Diseased thinking: dissolving questions about disease

Re­lated to: Dis­guised Queries, Words as Hid­den In­fer­ences, Dis­solv­ing the Ques­tion, Eight Short Stud­ies on Excuses

To­day’s ther­a­peu­tic ethos, which cel­e­brates cur­ing and dis­par­ages judg­ing, ex­presses the liberal dis­po­si­tion to as­sume that crime and other prob­le­matic be­hav­iors re­flect so­cial or biolog­i­cal cau­sa­tion. While this ab­solves the in­di­vi­d­ual of re­spon­si­bil­ity, it also strips the in­di­vi­d­ual of per­son­hood, and moral dignity

-- Ge­orge Will, town­hall.com

Sandy is a mor­bidly obese woman look­ing for ad­vice.

Her hus­band has no sym­pa­thy for her, and tells her she ob­vi­ously needs to stop eat­ing like a pig, and would it kill her to go to the gym once in a while?

Her doc­tor tells her that obe­sity is pri­mar­ily ge­netic, and recom­mends the diet pill orli­s­tat and a con­sul­ta­tion with a sur­geon about gas­tric by­pass.

Her sister tells her that obe­sity is a perfectly valid lifestyle choice, and that fat-ism, equiv­a­lent to racism, is so­ciety’s way of keep­ing her down.

When she tells each of her friends about the opinions of the oth­ers, things re­ally start to heat up.

Her hus­band ac­cuses her doc­tor and sister of ab­solv­ing her of per­sonal re­spon­si­bil­ity with feel-good plat­i­tudes that in the end will only pre­vent her from get­ting the willpower she needs to start a real diet.

Her doc­tor ac­cuses her hus­band of ig­no­rance of the real causes of obe­sity and of the most effec­tive treat­ments, and ac­cuses her sister of le­gi­t­imiz­ing a dan­ger­ous health risk that could end with Sandy in hos­pi­tal or even dead.

Her sister ac­cuses her hus­band of be­ing a jerk, and her doc­tor of try­ing to med­i­cal­ize her be­hav­ior in or­der to turn it into a “con­di­tion” that will keep her on pills for life and make lots of money for Big Pharma.

Sandy is fic­tional, but similar con­ver­sa­tions hap­pen ev­ery day, not only about obe­sity but about a host of other marginal con­di­tions that some con­sider char­ac­ter flaws, oth­ers dis­eases, and still oth­ers nor­mal vari­a­tion in the hu­man con­di­tion. At­ten­tion deficit di­s­or­der, in­ter­net ad­dic­tion, so­cial anx­iety di­s­or­der (as one skep­tic said, didn’t we used to call this “shy­ness”?), al­co­holism, chronic fa­tigue, op­po­si­tional defi­ant di­s­or­der (“didn’t we used to call this be­ing a teenager?“), com­pul­sive gam­bling, ho­mo­sex­u­al­ity, Asperg­ers’ syn­drome, an­ti­so­cial per­son­al­ity, even de­pres­sion have all been placed in two or more of these cat­e­gories by differ­ent peo­ple.

Sandy’s sister may have a point, but this post will con­cen­trate on the de­bate be­tween her hus­band and her doc­tor, with the un­der­stand­ing that the same tech­niques will ap­ply to eval­u­at­ing her sister’s opinion. The dis­agree­ment be­tween Sandy’s hus­band and doc­tor cen­ters around the idea of “dis­ease”. If obe­sity, de­pres­sion, al­co­holism, and the like are dis­eases, most peo­ple de­fault to the doc­tor’s point of view; if they are not dis­eases, they tend to agree with the hus­band.

The de­bate over such marginal con­di­tions is in many ways a de­bate over whether or not they are “real” dis­eases. The usual sur­face level ar­gu­ments trot­ted out in fa­vor of or against the propo­si­tion are gen­er­ally in­con­clu­sive, but this post will ap­ply a host of tech­niques pre­vi­ously dis­cussed on Less Wrong to illu­mi­nate the is­sue.

What is Disease?

In Dis­guised Queries , Eliezer demon­strates how a word refers to a cluster of ob­jects re­lated upon mul­ti­ple axes. For ex­am­ple, in a com­pany that sorts red smooth translu­cent cubes full of vana­dium from blue furry opaque eggs full of pal­la­dium, you might in­vent the word “rube” to des­ig­nate the red cubes, and an­other “blegg”, to des­ig­nate the blue eggs. Both words are use­ful be­cause they “carve re­al­ity at the joints”—they re­fer to two com­pletely sep­a­rate classes of things which it’s prac­ti­cally use­ful to keep in sep­a­rate cat­e­gories. Cal­ling some­thing a “blegg” is a quick and easy way to de­scribe its color, shape, opac­ity, tex­ture, and chem­i­cal com­po­si­tion. It may be that the odd blegg might be pur­ple rather than blue, but in gen­eral the char­ac­ter­is­tics of a blegg re­main suffi­ciently cor­re­lated that “blegg” is a use­ful word. If they weren’t so cor­re­lated—if blue ob­jects were equally likely to be pal­la­dium-con­tain­ing-cubes as vana­dium-con­tain­ing-eggs, then the word “blegg” would be a waste of breath; the char­ac­ter­is­tics of the ob­ject would re­main just as mys­te­ri­ous to your part­ner af­ter you said “blegg” as they were be­fore.

“Disease”, like “blegg”, sug­gests that cer­tain char­ac­ter­is­tics always come to­gether. A rough sketch of some of the char­ac­ter­is­tics we ex­pect in a dis­ease might in­clude:

1. Some­thing caused by the sorts of thing you study in biol­ogy: pro­teins, bac­te­ria, ions, viruses, genes.

2. Some­thing in­vol­un­tary and com­pletely im­mune to the op­er­a­tions of free will

3. Some­thing rare; the vast ma­jor­ity of peo­ple don’t have it

4. Some­thing un­pleas­ant; when you have it, you want to get rid of it

5. Some­thing dis­crete; a graph would show two widely sep­a­rate pop­u­la­tions, one with the dis­ease and one with­out, and not a nor­mal dis­tri­bu­tion.

6. Some­thing com­monly treated with sci­ence-y in­ter­ven­tions like chem­i­cals and ra­di­a­tion.

Cancer satis­fies ev­ery one of these crite­ria, and so we have no qualms what­so­ever about clas­sify­ing it as a dis­ease. It’s a type spec­i­men, the spar­row as op­posed to the os­trich. The same is true of heart at­tack, the flu, di­a­betes, and many more.

Some con­di­tions satisfy a few of the crite­ria, but not oth­ers. Dwarfism seems to fail (5), and it might get its sta­tus as a dis­ease only af­ter stud­ies show that the sup­posed dwarf falls way out of nor­mal hu­man height vari­a­tion. De­spite the best efforts of tran­shu­man­ists, it’s hard to con­vince peo­ple that ag­ing is a dis­ease, partly be­cause it fails (3). Cal­ling ho­mo­sex­u­al­ity a dis­ease is a poor choice for many rea­sons, but one of them is cer­tainly (4): it’s not nec­es­sar­ily un­pleas­ant.

The marginal con­di­tions men­tioned above are also in this cat­e­gory. Obe­sity ar­guably sort-of-satis­fies crite­ria (1), (4), and (6), but it would be pretty hard to make a case for (2), (3), and (5).

So, is obe­sity re­ally a dis­ease? Well, is Pluto re­ally a planet? Once we state that obe­sity satis­fies some of the crite­ria but not oth­ers, it is mean­ingless to talk about an ad­di­tional fact of whether it “re­ally de­serves to be a dis­ease” or not.

If it weren’t for those pesky hid­den in­fer­ences...

Hid­den In­fer­ences From Disease Concept

The state of the dis­ease node, mean­ingless in it­self, is used to pre­dict sev­eral other nodes with non-em­piri­cal con­tent. In English: we make value de­ci­sions based on whether we call some­thing a “dis­ease” or not.

If some­thing is a real dis­ease, the pa­tient de­serves our sym­pa­thy and sup­port; for ex­am­ple, can­cer suffer­ers must uni­ver­sally be de­scribed as “brave”. If it is not a real dis­ease, peo­ple are more likely to get our con­dem­na­tion; for ex­am­ple Sandy’s hus­band who calls her a “pig” for her in­abil­ity to con­trol her eat­ing habits. The differ­ence be­tween “shy­ness” and “so­cial anx­iety di­s­or­der” is that peo­ple with the first get called “weird” and told to man up, and peo­ple with the sec­ond get spe­cial priv­ileges and the sym­pa­thy of those around them.

And if some­thing is a real dis­ease, it is so­cially ac­cept­able (maybe even man­dated) to seek med­i­cal treat­ment for it. If it’s not a dis­ease, med­i­cal treat­ment gets de­rided as a “quick fix” or an “ab­di­ca­tion of per­sonal re­spon­si­bil­ity”. I have talked to sev­eral doc­tors who are un­com­fortable sug­gest­ing gas­tric by­pass surgery, even in peo­ple for whom it is med­i­cally in­di­cated, be­cause they be­lieve it is morally wrong to turn to medicine to solve a char­ac­ter is­sue.

While a con­di­tion’s sta­tus as a “real dis­ease” ought to be mean­ingless as a “hang­ing node” af­ter the sta­tus of all other nodes have been de­ter­mined, it has ac­quired poli­ti­cal and philo­soph­i­cal im­pli­ca­tions be­cause of its role in de­ter­min­ing whether pa­tients re­ceive sym­pa­thy and whether they are per­mit­ted to seek med­i­cal treat­ment.

If we can de­ter­mine whether a per­son should get sym­pa­thy, and whether they should be al­lowed to seek med­i­cal treat­ment, in­de­pen­dently of the cen­tral node “dis­ease” or of the crite­ria that feed into it, we will have suc­cess­fully unasked the ques­tion “are these marginal con­di­tions real dis­eases” and cleared up the con­fu­sion.

Sym­pa­thy or Con­dem­na­tion?

Our at­ti­tudes to­ward peo­ple with marginal con­di­tions mainly re­flect a de­on­tol­o­gist liber­tar­ian (liber­tar­ian as in “free will”, not as in “against gov­ern­ment”) model of blame. In this con­cept, peo­ple make de­ci­sions us­ing their free will, a spiritual en­tity op­er­at­ing free from biol­ogy or cir­cum­stance. Peo­ple who make good de­ci­sions are in­trin­si­cally good peo­ple and de­serve good treat­ment; peo­ple who make bad de­ci­sions are in­trin­si­cally bad peo­ple and de­serve bad treat­ment. But peo­ple who make bad de­ci­sions for rea­sons that are out­side of their free will may not be in­trin­si­cally bad peo­ple, and may there­fore be ab­solved from de­serv­ing bad treat­ment. For ex­am­ple, if a nor­mally peace­ful per­son has a brain tu­mor that af­fects ar­eas in­volved in fear and ag­gres­sion, they go on a crazy kil­ling spree, and then they have their brain tu­mor re­moved and be­come a peace­ful per­son again, many peo­ple would be will­ing to ac­cept that the kil­ling spree does not re­flect nega­tively on them or open them up to de­serv­ing bad treat­ment, since it had biolog­i­cal and not spiritual causes.

Un­der this model, de­cid­ing whether a con­di­tion is biolog­i­cal or spiritual be­comes very im­por­tant, and the ra­tio­nale for wor­ry­ing over whether some­thing “is a real dis­ease” or not is plain to see. Without figur­ing out this ex­tremely difficult ques­tion, we are at risk of ei­ther blam­ing peo­ple for things they don’t de­serve, or else let­ting them off the hook when they com­mit a sin, both of which, to liber­tar­ian de­on­tol­o­gists, would be ter­rible things. But de­ter­min­ing whether marginal con­di­tions like de­pres­sion have a spiritual or biolog­i­cal cause is difficult, and no one knows how to do it re­li­ably.

Deter­minist con­se­quen­tial­ists can do bet­ter. We be­lieve it’s biol­ogy all the way down. Separat­ing spiritual from biolog­i­cal ill­nesses is im­pos­si­ble and un­nec­es­sary. Every con­di­tion, from brain tu­mors to poor taste in mu­sic, is “biolog­i­cal” in­so­far as it is en­coded in things like cells and pro­teins and fol­lows laws based on their struc­ture.

But de­ter­minists don’t just ig­nore the very im­por­tant differ­ences be­tween brain tu­mors and poor taste in mu­sic. Some biolog­i­cal phe­nom­ena, like poor taste in mu­sic, are en­coded in such a way that they are ex­tremely vuln­er­a­ble to what we can call so­cial in­fluences: praise, con­dem­na­tion, in­tro­spec­tion, and the like. Other biolog­i­cal phe­nom­ena, like brain tu­mors, are com­pletely im­mune to such in­fluences. This al­lows us to de­velop a more use­ful model of blame.

The con­se­quen­tial­ist model of blame is very differ­ent from the de­on­tolog­i­cal model. Be­cause all ac­tions are biolog­i­cally de­ter­mined, none are more or less meta­phys­i­cally blame­wor­thy than oth­ers, and none can mark any­one with the meta­phys­i­cal sta­tus of “bad per­son” and make them “de­serve” bad treat­ment. Con­se­quen­tial­ists don’t on a pri­mary level want any­one to be treated badly, full stop; thus is it writ­ten: “Sad­dam Hus­sein doesn’t de­serve so much as a stubbed toe.” But if con­se­quen­tial­ists don’t be­lieve in pun­ish­ment for its own sake, they do be­lieve in pun­ish­ment for the sake of, well, con­se­quences. Hurt­ing bank rob­bers may not be a good in and of it­self, but it will pre­vent banks from be­ing robbed in the fu­ture. And, one might in­fer, al­though al­co­holics may not de­serve con­dem­na­tion, so­cietal con­dem­na­tion of al­co­holics makes al­co­holism a less at­trac­tive op­tion.

So here, at last, is a rule for which dis­eases we offer sym­pa­thy, and which we offer con­dem­na­tion: if giv­ing con­dem­na­tion in­stead of sym­pa­thy de­creases the in­ci­dence of the dis­ease enough to be worth the hurt feel­ings, con­demn; oth­er­wise, sym­pa­thize. Though the rule is based on philos­o­phy that the ma­jor­ity of the hu­man race would dis­avow, it leads to in­tu­itively cor­rect con­se­quences. Yel­ling at a can­cer pa­tient, shout­ing “How dare you al­low your cells to di­vide in an un­con­trol­led man­ner like this; is that the way your mother raised you??!” will prob­a­bly make the pa­tient feel pretty awful, but it’s not go­ing to cure the can­cer. Tel­ling a lazy per­son “Get up and do some work, you worth­less bum,” very well might cure the laz­i­ness. The can­cer is a biolog­i­cal con­di­tion im­mune to so­cial in­fluences; the laz­i­ness is a biolog­i­cal con­di­tion sus­cep­ti­ble to so­cial in­fluences, so we try to so­cially in­fluence the laz­i­ness and not the can­cer.

The ques­tion “Do the obese de­serve our sym­pa­thy or our con­dem­na­tion,” then, is ask­ing whether con­dem­na­tion is such a use­ful treat­ment for obe­sity that its util­ity out­weights the di­su­til­ity of hurt­ing obese peo­ple’s feel­ings. This ques­tion may have differ­ent an­swers de­pend­ing on the par­tic­u­lar obese per­son in­volved, the par­tic­u­lar per­son do­ing the con­demn­ing, and the availa­bil­ity of other meth­ods for treat­ing the obe­sity, which brings us to...

The Ethics of Treat­ing Marginal Conditions

If a con­di­tion is sus­cep­ti­ble to so­cial in­ter­ven­tion, but an effec­tive biolog­i­cal ther­apy for it also ex­ists, is it okay for peo­ple to use the biolog­i­cal ther­apy in­stead of figur­ing out a so­cial solu­tion? My gut an­swer is “Of course, why wouldn’t it be?“, but ap­par­ently lots of peo­ple find this con­tro­ver­sial for some rea­son.

In a liber­tar­ian de­on­tolog­i­cal sys­tem, throw­ing biolog­i­cal solu­tions at spiritual prob­lems might be dis­re­spect­ful or de­hu­man­iz­ing, or a band-aid that doesn’t af­fect the deeper prob­lem. To some­one who be­lieves it’s biol­ogy all the way down, this is much less of a con­cern.

Others com­plain that the ex­is­tence of an easy med­i­cal solu­tion pre­vents peo­ple from learn­ing per­sonal re­spon­si­bil­ity. But here we see the sta­tus-quo bias at work, and so can ap­ply a prefer­ence re­ver­sal test. If peo­ple re­ally be­lieve learn­ing per­sonal re­spon­si­bil­ity is more im­por­tant than be­ing not ad­dicted to heroin, we would ex­pect these peo­ple to sup­port de­liber­ately ad­dict­ing schoolchil­dren to heroin so they can de­velop per­sonal re­spon­si­bil­ity by com­ing off of it. Any­one who dis­agrees with this some­what shock­ing pro­posal must be­lieve, on some level, that hav­ing peo­ple who are not ad­dicted to heroin is more im­por­tant than hav­ing peo­ple de­velop what­ever mea­sure of per­sonal re­spon­si­bil­ity comes from kick­ing their heroin habit the old-fash­ioned way.

But the most con­vinc­ing ex­pla­na­tion I have read for why so many peo­ple are op­posed to med­i­cal solu­tions for so­cial con­di­tions is a sig­nal­ing ex­pla­na­tion by Robin Hans...wait! no!...by Katja Grace. On her blog, she says:

...the situ­a­tion re­minds me of a pat­tern in similar cases I have no­ticed be­fore. It goes like this. Some peo­ple make per­sonal sac­ri­fices, sup­pos­edly to­ward solv­ing prob­lems that don’t threaten them per­son­ally. They sort re­cy­cling, buy free range eggs, buy fair trade, cam­paign for wealth re­dis­tri­bu­tion etc. Their ac­tions are seen as vir­tu­ous. They see those who don’t join them as un­car­ing and im­moral. A more effi­cient solu­tion to the prob­lem is sug­gested. It does not re­quire per­sonal sac­ri­fice. Peo­ple who have not pre­vi­ously sac­ri­ficed sup­port it. Those who have pre­vi­ously sac­ri­ficed ob­ject on grounds that it is an ex­cuse for peo­ple to get out of mak­ing the sac­ri­fice. The sup­posed in­stru­men­tal ac­tion, as the visi­ble sign of car­ing, has be­come vir­tu­ous in its own right. Solv­ing the prob­lem effec­tively is an at­tack on the moral peo­ple.

A case in which some peo­ple eat less en­joy­able foods and ex­er­cise hard to avoid be­com­ing obese, and then cam­paign against a pill that makes avoid­ing obe­sity easy demon­strates some of the same prin­ci­ples.

There are sev­eral very rea­son­able ob­jec­tions to treat­ing any con­di­tion with drugs, whether it be a clas­si­cal dis­ease like can­cer or a marginal con­di­tion like al­co­holism. The drugs can have side effects. They can be ex­pen­sive. They can build de­pen­dence. They may later be found to be place­bos whose effi­cacy was over­hyped by dishon­est phar­ma­ceu­ti­cal ad­ver­tis­ing.. They may raise eth­i­cal is­sues with chil­dren, the men­tally in­ca­pac­i­tated, and other peo­ple who can­not de­cide for them­selves whether or not to take them. But these is­sues do not mag­i­cally be­come more dan­ger­ous in con­di­tions typ­i­cally re­garded as “char­ac­ter flaws” rather than “dis­eases”, and the same good-enough solu­tions that work for can­cer or heart dis­ease will work for al­co­holism and other such con­di­tions (but see here).

I see no rea­son why peo­ple who want effec­tive treat­ment for a con­di­tion should be de­nied it or stig­ma­tized for seek­ing it, whether it is tra­di­tion­ally con­sid­ered “med­i­cal” or not.


Peo­ple com­monly de­bate whether so­cial and men­tal con­di­tions are real dis­eases. This mas­quer­ades as a med­i­cal ques­tion, but its im­pli­ca­tions are mainly so­cial and eth­i­cal. We use the con­cept of dis­ease to de­cide who gets sym­pa­thy, who gets blame, and who gets treat­ment.

In­stead of con­tin­u­ing the fruitless “dis­ease” ar­gu­ment, we should ad­dress these ques­tions di­rectly. Tak­ing a de­ter­minist con­se­quen­tial­ist po­si­tion al­lows us to do so more effec­tively. We should blame and stig­ma­tize peo­ple for con­di­tions where blame and stigma are the most use­ful meth­ods for cur­ing or pre­vent­ing the con­di­tion, and we should al­low pa­tients to seek treat­ment when­ever it is available and effec­tive.