like "imagining a carrot". A useful response to noticing that I'm curious about what my boyfriend is thinking would be "check to see if he's busy" and then "say, 'What are you thinking about?'"

See, this "noticing" thing sounds boringly simple at first, and not worth much consideration in the art of rationality. Even in his original "noticing confusion" post, Eliezer really talked more about recognizing the implications of confusion than about the noticing itself.

Noticing is more complicated than it seems at first, and it's easy to mix it up with responding. There's a whole sub-art to noticing, and I really think that deliberate practice is making me better at it. Responses can be hard. It's essential to make noticing as effortless as possible. Then you can break the noticing and the responding apart, and you can recognize reality even before you know what to do with it.

New Comment
16 comments, sorted by Click to highlight new comments since:

I care very much about phenomenology. At the moment I'm writing an Android app that's supposed to teach the user to perceive phonemes that he didn't perceive at first. It uses the spaced repetition principle but optimizes the algorithm for multiple choice questions.

I see that I do have weaknesses perceiving those phonemes in the English language that are generally hard to perceive for Germans (which is my native language). I learned a lot of my English on the internet via the medium of text and generally feel really comfortable about the language, so it's interesting to tackle that issue.

In another project spaced repetition project I used Anki to learn to distinguish color that he didn't distinguish beforehand.

I do not at all mean to say that you should go read Husserl and Heidegger. Despite their apparent potential for unprecedented clarity, the phenomenologists, without exception, seem to revel in obfuscation.

If you train yourself in perception you start to be able to make distinctions that you couldn't make beforehand. If you express yourself and speak about those distinctions to a person who's not able to make those distinctions he can't understand you and it might seem like you are engaging in obfuscating.

On that topic I can recommend reading about phenomenological primitives. The term comes from people thinking about teaching physics and those people generally speak in a language that should be easy to understand for someone with LW background.,_1983 is very much worth reading.

Somatics is another field that full of phenomenological investigation. It's about noticing what goes on in your body. Being able to notice that a given thought triggers a fear based freeze reaction in yourself that makes your breathing shallow is very useful. It allows you to return to normal breathing before you think more about the thought.

Somatics also has a lot of prior art that's interesting when you want train your abilities of perception.

See, this "noticing" thing sounds boringly simple at first, and not worth much consideration in the art of rationality.

Want concepts to be exiting instead of boring is a classic failure mode of a lot of rationalists. It's a key to understand that just because something isn't exciting doesn't mean that it's useless. "Boringly simple" also often means: I think I understand topic X and I feel really uncomfortable about investigating the issue further.

Simple knowledge is good knowledge because you can build on it. It's reliable. Complex intellectual arguments are usually not very reliable. On the other hand engaging in them is much more entertaining and for a lot of rationalists engaging in intellectual debates is their favorite form of entertainment. There's also nothing wrong about debating ideas for entertainment but you shouldn't let it keep you from also looking at the issues that feel uncomfortable and that aren't entertaining.

In another project spaced repetition project I used Anki to learn to distinguish color that he didn't distinguish beforehand.

I think I managed to do this when learning flags, with Chad and Romania. It seemed like I got to the point where I could reliably distinguish their flags on my phone, whereas when I started, I did no better than chance. I did consciously explain this to somebody else as something interesting, but now that I think about it, I failed to find it as interesting as I should have, because the idea that seeing a card a few times on Anki can increase my phenomenal granularity or decrease the amount of phenomenal data that my brain throws away, is pretty amazing.

A while ago I also learned country flags via Anki. While the flags in Wikipedia are different I'm not sure that the flags of Chad and Romania are different in reality. German law for example simple says that the colors of the flag are red, gold and black. It doesn't specify the exact shade of red and different flag producers might produce slightly different shades of red.

Having phenomenal granularity for distinguishing different flags is also not that useful in real life. I think the key question is: "What are areas where having more phenomenal granularity actually matters?"

Examples that I have found are:

Audio: Phonemes, pitch of musical notes, duration of musical notes

Visual: Colors, Speed Reading

Kinesthetic: A lot of interesting stuff in somatics. Apart from that heartrate, breathing rate and things that are more difficult to label. Emotions are very important because noticing your emotions affect your reasoning, whether or not you are aware of them.

Taste: Recognise different spices. Tim Ferriss writes about training that skill in 4-Hour Body.

Mental: Credence, time intervals

Your blog posts are great!

Re: your question on Facebook about relative upvotes between this and your "Deferring" post.

The thesis of this post is the last paragraph. I had to read this whole long-ish before finding out what your point was. It wasn't a bad point, but if you're going to keep me interested in hearing about you driving around doing errands and noticing roofs, then I should know in advance what the intended lesson of the post is. I would have found the post much improved if some version the "Really About" section had come first, rather than (or, better yet, in addition to) last.

In the "Deferring" post, the thesis of your post was the first sentence.

My professor's favorite advice for giving presentations:

Tell 'em what you're gonna tell 'em, then tell 'em, then tell 'em what you told 'em.

I'm not party to the FB post in question, but:

Defer had obvious and extremely useful applications and little to no lag time to try them, whereas training habits is inherently non-instantaneous.

Defer used social anxiety as an example (which I think many readers have experienced to one degree or another), whereas Notice is about ... red barn roofs?

Defer had a super-catchy title, and may have gotten more clicks in the first place.

As an aside, part of me hopes that Notice and Defer constitute some kind of A/B testing, and that Brienne is figuring out how to make Really Awesome Posts, and is going to do so :D

Interesting. Reading thru it I thought that there must be some mental discipline traditions about such. I googled and this is what I tunred up:

This post is brilliant.

(Sensations of potential are fascinating to me. I noticed a few weeks ago that after memorizing a list of names and faces, I could predict in the first half second of seeing the face whether or not I'd be able to retrieve the name in the next five seconds. Before I actually retrieved the name. What??? I don't know either.)

Right! When telling people about Anki, I often mention the importance of not self-deluding about whether one knows the answer. But sometimes I also mention how I mark a card as 'Easy' before I've retrieved or subvocalized the answer. It definitely felt like the latter was not self-delusion (especially when Anki was asking me what the capital of the UK was, say). But I felt unable to communicate why it was not self-delusion, and worried that without the other person understanding that mental phenomenon, they would think I was self-deluding and conclude that self-delusion is actually okay after all.

I vaguely noticed that awkwardness to some degree, but I still need to work on the skill of noticing such impasses and verbalizing them. And I certainly wasn't conscious enough of it, or didn't dwell enough on it, to think more about noticing.

It definitely felt like the latter was not self-delusion (especially when Anki was asking me what the capital of the UK was, say). But I felt unable to communicate why it was not self-delusion,

Wild speculation: it's possible to notice that a node in a representational graph is well-connected and thus likely to be close to another node, without following any actual edges (this is very close to a general metric of familiarity but doesn't actually require representing that metric). Something similar might be going on in your head: you haven't retrieved what the capital of the UK is yet, but you know you know a lot about the UK.

This matches my experience extremely well.

But sometimes I also mention how I mark a card as 'Easy' before I've retrieved or subvocalized the answer. It definitely felt like the latter was not self-delusion (especially when Anki was asking me what the capital of the UK was, say).

For that reason I have set all my Anki cards for typing. If you actually type the city name and you get it wrong you notice. Even when I already pressed "easy" Anki allows going back via Crtl+z.

That does happen frequently enough for my for me to think that you are probably sometimes deluding yourself. There are cards where you think you know the right answer but get the card wrong.

It has the added bonus of training typing speed ;) I still have an average answer speed of 16 cards/minute over the last month so I don't think it slows me down much.

I found typing to be a massive deterrent personally. Lots of my Anki is done in bed or on trains on my phone, and I found Memrise (on a laptop) much less compelling and harder to get myself to do than Anki because of all the typing, multiple choice, and drag-n-drop (and it would switch between those which would break my focus). I don't want to have to type 'London' when I'm asked what the capital of the UK is or click it on a multiple choice. Maybe if it were just typing on a fully-fledged computer, like you describe, it wouldn't be so bad?

I still don't think I self-deluded to any actionable extent, but I probably should mention that sometimes I would mark a card as Easy, see the answer and Just Know the answer was different from what I would have answered, undo, and mark the card as Again. I can see how you'd be much more confident I was self-deluding without that detail, which I forgot.

In my personal experience with Memrise is that it does a lot of overtesting for new words. It shows you a card that you have seen the first time and marked as true again in the same session.

I agree that dragging and clicking on multiple choice items don't work well on the laptop. Typing doesn't work well on phones. Multitouch is simple a completely different way of interacting with a device.

On a multitouch device you would want ideally to have to a map and simple click on the country on the map to select it. Speed Anatomy does that really well but it doesn't do spaced repetition. At the moment I'm working on getting binary choices on phones right and afterwards I will go to challenges such as clicking on items on a map.

As those kinds of answers can be scored automatically I'm also getting rid of self evaluation. I want to instead replace it with calculating confidence in the card via things like pressure, time taking to answer and where a button get's pressed. If you want you can then click all buttons for card where you are sure on the top and all buttons where you are unsure on the bottom and the App will automatically learn your pattern.

Smart users who want to tell the App their confidence (so that the app calculates intervals better) can and the average user that doesn't care isn't distracted and the app might even find unconscious patterns.

I experienced a far less conscious and intentional version of noticing reflexively throughout my childhood. Specifically, I became very highly attuned to the act of stepping on cracks in pavings in response to the schoolyard rhyme "Don't step on the crack or you'll break your Momma's back." I never labored under the delusion that there was some mystical force which would cause gross harm to my mother if I did (or didn't) step on a crack--it was more of a game. A game which lasted from early in elementary school through puberty. I have other gamified (if immature) examples of passive noticing--the game) comes to mind. (Apologies if anyone still cares about the game, by the way). Now, these parallels are shallow in that I wasn't meta-noticing as Brienne was. But it does lend a concept I'll find useful in applying the principles of noticing and meta-noticing: namely, the act of gamification. I question, however, whether gamification lends itself to moving the intention from conscious searching to subconscious noticing.


A wonderful skill, and it makes your life so much fuller. Even when you don't want it. I remember a day in the wild when my supervisor sent me out to measure some orchids (Dactylorhiza incarnata) in a population we had been studying for a few years. It was a very hot day, and I really wanted to come home ASAP. So when I saw a Dactylorhiza with two basal leaves, I didn't measure it and moved on. Soon, there was one with two basal leaves and delicate pale yellow flowers - but still definitely an orchid. OK, I started thinking that maybe it was time for a shower. Then after a while there were two together in a way that no Dactylorhiza I'd ever seen did. I was fed up and called my supervisor who 'almost fell out of bed' and said it was Liparis loeselii which was thought to be extinct in the area.