Curiosity: A Greedy Feeling

post by AllAmericanBreakfast · 2020-04-11T04:38:09.544Z · score: 41 (12 votes) · LW · GW · 7 comments

Contents

  What is intrinsic motivation?
  Making yourself curious
  Curiosity is not a cure-all: deliberate practice is still hard
None
7 comments
But what does the phrase “scientifically explicable” mean? It means that someone else knows how the light bulb works. - Eliezar Yudkowsky [LW · GW]
In principle, I agree with the notion that it is unforgivable to not want to know, and not want to improve your map to match the territory.  However, even the most curious person in the world cannot maintain equal curiosity about all things, and even if they could there are limits on time and energy. - Elizabeth [LW · GW]
Rarely am I pulled by a specific desire to know. And most of my learning happens at those rare times. - Anna Salamon [LW · GW]

Which question immediately sparks your specific desire to know?

1) Which enzymes are directly involved in the human transcription of lactase?

2) How do cells manufacture proteins?

3) Which theory is the best account of abiogenesis?

4) Is life inherently valuable?

5) If we could grow artificial human organs, would it be possible to cure aging by periodically replacing the worn-out parts of our bodies?

Without some research, I could not supply more than a rudimentary answer for any of these questions. But for me personally, #5 is the only one that sparks my curiosity.

Why? The first three questions represent different strata in the lake of systematized knowledge. If I ever needed an answer, I could look up fine-grained information on each of them. The fourth question is bottomless, and I know in advance that there will be no definitive answer. None of them suggest a pressing reason I'd want to know.

The last question is different. I can immediately see its practical and scholarly significance. Breaking it down into an array of tractable research questions, then determining what is known and unknown, would be an adventure.

In pursuing my inquiry to the limits of scientific knowledge, I would eventually arrive at strange questions, with a void where the answers should be. I'd never normally think to ask these strange questions, nor care particularly about the answers. Now they would be imbued with the significance of my original purpose.

Curiosity must have been a useful sensation for our ancestors, but a costly one. It motivates action, sometimes patterns of action that are tiring, dangerous, and require the efforts of the whole tribe. Learning is costly. So evolution shaped our curiosity into a getting feeling, a greedy feeling.

What is intrinsic motivation?

That makes curiosity sound like it's fueled by extrinsic motivations, which seems counterintuitive.

One definition of intrinsic motivation is:

Intrinsic work motivation is a cognitive state reflecting the extent to which the worker attributes the force of his or her task behaviors to outcomes derived from the task per se; that is, from outcomes which are not mediated by a source external to the task-person situation. Such a state of motivation can be characterized as a self-fulfilling experience.

OK, this needs some unpacking. Here are two different scenarios:

In the first case, whether or not BizCorp has an e-commerce website, John's getting paid. The outcome of his work, his reward, is "mediated by a source external to the task-person situation."

In the second case, John is not going to get his widget sold without an e-commerce site. His monetary reward, and the satisfaction of bringing a useful new widget to the world, depends directly on creating an e-commerce site. If he finds the creation of the site enjoyable, so much the better.

In the case of curiosity, the reward at the end, tangible or envisioned, will only be available when the curiosity has been satisfied, the quest completed. If you're trying to cure aging, you don't get the reward of immortality unless you succeed. And just think of how much more there would be to learn and experience with that achieved!

If you feel inspired by that objective, even knowing that you personally might die before the investment in research pays back society, then I believe the goal of immortality would constitute an intrinsic reward. So would the pleasure of discovery as you worked in the lab.

So curiosity can be a greedy feeling, a getting feeling, and still stem from intrinsic motivation.

Making yourself curious

Back in 2012, lukeprog wrote Get Curious [LW · GW]. It included visualization, meditation, and brainstorming activities for when you should be curious, but aren't. It follows the narrative in which some inner drive is weak, and we have to conjure up some mysterious inner force to overcome the deficiency.

I believe that lack of curiosity in a seemingly-important question does not stem from lack of drive or laziness.

I think it's that effective curiosity is the synthesis of playful, childlike imaginative interest in a topic we don't understand, and our skillful, adult capacities in an activity we're experienced with. If we don't practice synthesizing these two sides of ourselves, or worse, if we only practice one of these potentials and neglect the other, we'll lose our sense of curiosity.

We'll be left as either skillful but stuck, or imaginative but incapable: a Ben Wyatt or an Andy Dwyer, if you watch Parks & Rec. Part of the arc of both characters is a reconnection with the other half of themselves. Ben Wyatt comes to understand that even though his imagination got him into trouble at an early age, now that he's a responsible and focused adult, allowing his imagination back in will allow him to accomplish amazing things and make his life far more full and rich than it could ever be as a small-town Indiana budget-slasher. Andy Dwyer discovers that when his natural playfulness find a structured outlet as a children's musician, it brings him not just money and respect, but a larger vision for his creativity and a chance to be a leader.

I think the reason why we subscribe to the flagging-drive model of curiosity is that life forces learning upon us. I'm in a calculus course right now, and boy, it would be a lot easier to master that body of knowledge if I was hungry to know it. If only there was a way to make myself feel charged up to learn calculus every single day.

I don't know how to do that.

But if the problem isn't forcing yourself to want to learn any arbitrary body of knowledge you choose, but stimulating your passion for investigation, then there is a way forward.

1) Imagine an ability or set of knowledge or invention or experience that would absolutely delight and amaze you right now if you could access it. I can think of lots of examples for myself:

2) Pick one, and write down some ideas for how we could potentially achieve that goal, based on what you know right now. For being immortal or very long-lived, I'd write:

3) Now start researching the efforts that are underway for #2 (almost certainly, there are some) and try to figure out which seems most practical. Keep a sense of all the sub-questions you might have about your original imagination. Do you need to investigate deeper into one of them, or transition to fleshing out your knowledge on a different sub-question? Always be motivated in your research by the desire to know how to do the impossible, by your desire to turn a fantasy into reality.

These have a strong relationship with lukeprog's three steps for getting curious:

1) Feel that you don't already know the answer.

2) Want to know the answer.

3) Sprint headlong into reality.

But lukeprog's writing is predicated on the idea that you feel as if you know the answer, don't really want to know, and feel no drive to discover it. And this is all because you've selected an arbitrary topic - perhaps a topic life forced on you, or one your peers think is important, or one you feel you ought to be concerned with - and tried to make yourself get curious about it.

If instead, you start with an imaginative vision that you'd genuinely, immediately feel very excited about if it were realized, the basic circumstances are changed. I know right from the start that I've got no idea how to be immortal and have no idea whether or not progressive organ transplants might accomplish this. I do want to know the answer - that would be fascinating - and I can already think of the first questions I would look up. Googling "could organ transplants make us immortal" would be my first step. This is the sprinting headlong into reality.

Curiosity is not a cure-all: deliberate practice is still hard

What if I am trying to learn calculus, though, and want that hit of curiosity to drive me to study?

I already know that it will be helpful for my studies in grad school. But that's a fairly extrinsic motivation. Oh sure, I believe in an abstract way that calculus will eventually be directly useful for my intended course of study in computational biology. But I also have no ability to imagine how. I know I don't know enough to figure out how to apply it right now.

Games are generally a good way to get your brain activated for an arbitrary goal. So are stories. But part of what makes a game or a story good is that they make the challenges or the language interesting. Shoehorning in the solving of equations is a massive design challenge. You end up with a game that doesn't entertain and a homework assignment that's inefficient.

The best I've ever been able to do is to relate to my studies a little bit like it's a sport.

I get into the zone. I see how quickly, how effortlessly, how accurately I can answer problems. How fast can I read this chapter? What's the craziest organic molecule I can draw and then name?

I try to invent my own problems. I rewrite passages in my own words. I don't just try to answer the exercises and remember the concepts. I try to remember what the exercise questions were, or even to make up my own exercise questions. It's the next best thing to tutoring someone else.

I'm not being guided by curiosity. I'm being motivated by competitiveness and restlessness and the desire to test myself, to push the limits. When I'm at my best in my studies, I'm acting more like a fiercely competitive soccer player, or maybe like the stereotype of a hot-shot fighter pilot.

So what keeps a soccer player or a fighter pilot focused on their practice?

One of my skills is playing classical piano music. Some of the most challenging pieces I ever played are J'eux D'eau, by Ravel, and the Ballades of Chopin. I remember what motivated me there.

There was a measure of curiosity here - interest in these pieces, the desire to make them my own. Also of competitiveness, the desire to play some of the most difficult music in the classical canon, or to impress and delight others with my ability

The second remains as I study calculus. It's partly the desire to master math that others find intimidating that motivates me to study calculus. Or to be able to help others with complex, important projects that need a "math guy."

That sense of curiosity is missing to some extent. When I start my third quarter of calculus, I'm not going to know in advance what it entails. There's no way to effortlessly understand as somebody else demonstrates it, the way I can listen to a recording of a Ravel piece I'm considering learning. The doing is identical to the observing. Math is not a spectator sport. That's probably why it's so hard to get people interested in learning it.

My guess is that doing a math assignment can be a thrill, just as much as scrimmage on a soccer team. And that this sense of thrill can be cultivated.

But if you approach a math assignment as though it's a task to complete in order to get a grade, or a tool you're using to force new neural connections in your brain, you won't be paying attention to that sense of thrill.

Where does it come from?

When you're little, playing dodge ball, the game's in motion whether you're ready or not. If the ball hits you, you're out. There's no re-do. It's not a big deal, but secretly, you want to be perfect at dodge ball. You want to hit the other kids with every throw, and never be hit yourself. And you want to do all that while time, and the game, run on.

Can your math assignment be the same way?

Can you connect with time as you work on calculus the same way you connect with it in a game of dodge ball? Can you approach organization on the page, accurate algebra, progress through the substitutions and graphing, with the same tenacity and joy you brought to targeting Johnny on the other team with your rubber ball of death?

Can you find a thrill in the right answer, the same way you felt a thrill when you heard the smack as Johnny got hit by your throw? Can you take delight in noticing you almost dropped a negative sign? Can you approach realizing you missed a step and have to redo the problem the same way you'd have felt getting hit by Johnny when your back was turned - not as frustrated at getting out as you were eager to get back in and seek vengeance?

Math is a lonely game, played against one.

And maybe the idea that a calculus assignment would be as exciting as a childhood game of dodgeball is a pipe dream.

But think about what actually goes into being a fighter pilot. Learning a lot of controls. Switches and dials and messing around with a joystick. You're not running around, you're sitting in a chair, seatbelt on. The wind is not in your hair. If it was the feeling of spinning and high speeds that you loved, a roller coaster would be a safer and easier way to get that sensation than joining the damned air force.

And yet people compete fiercely and work their asses off to be fighter pilots. They're all by themselves up in that plane. It's a lonely game, almost always played against one.

Likewise marksmanship. Likewise the piano. Likewise computer programming. Likewise SCUBA diving.

I think the people who stick with these activities must do a few things:

This quarter, I'm going to see if I can make my calculus and chemistry classes less like toil and more like soccer or piano practice used to be. I don't know what that means yet, or how in practice I can achieve that. But that impossible-seeming goal - what if calculus and chemistry could be as thrilling and joyful as sports or music - that's the first step in activating curiosity. I think I'm on the right track.

7 comments

Comments sorted by top scores.

comment by NaiveTortoise (An1lam) · 2020-04-11T14:27:46.883Z · score: 11 (5 votes) · LW(p) · GW(p)

Great post!

The first thing worth noting is that, as you mentioned, school is sort of an artificial environment which often penalizes following your intrinsic curiosity. Taking that into account, I'm still quite uncertain about the degree to which I'm able to develop expertise purely through intrinsic motivation.

On the one hand, I've seen some comments / posts on LessWrong / Twitter arguing some subset of:

  • For intellectual activities like math & programming, if you're not motivated by curiosity you're doing it wrong.
  • Children naturally master things they're curious about but then school ruins that drive by forcing you to do things you don't want to do.

On the other hand, if you read the "expertise literature" (Ericsson on Deliberate Practice, The Mundanity of Excellence), there's a big focus on consistently doing uncomfortable things that you're not going to be intrinsically motivated to do in the moment. Now, this literature tends to focus on more skill-based, often physical, "rote" activities like swimming, violin, etc. Chess is probably the closest thing to an intellectual pursuit that's been studied by academics through the lens of deliberate practice. That said, Cal Newport (a Computer Science professor and sort of pop science writer), has written extensively about his belief and experience applying deliberate practice to knowledge work and I think he makes good arguments that fit with my personal experience.

Finally, my personal experience has also provided mixed evidence for the two conflicting hypotheses. I do learn much better when I'm intrinsically curious about what I'm learning and as a result spend time idly thinking about it. But, I also find that in order to really learn something challenging well, especially early on, I sometimes need to force myself to sit my butt down at the chair and practice.

As a concrete example, I used to not be that intrinsically interested in algorithms as an area beyond the basic ones that everyone knows about (BFS, DFS, LRU cache, etc.). As a result, when I decided I wanted to prepare for interviews before looking for a new job, I mostly forced myself to sit down and grind through a bunch of algorithms-y problems. This was not that fun at first because it was quite hard and even when I was curious about the answers to, in particular, the harder problems, I wouldn't be able to get them or would take a long time to. Also, it was originally driven by extrinsic motivation. However things changed as time went on. Once I reached a certain level of proficiency, I grew to like practicing and learning about algorithms more because I worried less about constantly failing. My two take-aways from this experience were:

  • A sense of inadequacy is a big obstacle to intrinsic motivation.
  • It's much easier to be curious when you're already somewhat competent.

In the end, I think both the "perfect practice makes perfect" and "harness the desire to know" perspectives have something to contribute towards developing expertise, but I personally am not yet at a stage where I can learn really hard things without doing some amount of "forcing" myself.

comment by NaiveTortoise (An1lam) · 2020-04-11T14:34:40.775Z · score: 5 (3 votes) · LW(p) · GW(p)

I just re-read your post one more time and want to comment that I can strongly relate to this section in particular:

The best I've ever been able to do is to relate to my studies a little bit like it's a sport.

I get into the zone. I see how quickly, how effortlessly, how accurately I can answer problems. How fast can I read this chapter? What's the craziest organic molecule I can draw and then name?

I try to invent my own problems. I rewrite passages in my own words. I don't just try to answer the exercises and remember the concepts. I try to remember what the exercise questions were, or even to make up my own exercise questions. It's the next best thing to tutoring someone else.

I'm not being guided by curiosity. I'm being motivated by competitiveness and restlessness and the desire to test myself, to push the limits. When I'm at my best in my studies, I'm acting more like a fiercely competitive soccer player, or maybe like the stereotype of a hot-shot fighter pilot.

comment by TurnTrout · 2020-04-11T18:50:35.469Z · score: 2 (1 votes) · LW(p) · GW(p)

ETA: started typing up thoughts, but accidentally published early. May finish typing up thoughts later.

comment by NaiveTortoise (An1lam) · 2020-04-11T21:24:19.271Z · score: 2 (2 votes) · LW(p) · GW(p)

Thoughts much appreciated :) !

comment by G Gordon Worley III (gworley) · 2020-04-11T22:41:47.421Z · score: 6 (2 votes) · LW(p) · GW(p)
They're in the habit of defining small but meaningful projects for themselves, breaking those projects down into steps, and practicing the steps until the whole project attains a smooth and satisfying flow.
They're looking not just for goal-oriented success but for little, intuitive, hard-to-explain ways of finding a thrill in the practice itself. They make little games for themselves that would seem utterly neurotic and silly to other people. And those self-imposed, idiosyncratic games are actually an important part of the lived experience of practicing for these people.
They look for opportunities to teach and show off their skills.

I feel like these are all kind of true, but also slightly askew of my reality.

Sometimes it feels like it was overdetermined that I would become a programmer. I started when I was 8, kept coming back to it, and had this persistent drive to make computers do things and to build things out of computers. Looking back it didn't make a lot of sense; I had some unjustified belief that COMPUTERS!!! were the answer to all my problems.

And then there was the reality of becoming a programmer. The long trail of abandoned projects, the procrastination, the vast difficulty of getting the computer to do what I wanted it to do, the constant failure. It was like getting punched in the face over and over, and yet I kept coming back for more until I finally learned one day after decades of effort how to punch back. If you had looked at me from the inside you would have not seen what looked like a healthy way to become an expert at something.

So I don't know, I feel like there are a lot of pieces to the puzzle, and sometimes things work out in spite of conditions that seem like they should lead to failure, or that combine in weird ways to work despite lots of forces working against mastery.

comment by johnswentworth · 2020-04-11T16:23:59.642Z · score: 4 (3 votes) · LW(p) · GW(p)
I'm in a calculus course right now, and boy, it would be a lot easier to master that body of knowledge if I was hungry to know it. If only there was a way to make myself feel charged up to learn calculus every single day.

I recommend an intro physics class. Calculus and mechanics are subjects best learned side-by-side; that's mainly why calculus was invented in the first place.

(Personally, I took a physics class before calculus, and when I learned calculus I had this perfect calculus-shaped hole in my head just waiting to be filled in.)

comment by NaiveTortoise (An1lam) · 2020-04-11T21:26:17.404Z · score: 2 (2 votes) · LW(p) · GW(p)

A little trickier because it tends to require more pre-requisites like linear algebra but I feel similarly about the optimization-focused aspects of Machine Learning.