# Probabilistic decision-making as an anxiety-reduction technique

One prob­lem I run into when mak­ing de­ci­sions that im­pact my life sub­stan­tially (say, mak­ing a big pur­chase) is that I spend a lot of time star­ing at de­ci­sion be­com­ing in­creas­ingly anx­ious. Usu­ally, this is ac­com­panied by only very minor amounts of in­for­ma­tion be­ing ac­quired.

There is a well-known solu­tion to this: flip a coin.

Flip­ping a coin is nice in that it al­lows you to sub­stan­tially com­press the amount of time dur­ing which you are anx­ious. How­ever, it is not with­out its flaws.

1. Flip­ping a coin does not al­low you choose among more than two op­tions. That’s of­ten enough, but more of­ten than not, it’s not enough.

2. Flip­ping a coin does not al­low you to in­clude any of the in­for­ma­tion that is per­ti­nent to the de­ci­sion you are mak­ing.

So while flip­ping a coin is nice be­cause it al­lows you to skip lengthy pointless fret­ting, it also does not re­ally lead to an op­ti­mal out­come. Now of course, if you’re at that point, you don’t already know the op­ti­mal out­come. Other­wise you’d just do that. Right? If you know what you’re go­ing to do at the end of the fret­ting, just do it. The prob­lem is that you don’t know.

So I’m propos­ing the fol­low­ing ap­proach:

1. Es­ti­mate the prob­a­bil­ity that you will end up se­lect­ing each of the op­tions.

2. Pick one of the out­comes ran­domly in ac­cor­dance with the prob­a­bil­ity you as­signed to it.

e.g. I’m try­ing to pur­chase pants. Either black or navy. They cost a few hun­dreds of dol­lars so it is an anx­iety-gen­er­at­ing pur­chase. I’ve an­a­lyzed the pros-and-cons of the two pairs of pants and at this point, I’m just wor­ry­ing about whether or not I’m mak­ing the best choice.

Step 1. I as­sign a 0.4 prob­a­bil­ity to get­ting the black pants. A 0.3 prob­a­bil­ity that I will get the navy pants. And a 0.3 prob­a­bil­ity that I will aban­don the pur­chase.

Step 2. I com­pute the in­ter­vals: 0-0.4 is black, 0.4-0.7 is navy, 0.7-1.0 is no pur­chase.

Step 3. I gen­er­ate a ran­dom num­ber. I just search Google for “ran­dom num­ber be­tween 1 and 100” and di­vide the re­sult by 100.

Step 4. Com­pare the re­sult to the in­ter­vals to figure out what I just de­cided.

Step 5. Im­ple­ment the de­ci­sion.

I figure I can’t be the first per­son to have thought of this. If you know of prior art on this, let me know.

No nominations.
No reviews.
• I’m con­fused; why don’t you just pick black with prob­a­bil­ity 100%? (As­sum­ing that your util­ity is 1 if you make the on-fur­ther-re­flec­tion-cor­rect choice, 0 else.)

This isn’t the same thing as know­ing what you’re go­ing to think at the end of your fret­ting—you still don’t know—but the cor­rect re­sponse to un­cer­tainty is not half speed, and just pick­ing the 40%-to-be-right op­tion all of the time is the ex­pec­tancy-max­i­miz­ing re­sponse to un­cer­tainty.

• I also use a sim­ple ver­sion of this, with a key ex­tra step at the end:

1) have a de­ci­sion you are un­sure about. 2) perform ran­domi­sa­tion (I usu­ally just use a coin). 3) no­tice how the out­come makes you feel. If you find that you wish the coin landed the other way, over­ride the de­ci­sion and do what you se­cretly wanted to do all along.

You might think the third step defeats the pur­pose of the ex­er­cise, but so long as you ac­tu­ally com­mit to fol­low­ing the ran­domi­sa­tion most of the time, it gives you di­rect ac­cess to very use­ful in­for­ma­tion. It also sets up the right in­cen­tive, wherein you never re­ally need to work your willpower against your de­sires (ex­cept, I guess, the de­sire to de­liber­ate more).

I mostly use this for a slightly differ­ent use case – in­con­se­quen­tial de­ci­sions like where to eat or small pur­chases, where tak­ing a lot of time to op­ti­mise isn’t worth it. Your mileage may vary with more im­por­tant de­ci­sions, but I see no rea­son in prin­ci­ple this couldn’t work.

• Never leave the house with­out your d20 :-P

But I agree with you. This seems a sim­ple way to do some­thing like satis­fic­ing. Avoid­ing the great com­pu­ta­tional cost of an op­ti­mal de­ci­sion.

In terms of prior art that is prob­a­bly the field you want to ex­plore: https://​​en.m.wikipe­dia.org/​​wiki/​​Satisficing

• Check out the dice­man. Things es­ca­lated quickly...