Practical Advice Backed By Deep Theories

Once upon a time, Seth Roberts took a Euro­pean va­ca­tion and found that he started los­ing weight while drink­ing un­fa­mil­iar-tast­ing caloric fruit juices.

Now sup­pose Roberts had not known, and never did know, any­thing about metabolic set points or fla­vor-calorie as­so­ci­a­tions—all this high-falutin’ sci­en­tific ex­per­i­men­tal re­search that had been done on rats and oc­ca­sion­ally hu­mans.

He would have posted to his blog, “Gosh, ev­ery­one! You should try these amaz­ing fruit juices that are mak­ing me lose weight!” And that would have been the end of it. Some peo­ple would have tried it, it would have worked tem­porar­ily for some of them (un­til the fla­vor-calorie as­so­ci­a­tion kicked in) and there never would have been a Shangri-La Diet per se.

The ex­ist­ing Shangri-La Diet is visi­bly in­com­plete—for some peo­ple, like me, it doesn’t seem to work, and there is no ap­par­ent rea­son for this or any logic per­mit­ting it. But the rea­son why as many peo­ple have benefited as they have—the rea­son why there was more than just one more blog post de­scribing a trick that seemed to work for one per­son and didn’t work for any­one else—is that Roberts knew the ex­per­i­men­tal sci­ence that let him in­ter­pret what he was see­ing, in terms of deep fac­tors that ac­tu­ally did ex­ist.

One of the pieces of ad­vice on OB/​LW that was fre­quently cited as the most im­por­tant thing learned, was the idea of “the bot­tom line”—that once a con­clu­sion is writ­ten in your mind, it is already true or already false, already wise or already stupid, and no amount of later ar­gu­ment can change that ex­cept by chang­ing the con­clu­sion. And this ties di­rectly into an­other oft-cited most im­por­tant thing, which is the idea of “en­g­ines of cog­ni­tion”, minds as map­ping en­g­ines that re­quire ev­i­dence as fuel.

If I had merely writ­ten one more blog post that said, “You know, you re­ally should be more open to chang­ing your mind—it’s pretty im­por­tant—and oh yes, you should pay at­ten­tion to the ev­i­dence too.” And this would not have been as use­ful. Not just be­cause it was less per­sua­sive, but be­cause the ac­tual op­er­a­tions would have been much less clear with­out the ex­plicit the­ory back­ing it up. What con­sti­tutes ev­i­dence, for ex­am­ple? Is it any­thing that seems like a force­ful ar­gu­ment? Hav­ing an ex­plicit prob­a­bil­ity the­ory and an ex­plicit causal ac­count of what makes rea­son­ing effec­tive, makes a large differ­ence in the force­ful­ness and im­ple­men­ta­tional de­tails of the old ad­vice to “Keep an open mind and pay at­ten­tion to the ev­i­dence.”

It is also im­por­tant to re­al­ize that causal the­o­ries are much more likely to be true when they are picked up from a sci­ence text­book than when in­vented on the fly—it is very easy to in­vent cog­ni­tive struc­tures that look like causal the­o­ries but are not even an­ti­ci­pa­tion-con­trol­ling, let alone true.

This is the sig­na­ture style I want to con­vey from all those posts that en­tan­gled cog­ni­tive sci­ence ex­per­i­ments and prob­a­bil­ity the­ory and episte­mol­ogy with the prac­ti­cal ad­vice—that prac­ti­cal ad­vice ac­tu­ally be­comes prac­ti­cally more pow­er­ful if you go out and read up on cog­ni­tive sci­ence ex­per­i­ments, or prob­a­bil­ity the­ory, or even ma­te­ri­al­ist episte­mol­ogy, and re­al­ize what you’re see­ing. This is the brand that can dis­t­in­guish LW from ten thou­sand other blogs pur­port­ing to offer ad­vice.

I could tell you, “You know, how much you’re satis­fied with your food prob­a­bly de­pends more on the qual­ity of the food than on how much of it you eat.” And you would read it and for­get about it, and the im­pulse to finish off a whole plate would still feel just as strong. But if I tell you about scope in­sen­si­tivity, and du­ra­tion ne­glect and the Peak/​End rule, you are sud­denly aware in a very con­crete way, look­ing at your plate, that you will form al­most ex­actly the same ret­ro­spec­tive mem­ory whether your por­tion size is large or small; you now pos­sess a deep the­ory about the rules gov­ern­ing your mem­ory, and you know that this is what the rules say. (You also know to save the dessert for last.)

I want to hear how I can over­come akra­sia—how I can have more willpower, or get more done with less men­tal pain. But there are ten thou­sand peo­ple pur­port­ing to give ad­vice on this, and for the most part, it is on the level of that al­ter­nate Seth Roberts who just tells peo­ple about the amaz­ing effects of drink­ing fruit juice. Or ac­tu­ally, some­what worse than that—it’s peo­ple try­ing to de­scribe in­ter­nal men­tal lev­ers that they pul­led, for which there are no stan­dard words, and which they do not ac­tu­ally know how to point to. See also the illu­sion of trans­parency, in­fer­en­tial dis­tance, and dou­ble illu­sion of trans­parency. (No­tice how “You over­es­ti­mate how much you’re ex­plain­ing and your listen­ers over­es­ti­mate how much they’re hear­ing” be­comes much more force­ful as ad­vice, af­ter I back it up with a cog­ni­tive sci­ence ex­per­i­ment and some evolu­tion­ary psy­chol­ogy?)

I think that the ad­vice I need is from some­one who reads up on a whole lot of ex­per­i­men­tal psy­chol­ogy deal­ing with willpower, men­tal con­flicts, ego de­ple­tion, prefer­ence re­ver­sals, hy­per­bolic dis­count­ing, the break­down of the self, pi­coeco­nomics, etcetera, and who, in the pro­cess of over­com­ing their own akra­sia, man­ages to un­der­stand what they did in truly gen­eral terms—thanks to ex­per­i­ments that give them a vo­cab­u­lary of cog­ni­tive phe­nom­ena that ac­tu­ally ex­ist, as op­posed to phe­nom­ena they just made up. And more­over, some­one who can ex­plain what they did to some­one else, thanks again to the ex­per­i­men­tal and the­o­ret­i­cal vo­cab­u­lary that lets them point to repli­ca­ble ex­per­i­ments that ground the ideas in very con­crete re­sults, or math­e­mat­i­cally clear ideas.

Note the grade of in­creas­ing difficulty in cit­ing:

  • Con­crete ex­per­i­men­tal re­sults (for which one need merely con­sult a pa­per, hope­fully one that re­ported p < 0.01 be­cause p < 0.05 may fail to repli­cate)
  • Causal ac­counts that are ac­tu­ally true (which may be most re­li­ably ob­tained by look­ing for the the­o­ries that are used by a ma­jor­ity within a given sci­ence)

  • Math val­idly in­ter­preted (on which I have trou­ble offer­ing use­ful ad­vice be­cause so much of my own math tal­ent is in­tu­ition that kicks in be­fore I get a chance to de­liber­ate)

If you don’t know who to trust, or you don’t trust your­self, you should con­cen­trate on ex­per­i­men­tal re­sults to start with, move on to think­ing in terms of causal the­o­ries that are widely used within a sci­ence, and dip your toes into math and episte­mol­ogy with ex­treme cau­tion.

But prac­ti­cal ad­vice re­ally, re­ally does be­come a lot more pow­er­ful when it’s backed up by con­crete ex­per­i­men­tal re­sults, causal ac­counts that are ac­tu­ally true, and math val­idly in­ter­preted.