Outside the Laboratory

“Out­side the lab­o­ra­tory, sci­en­tists are no wiser than any­one else.” Some­times this proverb is spo­ken by sci­en­tists, humbly, sadly, to re­mind them­selves of their own fal­li­bil­ity. Some­times this proverb is said for rather less praise­wor­thy rea­sons, to de­value un­wanted ex­pert ad­vice. Is the proverb true? Prob­a­bly not in an ab­solute sense. It seems much too pes­simistic to say that sci­en­tists are liter­ally no wiser than av­er­age, that there is liter­ally zero cor­re­la­tion.

But the proverb does ap­pear true to some de­gree, and I pro­pose that we should be very dis­turbed by this fact. We should not sigh, and shake our heads sadly. Rather we should sit bolt up­right in alarm. Why? Well, sup­pose that an ap­pren­tice shep­herd is la­bo­ri­ously trained to count sheep, as they pass in and out of a fold. Thus the shep­herd knows when all the sheep have left, and when all the sheep have re­turned. Then you give the shep­herd a few ap­ples, and say: “How many ap­ples?” But the shep­herd stares at you blankly, be­cause they weren’t trained to count ap­ples—just sheep. You would prob­a­bly sus­pect that the shep­herd didn’t un­der­stand count­ing very well.

Now sup­pose we dis­cover that a Ph.D. economist buys a lot­tery ticket ev­ery week. We have to ask our­selves: Does this per­son re­ally un­der­stand ex­pected util­ity, on a gut level? Or have they just been trained to perform cer­tain alge­bra tricks?

One thinks of Richard Feyn­man’s ac­count of a failing physics ed­u­ca­tion pro­gram:

“The stu­dents had mem­o­rized ev­ery­thing, but they didn’t know what any­thing meant. When they heard ‘light that is re­flected from a medium with an in­dex’, they didn’t know that it meant a ma­te­rial such as wa­ter. They didn’t know that the ‘di­rec­tion of the light’ is the di­rec­tion in which you see some­thing when you’re look­ing at it, and so on. Every­thing was en­tirely mem­o­rized, yet noth­ing had been trans­lated into mean­ingful words. So if I asked, ‘What is Brew­ster’s An­gle?’ I’m go­ing into the com­puter with the right key­words. But if I say, ‘Look at the wa­ter,’ noth­ing hap­pens—they don’t have any­thing un­der ‘Look at the wa­ter’!”

Sup­pose we have an ap­par­ently com­pe­tent sci­en­tist, who knows how to de­sign an ex­per­i­ment on N sub­jects; the N sub­jects will re­ceive a ran­dom­ized treat­ment; blinded judges will clas­sify the sub­ject out­comes; and then we’ll run the re­sults through a com­puter and see if the re­sults are sig­nifi­cant at the 0.05 con­fi­dence level. Now this is not just a rit­u­al­ized tra­di­tion. This is not a point of ar­bi­trary eti­quette like us­ing the cor­rect fork for salad. It is a rit­u­al­ized tra­di­tion for test­ing hy­pothe­ses ex­per­i­men­tally. Why should you test your hy­poth­e­sis ex­per­i­men­tally? Be­cause you know the jour­nal will de­mand so be­fore it pub­lishes your pa­per? Be­cause you were trained to do it in col­lege? Be­cause ev­ery­one else says in uni­son that it’s im­por­tant to do the ex­per­i­ment, and they’ll look at you funny if you say oth­er­wise?

No: be­cause, in or­der to map a ter­ri­tory, you have to go out and look at the ter­ri­tory. It isn’t pos­si­ble to pro­duce an ac­cu­rate map of a city while sit­ting in your liv­ing room with your eyes closed, think­ing pleas­ant thoughts about what you wish the city was like. You have to go out, walk through the city, and write lines on pa­per that cor­re­spond to what you see. It hap­pens, in mi­ni­a­ture, ev­ery time you look down at your shoes to see if your shoelaces are un­tied. Pho­tons ar­rive from the Sun, bounce off your shoelaces, strike your retina, are trans­duced into neu­ral firing fre­quences, and are re­con­structed by your vi­sual cor­tex into an ac­ti­va­tion pat­tern that is strongly cor­re­lated with the cur­rent shape of your shoelaces. To gain new in­for­ma­tion about the ter­ri­tory, you have to in­ter­act with the ter­ri­tory. There has to be some real, phys­i­cal pro­cess whereby your brain state ends up cor­re­lated to the state of the en­vi­ron­ment. Rea­son­ing pro­cesses aren’t magic; you can give causal de­scrip­tions of how they work. Which all goes to say that, to find things out, you’ve got to go look.

Now what are we to think of a sci­en­tist who seems com­pe­tent in­side the lab­o­ra­tory, but who, out­side the lab­o­ra­tory, be­lieves in a spirit world? We ask why, and the sci­en­tist says some­thing along the lines of: “Well, no one re­ally knows, and I ad­mit that I don’t have any ev­i­dence—it’s a re­li­gious be­lief, it can’t be dis­proven one way or an­other by ob­ser­va­tion.” I can­not but con­clude that this per­son liter­ally doesn’t know why you have to look at things. They may have been taught a cer­tain rit­ual of ex­per­i­men­ta­tion, but they don’t un­der­stand the rea­son for it—that to map a ter­ri­tory, you have to look at it—that to gain in­for­ma­tion about the en­vi­ron­ment, you have to un­dergo a causal pro­cess whereby you in­ter­act with the en­vi­ron­ment and end up cor­re­lated to it. This ap­plies just as much to a dou­ble-blind ex­per­i­men­tal de­sign that gath­ers in­for­ma­tion about the effi­cacy of a new med­i­cal de­vice, as it does to your eyes gath­er­ing in­for­ma­tion about your shoelaces.

Maybe our spiritual sci­en­tist says: “But it’s not a mat­ter for ex­per­i­ment. The spirits spoke to me in my heart.” Well, if we re­ally sup­pose that spirits are speak­ing in any fash­ion what­so­ever, that is a causal in­ter­ac­tion and it counts as an ob­ser­va­tion. Prob­a­bil­ity the­ory still ap­plies. If you pro­pose that some per­sonal ex­pe­rience of “spirit voices” is ev­i­dence for ac­tual spirits, you must pro­pose that there is a fa­vor­able like­li­hood ra­tio for spirits caus­ing “spirit voices”, as com­pared to other ex­pla­na­tions for “spirit voices”, which is suffi­cient to over­come the prior im­prob­a­bil­ity of a com­plex be­lief with many parts. Failing to re­al­ize that “the spirits spoke to me in my heart” is an in­stance of “causal in­ter­ac­tion”, is analo­gous to a physics stu­dent not re­al­iz­ing that a “medium with an in­dex” means a ma­te­rial such as wa­ter.

It is easy to be fooled, per­haps, by the fact that peo­ple wear­ing lab coats use the phrase “causal in­ter­ac­tion” and that peo­ple wear­ing gaudy jew­elry use the phrase “spirits speak­ing”. Dis­cus­sants wear­ing differ­ent cloth­ing, as we all know, de­mar­cate in­de­pen­dent spheres of ex­is­tence—“sep­a­rate mag­is­te­ria”, in Stephen J. Gould’s im­mor­tal blun­der of a phrase. Ac­tu­ally, “causal in­ter­ac­tion” is just a fancy way of say­ing, “Some­thing that makes some­thing else hap­pen”, and prob­a­bil­ity the­ory doesn’t care what clothes you wear.

In mod­ern so­ciety there is a preva­lent no­tion that spiritual mat­ters can’t be set­tled by logic or ob­ser­va­tion, and there­fore you can have what­ever re­li­gious be­liefs you like. If a sci­en­tist falls for this, and de­cides to live their ex­tral­ab­o­ra­to­rial life ac­cord­ingly, then this, to me, says that they only un­der­stand the ex­per­i­men­tal prin­ci­ple as a so­cial con­ven­tion. They know when they are ex­pected to do ex­per­i­ments and test the re­sults for statis­ti­cal sig­nifi­cance. But put them in a con­text where it is so­cially con­ven­tional to make up wacky be­liefs with­out look­ing, and they just as hap­pily do that in­stead.

The ap­pren­tice shep­herd is told that if “seven” sheep go out, and “eight” sheep go out, then “fif­teen” sheep had bet­ter come back in. Why “fif­teen” in­stead of “four­teen” or “three”? Be­cause oth­er­wise you’ll get no din­ner tonight, that’s why! So that’s pro­fes­sional train­ing of a kind, and it works af­ter a fash­ion—but if so­cial con­ven­tion is the only rea­son why seven sheep plus eight sheep equals fif­teen sheep, then maybe seven ap­ples plus eight ap­ples equals three ap­ples. Who’s to say that the rules shouldn’t be differ­ent for ap­ples?

But if you know why the rules work, you can see that ad­di­tion is the same for sheep and for ap­ples. Isaac New­ton is justly revered, not for his out­dated the­ory of grav­ity, but for dis­cov­er­ing that—amaz­ingly, sur­pris­ingly—the ce­les­tial planets, in the glo­ri­ous heav­ens, obeyed just the same rules as fal­ling ap­ples. In the macro­scopic world—the ev­ery­day an­ces­tral en­vi­ron­ment—differ­ent trees bear differ­ent fruits, differ­ent cus­toms hold for differ­ent peo­ple at differ­ent times. A gen­uinely unified uni­verse, with sta­tion­ary uni­ver­sal laws, is a highly coun­ter­in­tu­itive no­tion to hu­mans! It is only sci­en­tists who re­ally be­lieve it, though some re­li­gions may talk a good game about the “unity of all things”.

As Richard Feyn­man put it:

“If we look at a glass closely enough we see the en­tire uni­verse. There are the things of physics: the twist­ing liquid which evap­o­rates de­pend­ing on the wind and weather, the re­flec­tions in the glass, and our imag­i­na­tions adds the atoms. The glass is a dis­til­la­tion of the Earth’s rocks, and in its com­po­si­tion we see the se­cret of the uni­verse’s age, and the evolu­tion of the stars. What strange ar­ray of chem­i­cals are there in the wine? How did they come to be? There are the fer­ments, the en­zymes, the sub­strates, and the prod­ucts. There in wine is found the great gen­er­al­iza­tion: all life is fer­men­ta­tion. No­body can dis­cover the chem­istry of wine with­out dis­cov­er­ing, as did Louis Pas­teur, the cause of much dis­ease. How vivid is the claret, press­ing its ex­is­tence into the con­scious­ness that watches it! If our small minds, for some con­ve­nience, di­vide this glass of wine, this uni­verse, into parts — physics, biol­ogy, ge­ol­ogy, as­tron­omy, psy­chol­ogy, and so on — re­mem­ber that Na­ture does not know it! So let us put it all back to­gether, not for­get­ting ul­ti­mately what it is for. Let it give us one more fi­nal plea­sure: drink it and for­get it all!”

A few re­li­gions, es­pe­cially the ones in­vented or re­fur­bished af­ter Isaac New­ton, may pro­fess that “ev­ery­thing is con­nected to ev­ery­thing else”. (Since there is a triv­ial iso­mor­phism be­tween graphs and their com­ple­ments, this profound wis­dom con­veys ex­actly the same use­ful in­for­ma­tion as a graph with no edges.) But when it comes to the ac­tual meat of the re­li­gion, prophets and priests fol­low the an­cient hu­man prac­tice of mak­ing ev­ery­thing up as they go along. And they make up one rule for women un­der twelve, an­other rule for men over thir­teen; one rule for the Sab­bath and an­other rule for week­days; one rule for sci­ence and an­other rule for sor­cery...

Real­ity, we have learned to our shock, is not a col­lec­tion of sep­a­rate mag­is­te­ria, but a sin­gle unified pro­cess gov­erned by math­e­mat­i­cally sim­ple low-level rules. Differ­ent build­ings on a uni­ver­sity cam­pus do not be­long to differ­ent uni­verses, though it may some­times seem that way. The uni­verse is not di­vided into mind and mat­ter, or life and non­life; the atoms in our heads in­ter­act seam­lessly with the atoms of the sur­round­ing air. Nor is Bayes’s The­o­rem differ­ent from one place to an­other.

If, out­side of their spe­cial­ist field, some par­tic­u­lar sci­en­tist is just as sus­cep­ti­ble as any­one else to wacky ideas, then they prob­a­bly never did un­der­stand why the sci­en­tific rules work. Maybe they can par­rot back a bit of Pop­pe­rian falsifi­ca­tion­ism; but they don’t un­der­stand on a deep level, the alge­braic level of prob­a­bil­ity the­ory, the causal level of cog­ni­tion-as-ma­chin­ery. They’ve been trained to be­have a cer­tain way in the lab­o­ra­tory, but they don’t like to be con­strained by ev­i­dence; when they go home, they take off the lab coat and re­lax with some com­fortable non­sense. And yes, that does make me won­der if I can trust that sci­en­tist’s opinions even in their own field—es­pe­cially when it comes to any con­tro­ver­sial is­sue, any open ques­tion, any­thing that isn’t already nailed down by mas­sive ev­i­dence and so­cial con­ven­tion.

Maybe we can beat the proverb—be ra­tio­nal in our per­sonal lives, not just our pro­fes­sional lives. We shouldn’t let a mere proverb stop us: “A witty say­ing proves noth­ing,” as Voltaire said. Maybe we can do bet­ter, if we study enough prob­a­bil­ity the­ory to know why the rules work, and enough ex­per­i­men­tal psy­chol­ogy to see how they ap­ply in real-world cases—if we can learn to look at the wa­ter. An am­bi­tion like that lacks the com­fortable mod­esty of be­ing able to con­fess that, out­side your spe­cialty, you’re no bet­ter than any­one else. But if our the­o­ries of ra­tio­nal­ity don’t gen­er­al­ize to ev­ery­day life, we’re do­ing some­thing wrong. It’s not a differ­ent uni­verse in­side and out­side the lab­o­ra­tory.

Ad­den­dum: If you think that (a) sci­ence is purely log­i­cal and there­fore op­posed to emo­tion, or (b) that we shouldn’t bother to seek truth in ev­ery­day life, see “Why Truth?” For new read­ers, I also recom­mend “Twelve Virtues of Ra­tion­al­ity.