London LW CoZE exercise report

(Cross-posted to my shiny new blog.)

Hu­man brains are bad at eval­u­at­ing con­se­quences. Some­times we want to do some­thing, and log­i­cally we’re pretty sure we won’t die or any­thing, but our lizard hind­brains are scream­ing at us to flee. Com­fort Zone Ex­pan­sion (CoZE) is an ex­er­cise that CFAR teaches to get our lizard hind­brains to ac­cept that what we’re do­ing is ac­tu­ally pretty safe.

Roughly it in­volves two steps. One: do some­thing that makes our lizard hind­brains get pretty antsy. Two: don’t get eaten as a re­sult.

I or­ganised a CoZE ex­er­cise for LessWrong Lon­don on Sept­meber 1st. We had a to­tal of eight par­ti­ci­pants, I think I was the only one who’d done any struc­tured CoZE be­fore.

My plan was: we meet at 11am, have a dis­cus­sion about CoZE, play some im­prov games to warm up, and then head to the nearby mall for 1.30 pm. In re­al­ity, the dis­cus­sion started closer to 12pm, with some peo­ple show­ing up part way through or not un­til it was finished.

After finish­ing the dis­cus­sion, we didn’t end up do­ing any im­prov games. We also be­came slightly di­s­or­ganised; we agreed on a meet­ing time an hour and a half in the fu­ture, but then our group didn’t re­ally split up un­til about twenty min­utes af­ter that. I could have han­dled this bet­ter. We got dis­tracted by the need for lunch, which I could have made spe­cific plans for. (Ideally I would have started the dis­cus­sion af­ter lunch, but shops close early on Sun­day.)


My solo ac­tivi­ties went con­sid­er­ably less well than I’d ex­pected. My first thing was to ask a ven­dor in the walk­way for some free choco­late, which an­noyed her more than I’d ex­pected. Maybe she gets asked that a lot? It was kind of dis­cour­ag­ing.

After that I wanted to go into a per­fume shop and ask for help with scent, be­cause I don’t know any­thing about it. I wan­dered past the front a cou­ple of times, de­cid­ing to go when the shop was nearly empty, but then when that hap­pened I still chick­ened out. That, too, was kind of dis­cour­ag­ing.

Then I de­cided to get back in state by do­ing some­thing that seemed easy: mak­ing eye con­tact with peo­ple and smil­ing. It turns out that “mak­ing eye-con­tact” is a two-player game, and no­body else was play­ing. After some min­utes of that I just gave up for the day.

In my defense: I had a cold that day and was feel­ing a bit shitty (while wan­der­ing near the per­fume shop I got a nose-bleed, and had to di­vert to the bath­room tem­porar­ily), and that might have drained my en­ergy. I did also get a few minor vic­to­ries in. The most no­table is that I did some pull-ups on some scaf­fold­ing out­side, and some­one walk­ing past said some­thing en­courag­ing like “awe­some”. (I’d like to do these more of­ten, but the first time I tried there was ick on the scaf­fold­ing. There wasn’t this time, so I should col­lect more data.)

[[I spoke with Critch from CFAR a few days af­ter­wards, and he gave me a new per­spec­tive: if I go in ex­pect­ing peo­ple to re­spond well to me, then when they don’t, that’s go­ing to bother me. If I go in ex­pect­ing to an­noy peo­ple, but re­mem­ber­ing that an­noy­ing peo­ple, while bad, doesn’t cor­re­spond to any se­ri­ous con­se­quences, then it’s go­ing to be eas­ier to han­dle. For any given in­ter­ac­tion, I should be try­ing to make it go well, but I should choose the in­ter­ac­tions such that they won’t all go well. (Anal­ogy: in a game of Go, the stronger player might give a hand­i­cap to the weaker player, but once play starts they’ll do their best to win.)

He also gave me a po­ten­tial way to avoid chick­en­ing out: if I imag­ine my­self do­ing some­thing, and then I try to do it and it turns out to be scary, then that feels like new in­for­ma­tion and a rea­son to ac­tu­ally not do it. If I imag­ine my­self do­ing some­thing, *be­ing scared and do­ing it any­way*, then when it turns out to be scary, that no longer counts as an ex­cuse. I haven’t had a chance to try this yet.]]

Other peo­ple had more suc­cess. We’d primed our­selves by talk­ing about star­ing con­tests a lot pre­vi­ously, so a few peo­ple asked strangers for those. I think only one stranger ac­cepted. Try­ing to get high-fives was also com­mon; one per­son ob­served that he some­times does that any­way, and has a much higher suc­cess rate than he did in the mall. One per­son went into a high-end lin­gerie store and asked what he could buy on a bud­get of £20 (an­swer: noth­ing). And of course there were sev­eral other suc­cesses that I’ve for­got­ten. I got the im­pres­sion that most peo­ple did bet­ter than me.

There was in­ter­est in do­ing this again. At the time I was hes­i­tant but re­al­ised that I would prob­a­bly be­come less hes­i­tant with time. I’ve now reached a point where I, too, would quite like to do it again. We haven’t got any spe­cific plans yet.

Things to take away:

  • Have some well-defined way to tran­si­tion from “about-to-start” to “started”.

  • Hav­ing an au­di­ence makes some things much eas­ier. This is po­ten­tially a way to es­ca­late difficulty slowly.

  • When I did CoZE at the CFAR work­shop I was hang­ing out with some­one, who en­couraged me to ac­tu­ally do things as well as pro­vided an au­di­ence. At the time I won­dered whether the fact that we were to­gether meant I did less, but I did far more with her than by my­self.

  • We didn’t do any­thing be­fore­hand to get in state, like im­prov games. I can’t say whether this would have helped, but it seems like it might have done.

  • We had a non­par­ti­ci­pant vol­un­teer to be a meet­ing point if par­ti­ci­pants wanted to finish early. If she hadn’t been there, I might not have quit twenty min­utes early, but given that I did, it was nice to be able to hang out. This might be a good way to en­able longer ses­sions.

An objection

Dur­ing the dis­cus­sion about ethics, some­one brought up an ob­jec­tion, which I think cashes out as: there are good places to ex­pand our com­fort zones into, there are bad places, and there are lame places. A lot of the stuff on the recom­mended list of ac­tivi­ties is lame (do we re­ally need to be bet­ter at ask­ing peo­ple for star­ing con­tests?), and it’s not clear how much it gen­er­al­ises to good stuff. Un­der the novice-driver line of thought, both­er­ing peo­ple is an ac­cept­able cost of CoZE; but if we’re us­ing CoZE for lame things, the benefits be­come small, and maybe it’s no longer worth the cost.

I guess this boils down to a ques­tion of how much the lame stuff gen­er­al­ises. I’m op­ti­mistic; for ex­am­ple, it seems to me that a lot of the lame stuff is go­ing to be over­com­ing a feel­ing of “I don’t want to bother this per­son”, which is also pre­sent in the good stuff, so that par­tic­u­lar fea­ture should gen­er­al­ise. (It may also be that the lame stuff is dom­i­nated by the good stuff, so there’s no rea­son to ever prac­tice any­thing lame; but that seems a suffi­ciently com­pli­cated hy­poth­e­sis to be low prior.)

(There’s also a ques­tion of whether or not peo­ple are ac­tu­ally both­ered by strangers ask­ing for star­ing con­tests. My ini­tial as­sump­tion was not, but af­ter do­ing the ex­er­cise, I’m not sure.)