Instrumental Rationality 1: Starting Advice

[In­stru­men­tal Ra­tion­al­ity Se­quence 17. Re­post from LW]

[This sec­tion goes over 4 con­cepts that I think are im­por­tant to keep in mind be­fore we start the other stuff. We go over car­ing about the ob­vi­ous, look­ing for ways to ap­ply ad­vice in the real world, prac­tic­ing well, and hold­ing re­al­is­tic ex­pec­ta­tions.]

In Defense of the Ob­vi­ous:

[As ad­ver­tised.]

A lot of the things I’m go­ing to go over in this se­quence are some­times go­ing to sound ob­vi­ous, bor­ing, re­dun­dant, or down­right tau­tolog­i­cal. I’m here to con­vince you that you should try to listen to such ad­vice any­way, even if it sounds stupidly ob­vi­ous.

First off, our brains don’t always see all the con­nec­tions at once. Thus, even if some given ad­vice is ap­par­ently ob­vi­ous, you still might be learn­ing new things.

For ex­am­ple, say I told you, “If you want to ex­er­cise more, then you should prob­a­bly ex­er­cise more. Once you do that, you’ll be­come the type of per­son who ex­er­cises more, and then you’ll likely ex­er­cise more.”

The above ad­vice might sound pretty silly, but keep in mind that our men­tal cat­e­gories for “ex­er­cise” and “per­sonal iden­tity” might be in differ­ent places. Sure, it’s tau­tolog­i­cally true that some­one who ex­er­cises be­comes a per­son who ex­er­cises more. But if you’re not ex­plic­itly think­ing in terms of how your ac­tions change who you are, which is what the tau­tol­ogy is point­ing at, then you’ve still learned some­thing new.

Hu­mans are of­ten weirdly in­con­sis­tent with our men­tal buck­ets—things that log­i­cally seem like they “should” be lumped to­gether by log­i­cal im­pli­ca­tion of­ten aren’t.

By pay­ing at­ten­tion to even tau­tolog­i­cal ad­vice like this, you’re able to form new con­nec­tions in your brain and link new men­tal cat­e­gories to­gether, per­haps dis­cov­er­ing new in­sights that you “already knew”.

Se­condly, ob­vi­ous ad­vice is of­ten used as a la­bel for what ev­ery­one know works. If your brain is pat­tern-match­ing some­thing as “bor­ing ad­vice” or “ob­vi­ous”, you’ve likely heard it be­fore many times be­fore.

For ex­am­ple, you can prob­a­bly guess the top 5 things on any “How to be Pro­duc­tive” list—make a sched­ule, re­move dis­trac­tions, take breaks, etc. etc. You can al­most feel your brain roll its metaphor­i­cal eyes at such dreary, well-worn ad­vice.

But if you’ve heard these things re­peated many times be­fore, this is also good rea­son to sus­pect that, at least for a lot of peo­ple, it ac­tu­ally works. Mean­ing that if you aren’t tak­ing such ad­vice already, you can prob­a­bly get a boost by do­ing so.

If you just did those top 5 things, you’d prob­a­bly already be quite the pro­duc­tive per­son.

The trick, then, is ac­tu­ally do­ing them.

Lastly, it can be easy to dis­count ob­vi­ous ad­vice when you’ve seen too much of it. When you’re bom­barded with bor­ing-seem­ing ad­vice from all an­gles, it’s easy to be­come de­sen­si­tized.

What I mean is that it’s pos­si­ble to dis­miss ob­vi­ous ad­vice out­right be­cause it sounds way too sim­ple. “This can’t pos­si­bly work,” your brain might say, “the se­cret to get­ting things done must be more com­plex than that!”

In philos­o­phy, there’s this idea of a “he­do­nic tread­mill”, an idea based off our hu­man in­cli­na­tion to com­pare ex­pe­riences rel­a­tively. For ex­am­ple, if you have some­thing tasty, like a choco­late bar, then you’ll need some­thing even tastier the next time, like a tiramisu cake. A more stereo­typ­i­cal ex­am­ple might be a drug user seek­ing out an even stronger high for their sec­ond ex­pe­rience.

The point is that, as you are ex­posed to more and more plea­sures, you find your­self on a tread­mill climb to seek out ever more deli­cious ex­pe­riences (be­cause, by com­par­i­son, ev­ery­thing else will seem dull).

There’s some­thing akin to the he­do­nic tread­mill hap­pen­ing here where, af­ter hav­ing been ex­posed to all the “nor­mal” ad­vice, you start to seek out deeper and deeper ideas in search of some sort of men­tal high. What hap­pens is that you be­come a kind of self-help junkie.

As a self-help junkie, you end up adopt­ing quite the con­trar­ian stance—you re­ject the typ­i­cal idea of ad­vice on grounds of its ob­vi­ous­ness alone. There’s a cer­tain aes­thetic to be­ing the cool kid who knows that sim­ple ad­vice isn’t enough to solve their com­plex, multi-faceted prob­lems.

You can end up crav­ing the bleed­ing edge of crazy ideas be­cause liter­ally noth­ing else seems worth­while. You might end up dis­miss­ing ob­vi­ous helpful ideas sim­ply be­cause they’re not paradigm-crush­ing, mind-blow­ing, or men­tally stim­u­lat­ing enough.

If this de­scribes, might I tempt you with the meta-con­trar­ian point of view?

Here’s this for a crazy idea: One of the se­crets to win­ning at life is look­ing at ob­vi­ous ad­vice, ac­knowl­edg­ing that it’s ob­vi­ous, and then do­ing it any­way.

(That’s right, you can join the even cooler group of kids who scoff at those who scoff at the ob­vi­ous!)

You can both say, “Hey, this is pretty sim­ple stuff I’ve heard a thou­sand times be­fore,” as well as say, “Hey, this is pretty use­ful stuff I should shut up and do any­way even if it sounds sim­ple be­cause I’m smart and I rec­og­nize the value here.”

At some point, be­ing more so­phis­ti­cated than the so­phis­ti­cates means be­ing able the grasp the idea that not all things have to be hy­per com­plex. Of­ten­times, the trick to get­ting some­thing done is sim­ply to get started.

Be­cause some things in life re­ally are ob­vi­ous.


Hunt­ing for Prac­ti­cal­ity:

[This is about look­ing for ways to have any ad­vice you read be ac­tu­ally use­ful, by hav­ing it ap­ply to the real world. ]

Imag­ine some­one try­ing to ex­plain ex­actly what the mi­to­chon­dria does in the cell, and con­trast that to some­one try­ing to score points in a game of bas­ket­ball.

Some­one could take classes to learn how to get bet­ter at each of those two things.

Yet, there’s some­thing clearly differ­ent about what each per­son is try­ing to do, even if we lumped both un­der the la­bel of “learn­ing”.

It turns out there are roughly two types of knowl­edge you can learn: declar­a­tive and pro­ce­du­ral knowl­edge.

Declar­a­tive knowl­edge is like the stu­dent try­ing to puz­zle out the mi­to­chon­dria ques­tion; it’s about what you know. It’s about how your con­cepts link to one an­other, like how you can know that Paris is the cap­i­tal city of France, or that it takes about 20 min­utes to walk a mile.

In con­trast, pro­ce­du­ral knowl­edge, like the fledgling bas­ket­ball player, is about what you do. It’s about how you ac­tu­ally carry out cer­tain ac­tions, like how you learn to throw a fris­bee well, or how to ride a bike.

I bring up this di­vide be­cause many of the tech­niques in in­stru­men­tal ra­tio­nal­ity will feel like declar­a­tive knowl­edge, but they’ll ac­tu­ally be much more pro­ce­du­ral in na­ture.

For ex­am­ple, say you’re read­ing an es­say on mo­ti­va­tion, and you read about how “Mo­ti­va­tion = En­ergy to do the thing + a Re­minder to do the thing + Time to do the thing = E+R+T”.

What’ll likely hap­pen is that your brain will form a new set of men­tal nodes that con­nects “mo­ti­va­tion” to “E+R+T”. This would be great if I ended up quizzing you “What does mo­ti­va­tion equal?” where­upon you’d cor­rectly an­swer “E+R+T”.

But that’s not the point here!

The point is to have the equa­tion ac­tu­ally cash out into the real world and pos­i­tively af­fect your ac­tions. If in­for­ma­tion isn’t chang­ing you view or act, then you’re prob­a­bly not ex­tract­ing all the value you can.

What that of­ten means is figur­ing out the an­swer to this ques­tion:

“How do I see my­self tak­ing differ­ent ac­tions as a re­sult of hav­ing learned this in­for­ma­tion?”

With that in mind, maybe you gen­er­ate some ex­am­ples and make a list in re­sponse to the ques­tion.

Your list of real-world ac­tions might end up look­ing like:

  1. Re­mem­ber­ing to stay hy­drated more of­ten (En­ergy)

  2. Us­ing more Post-It notes as memos (Re­minder)

  3. Start us­ing Google Cal­en­dar to block out chunks of time (Time).

The point is to be always on the look­out for ways to see how you can use what you’re learn­ing to in­form your ac­tions. Learn­ing about all these things is only use­ful if you can find ways to ap­ply them.

As we go for­ward, I’ll try to give con­crete and use­ful ex­am­ples for all the ideas we go over, but you want to find ways to move past the sim­ple pat­tern-match­ing. You want to do more than have empty boxes that link con­cepts to­gether. It’s im­por­tant to have those boxes linked up to ways you can do bet­ter in the real world.

You want to ac­tu­ally put in some effort try­ing to an­swer ques­tion of prac­ti­cal­ity.

Knowl­edge might be power, but you also of­ten need to act on it.


Ac­tu­ally Prac­tic­ing:

[This is about know­ing the nu­ances of lit­tle steps be­hind any sort of self-im­prove­ment skill you learn, and how those lit­tle steps are im­por­tant when learn­ing the whole.]

So on one level, us­ing knowl­edge from in­stru­men­tal ra­tio­nal­ity is about how you take declar­a­tive-seem­ing in­for­ma­tion and find ways to uti­lize it in real world ac­tions.

That’s definitely im­por­tant.

Another thing, though, is to note that the very skill of “Gen­er­at­ing Ex­am­ples”—the thing you did in the above es­say to even figure out which ac­tions can fit in the above equa­tion to fill in the blanks of E, R, and T—is it­self a men­tal habit that re­quires pro­ce­du­ral knowl­edge.

What I mean is that there’s a sub­tler thing that’s hap­pen­ing in­side your head when you try to come up with ex­am­ples—your brain is do­ing some­thing—and this “some­thing” is im­por­tant.

It’s im­por­tant, I claim, be­cause if we peer a lit­tle more deeply at what it means for your brain to gen­er­ate ex­am­ples, we’ll come away with a list of steps that will feel a lot like some­thing a brain can do, a prime ex­am­ple of pro­ce­du­ral knowl­edge.

For ex­am­ple, we can imag­ine a ma­gi­cian try­ing to learn a card trick. They go through the steps. First they need to spread the cards. Then comes the se­cret move. Fi­nally comes the mag­i­cal re­veal of the se­lected card in the ma­gi­cian’s pocket.

Even though the au­di­ence ex­pe­riences the whole trick as one mag­i­cal unit, the ma­gi­cian knows that it’s re­ally made up of all those lit­tle steps.

Like­wise, we can ap­ply the same anal­ogy to things we’ll learn in in­stru­men­tal ra­tio­nal­ity, to the men­tal habits we’ll go over.

If we spent some time re­ally look­ing at the lit­tle steps of com­ing up with ex­am­ples, we could de­scribe it in de­tail. The skill of Gen­er­at­ing Ex­am­ples, with a re­duc­tion­ist view, might look some­thing like this:

Tech­nique: Gen­er­at­ing Ex­am­ples

  1. Imag­ine the “skele­ton” of the con­cept you are try­ing to fit an ex­am­ple to.
    EX: Anne is try­ing to come up with an ex­am­ple of what is a Sys­tem 1 pro­cess. She knows it’s about fast and some­times mis­taken think­ing, so she uses that as the “frame” to search for ex­am­ples.

  2. Look for things in your ev­ery­day life that fit.
    EX: Anne thinks about things in her daily rou­tine which don’t re­quire much thought. “Maybe brush­ing my teeth?” she won­ders.

  3. Think about books, movies, or other pop cul­ture if real life doesn’t prove fruit­ful.
    EX: Anne thinks about a movie about a char­ac­ter who always gets into trou­ble be­cause of their quick wit and fast tongue. “Hmm, maybe also the sort of so­cial re­sponses we give count as S1 re­sponses?” she thinks.


The idea here is to de­scribe any men­tal skill with enough gran­u­lar­ity and de­tail, at the 5 sec­ond level. What I mean by that is that you should be able to de­scribe such that you’d both be able to go through the same steps a sec­ond time and teach some­one else.

That means hav­ing a very deep un­der­stand­ing of ex­actly what lit­tle steps you’re go­ing through in your head to pro­duce the skill.

Now of course most of us already know how to gen­er­ate ex­am­ples, so the above “tech­nique” for­mu­la­tion might seem a lit­tle alien, as it’s already some­thing we do with­out much ex­plicit, con­scious in­put.

How­ever, when we move on to more novel and com­plex habits of mind, tech­niques that in­volve mov­ing your brain in new ways, then hav­ing a good un­der­stand­ing that these steps are things you do be­comes quite im­por­tant.

A bas­ket­ball player doesn’t strongly im­prove just by watch­ing the NBA. Like­wise, figur­ing out this in­stru­men­tal ra­tio­nal­ity stuff is not a spec­ta­tor sport ei­ther. You need to re­ally go through the men­tal mo­tions in your own head and act on them.

This is what I mean when I say that men­tal habits are pro­ce­du­ral.

In ad­di­tion to figur­ing out the prac­ti­cal­ity and the lit­tle pieces, it’s im­por­tant to find op­por­tu­ni­ties to use these differ­ent skills.

You can’t get bet­ter at do­ing some­thing un­less you, y’know, ac­tu­ally do it.


Real­is­tic Ex­pec­ta­tions:

[An es­say about hav­ing re­al­is­tic ex­pec­ta­tions and look­ing past po­ten­tially harm­ful fram­ing effects.]

There’s this ten­dency to get frus­trated with learn­ing men­tal tech­niques af­ter just a few days. I think this is be­cause peo­ple miss the declar­a­tive vs pro­ce­du­ral dis­tinc­tion.

(But you hope­fully won’t fall prey to it be­cause we’ve cov­ered the dis­tinc­tion now.)

Once we liken the anal­ogy to be more like that play­ing a sport, it be­comes much eas­ier to see that any ex­pec­ta­tion of im­me­di­ately learn­ing a men­tal habit is rather silly—af­ter all, no one ex­pects to mas­ter ten­nis in just a week.

So, when it comes to try­ing to con­figure your ex­pec­ta­tions, I sug­gest that you try to renor­mal­ize your ex­pec­ta­tions by treat­ing learn­ing men­tal habits more like learn­ing a sport.

Keep that as an anal­ogy, and you’ll likely get fairly well-cal­ibrated ex­pec­ta­tions for learn­ing all this stuff.

Still, what, then, might be a re­al­is­tic time frame for learn­ing?

We’ll go over habits in far more de­tail in a later sec­tion, but a rough num­ber for now is ap­prox­i­mately two months. You can ex­pect that, on av­er­age, it’ll take you about 66 days to in­grain a new habit.

(There’ll be a lot more on habits in Part 4.)

Similarly, in­stru­men­tal ra­tio­nal­ity (prob­a­bly) won’t make you a god.

Dis­ap­point­ing, I know.

Still, in my ex­pe­rience, study­ing these ar­eas has been su­per use­ful, which is why I’m writ­ing this at all. Your own mileage will vary de­pend­ing where you are right now, but this serves as the gen­eral dis­claimer to keep your ex­pec­ta­tions within the bound of re­al­ity.

Here, the main point is that, even though men­tal habits don’t seem like they should be more similar to play­ing a sport, they re­ally are. There’s some­thing here about how first im­pres­sions can be rather de­ceiv­ing.

For ex­am­ple, a typ­i­cal trap I some­times fall into is miss­ing the dis­tinc­tion be­tween “the­o­ret­i­cally pos­si­ble” and “re­al­is­tic”.

I end up look­ing at the sup­posed 24 hours available to me ev­ery­day and then beat­ing my­self up for not be­ing able to har­ness all 24 hours to do pro­duc­tive work. “After all,” my brain says, “that’s all time you could be us­ing to do things!”

But such a fram­ing of the situ­a­tion is in­ac­cu­rate; things like sleep and eat­ing are of­ten very es­sen­tial to max­i­miz­ing pro­duc­tivity for the rest of the hours! Just be­cause it looks like I could “in the­ory” get ad­di­tional work time, self-care is also an im­por­tant fac­tor that we eas­ily miss!

So when div­ing in and prac­tic­ing, try to look a lit­tle deeper when set­ting your ex­pec­ta­tions. Bias to­wards pes­simism. No one likes to hear it, but the chances of you ac­tu­ally turn­ing your life around at any given mo­ment are likely slim.

That’s just how things work. Dis­ap­point­ing, for sure, but that’s all the more rea­son to be sus­pi­cious if you’ve got too rosy ex­pec­ta­tions of how things will turn out.

First glances tend to be de­ceiv­ing.


Next essay