The Substitution Principle

Par­tial re-in­ter­pre­ta­tion of: The Curse of Iden­ti­ty
Also re­lated to: Hu­mans Are Not Au­to­mat­i­cally Strate­gic, The Affect Heuris­tic, The Plan­ning Fal­lacy, The Availa­bil­ity Heuris­tic, The Con­junc­tion Fal­lacy, Urges vs. Goals, Your In­ner Google, sig­nal­ing, etc...

What are the best ca­reers for mak­ing a lot of money?

Maybe you’ve thought about this ques­tion a lot, and have re­searched it enough to have a well-formed opinion. But the chances are that even if you hadn’t, some sort of an an­swer popped into your mind right away. Doc­tors make a lot of money, maybe, or lawyers, or bankers. Rock stars, per­haps.

You prob­a­bly re­al­ize that this is a difficult ques­tion. For one, there’s the ques­tion of who we’re talk­ing about. One per­son’s strengths and weak­nesses might make them more suited for a par­tic­u­lar ca­reer path, while for an­other per­son, an­other ca­reer is bet­ter. Se­cond, the ques­tion is not clearly defined. Is a ca­reer with a small chance of mak­ing it rich and a large chance of re­main­ing poor a bet­ter op­tion than a ca­reer with a large chance of be­com­ing wealthy but no chance of be­com­ing rich? Third, who­ever is ask­ing this ques­tion prob­a­bly does so be­cause they are think­ing about what to do with their lives. So you prob­a­bly don’t want to an­swer on the ba­sis of what ca­reer lets you make a lot of money to­day, but on the ba­sis of which one will do so in the near fu­ture. That re­quires tricky tech­nolog­i­cal and so­cial fore­cast­ing, which is quite difficult. And so on.

Yet, de­spite all of these un­cer­tain­ties, some sort of an an­swer prob­a­bly came to your mind as soon as you heard the ques­tion. And if you hadn’t con­sid­ered the ques­tion be­fore, your an­swer prob­a­bly didn’t take any of the above com­pli­ca­tions into ac­count. It’s as if your brain, while gen­er­at­ing an an­swer, never even con­sid­ered them.

The thing is, it prob­a­bly didn’t.

Daniel Kah­ne­man, in Think­ing, Fast and Slow, ex­ten­sively dis­cusses what I call the Sub­sti­tu­tion Prin­ci­ple:

If a satis­fac­tory an­swer to a hard ques­tion is not found quickly, Sys­tem 1 will find a re­lated ques­tion that is eas­ier and will an­swer it. (Kah­ne­man, p. 97)

Sys­tem 1, if you re­call, is the quick, dirty and par­allel part of our brains that ren­ders in­stant judge­ments, with­out think­ing about them in too much de­tail. In this case, the ac­tual ques­tion that was asked was ”what are the best ca­reers for mak­ing a lot of money”. The ques­tion that was ac­tu­ally an­swered was ”what ca­reers have I come to as­so­ci­ate with wealth”.

Here are some other ex­am­ples of sub­sti­tu­tion that Kah­ne­man gives:

  • How much would you con­tribute to save an en­dan­gered species? be­comes How much emo­tion do I feel when I think of dy­ing dolphins?

  • How happy are you with your life these days? be­comes What is my mood right now?

  • How pop­u­lar will the pres­i­dent be six months from now? be­comes How pop­u­lar is the pres­i­dent right now?

  • How should fi­nan­cial ad­vi­sors who prey on the el­derly be pun­ished? be­comes How much anger do I feel when I think of fi­nan­cial preda­tors?

    All things con­sid­ered, this heuris­tic prob­a­bly works pretty well most of the time. The eas­ier ques­tions are not mean­ingless: while not com­pletely ac­cu­rate, their an­swers are still gen­er­ally cor­re­lated with the cor­rect an­swer. And a lot of the time, that’s good enough.

    But I think that the Sub­sti­tu­tion Prin­ci­ple is also the mechanism by which most of our bi­ases work. In The Curse of Iden­tity, I wrote:

    In each case, I thought I was work­ing for a par­tic­u­lar goal (be­come ca­pa­ble of do­ing use­ful Sin­gu­lar­ity work, ad­vance the cause of a poli­ti­cal party, do use­ful Sin­gu­lar­ity work). But as soon as I set that goal, my brain au­to­mat­i­cally and in­visi­bly re-in­ter­preted it as the goal of do­ing some­thing that gave the im­pres­sion of do­ing pres­ti­gious work for a cause (spend­ing all my wak­ing time work­ing, be­ing the spokesman of a poli­ti­cal party, writ­ing pa­pers or do­ing some­thing else few oth­ers could do).

    As Anna cor­rectly pointed out, I re­sorted to a sig­nal­ing ex­pla­na­tion here, but a sig­nal­ing ex­pla­na­tion may not be nec­es­sary. Let me re­word that pre­vi­ous gen­er­al­iza­tion: As soon as I set a goal, my brain asked it­self how that goal might be achieved, re­al­ized that this was a difficult ques­tion, and sub­sti­tuted it with an eas­ier one. So ”how could I ad­vance X” be­came ”what are the kinds of be­hav­iors that are com­monly as­so­ci­ated with ad­vanc­ing X”. That my brain hap­pened to pick the most pres­ti­gious ways of ad­vanc­ing X might be sim­ply be­cause pres­tige is of­ten cor­re­lated with achiev­ing a lot.

    Does this ex­clude the sig­nal­ing ex­pla­na­tion? Of course not. My be­hav­ior is prob­a­bly still driven by sig­nal­ing and sta­tus con­cerns. One of the mechanisms by which this works might be that such con­sid­er­a­tions get dis­pro­por­tionately taken into ac­count when choos­ing a heuris­tic ques­tion. And a lot of the ex­am­ples I gave in The Curse of Iden­tity seem hard to jus­tify with­out a sig­nal­ing ex­pla­na­tion. But sig­nal­ing need not to be the sole ex­pla­na­tion. Our brains may just re­sort to poor heuris­tics a lot.

    Some other bi­ases and how the Sub­sti­tu­tion Prin­ci­ple is re­lated to them (many of these are again bor­rowed from Think­ing, Fast and Slow):

    The Plan­ning Fal­lacy: ”How much time will this take” be­comes some­thing like ”How much time did it take for me to get this far, and many times should that be mul­ti­plied to get to com­ple­tion.” (Doesn’t take into ac­count un­ex­pected de­lays and in­ter­rup­tions, wan­ing in­ter­est, etc.)

    The Availa­bil­ity Heuris­tic: ”How com­mon is this thing” or ”how fre­quently does this hap­pen” be­comes ”how eas­ily do in­stances of this come to mind”.

    Over-es­ti­mat­ing your own share of house­hold chores: ”What frac­tion of chores have I done” be­comes ”how many chores do I re­mem­ber do­ing, as com­pared to the amount of chores I re­mem­ber my part­ner do­ing.” (You will nat­u­rally re­mem­ber more of the things that you’ve done than that some­body else has done, pos­si­bly when you weren’t even around.)

    Be­ing in an emo­tion­ally ”cool” state and over-es­ti­mat­ing your de­gree of con­trol in an emo­tion­ally ”hot” state (an­gry, hun­gry, sex­u­ally aroused, etc.): ”How well could I re­sist do­ing X in that state” be­comes ”how easy does re­sist­ing X feel like now”.

    The Con­junc­tion Fal­lacy: ”What’s the prob­a­bil­ity that Linda is a fem­i­nist” be­comes ”how rep­re­sen­ta­tive is Linda of my con­cep­tion of fem­i­nists”.

    Peo­ple vot­ing for poli­ti­ci­ans for seem­ingly ir­rele­vant rea­sons: ”How well would this per­son do his job as a poli­ti­cian” be­comes ”how much do I like this per­son.” (A bet­ter heuris­tic than you might think, con­sid­er­ing that we like peo­ple who like us, owe us fa­vors, re­sem­ble us, etc. - in the an­ces­tral en­vi­ron­ment, sup­port­ing the leader you liked the most was prob­a­bly a pretty good proxy for sup­port­ing the leader who was most likely to aid you in re­turn.)

    And so on.

    The im­por­tant point is to learn to rec­og­nize the situ­a­tions where you’re con­fronting a difficult prob­lem, and your mind gives you an an­swer right away. If you don’t have ex­ten­sive ex­per­tise with the prob­lem – or even if you do – it’s likely that the an­swer you got wasn’t ac­tu­ally the an­swer to the ques­tion you asked. So be­fore you act, stop to con­sider what heuris­tic ques­tion your brain might ac­tu­ally have used, and whether it makes sense given the situ­a­tion that you’re think­ing about.

    This in­volves three skills: first rec­og­niz­ing a prob­lem as a difficult one, then figur­ing out what heuris­tic you might have used, and fi­nally com­ing up with a bet­ter solu­tion. I in­tend to de­velop some­thing on how to task­ify those skills, but if you have any ideas for how that might be achieved, let’s hear them.