# RickJS comments on So You Think You’re a Bayesian? The Natural Mode of Probabilistic Reasoning

• Thanks, Matt!

That’s a nice ed­u­ca­tional post.

I want to pick a nit, not with you, but with Gigeren­zer and ” … the con­junc­tion fal­lacy can be miti­gated by chang­ing the word­ing of the ques­tion … ” Un­for­tu­nately, in real life, the prob­lems come at you the way they do, and you need to learn to deal with it.

I say that ra­tio­nal think­ing looks like this: pen­cil ap­plied to pa­per. Or a spread­sheet or other de­ci­sion sup­port pro­gram in use. We can’t do this stuff in our heads. At least I can’t. Evolu­tion didn’t de­liver ar­ith­metic, much less ra­tio­nal­ity. We teach ar­ith­metic to kids, slowly and painstak­ingly. We had bet­ter start teach­ing them ra­tio­nal­ity. Slowly and painstak­ingly, not like a 1-hour also-men­tioned.

And, since I have my spread­sheet pro­gram open, I will in­deed con­vert prob­a­bil­ities into fre­quen­cies and look at the world both ways, so my au­to­matic pro­ces­sors can par­ti­ci­pate. But, I only trust the an­swers on the screen. My brain lies to me too of­ten.

Once again, thanks Matt. Well done!

• I thought that a ma­jor point of heuris­tics and bi­ases pro­gram, at least for eco­nomics, was that they were sys­tem­atic and in a sense “baked-in” as de­fault. If these er­rors are ar­ti­facts of tweaks/​word­ing then that re­ally un­der­mines hope of the­o­ret­i­cal ex­ten­sion. The value of this kind of knowl­edge be­comes lop­sided to­wards mar­keters, ma­gi­ci­ans, and oth­ers try­ing to ma­nipu­late or trick peo­ple more effec­tively.

On the other hand, I think the idea of us­ing the er­ror data as clues as to neu­ral ar­chi­tec­ture and func­tion­ing is great! It seems that neu­ro­science-clus­tered re­search is fo­cused mostly bot­tom-up and rarely takes in­spira­tion from the other di­rec­tion.

We can’t do this stuff in our heads. At least I can’t. Evolu­tion didn’t de­liver ar­ith­metic, much less ra­tio­nal­ity. We teach ar­ith­metic to kids, slowly and painstak­ingly. We had bet­ter start teach­ing them ra­tio­nal­ity. Slowly and painstak­ingly, not like a 1-hour also-men­tioned.

This raises an in­ter­est­ing point. We can do ar­ith­metic in our heads, some of us more spec­tac­u­larly than oth­ers. Do you mean to say that there is no way to em­ploy/​train our brains to do ra­tio­nal think­ing more effec­tively and in­tu­itively?

I had always hoped that we could at least shape our in­tu­ition enough to give us a sense for situ­a­tions where it would be bet­ter to calcu­late—though it’s costly and slower. We do not always have our tools (al­though I guess in the fu­ture this is less and less likely).

• “Do you mean to say that there is no way to em­ploy/​train our brains to do ra­tio­nal think­ing more effec­tively and in­tu­itively?”

I don’t don’t know whether RickJS meant to say that or not. But this blog post sug­gests to me a way for­ward: when­ever con­fronted with ques­tions about like­li­hood or prob­a­bil­ity, con­sciously step back and as­sess whether a fre­quen­tist anal­y­sis is pos­si­ble. Use that ap­proach if it is. If not, shift to­ward Bayesian views. But in ei­ther case, also ask: can I re­ally com­pute this ac­cu­rately, or is it too com­plex? Some things you can do well enough in your head, es­pe­cially when perfect ac­cu­racy isn’t nec­es­sary (or even pos­si­ble). Some things you can’t.

Maybe if you started kids in their ju­nior year in high school, they might be pretty skil­led at tel­ling which was which (of the four pos­si­bil­ities in­her­ent in what I out­line above) by the end of their se­nior year.