The Principled Intelligence Hypothesis

Link post

I have been read­ing the thought pro­vok­ing Elephant in the Brain, and will prob­a­bly have more to say on it later. But if I un­der­stand cor­rectly, a dom­i­nant the­ory of how hu­mans came to be so smart is that they have been in an end­less cat and mouse game with them­selves, mak­ing norms and pun­ish­ing vi­o­la­tions on the one hand, and clev­erly cheat­ing their own norms and ex­cus­ing them­selves on the other (the ‘So­cial Brain Hy­poth­e­sis’ or ‘Machi­avel­lian In­tel­li­gence Hy­poth­e­sis’). In­tel­li­gence pur­port­edly evolved to get our­selves off the hook, and our abil­ity to con­struct rocket ships and proofs about large prime num­bers are just a lucky side product.

As a per­son who is both un­usu­ally smart, and who spent the last half hour wish­ing the seat­belt sign would go off so they could per­mis­si­bly use the re­stroom, I feel like there is some ten­sion be­tween this the­ory and re­al­ity. I’m not the only un­usu­ally smart per­son who hates break­ing rules, who wishes there were more rules tel­ling them what to do, who in­ces­santly makes up rules for them­selves, who in­ten­tion­ally steers clear of bor­der­line cases be­cause it would be so an­noy­ing to think about, and who wishes the nom­i­nal rules were po­liced pre­dictably and ac­tu­ally re­flected ex­pected be­hav­ior. This is a whole stereo­type of per­son.

But if in­tel­li­gence evolved for the prime pur­pose of evad­ing rules, shouldn’t the smartest peo­ple be best at nav­i­gat­ing rule eva­sion? Or at least re­li­ably non-ter­rible at it? Shouldn’t they be the most delighted to find them­selves in situ­a­tions where the rules were am­bigu­ous and the real situ­a­tion didn’t match the claimed rules? Shouldn’t the peo­ple who are best at mak­ing rocket ships and proofs also be the best at mak­ing ex­cuses and calcu­lat­edly risky norm-vi­o­la­tions? Why is there this stereo­type that the more you can make rocket ships, the more likely you are to break down cry­ing if the so­cial rules about when and how you are al­lowed to make rocket ships are am­bigu­ous?

It could be that these nerds are rare, yet salient for some rea­son. Maybe such peo­ple are funny, not rep­re­sen­ta­tive. Maybe the smartest peo­ple are ac­tu­ally savvy. I’m told that there is at least a pos­i­tive cor­re­la­tion be­tween so­cial skills and other in­tel­lec­tual skills.

I offer a differ­ent the­ory. If the hu­man brain grew out of an end­less cat and mouse game, what if the thing we tra­di­tion­ally think of as ‘in­tel­li­gence’ grew out of be­ing the cat, not the mouse?

The skill it takes to ap­ply ab­stract the­o­ries across a range of do­mains and to no­tice places where re­al­ity doesn’t fit sounds very much like polic­ing norms, not break­ing them. The love of con­sis­tency that fuels unify­ing the­o­ries sounds a lot like the one that in­sists on fair ap­pli­ca­tion of laws, and so­cial codes that can ap­ply in ev­ery cir­cum­stance. Math is ba­si­cally just the con­struc­tion of a bunch of rules, and then end­less spec­u­la­tion about what they im­ply. A ma­jor ob­ject of sci­ence is even called dis­cov­er­ing ‘the laws of na­ture’.

Rules need to gen­er­al­ize across a lot of situ­a­tions—you will have a ter­rible time as rule-en­forcer if you see ev­ery situ­a­tion as hav­ing new, ad-hoc ap­pro­pri­ate be­hav­ior. We wouldn’t even call this hav­ing a ‘rule’. But more to the point, when peo­ple bring you their ex­cuses, if your rule doesn’t already im­ply an im­mov­able po­si­tion on ev­ery case you have never imag­ined, then you are open to ac­cept­ing ex­cuses. So you need to see the one law man­i­fest ev­ery­where. I posit that tech­ni­cal in­tel­li­gence comes from the drive to make these gen­er­al­iza­tions, not the drive to thwart them.

On this the­ory, prob­a­bly some other as­pects of hu­man skill are for evad­ing norms. For in­stance, per­haps so­cial or emo­tional in­tel­li­gence (I hear these are things, but will not pre­tend to know much about them). If norm-polic­ing and norm-evad­ing are some­what differ­ent ac­tivi­ties, we might ex­pect to have at least two sys­tems that are en­gorged by this end­less strug­gle.

I think this would solve an­other prob­lem: if we came to have in­tel­li­gence for cheat­ing each other, it is un­clear why gen­eral in­tel­li­gence per se is is the an­swer to this, but not to other prob­lems we have ever had as an­i­mals. Why did we get men­tal skills this time rather than ear­lier? Like that time we were com­pet­ing over eat­ing all the plants, or es­cap­ing preda­tors bet­ter than our cous­ins? This isn’t the only time that a species was in fierce com­pe­ti­tion against them­selves for some­thing. In fact that has been hap­pen­ing for­ever. Why didn’t we de­velop in­tel­li­gence to com­pete against each other for food, back when we lived in the sea? If the the­ory is just ‘there was strong com­pet­i­tive pres­sure for some­thing that will help us win, so out came in­tel­li­gence’, I think there is a lot left un­ex­plained. Espe­cially since the thing we most want to ex­plain is the space­ship stuff, that on this the­ory is a ran­dom side effect any­way. (Note: I may be mi­s­un­der­stand­ing the usual the­ory, as a re­sult of know­ing al­most noth­ing about it.)

I think this Prin­ci­pled In­tel­li­gence Hy­poth­e­sis does bet­ter. Track­ing gen­eral prin­ci­ples and spot­ting de­vi­a­tions from them is close to what sci­en­tific in­tel­li­gence is, so if we were com­pet­ing to do this (against peo­ple seek­ing to thwart us) it would make sense that we ended up with good the­ory-gen­er­al­iz­ing and de­vi­a­tion-spot­ting en­g­ines.

On the other hand, I think there are sev­eral rea­sons to doubt this the­ory, or de­tails to re­solve. For in­stance, while we are be­ing un­nec­es­sar­ily norm-abid­ing and go­ing with anec­do­tal ev­i­dence, I think I am ac­tu­ally pretty great at mak­ing up ex­cuses, if I do say so. And I feel like this rests on is the same skill as ‘analo­gize one thing to an­other’ (my be­ing here to hide from a party could just as well be in­ter­preted as my be­ing here to look for the drinks, much as the econ­omy could also be in­ter­preted as a kind of ner­vous sys­tem), which seems like it is quite similar to the skill of mak­ing up sci­en­tific the­o­ries (these five ob­ser­va­tions be­ing true is much like the­ory X ap­ply­ing in gen­eral), though ar­guably not the skill of mak­ing up sci­en­tific the­o­ries well. So this is ev­i­dence against smart peo­ple be­ing bad at norm eva­sion in gen­eral, and against norm eva­sion be­ing a differ­ent kind of skill to norm en­force­ment, which is about gen­er­al­iz­ing across cir­cum­stances.

Some other out­side view ev­i­dence against this the­ory’s cor­rect­ness is that my friends all think it is wrong, and I know noth­ing about the rele­vant liter­a­ture. I think it could also do with some in­side view de­tails – for in­stance, how ex­actly does any crea­ture ever benefit from en­forc­ing norms well? Isn’t it a bit of a tragedy of the com­mons? If norm eva­sion and norm polic­ing skills vary in a pop­u­la­tion of agents, what hap­pens over time? But I thought I’d tell you my rough thoughts, be­fore I set this aside and fail to look into any of those de­tails for the in­definite fu­ture.

No nominations.
No reviews.