What is moral foundation theory good for?

I’ve seen Jonathan Haidt men­tioned on Less Wrong a few times, and so when I saw an ar­ti­cle about (in part) Haidt’s new book el­se­where, I thought it would be an in­ter­est­ing read. It was, but not for the rea­sons I ex­pected. Per­haps it is un­fair to judge Haidt be­fore I have read the book, but the quotes in the ar­ti­cle re­veal some se­ri­ously sloppy think­ing.

Haidt be­lieves that there are at least six sources of moral val­ues; the first five are harm/​car­ing, fair­ness, loy­alty, au­thor­ity, sanc­tity/​dis­gust. Liberty was re­cently added to the list, but doesn’t seem to have made it into this ar­ti­cle. He claims that liber­als (in the Amer­i­can sense), care mostly (or only) only about the harm and fair­ness val­ues, while con­ser­va­tives care about all five. I my­self am a one-foun­da­tion per­son, since I con­sider un­fair­ness ei­ther a spe­cial case of harm, or a good heuris­tic for where harm is likely to oc­cur; my views are ap­par­ently so rare that they haven’t come up on Haidt’s sur­vey, and I haven’t met any­one else who has re­ported a score like mine.

While Haidt de­scribes him­self as a “cen­trist”, he ar­gues that “you need loy­alty, au­thor­ity and sanc­tity to run a de­cent so­ciety.” There are at least three ways that this claim can be read:

(1) Haidt’s per­sonal moral foun­da­tions ac­tu­ally in­clude all five bases, so this is a tau­tol­ogy; of course some­one who thinks loy­alty is fun­da­men­tal will think a so­ciety with­out loy­alty is not de­cent. From the tenor of the ar­ti­cle, this is at least psy­cholog­i­cally plau­si­ble.

(2) The three non-uni­ver­sal val­ues can be jus­tified in terms of the com­mon val­ues. This is the in­ter­pre­ta­tion that seems to be sup­ported by some parts of the ar­ti­cle, but it has its own is­sues.

(3) Haidt can­not tell the differ­ence be­tween (1) and (2). Most of the ar­ti­cle makes this claim en­tirely plau­si­ble.

Here’s one ex­am­ple of Haidt’s moral con­fu­sion:

“In In­dia, where he performed field stud­ies early in his pro­fes­sional ca­reer, he en­coun­tered a so­ciety in some ways pa­tri­ar­chal, sex­ist and illiberal. Yet it worked and the peo­ple were lovely.”

First, was Haidt sur­prised to find peo­ple with differ­ent poli­tics than his to be per­son­able? Had he liter­ally never met a con­ser­va­tive be­fore?

Se­cond, what does it mean to say that the so­ciety “worked”, or that the peo­ple were “lovely”? In­dian so­ciety priv­ileges men and cer­tain castes over women and other castes. I say this not to den­i­grate In­dia speci­fi­cally, since there’s no so­ciety in which women are treated equally to men, but to ex­plain that In­dia does have se­ri­ous prob­lems. Liter­acy rates among women are 68% of that of men, to pick a ran­dom statis­tic. And, of course, vi­o­lence against women is en­demic. Haidt re­ports that he “dined with men whose wives silently served us and then re­treated to the kitchen.” What does he sup­pose would have hap­pened if one day one of those women re­fused to serve, or even, af­ter serv­ing, sat down at the table to join the dis­cus­sion?

Of course, even this is an up­per-class con­cern; lower class In­dian women are far more likely to work out­side the home, in or­der to sur­vive. Ap­par­ently in some parts of In­dia, pub­lic toi­lets charge women (who can ill af­ford it) but not men. And I can only as­sume that the situ­a­tion was worse when Haidt was there, at least a decade ago.

Haidt ra­tio­nal­izes this by say­ing, “I was able to see a moral world in which fam­i­lies, not in­di­vi­d­u­als, are the ba­sic unit of so­ciety...”. Per­haps this is the story that they tell (and per­haps they even be­lieve it). But his­tory shows that when women can find al­ter­na­tives, they don’t choose to live like this. So there is both a harm and a fair­ness con­cern here. Haidt, hav­ing seen the loy­alty/​au­thor­ity story, comes to ig­nore the harm/​fair­ness story. He fol­lows this by an anec­dote fo­cus­ing on the harm caused by in­di­vi­d­u­al­ism, since he is ap­par­ently in­ca­pable of jus­tify­ing the non-uni­ver­sal foun­da­tions on their own terms.

Here’s an­other case of this con­fu­sion. Haidt claimed that among street chil­dren in Brazil, the “most dan­ger­ous per­son in the world is mom’s boyfriend. When women have a suc­ces­sion of men com­ing through, their daugh­ters will get raped,” he says. “The right is right to be sound­ing the alarm about the de­cline of mar­riage, and the left is wrong to say, ‘Oh, any kind of fam­ily is OK.’ It’s not OK.”

In this in­stance, Haidt is switch­ing the goal­posts. His moral foun­da­tion test is de­signed to iso­late the five foun­da­tions. But here, there is clearly harm in ad­di­tion to any vi­o­la­tion of tra­di­tion. He doesn’t ex­actly say which non-harm foun­da­tion he wants to in­voke here—that is, what the moth­ers’ vi­o­la­tion is. Im­pu­rity is the only plau­si­ble choice. This, of course, brings to the front one of the most com­mon real effects of the “pu­rity” foun­da­tion: to dis­em­power women.

I should add that there is no cita­tion on this data; it also doesn’t seem to ap­pear in the book (at least, not that I could find via Google Books). A quick glance through Google does not re­veal a plau­si­ble source for this. So where did he get it from? Prob­a­bly not via di­rect ob­ser­va­tion (how would he have ob­served these rapes?). He must have heard it from Brazili­ans. Well, if that’s true, then these Brazilian women must know it. And since no­body wants their daugh­ter to get raped, this must mean that they have a very good rea­son for invit­ing these men in—maybe the al­ter­na­tive is star­va­tion. Re­call that we’re talk­ing about “street chil­dren” here. I just can’t imag­ine a woman say­ing, “yeah, he’s go­ing to rape my daugh­ter, but I re­ally love him!” But I think it’s ac­tu­ally more likely that this is just the sort of ru­mor that the Catholic Church would want to spread, to com­bat un­mar­ried co­hab­ita­tion. It gets its memetic strength from blame-shift­ing/​just-wor­ldism: “If you didn’t want your daugh­ter to get raped, why did you shack (liter­ally?) up with this guy?”

It’s true that there are dan­gers from non-re­lated men, as Sarah Blaf­fer Hrdy dis­cusses in _Mother Na­ture: Ma­ter­nal In­stincts and How They Shape the Hu­man Species; there are also po­ten­tial benefits. Hrdy’s book (which I haven’t finished read­ing yet) dis­cusses both, and also vastly com­pli­cates the view of what “tra­di­tional” fam­ily is. She pre­sents mul­ti­ple equil­ibria, some more com­mon among farm­ers and oth­ers more com­mon among for­agers (to use Robin Han­son’s lan­guage). A Brazilian shan­ty­town doesn’t re­ally fit well into ei­ther frame­work, so it’s un­clear whether norms adapted for ei­ther would be effec­tive.

So does Haidt be­lieve that non­tra­di­tional fam­i­lies are wrong be­cause they vi­o­late pu­rity? Or be­cause they’re harm­ful? The stan­dard con­ser­va­tive re­ply to this is that our tra­di­tions evolved be­cause they were use­ful (i.e. pre­vented harm), and to erase the tra­di­tions with­out un­der­stand­ing the value that they pro­vided is an mis­take. This is put in a delight­fully pa­tron­iz­ing way by Ch­ester­ton—no­tice how he will “al­low” you to clear away a tra­di­tion as though it were his de­ci­sion to make.

And it is in fact rel­a­tively easy to come up with evolu­tion­ary psy­chol­ogy just-so sto­ries as rea­sons for why loy­alty, au­thor­ity, and pu­rity would have been use­ful in the an­ces­tral en­vi­ron­ment. (The same is true of fair­ness). Author­ity, for in­stance, might help with col­lec­tive de­ci­sion mak­ing. Maybe it’s best for the tribe to go take the left fork, and it might be bet­ter to take the right fork. But it is al­most always bet­ter for them all to take the same fork, than it is to split up. If there’s one tribal leader, then they can make that de­ci­sion and have oth­ers agree with it. This isn’t a case of group se­lec­tion; ev­ery in­di­vi­d­ual of the group benefits from co­or­di­na­tion. I de­scribe this as a “just-so story” here be­cause it would be ex­tremely difficult to find ev­i­dence for whether in fact a spe­cific moral in­tu­ition evolved for a spe­cific rea­son. Haidt’s book ap­par­ently pre­sents some of these ar­gu­ments in the con­text of group se­lec­tion, but in this par­tic­u­lar ex­am­ple, group se­lec­tion (or even kin se­lec­tion or re­cip­ro­cal al­tru­ism) isn’t a nec­es­sary part of the hy­poth­e­sis; treat­ing groups as part of the en­vi­ron­ment (rather than as the unit of evolu­tion) is suffi­cient.

Mo­ral foun­da­tions the­ory is per­haps use­ful de­scrip­tively, in that, if it were shown to be some­thing be­yond a just-so story, it would ex­plain why there are five (or six, or more) foun­da­tions as op­posed to one or two. It is, how­ever, miss­ing a piece: why are there peo­ple who don’t share all five foun­da­tions? The evolu­tion­ary ar­gu­ment is not use­ful pre­scrip­tively, be­cause evolu­tion only cares about harm (and only cer­tain kinds of harm), and once we de­cide to see moral ques­tions in terms of harm, then ques­tions of ac­tual harm can screen off the other evolved heuris­tics. Yes, hu­mans are Adap­ta­tion-Ex­e­cuters, not Fit­ness-Max­i­miz­ers. So there are lots of cases where we fol­low our evolved in­tu­itions rather than the pres­sures that se­lected for those in­tu­itions. But we are also ap­par­ently adapted to con­tem­plate moral philos­o­phy. So when we find our­selves jus­tify­ing an evolved in­tu­ition A in terms of an­other evolved in­tu­ition B, we might con­sider B more fun­da­men­tal. And if there are cases where A isn’t ex­plain­able in terms of B, five-foun­da­tion peo­ple just get stuck. This, per­haps does ex­plain the one- or two-foun­da­tion view; it’s what hap­pens when you ask “why?” once, and throw out ev­ery­thing that doesn’t ac­tu­ally have an an­swer. When you ask a sec­ond time, you’re get­ting into the realm of meta-ethics. In­stru­men­tal five-foun­da­tion peo­ple (such as Haidt, prob­a­bly), wouldn’t get stuck—but they would fall back to harm.

Maybe there’s an­other ar­gu­ment for the three non-uni­ver­sal foun­da­tions, but Haidt doesn’t make it. Does he feel that, by defin­ing some­thing as a “foun­da­tion”, it doesn’t need an ar­gu­ment? But if so, why does he keep reach­ing for harm as an ex­pla­na­tion?

As a de­scrip­tive the­ory, Haidt’s moral foun­da­tion frame­work helps ex­plain some of the differ­ing moral val­ues peo­ple have. Haidt seems to wrongly in­ter­pret it as a use­ful pre­scrip­tive tool. How­ever he has not pre­sented any rea­son to think that it is, in fact, use­ful pre­scrip­tively, and has pre­sented sev­eral rea­sons to doubt it.

[Added later:]

None of this is to say that there are no rea­sons to be con­ser­va­tive. You could be con­ser­va­tive in­stru­men­tally (as Haidt seems to be), or you could be con­ser­va­tive be­cause you re­ally do con­sider all five bases to be in­her­ently valuable (you could also do both at once, but that should make you slightly sus­pi­cious that you’re ra­tio­nal­iz­ing). There’s no in­her­ent prob­lem with ei­ther of those. Haidt’s prob­lem is that he wants to have it both ways; he want to pre­sent the non-uni­ver­sal foun­da­tions as in­her­ently valuable, but all his ac­tual ar­gu­ments are about their in­stru­men­tal value.