Empirical claims, preference claims, and attitude claims

What do the fol­low­ing state­ments have in com­mon?

  • “At­las Shrugged is the best book ever writ­ten.”

  • “You break it, you buy it.”

  • “Earth is the most in­ter­est­ing planet in the so­lar sys­tem.”

My an­swer: None of them are falsifi­able claims about the na­ture of re­al­ity. They’re all closer to what one might call “opinions”. But what is an “opinion”, ex­actly?

There’s already been some dis­cus­sion on Less Wrong about what ex­actly it means for a claim to be mean­ingful. This post fo­cuses on the nega­tive defi­ni­tion of mean­ing: what sort of state­ments do peo­ple make where the pri­mary con­tent of the state­ment is non-em­piri­cal? The idea here is similar to the idea be­hind anti-virus soft­ware: Even if you can’t rigor­ously de­scribe what pro­grams are safe to run on your com­puter, there still may be util­ity in keep­ing a database of pro­grams that are known to be un­safe.

Why is it use­ful to be able to be able to flag non-em­piri­cal claims? Well, for one thing, you can be­lieve what­ever you want about them! And it seems likely that this pat­tern-match­ing ap­proach works bet­ter for flag­ging them than a more con­struc­tive defi­ni­tion.

But first, a bit on the philos­o­phy of non-em­piri­cal claims.

Let’s take a typ­i­cal opinion state­ment: “Justin Bie­ber sucks”. There are a few ways we could in­ter­pret this as short­hand for a differ­ent claim. For ex­am­ple, maybe what the speaker re­ally means is “I pre­fer not to listen to Justin Bie­ber’s mu­sic.” (Prefer­ence claim.) Or maybe what the speaker re­ally means is “Of the peo­ple who have heard songs by Justin Bie­ber, the ma­jor­ity pre­fer not to listen to his mu­sic.” (Em­piri­cal claim.)

I don’t think short­hand in­ter­pre­ta­tions like these are ac­cu­rate for most peo­ple who claim that JB sucks. In­stead, I sus­pect most peo­ple who ar­gue this are com­mu­ni­cat­ing some com­bi­na­tion of (a) nega­tive af­fect to­wards JB and (b) tribal af­fili­a­tion with fel­low JB haters. I’ve taken to refer­ring to state­ments like these, that are nei­ther prefer­ence claims nor em­piri­cal claims, as “at­ti­tude claims”.

This ex­am­ple doesn’t mean that all “X sucks” style claims are at­ti­tude claims. Take the claim “Win­dows sucks”. It does seem plau­si­ble that some­one who said this could be per­suaded that their claim was false through em­piri­cal ev­i­dence—e.g. by a meta-anal­y­sis that com­pared Win­dows worker pro­duc­tivity fa­vor­ably to worker pro­duc­tivity us­ing other op­er­at­ing sys­tems.

So if some­one says Win­dows sucks, then whether their claim is em­piri­cal, at­ti­tu­di­nal, or (most likely) some mix­ture de­pends on what’s go­ing on in their head. You may be able to clas­sify the claim with fur­ther con­ver­sa­tion, how­ever. If they say “even if users are hap­piest and most pro­duc­tive us­ing Win­dows, it still sucks!”, that sug­gests the claim is al­most en­tirely at­ti­tu­di­nal.

At­ti­tude claims taxonomy

I’ve been writ­ing down at­ti­tude claims I think of or come across in my note­book. Here’s some that I’ve seen so far. Hope­fully they’ll serve as good train­ing data for your in­ter­nal clas­sifier.

Not all of the ex­am­ples I’ve found fit neatly in to one of these cat­e­gories (e.g. “I can do any­thing I want”), and it’s pretty com­mon to find claims that seem like mix­tures of at­ti­tude and fact/​prefer­ence state­ments. For ex­am­ple, if some­one says “Be­ing out­ra­geous is the best way to be”, are they say­ing “I pre­fer to be out­ra­geous” or “Yay out­ra­geous­ness”? Prob­a­bly a bit of both.

What at­ti­tudes should I have?

That’s a “should” ques­tion, i.e. a ques­tion about so­cial rules. Un­less you meant it as a short­hand for a ques­tion about how best to achieve some goal, e.g. “What at­ti­tudes should I have in or­der to best achieve my prefer­ences?” Then it be­comes an em­piri­cal ques­tion.

I sus­pect that most peo­ple can bet­ter achieve their prefer­ences by con­sciously choos­ing and adopt­ing at­ti­tudes rather than go­ing with what­ever de­faults they grew up with or are preva­lent within their so­cial group. At­ti­tude hack­ing is not triv­ial, so you might want to find a friend to adopt your preferred at­ti­tude with. (This isn’t anti-epistemic group­think as long as you’re do­ing this for at­ti­tudes only and not for facts.)

Are at­ti­tudes bad?

That’s an at­ti­tude ques­tion.

I think to best achieve your prefer­ences, it’s likely op­ti­mal to take some at­ti­tudes se­ri­ously, e.g. Jon Ka­bat-Zinn: “as long as you are breath­ing, there is more right with you than there is wrong, no mat­ter how ill or how hope­less you may feel”, or Eliezer Yud­kowsky: “prob­a­bil­ity the­ory is also a kind of Author­ity and I try to be ruled by it as much as I can man­age.”

Un­for­tu­nately, I haven’t man­aged to take any at­ti­tude claims as se­ri­ously ever since I re­al­ized that they’re ba­si­cally just made up. (Which is it­self an at­ti­tude state­ment of the af­fect type, about the im­por­tance of at­ti­tudes.) But I’ve also felt more free to “cheat” and mod­ify my at­ti­tudes di­rectly in or­der to op­ti­mize for my prefer­ences.

Will point­ing out that so­cial rules are so­cial rules make peo­ple less likely to take them se­ri­ously? Prob­a­bly. The ideas in this post are dan­ger­ous knowl­edge that shouldn’t be spread be­yond ra­tio­nal­ist cir­cles.

If you’re like me, you may get kind of squeamish con­sum­ing at­ti­tude-heavy me­dia (which is also pro­duced by ra­tio­nal­ists, by the way; see Paul Gra­ham or Ju­lia Galef). That’s an at­ti­tude.

Con­nec­tion with Non­vi­o­lent Communication

Em­piri­cal claim: If you re­strict your­self to em­piri­cal claims and prefer­ence claims when you have an ar­gu­ment, you and the peo­ple you ar­gue with will be more pleased with the out­come of your ar­gu­ments.

Non­vi­o­lent Com­mu­ni­ca­tion is a philos­o­phy that recom­mends re­plac­ing at­ti­tude claims like “You’re an awful neigh­bor” or “It’s your fault I can’t get to sleep” with em­piri­cal claims, prefer­ence claims, and re­quests: “Your mu­sic is play­ing very loudly (fact). I’m hav­ing a hard time sleep­ing (fact). I’d re­ally like to be able to get to sleep (prefer­ence). Could you turn down the vol­ume?” Pre­sum­ably this works be­cause (a) ar­gu­ments over em­piri­cal claims are some­times ac­tu­ally re­solved and (b) if you share prefer­ences in­stead of blud­geon­ing peo­ple with so­cial rules, they’re more likely to em­pathize with you and do things to make you happy.

More thoughts

After crys­tal­liz­ing the fact/​at­ti­tude dis­tinc­tion, I started try­ing to ap­ply self-skep­ti­cism to em­piri­cal claims only, and just ig­nor­ing at­ti­tude claims I didn’t like. (“That’s just, like, your opinion, man.”) Care­fully con­sid­er­ing un­com­fortable em­piri­cal claims is a habit that will im­prove my model of the world, thereby helping me achieve my prefer­ences. (That’s what it’s all about, right?) Care­fully con­sid­er­ing un­com­fortable at­ti­tude claims, not so much, ex­cept maybe if they’re from peo­ple with whom I have val­ued re­la­tion­ships that I want to de­bug.

Does this post de­scribe an at­ti­tude? I ac­tu­ally put it and other af­fect-free clas­sifi­ca­tion schemes in to a fourth cat­e­gory: that of a “cog­ni­tive tool”, like a de­scrip­tion of an al­gorithm, that you can take or leave as you wish.