So, you've considered your past experiences and your motivations, and you've got a decent idea of the effort required of you: Six to eight hours of solid work before you start showing improvement, and about twenty hours total before you start to exhaust the low hanging fruit. You want to learn to draw. What exactly does that entail?

A lot of things, really. There's probably hundreds of subskills, techniques and bits of knowledge that go into creating a quality drawing. But I think they cluster into three main categories:

  • Observation
  • Technical Skill
  • "Instilling Energy and Weight"

Each of these skills is developed with different exercises, requiring different mindsets. Switching between those exercises can be difficult. Studying any of them can produce something that is interesting to look at, but ultimately you want to integrate them into a single, fluid mental process. You'll need to develop some competence in each of them first. As you begin, the most important thing to remember is that *learning* to draw is not the same as actually drawing. To improve as quickly as possible you may have to set aside the reasons you wanted to draw in the first place. Don't worry - you'll achieve those terminal goals in time.

In this article I'll briefly discuss each of those skill clusters, and how I believe they relate to rationality.


The ability to see reality, as it truly is. "Reality" refers both to the physical objects and light that you're looking at, as well as to your perception of your own mental processes and how they should be interpreting those physical objects. If you have previous rationality training, this is where I expect it to benefit you the most. Your model of reality will be flawed, and you'll need to fix it. Existing ability to notice biases and counter them should give you an advantage.

The advantage *may* come in reduced study time (I'd need a lot of data to confirm this) but mostly of the advantage will come from willingness to actually study effectively in the first place (this is a genuinely big deal). With minimal instruction, you could probably figure out where your biases lie and how to fix them. Some knowledge of cognitive science might even give you insight as to where and why they might be flawed. I'd be interested if someone without much drawing experience attempted to predict their biases in advance, once before having drawn at all in the recent past, and once again after one or two drawings.

It would still take a long time to do that, which is why in the next article I'm going to give away a lot of common answers. As a bonus rationality exercise you may want to predict those answers in advance. (For the benefit of others, leave your answers encrypted in the comments)

I'm not sure how to test for it, but I believe the skills here can transfer to other domains, if one deliberately made the attempt.

Technical Skill:

The ability to control your pencil, moving it in the direction you want it to go, applying different amounts of pressure to make it darker or lighter, thicker or thinner. As your observation and technical skill improve together, you'll learn to identify important parts of reality, and use pencil techniques to emphasize them properly. It's important to develop a minimal threshold of technical skill. But afterwards, it can wait until you've gotten a solid understanding of Observation, Energy and Weight.

Nine thousand hours practicing technical skill is what makes the difference between a competent amateur and a professional. A few of those hours will yield low hanging fruit, but not many. Rationality won't be particularly relevant here. I'll be doing one article that covers the basics, and later on I'll link some online tutorials that may be useful *after* you've put in an initial twenty hours or so.

Instilling Energy and Weight

This is the most mysterious of the skill clusters. It's (fairly) obvious to a lay person that they need to look at things and practice moving their pencil in order to improve at drawing. That's what high school art students focus on. But their drawings feel flat, and lifeless. They'll copy something from a photo and there won't be anything obviously wrong. It's just... boring somehow.

The problem was that the technical skill they developed focused on small, slow, precise movements. To instill energy and weight, you need to be able to draw big, bold lines, and to draw them quickly. They require a kind of hand-eye coordination (and more importantly, arm-eye coordination) that you've probably never developed. The first big, bold lines you draw will look hopelessly wrong against reality. You need to keep practicing, until your entire arm can do what your brain wants it to. Until then, much of the technical skill you acquire won't be implemented correctly.

The harder part is the that to instill energy and weight, you may sometimes need to NOT draw the reality you see.[1] As this skill develops alongside technical knowledge, kinesthetic skill and observational ability, you'll learn how to draw lines that differ from reality to make your drawing more interesting and creative, without sacrificing a realistic look that captures the original likeness. And you'll need to start doing this with a part of your brain so small and underdeveloped that it doesn't even show up on your model of yourself.

Your rationality training will probably not help you, because it doesn't directly prepare for this sort of thing. The actual skills you're developing here are kinesthetic, not high level cognition. But I think preparing for this sort of thing is a kind of instrumental rationality technique we *should* be working on. This weirdness you feel as you practice, the certainty that your teacher is screwing with you and that you should go back to the comfortable ways you're familiar with... this is the feeling of your model being _wrong_, and butting up uncomfortably against reality of how your brain works. Not just getting a fact wrong. Not even failing to ask the right questions in the first place, but lacking the sense capabilities needed to even interact with the part of reality you needed to see in order to ask those questions. It's having your decision-making algorithm *be* wrong on a gut level that can't really be explained, only experienced.[2]

I think this is among the more important experiences that drawing offers[3], if the effort is made to understand it. Obviously, not every "wrong" feeling you ever have will stem from a model-reality mismatch. Sometimes when something feels wrong, it really is wrong. "Wrong" feelings are still evidence. But they are not absolute evidence.

Yes, I know you already "knew" that. But unless you've already had a similar experience, you probably didn't really understand it on a gut level. Being able to recognize when you're approaching things is a fundamentally wrong way, and be okay with it, is an incredibly important skill. And because each way of being fundamentally wrong is slightly different, we could use a variety of examples in order to better understand new ones in the future.

Existing Less Wrong literature has two areas that I consider to have helped build this skill for me. One was the overall "woah" moment when I realized I should actually incorporate the Singularity into my model of the world (Even this doesn't really count, because the "that can't possibly be right" feeling was accompanied by "it would be sooo cooool if it were!" instead of "ugh this is painful to think about). The other was the emphasis that quantum physics isn't weird, YOU are. Those realizations may have more serious consequences on how I interpret the world, but the experiences themselves were weird on a similar level. Plenty of other content here talks about this sort of thing, but there's a limited number of ways to actually experience being wrong in this way. I think we need more of that.

Somewhere in your brain is an invisible mental-muscle that you never knew existed.[4] You'll need to trust that it exists, and start practicing with it. Eventually you'll be able to "see" that part of your mind, and you'll incorporate it into your model of yourself.

And then you can learn how to make interesting depictions of reality.

Perhaps more importantly, you can learn to create Art™.

[1] Many readers may notice that this statement... really makes no sense. That's because it's a simplification. I'll explain the more complex rule that actually governs it in a later post. Feel free to speculate in the comments about why I chose to simplify it that way.

[2] I know that's three posts in a row where I hint at the same mysterious thing. I'm building it up because it's important and needs to be emphasized. You're still not going to really understand it until a teacher makes you do it and helps explain some kinesthetic things that I'm going to have a hard time communicating through written text. Three posts from now I'll attempt to explain it well enough that you'll understand the teacher the first or second time e explains it, rather then after many frustrating hours.

[3] I do NOT recommend people learn to draw purely to experience this feeling of wrongness. In order for any of this to work you need to be instrinsically motivated.

[4] This may technically be untrue, if you HAVE already become aware of this mental-muscle and used it in some fashion. It may not even strike you as an odd thing, if you've had it for a while. But I'm willing to bet it'll be a new thing for most people here. I actually think we'll have a higher percentage of people who have a particularly hard time working with it.

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

I've been drawing my whole life, and never thought to apply the ideas I learn here to the lessons on observing the world that my art teacher taught me. I read the previous two posts and the posts on talent (the one linked), and it's very interesting.

When it comes to gesture drawings, one of the main problems the students (including myself) found, was trusting their eye, and not their memory of what they see. When you plan on drawing your favorite coffee mug, you can picture it in your mind, but it is very unlikely that your memory will perfectly preserve every curve, every detail on that mug. Not only that, but the person drawing needs to scratch the idea of drawing objects entirely.

What happens to the kitchen table the mug is sitting on? The mug does not float in empty white space! There is the wall, the kitchen table, whatever knickknacks you have on your table, that is, if your coffee mug is on your kitchen table at all.

The way I learned to draw was to not look at the paper while drawing. Instead, my class and I were told to look at the scene we wanted to draw, to let our eyes travel around the objects and empty spaces in between and let our hands move over the paper. This process has a much slower learning curve, but the improvements were substantial. We were told to not focus on objects, but on the shapes the objects make.

When you draw a chair, you are not drawing a chair from your memory, nor the object "chair." It's darker lines, lighter lines, the space between those lines, and shapes.

This might be a bit challenging for a beginner, but it might be worthwhile to consider.

Yeah, looking at the reality is incredibly important. There's a followup post on observation now that deals with some of the issues you describe.

Did you learn to draw well (i.e. create a finished product you're happy with) without looking at the paper at all? If so, how long did it take you. (That's actually what I'm currently learning how to do. I don't think it's necessary to be completely strict about it, for developing general skill. But I also want to be able to draw people while I'm talking to them, and its a prerequisite for that.)

We would usually take our sketches that we made in class and find which ones we liked the most, then go back to where we drew them and create a composition out of them. However, I do remember that in sophomore year, we had to do a charcoal drawing of a still life and we had to do it without looking at the paper at all. The results varied across the class. I found that mine was not terrible, but I was not particularly pleased with it. I think the reasons had more to do with the charcoal and less to do with the technique, though. I always had an issue with working in charcoal. Regardless, I remember the class starting to work without looking at the paper in freshman year throughout sophomore year. We picked it up again in senior year a bit, doing gesture drawings of people, but did not focus on it much, since most of senior year was self-teaching and solo projects.

The class was 5 times a week for 45 minutes each session.


I'm really itching to try this out! ;)

(Consider this as a word of encouragement. I'll to think about my predictions and will post them here if I come up with anything useful. But, in the time being I wanted to say at least this much.)

The "quantum mechanics isn't weird, you are" link is broken.

Thanks, fixed.

"Six to eight hours of solid work before you start showing improvement, and about twenty hours total before you start to exhaust the low hanging fruit."

Can we put this to the test? I'd like to see some people keep every drawing they make in the first twenty hours, scan them, and let us see how much improvement there is.

I'll volunteer for this, but I'm likely going to do my first 20 hours this week, using different practices than yours, or I will have to do it much later. The reason being that my previous learning experiences tell me that spread out, divided practice is diluted practice. I'm on break from school, so now would be a good time to do this.

I definitely want to test this as much as possible. Uploading every single drawing is probably impractical (we're talking 30 second gesture drawings... you do the math), not only because it's a lot for one person to upload, but it's a lot for people on the interwebs to bother rating. I am planning to post a few drawings from one volunteer to show what kind of improvement I'm talking about.

My initial 8 hour plan was based off a particular class I took in college. I'm pretty positive that it's not the optimal presentation of material, although it was a good starting place. I'd like to refine various ideas until I've figured out an ideal 8 hour class that I can give to random people and get good results as fast as possible. Ultimately, I think we'll be able to expect improvement to show sooner, although it'll take some experimentation and research.

Experimentation will be confounded by me getting better at teaching alongside me trying new things.

I agree about diluted practice. I was actually very happy and surprised when results turned out approximately in the timeframe I expected, since there was a multi-week period between the first and second session.

Good luck with your practice. What methods are you using?

"Uploading every single drawing is probably impractical (we're talking 30 second gesture drawings... you do the math), not only because it's a lot for one person to upload, but it's a lot for people on the interwebs to bother rating."

True, a few drawings from each time period will suffice. I plan to work digitally, since digital drawing is my end-goal, so uploading each drawing is more realistic for me than for pencil and paper practicers.

"What methods are you using?"

I'm planning on following The Natural Way To Draw by Kimon Nicolaides, which advises the first 15-20 hours to be split mostly between contour drawing and gesture drawing.

I'm still not convinced that drawing has any real relevance to rationality. To me drawing seems to mostly involve unconscious motor learning that does not generalize much to other domains. For most rationality related purposes I don't see any major difference from other motor-based, practice-requiring skills such as juggling, playing golf, playing an instrument or patting-your-head-while-rubbing-your-stomach.

Sure, in some sense you will have to "see reality, as it truly is", and sure "your model of reality will be flawed, and you'll need to fix it", but I think this is in a sense that is only trivially analogous to what we usually talk about when we talk rationality, and I think the same statements could be made about the other motor skills mentioned above. When you practice these kinds of skills some parts of your brain/body do develop better models of reality, but it doesn't seem like those parts are particularly conscious or that what these parts have learned will benefit other parts much.

I don't know much about the neuroscience of learning, but my own experience with these kinds of skills is that you simply get better by repeated practice and that the underlying basis for the improvement is highly unconscious and hidden away somewhere in the structure of whatever neural systems deal with these kind of motor activities. Your muscle memory is developing a better model of reality, but your muscle memory is probably very disconnected from whatever parts of your brain deal with rationality in the traditional LW sense (which of course is not necessarily conscious). Placing drawing in the same bucket as "grokking quantum physics or abandoning a religion" seems very far-fetched to me. My guess is that the underlying cognitive structures are very different.

I edited the article to explain what I meant a little better, and sound a little less grandiose. (I'm not entirely sure it's better this way - I have the old article saved, anyone who read both articles, let me know what you think of the difference). I think I may have implied a particular kind of "relevance" that I didn't intend to. Let me know if you still disagree. Also let me know if the disagreement stems from me misusing the word rationality, or from you disbelieving that drawing will have the effect that I say it will.

I think it's very easy to learn to draw without having the skills generalize, but I do think certain skills can generalize if you approach them in a certain way. (I'm trying to work out a testable prediction that would settle this, but I don't think we have very good tests that measure the relevant things in the first place)