One Man’s Modus Ponens

‘One man’s modus ponens is another man’s modus tollens’ is a saying in Western philosophy encapsulating a common response to a logical proof which generalizes the ‘reductio ad absurdum’ and consists of rejecting a premise based on an implied conclusion. I explain it in more detail, provide examples, and a Bayesian gloss.
philosophy, Bayes, insight-porn
2012-05-012020-03-23 finished certainty: highly likely importance: 6

A log­i­cal­ly-valid ar­gu­ment which takes the form of a may be in­ter­preted in sev­eral ways; a ma­jor one is to in­ter­pret it as a kind of , where by ‘prov­ing’ a con­clu­sion be­lieved to be false, one might in­stead take it as a which proves that one of the premises is false. This “Moorean shift” is aphorized as the , “One man’s modus po­nens is an­other man’s modus tol­lens”. The Moorean shift is a pow­er­ful coun­ter-ar­gu­ment which has been de­ployed against many skep­ti­cal & meta­phys­i­cal claims in phi­los­o­phy, where often the con­clu­sion is ex­tremely un­likely and lit­tle ev­i­dence can be pro­vided for the premises used in the proofs; and it is rel­e­vant to many other de­bates, par­tic­u­larly method­olog­i­cal ones.

In syl­lo­gis­tic log­ic, 2 of the sim­plest ar­gu­ment pat­terns used in valid rea­son­ing are the modus po­nens and the modus tol­lens. Both modus po­nens and modus tol­lens are ‘log­i­cally valid’, as they use uni­ver­sally ac­cepted rules of in­fer­ence to pro­ceed from given premises to a con­clu­sion, and are equiv­a­lent via a .

Di­a­grammed re­spec­tive­ly, modus po­nens:

  1. A
  2. A → B
  3. ∴ B

and modus tol­lens:

  1. A → B
  2. ¬B
  3. ∴ ¬A

These ar­gu­ments are log­i­cally cor­rect, but whether any given ar­gu­ment us­ing a modus is true is an en­tirely differ­ent ques­tion: an ar­gu­ment could be wrong be­cause it mis­ap­plies the rules of in­fer­ence (and the con­clu­sion does not ac­tu­ally fol­low), or be­cause the premises them­selves are wrong. (Logic is like pipes: a good set of pipes moves wa­ter around with­out let­ting the wa­ter within out, or let­ting things with­out in­—but it only moves wa­ter around, and can­not cre­ate wa­ter out of noth­ing.) Cer­tain con­tentions coun­te­nance nei­ther con­tra­dic­tion nor con­vic­tion.1

The Eco­nomic Ar­gu­ment (XKCD #808)

Given a modus po­nens proof of some­thing like the skep­ti­cal claim that there is no ex­ter­nal world (solip­sis­m), one can, rather than sim­ply re­ject­ing it out of hand while re­fus­ing to dis­cuss it fur­ther, or at­tempt­ing to find a flaw in the ap­pli­ca­tion of in­fer­ence rules which ren­ders the ar­gu­ment a non se­quitur (which usu­ally is­n’t there2), or strug­gling to find spe­cific strong ev­i­dence against any of the premises (which can be ex­tremely diffi­cult for ab­stract points), one can in­stead flip the ar­gu­ment on its head: given that one knows there is an ex­ter­nal world (solip­sism is not true), by modus tol­lens, the skep­ti­cal ar­gu­men­t’s premises about knowl­edge must then be false.3 As a proof is merely truth-p­re­serv­ing ma­chin­ery, it can­not cre­ate out­puts which are more true than its in­puts (GIGO); if the out­put is clearly false, then in­puts must be false. This re­sponse is closely re­lated to the , by putting at­ten­tion on the whole ar­gu­ment, and it can be con­sid­ered a flaw in uses of proofs by con­tra­dic­tion or the —how does one know the con­clu­sion re­ally is ab­surd and to re­ject one of the premises in­stead of per­haps ? Peo­ple may dis­agree greatly about some­thing be­ing ‘ab­surd’, and pre­sent­ing an ar­gu­ment might ‘back­fire’.4

Prob­a­bly the most fa­mous philoso­pher to use this spe­cific ar­gu­ment, and the rea­son it is called a “Moorean shift”, is Here is one hand” ar­gu­ment against var­i­ous skep­ti­cal ar­gu­ments (“Proof of an Ex­ter­nal World”/“A De­fence of Com­mon Sense”): as WP sum­ma­rizes it,

Moore ar­gues against ide­al­ism and skep­ti­cism to­ward the ex­ter­nal world on the grounds that skep­tics could not give rea­sons to ac­cept their meta­phys­i­cal premises that were more plau­si­ble to him than the rea­sons he had to ac­cept the com­mon sense claims about our knowl­edge of the world that skep­tics and ide­al­ists must de­ny. In other words, he is more will­ing to be­lieve that he has a hand than to be­lieve the premises of what he deems “a strange ar­gu­ment in a uni­ver­sity class­room.”

An Is­lamic ver­sion goes (, The Sub­lime Trea­sures):

Noth­ing can be soundly un­der­stood
If day­light it­self needs proof.

, con­sid­er­ing the skep­tic deny­ing truth with his ar­gu­ments, asks what would make us be­lieve any ar­gu­ments more than our own senses (William Ellery Leonard trans­la­tion)?

Thou’lt find
That from the senses first hath been cre­ate
Con­cept of truth, nor can the senses be
Re­butted. For cri­te­rion must be found
Wor­thy of greater trust, which shall de­feat
Through own au­thor­ity the false by true;
What, then, than these our senses must there be
Wor­thy a greater trust? Shall rea­son, sprung
From some false sense, pre­vail to con­tra­dict
Those sens­es, sprung as rea­son wholly is
From out of the sens­es?—­For lest these be true,
All rea­son also then is fal­si­fied.

de­scribed the con­flict of skep­ti­cal ar­gu­ments with “com­mon sense” & bur­dens of proof this way:

Has­n’t com­mon sense been wrong be­fore? Of course. But how do peo­ple show that a com­mon sense view is wrong? By demon­strat­ing a con­flict with other views even more firmly grounded in com­mon sense. The strongest sci­en­tific ev­i­dence can al­ways be re­jected if you’re will­ing to say, “Our senses de­ceive us” or “Mem­ory is never re­li­able” or “All the sci­en­tists have con­spired to trick us.” The only prob­lem with these fool­proof in­tel­lec­tual de­fenses is… that… they’re… ab­surd.

Since every­thing we learn about ‘logic’ or ‘meta­physics’ or ‘ne­ces­sity’ it­self de­rives from ex­pe­ri­ence, it is diffi­cult to see how one could ever be more con­fi­dent in the ab­stract claims and ar­gu­ments which make up a dis­proof of the ex­ter­nal world than in the premise that the ex­ter­nal world ex­ists, or other claims like the Eleatic dis­proofs of time/motion/change. (This has an easy in­ter­pre­ta­tion, in what one might call Bayesian in­for­mal logic, as re­flect­ing differ­ent prior prob­a­bil­i­ties in a Bayesian net­work.)

“Moorean shift” is not the most mem­o­rable phrase5, and at some point some­one coined the max­im, “One man’s modus po­nens is an­other man’s modus tol­lens.”, which is more self­-ex­plana­tory and has en­tered wider cir­cu­la­tion.

The maxim ver­sion is use­ful be­cause it re­minds us of the am­bi­gu­ity of any given ar­gu­ment, and that for pro­duc­tive dis­cus­sion, it is not enough to sim­ply pro­duce a log­i­cally valid ar­gu­ment, but we must con­sider to what ex­tent other peo­ple would agree with the premises. This point is­n’t al­ways ap­pre­ci­at­ed: when you have 2 con­tra­dict­ing claims or ar­gu­ments, only 1 can be cor­rect but the con­tra­dic­tion does­n’t tell you which one is cor­rect. You need to step out­side the ar­gu­ment and find ad­di­tional data or per­spec­tives. From :

A para­dox arises when two seem­ingly air­tight ar­gu­ments lead to con­tra­dic­tory con­clu­sion­s—­con­clu­sions that can­not pos­si­bly both be true. It’s sim­i­lar to adding a set of num­bers in a two-di­men­sional ar­ray and get­ting differ­ent an­swers de­pend­ing on whether you sum up the rows first or the columns. Since the cor­rect to­tal must be the same ei­ther way, the differ­ence shows that an er­ror must have been made in at least one of the two sets of cal­cu­la­tions. But it re­mains to dis­cover at which step (or steps) an er­ro­neous cal­cu­la­tion oc­curred in ei­ther or both of the run­ning sums. There are two ways to re­but an ar­gu­ment. We might call them ‘coun­ter­ing’ and ‘in­val­i­dat­ing’.

  • To counter an ar­gu­ment is to pro­vide an­other ar­gu­ment that es­tab­lishes the op­po­site con­clu­sion.
  • To in­val­i­date an ar­gu­ment, we show that there is some step in that ar­gu­ment that sim­ply does not fol­low from what pre­cedes it (or we show that the ar­gu­men­t’s premis­es—the ini­tial step­s—are them­selves false).

If an ar­gu­ment starts with true premis­es, and if every step in the ar­gu­ment does fol­low, then the ar­gu­men­t’s con­clu­sion must be true. How­ev­er, in­val­i­dat­ing an ar­gu­men­t—i­den­ti­fy­ing an in­cor­rect step some­where-does not show that the ar­gu­men­t’s con­clu­sion must be false. Rather, the in­val­i­da­tion merely re­moves that ar­gu­ment it­self as a rea­son to think the con­clu­sion true; the con­clu­sion might still be true for other rea­sons. There­fore, to firmly re­but an ar­gu­ment whose con­clu­sion is false, we must both in­val­i­date the ar­gu­ment and also present a coun­ter­ar­gu­ment for the op­po­site con­clu­sion.

In the case of a para­dox, in­val­i­dat­ing is es­pe­cially im­por­tant. Whichever of the con­tra­dic­tory con­clu­sions is in­cor­rect, we’ve al­ready got an ar­gu­ment to counter it—that’s what makes the mat­ter a para­dox in the first place! Pil­ing on ad­di­tional coun­ter­ar­gu­ments may (or may not) lead to help­ful in­sights, but the coun­ter­ar­gu­ments them­selves can­not suffice to re­solve the para­dox. What we must also do is in­val­i­date the ar­gu­ment for the false con­clu­sion-that is, we must show how that ar­gu­ment con­tains one or more steps that do not fol­low.

Fail­ing to rec­og­nize the need for in­val­i­da­tion can lead to frus­trat­ingly cir­cu­lar ex­changes be­tween pro­po­nents of the con­flict­ing po­si­tions. One side re­sponds to the oth­er’s ar­gu­ment with a coun­ter­ar­gu­ment, think­ing it a suffi­cient re­but­tal. The other side re­sponds with a coun­ter-coun­ter­ar­gu­men­t—per­haps even a rep­e­ti­tion of the orig­i­nal ar­gu­men­t—­think­ing it an ad­e­quate re­but­tal of the re­but­tal. This cy­cle may per­sist in­defi­nite­ly. With due at­ten­tion to the need to in­val­i­date as well as coun­ter, we can in­ter­rupt the cy­cle and achieve a more pro­duc­tive dis­cus­sion.

The dan­ger of the Moorean shift is that it be­comes a li­cense for fa­nati­cism; as (“A Philoso­pher De­fends Re­li­gion”) de­scribes at­tempt to save re­li­gion from fal­si­fi­ca­tion:

An athe­ist fa­mil­iar with bi­ol­ogy and med­i­cine has no rea­son to be­lieve the bib­li­cal story of the res­ur­rec­tion. But a Chris­t­ian who be­lieves it by faith should not, ac­cord­ing to Planti­nga, be dis­suaded by gen­eral bi­o­log­i­cal ev­i­dence. Planti­nga com­pares the differ­ence in jus­ti­fied be­liefs to a case where you are ac­cused of a crime on the ba­sis of very con­vinc­ing ev­i­dence, but you know that you did­n’t do it. For you, the im­me­di­ate ev­i­dence of your mem­ory is not de­feated by the pub­lic ev­i­dence against you, even though your mem­ory is not avail­able to oth­ers. Like­wise, the Chris­tian’s faith in the truth of the gospels, though un­avail­able to the athe­ist, is not de­feated by the sec­u­lar ev­i­dence against the pos­si­bil­ity of res­ur­rec­tion. Of course some­times con­trary ev­i­dence may be strong enough to per­suade you that your mem­ory is de­ceiv­ing you. Some­thing anal­o­gous can oc­ca­sion­ally hap­pen with be­liefs based on faith, but it will typ­i­cally take the form, ac­cord­ing to Planti­nga, of a change in in­ter­pre­ta­tion of what the Bible means. This tra­di­tion of in­ter­pret­ing scrip­ture in light of sci­en­tific knowl­edge goes back to Au­gustine, who ap­plied it to the ‘days’ of cre­ation. But Planti­nga even sug­gests in a foot­note that those whose faith in­cludes, as his does not, the con­vic­tion that the bib­li­cal chronol­ogy of cre­ation is to be taken lit­er­ally can for that rea­son re­gard the ev­i­dence to the con­trary as sys­tem­at­i­cally mis­lead­ing. One would think that this is a con­se­quence of his epis­te­mo­log­i­cal views that he would hope to avoid.

Nev­er­the­less, this is an im­por­tant ar­gu­ment to be fa­mil­iar with, as it is widely used, cor­rect in many cas­es, and is at the core of many method­olog­i­cal dis­cus­sions.



  • a Cre­ation­ist ar­gu­ment against evo­lu­tion runs:

    If a ra­tio­nal God is not re­spon­si­ble for hu­man minds, and in­stead they were cob­bled to­gether by un­guided evo­lu­tion­ary process­es, we should not ex­pect them to be trust­wor­thy. Since our minds are gen­er­ally trust­wor­thy, though, the evo­lu­tion­ary world­view must not be cor­rect.

  • on : “the mad dog nat­u­ral­ist: Alex Rosen­berg in­ter­viewed by Richard Mar­shall”

    3AM: “In the rather heated re­sponse to Jerry Fodor’s provo­ca­tions about nat­ural se­lec­tion your re­sponse was one of the few that rec­og­nized that he was onto some­thing. I want to quote you: ’ His modus tol­lens is a bi­ol­o­gist’s and cog­ni­tive sci­en­tist’s modus po­nens. As­sum­ing his ar­gu­ment is valid and the right con­clu­sion is not that Dar­win’s the­ory is mis­taken but that Jer­ry’s and any other non-Dar­win­ian ap­proach to the mind is wrong. That puts Jerry in good com­pa­ny, of course: Ein­stein’s.’ I don’t know if you’d agree, but it struck me that many of the re­sponses to Fodor’s ar­gu­ment got it wrong about why he was wrong (if he was). Why do you think Fodor wrong but in an in­ter­est­ing way?”

    Alex Rosen­berg: “When Ein­stein de­vel­oped the ob­jec­tion to quan­tum me­chan­ics in the’30s (the EPR thought ex­per­i­men­t), he had no idea he was ac­tu­ally for­mu­lat­ing the idea of en­tan­gle­ment and that his ob­jec­tion when tested would vin­di­cate quan­tum me­chan­ics 40 or 50 years lat­er. When Fodor ar­gued that nat­ural se­lec­tion can’t see prop­er­ties, and can’t pro­duce or­ganic sys­tems, for ex­am­ple brain­s—that re­spond to, rep­re­sent, reg­is­ter prop­er­ties, he thought he was pro­vid­ing a re­duc­tion ad ab­sur­dum of Dar­win­ian the­ory (the way Ein­stein thought he was pro­vid­ing a re­duc­tion ad ab­sur­dum of quan­tum the­o­ry.) The con­fir­ma­tion of the­o­rem by D’es­pag­nat’s ex­per­i­ments turned Ein­stein’s re­duc­tio into a modus tol­lens. I be­lieve that Fodor’s at­tempted re­duc­tio of Dar­win­ian the­ory is a modus tol­lens of rep­re­sen­ta­tion­al­ist the­o­ries of the mind, the­o­ries that ac­cord to the wet stuff, to neural states what Searle calls orig­i­nal in­ten­tion­al­i­ty. It’s an ar­gu­ment for about in­ten­tional con­tent. So Fodor is to­tally wrong abut Dar­win­ian the­o­ry, but his ar­gu­ment shows that we Dar­wini­ans (and all the physi­cists if I am right that Dar­win’s the­ory is just the 2d law in ac­tion among the macro­mol­e­cules) have to go elim­i­na­tivist about the brain.”

  • Derek Parfit, dis­cussing the ques­tion of why the uni­verse ex­ists rather than noth­ing, men­tions of the that:

    Of the range of pos­si­ble ini­tial con­di­tions, fewer than one in a bil­lion bil­lion would have pro­duced a Uni­verse with the com­plex­ity that al­lows for life. If this claim is true, as I shall here as­sume, there is some­thing that cries out to be ex­plained. Why was one of this tiny set also the one that ac­tu­ally ob­tained? On one view, this was a mere co­in­ci­dence. That is con­ceiv­able, since co­in­ci­dences hap­pen. But this view is hard to be­lieve, since, if it were true, the chance of this co­in­ci­dence oc­cur­ring would be be­low one in a bil­lion bil­lion.

    Oth­ers say: “The Big Bang was fine-tuned. In cre­at­ing the Uni­verse, God chose to make life pos­si­ble.” Athe­ists may re­ject this an­swer, think­ing it im­prob­a­ble that God ex­ists. But this prob­a­bil­ity can­not be as low as one in a bil­lion bil­lion. So even athe­ists should ad­mit that, of these two an­swers to our ques­tion, the one that in­vokes God is more likely to be true. This rea­son­ing re­vives one of the tra­di­tional ar­gu­ments for be­lief in God.

    Leav­ing aside the is­sue of to what ex­tent like­li­hoods of bil­lion bil­lions hap­pen and how many bits of ev­i­dence it would take, William Flesch sim­ply notes that

    Parfit seems to think that the prob­a­bil­ity that God ex­ists is greater than one in a bil­lion bil­lion, so that the ex­is­tence of God is more likely to be true than the ac­ci­den­tal ex­is­tence of a life-sup­port­ing uni­verse. But his stip­u­la­tion that he’s as­sum­ing that the claims of cur­rent cos­mol­ogy are true gives the game away. For even if you think that the odds that God ex­ists are greater than one in a bil­lion bil­lion, it’s dizzy­ingly more prob­a­ble that cos­mol­ogy has it wrong. (After all, sim­i­lar sorts of er­ror are not un­prece­dented in the his­tory of physic­s.) In fact, Parfit’s ar­gu­ment ought to em­bar­rass cos­mol­o­gists, not athe­ists. To para­phrase Parfit: cos­mol­o­gists may re­ject this an­swer, think­ing it im­prob­a­ble that their the­ory is wrong. But this prob­a­bil­ity can­not be as low as one in a bil­lion bil­lion. So even cos­mol­o­gists should ad­mit that, of these two an­swers to our ques­tion, the one that in­vokes sci­en­tific er­ror is more likely to be true.

  • An ex­am­ple from math­e­mat­ics by (“Vivid­ness in Math­e­mat­ics and Nar­ra­tive”, in Cir­cles Dis­turbed: The In­ter­play of Math­e­mat­ics and Nar­ra­tive), fo­cus­ing on the differ­ing cases of ra­tio­nal num­bers vs imag­i­nary num­bers:

    …a sug­ges­tion was made that are the math­e­mati­cian’s ver­sion of . I’m not sure I agree with that: when we give a proof by con­tra­dic­tion, we make it very clear that we are dis­cussing a coun­ter­fac­tu­al, so our words are in­tended to be taken at face val­ue. But per­haps this is not nec­es­sary. Con­sider the fol­low­ing pas­sage.

    There are those who would be­lieve that every poly­no­mial equa­tion with in­te­ger co­effi­cients has a ra­tio­nal so­lu­tion, a view that leads to some in­trigu­ing new ideas. For ex­am­ple, take the equa­tion x2 - 2 = 0. Let be a ra­tio­nal so­lu­tion. Then , from which it fol­lows that . The high­est power of 2 that di­vides p2 is ob­vi­ously an even pow­er, since if 2k is the high­est power of 2 that di­vides p, then 22_k_ is the high­est power of 2 that di­vides p2. Sim­i­lar­ly, the high­est power of 2 that di­vides 2_q_2 is an odd pow­er, since it is greater by 1 than the high­est power that di­vides q2. Since p2 and 2_q_2 are equal, there must ex­ist a pos­i­tive in­te­ger that is both even and odd. In­te­gers with this re­mark­able prop­erty are quite un­like the in­te­gers we are fa­mil­iar with: as such, they are surely wor­thy of fur­ther study.

    I find that it con­veys the ir­ra­tional­ity of √2 rather force­ful­ly. But could math­e­mati­cians afford to use this lit­er­ary de­vice? How would a reader be able to tell the differ­ence in in­tent be­tween what I have just writ­ten and the fol­low­ing su­per­fi­cially sim­i­lar pas­sage?

    There are those who would be­lieve that every poly­no­mial equa­tion has a so­lu­tion, a view that leads to some in­trigu­ing new ideas. For ex­am­ple, take the equa­tion x2 + 1 = 0. Let i be a so­lu­tion of this equa­tion. Then i2 + 1 = 0, from which it fol­lows that i^2 = −1. We know that i can­not be pos­i­tive, since then i2 would be pos­i­tive. Sim­i­lar­ly, i can­not be neg­a­tive, since i2 would again be pos­i­tive (be­cause the prod­uct of two neg­a­tive num­bers is al­ways pos­i­tive). And i can­not be 0, since 02 = 0. It fol­lows that we have found a num­ber that is not pos­i­tive, not neg­a­tive, and not ze­ro. Num­bers with this re­mark­able prop­erty are quite un­like the num­bers we are fa­mil­iar with: as such, they are surely wor­thy of fur­ther study.

    In­deed, how would a reader show the differ­ence—why do we ap­ply modus tol­lens when we ac­cept √2 must be ir­ra­tional but then ap­ply modus po­nens and ac­cept as be­ing real in some sense? Do we sim­ply ap­peal to the util­ity of us­ing i, and say with Wittgen­stein, “If a con­tra­dic­tion were now ac­tu­ally found in arith­metic—that would only prove that an arith­metic with such a con­tra­dic­tion in it could ren­der very good ser­vice; and it would be bet­ter for us to mod­ify our con­cept of the cer­tainty re­quired, than to say it would re­ally not yet have been a proper arith­metic.”6

  • if or ar­gu­ments sug­gest that it is moral to give away all one’s money to save many lives, does that re­fute al­tru­ism in gen­er­al? What about a re­jec­tion of em­pa­thy?

    For ex­am­ple, in re­sponse to the clas­sic demon­stra­tion of us­ing a dilemma of sav­ing birds from an oil slick, a com­men­ta­tor wrote:

    Scope in­sen­si­tiv­ity ex­ists, and to avoid the mis­takes caused by it, LW­ers say to shut up and mul­ti­ply—to take how much you’d do for one in­stance of what you care about, then mul­ti­ply the ex­pected util­ity even when the con­clu­sion may be coun­ter­in­tu­itive. For ex­am­ple, if you’d be will­ing to pay $3 to have the oil scrubbed off one nearby salient bird suffer­ing from an oil spill, you should be will­ing to pay more to have the oil scrubbed off >1 birds. But why not rea­son in the other di­rec­tion? In­stead of shut­ting up and mul­ti­ply­ing, why not shut up and di­vide? Sup­pose you’re not will­ing to do­nate any amount of money to save thou­sands of far­away birds. Then it would be ir­ra­tional for you to pay $3 to have the oil scrubbed off one salient bird. It’s true that it’s ir­ra­tional to both be will­ing to pay $3 to save one bird and not be will­ing to pay the same or more to save more birds. But from that alone, it does­n’t fol­low that you should do­nate >$3 to save more birds.

    Sim­i­lar­ly, ‘tor­ture vs dust specks’.

  • Pe­ter Van In­wa­gen (“Is It Wrong Every­where, Al­ways, and for Any­one to Be­lieve Any­thing on In­suffi­cient Ev­i­dence?”) ar­gues that it is morally fine to be­lieve in things (like gods) on in­suffi­cient ev­i­dence, be­cause if it was­n’t, that would mean that many non-re­li­gious be­liefs (with in­ad­e­quate ev­i­dence) would be morally wrong as well

  • Spencer Case, “Bear­ing Wit­ness: My Jour­ney Out of Mor­monism”:

    I con­tin­ued to dis­cuss my doubts with my dad and with my new bish­op, Bishop Ol­son, both of whom ad­mon­ished me to go on a mis­sion. At that point my de­par­ture would have been im­mi­nent. I re­call one phone con­ver­sa­tion with Bishop Ol­son in which he in­ad­ver­tently nudged me to part ways with the church. He said the fact that I was still in the church hav­ing these con­ver­sa­tions with him, seek­ing the truth, was proof that I re­ally did know that it was true. Oth­er­wise, what was the sense in my still go­ing to church? Why would I con­tinue seek­ing? He had a point. He ended the phone call with “See you in church this Sun­day.” I never went back.

  • if it is bad for Ko­re­ans to kill & eat dogs be­cause dogs can suffer and are in­tel­li­gent, does that mean it’s also bad to kill & eat pigs and many other an­i­mals and young hu­man in­fants?

  • if the ex­is­tence of God/souls means ar­ti­fi­cial in­tel­li­gence is im­pos­si­ble, then since ar­ti­fi­cial in­tel­li­gence in­creas­ingly looks pos­si­ble, does AI dis­prove gods/souls?

  • if it is im­per­mis­si­ble to at­tempt to make health­ier ba­bies by ge­netic se­lec­tion or en­gi­neer­ing (‘eu­gen­ics’), is it then im­per­mis­si­ble to use vac­ci­na­tion? What about PGD for ge­netic dis­or­ders, or abor­tion of fe­tuses with Down’s?

    “If we con­sider in­clu­sion and di­ver­sity to be a mea­sure of so­ci­etal pro­gress, then IQ screen­ing pro­pos­als are un­eth­i­cal,” says Lynn Mur­ray of Don’t Screen Us Out, a group that cam­paigns against pre­na­tal test­ing for Down’s syn­drome. “There must be wide con­sul­ta­tion.”

  • /: Ted Kaczyn­ski made many ob­ser­va­tions about tech­nol­ogy & civ­i­liza­tion that most would agree with, such as the even­tual takeover of Na­ture (), and con­cluded that as Na­ture is more valu­able than any­thing else, in­clud­ing hu­man well-be­ing, technology/civilization must be de­stroyed, rather than im­proved or ac­cel­er­ated

  • if, as ar­gues (Quan­tum Com­put­ing since Dem­ocri­tus, “Fun with the An­thropic Prin­ci­ple”), some highly con­tro­ver­sial ver­sions of weird s go poorly with Bayesian sta­tis­tics, does that dis­prove Bayesian sta­tis­tics?

  • , along with , ar­gued (in an an­thropic vein) that hu­man­ity must be of re­cent orig­in, given all the re­cent in­no­va­tions and the lack of recorded his­to­ry, con­sid­er­ing the ab­surd al­ter­na­tive

  • slav­ery:

    • Some ar­gu­ments given by et al 1853 in The Pro-Slav­ery Ar­gu­ment are strik­ing when quoted/paraphrased:

      “Fe­males are hu­man and ra­tio­nal be­ings. They may be found of bet­ter fac­ul­ties, and bet­ter qual­i­fied to ex­er­cise po­lit­i­cal priv­i­leges, and to at­tain the dis­tinc­tions of so­ci­ety, than many men; yet who com­plains of the or­der of so­ci­ety by which they are ex­cluded from them?” He says we have to func­tion with gen­eral rules that are good for so­ci­ety even if they vi­o­late the rights of in­di­vid­u­als (i.e., one 18 year old might be ca­pa­ble to hold po­lit­i­cal office but we still don’t al­low 18 year old­s).

      He goes on to list a bunch of ways in which so­ci­ety al­ready re­stricts rights and lib­erty (in­clud­ing an­i­mals) and why we think that’s fine and why it’s su­per nec­es­sary, and why are you sud­denly get­ting mad about slav­ery?

      …he weirdly ar­gues against util­i­tar­i­an­ism by say­ing how can you com­pare the plea­sure and suffer­ing of a man with cul­ti­vated and nu­anced taste to a man with dull and sim­ple taste?

      …“Who but a dri­v­el­ing fa­natic has thought of the ne­ces­sity of pro­tect­ing do­mes­tic an­i­mals from the cru­elty of their own­ers? And yet are not great and wan­ton cru­el­ties prac­tised on these an­i­mals?”

      …“[whip­ping] would be de­grad­ing to a free­man, who had the thoughts and as­pi­ra­tions of a free­man. In gen­er­al, it is not de­grad­ing to a slave, nor is it felt to be so. The evil is the bod­ily pain. Is it de­grad­ing to a child?”

    • was usu­ally em­ployed to jus­tify Amer­i­can slav­ery, point­ing to their docil­i­ty; but phre­nol­o­gist & abo­li­tion­ist ar­gued that phrenol­ogy proved that this proof of docil­ity showed that slav­ery could be abol­ished with­out the re­peat of prob­lems like the . (As no “war of ex­ter­mi­na­tion” took place after the Civil War when the slaves were freed, Combe ap­pears to have been cor­rect, if per­haps for the wrong rea­son­s…)

  • is crit­i­cized on the grounds that it is a highly ex­pen­sive way of cre­at­ing an­other pet cat or dog, while there are many an­i­mals that could be adopt­ed; one dog clon­er, Amy Vange­mert offers an in­ter­est­ing de­fense of her choice by anal­ogy to hu­man (non) adop­tion:

    She said: “I have had some se­ri­ous back­lash from peo­ple. A cou­ple of ac­quain­tances said I was wrong and it was in­hu­mane and there were so many dogs out there that need to be adopt­ed. But that’s like telling a mother that she should­n’t have her own child when there are chil­dren out there who need par­ents.”

  • An­to­nio Gar­cía Martínez crit­i­cizes the use of sta­tis­tics in Face­book ad­ver­tis­ing, not­ing that mod­els will in­evitably pick up cor­re­lates of var­i­ous sta­tuses like SES or race, and ‘dis­crim­i­nate’ based on this; he fol­lows the logic and notes that, since , the same facts which make Face­book ad­ver­tis­ing dis­crim­i­na­tory will make all other forms of ad­ver­tis­ing dis­crim­i­na­to­ry, such as ad­ver­tis­ing in only one mag­a­zine and not all pos­si­ble mag­a­zi­nes, and hence ad­ver­tis­ing must be even fur­ther banned or reg­u­lat­ed:

    It’s worth not­ing that if this reg­u­la­tory trend be­comes well es­tab­lished and more gen­er­al­ized, it could have im­pli­ca­tions way be­yond Face­book. Con­sider a mag­a­zine ad­ver­tiser who chooses to pub­li­cize se­nior ex­ec­u­tive po­si­tions in male-ori­ented but not in fe­male-ori­ented . Since mag­a­zine pub­lish­ers com­monly flaunt their spe­cific demos in sales pitch decks, it’s easy for ad­ver­tis­ers to seg­ment au­di­ences. Is that ad­ver­tiser vi­o­lat­ing the spirit of the law? I would say so. Should the gov­ern­ment en­force the law as they do with Face­book? Again, I would say so.

  • Is it im­moral to train self­-driv­ing cars on pub­lic roads when (as of 2019) they ap­pear to be only as safe as teenager or geri­atric car dri­vers?

    , who lives in test­ing hotspot Sun­ny­vale, fre­quently sees the cars on the road. He worked on them, too, as part of Google’s self­-driv­ing car project roughly a decade ago. Most ex­perts in the field say re­al-world test­ing is need­ed, he says, some­thing he agrees with. Tem­ple­ton says a small num­ber of crashes are ac­cept­able when con­sid­er­ing the even­tual over­all im­proved safety when hu­man dri­vers are off the roads. He com­pares it to teenagers learn­ing to dri­ve. “We ac­cept them dri­ving, with very high risk, be­cause it is the only way to turn them into safer mid­dle-aged dri­vers. And all we get out of that is one safer dri­ver,” he said. As au­tonomous ve­hi­cles are trained, “we get a mil­lion safer cars from a pro­to­type fleet of hun­dreds.”

    But John Joss, 85, does­n’t think the ro­bot dri­vers are that ma­ture. “They drive like ei­ther geri­atrics or 17-year-olds who have very lim­ited ex­pe­ri­ence of dri­ving,” said Joss, a mag­a­zine writer.

  • Re­ces­sions, sur­pris­ing­ly, ; does that mean re­ces­sions are good, or that de­creases in to­tal mor­tal­ity are bad (and the in­crease in deaths dur­ing boom times is good)?


  • Noriko: “Wow, you must have a real knack for it!”
  • Kazumi: “That’s not it, Miss Takaya! It takes hard work in or­der to achieve that.”
  • Noriko: “Hard work? You must have a knack for hard work, then!”7
  • Prob­a­bly the most fa­mous cur­rent ex­am­ple in sci­ence is the stil­l-con­tro­ver­sial , where sev­eral premises lead to an ex­per­i­men­tal­ly-fal­si­fied con­clu­sion, there­fore, by modus tol­lens, one of the premises is wrong—but which? There is no gen­eral agree­ment on which to re­ject, :

    • : re­jec­tion of the as­sump­tion of sta­tis­ti­cal in­de­pen­dence be­tween choice of mea­sure­ment & mea­sure­ment (ie. the uni­verse con­spires so the ex­per­i­menter al­ways just hap­pens to pick the ‘right’ thing to mea­sure and gets the right mea­sure­ment)
    • : re­jec­tion of as­sump­tion of lo­cal vari­ables, in fa­vor of uni­verse-wide vari­ables (ie. the uni­verse con­spires to link par­ti­cles, no mat­ter how dis­tant, to make the mea­sure­ment come out right)
    • : re­jec­tion of the speed of light as a lim­it, al­low­ing FTL/superluminal com­mu­ni­ca­tion (ie. the uni­verse con­spires to let two linked par­ti­cles com­mu­ni­cate in­stan­ta­neously to make the mea­sure­ment come out right)
    • : re­jec­tion of there be­ing a sin­gle mea­sure­ment in fa­vor of every pos­si­ble mea­sure­ment (ie. the uni­verse takes every pos­si­ble path, en­sur­ing it comes out right)
  • [Stephen Hawk­ing] tested the on 2009-06-28 by throw­ing a party & an­nounc­ing it later; he re­ported no one else at­tend­ed.8 Hawk­ing con­cluded that no time trav­el­ers at­tend­ed, that this was ev­i­dence for time trav­ellers not ex­ist­ing, and time travel be­ing im­pos­si­ble (con­sis­tent with ).

    How­ev­er, one could also con­clude that time trav­el­ers do not ex­ist be­cause hu­mans go ex­tinct be­fore in­vent­ing time trav­el, or that time trav­el­ers did at­tend the party in­vis­i­bly, or that they did be­cause Stephen Hawk­ing was a time trav­el­er!

  • ran­dom­ized con­trolled ex­per­i­ments (RCTs), par­tic­u­larly with blind­ing or pre­reg­is­tra­tion, es­pe­cially larger repli­ca­tions of small fa­mous cor­re­la­tional re­sults, typ­i­cally turn up much smaller or zero effects in med­i­cine, psy­chol­o­gy, and so­ci­ol­o­gy; the more rig­or­ous the ex­per­i­ment, the smaller the effect. The re­sponse to this is often to not ex­plain how merely flip­ping a coin can make gen­uine effects dis­ap­pear, but to at­tack the en­tire idea of RCTs/replication:

    • Rossi 1987 notes that a com­mon re­ac­tion in so­ci­ol­ogy to the fail­ure of many wel­fare or ed­u­ca­tion pro­grams, which ‘suc­ceeded’ when stud­ied at small scale or us­ing cor­re­la­tional data and then failed when tested with large ran­dom­ized ex­per­i­ments, is to deny that ran­dom­iza­tion or quan­ti­ta­tive mea­sure­ment are valid9

    • the Replication/Reproducibility cri­sis in psy­chol­ogy: Ja­son Mitchell ar­gues that the in­abil­ity of repli­ca­tors to con­firm the ‘stereo­type threat’ effect means that repli­ca­tion does­n’t work (and should­n’t be pub­lished):

      The re­cent spe­cial is­sue of So­cial Psy­chol­o­gy, for ex­am­ple, fea­tures one pa­per that suc­cess­fully re­pro­duced ob­ser­va­tions that Asian women per­form bet­ter on math­e­mat­ics tests when primed to think about their race than when primed to think about their gen­der. A sec­ond pa­per, fol­low­ing the same method­ol­o­gy, failed to find this effect (Moon & Roed­er, 2014); in fact, the 95% con­fi­dence in­ter­val does not in­clude the orig­i­nal effect size. These os­cil­la­tions should give se­ri­ous pause to fans of repli­cana. Ev­i­dent­ly, not all repli­ca­tors can gen­er­ate an effect, even when that effect is known to be re­li­able. On what ba­sis should we as­sume that other failed repli­ca­tions do not suffer the same un­spec­i­fied prob­lems that be­guiled Moon and Re­oder? The repli­ca­tion effort plainly suffers from a prob­lem of false neg­a­tives.

    • Mina Bis­sell, like­wise crit­i­ciz­ing repli­ca­tion ini­tia­tives be­cause bi­ol­ogy re­search is so frag­ile that they will not get the same re­sults (Gel­man com­men­tary):

      Many sci­en­tists use ep­ithe­lial cell lines that are ex­quis­itely sen­si­tive. The slight­est shift in their mi­croen­vi­ron­ment can al­ter the re­sult­s—­some­thing a new­comer might not spot. It is com­mon for even a sea­soned sci­en­tist to strug­gle with cell lines and cul­ture con­di­tions, and un­know­ingly in­tro­duce changes that will make it seem that a study can­not be re­pro­duced. Cells in cul­ture are often im­mor­tal be­cause they rapidly ac­quire epi­ge­netic and ge­netic changes. As such cells di­vide, any al­ter­ation in the me­dia or mi­croen­vi­ron­men­t—even if mi­nus­cule—­can trig­ger fur­ther changes that skew re­sults. Here are three ex­am­ples from my own ex­pe­ri­ence…

    • Tr­ish Green­halgh en­dorses a ban on RCTs be­cause of the null effects they keep find­ing:

      Here are some in­tel­lec­tual fal­lac­ies based on the more-re­search-is-needed as­sump­tion (I am sure read­ers will use the com­ments box to add more ex­am­ples).

      • De­spite dozens of ran­dom­ized con­trolled tri­als of self­-effi­cacy train­ing (the ‘ex­pert pa­tient’ in­ter­ven­tion) in chronic ill­ness, most peo­ple (e­spe­cially those with low so­cio-e­co­nomic sta­tus and/or low health lit­er­a­cy) still do not self­-man­age their con­di­tion effec­tive­ly. There­fore we need more ran­dom­ized tri­als of self­-effi­cacy train­ing.
      • De­spite con­flict­ing in­ter­pre­ta­tions (based largely on the value at­tached to ben­e­fits ver­sus those at­tached to harms) of the nu­mer­ous large, pop­u­la­tion-wide breast can­cer screen­ing stud­ies un­der­taken to date, we need more large, pop­u­la­tion-wide breast can­cer screen­ing stud­ies.
      • De­spite the al­most com­plete ab­sence of ‘com­plex in­ter­ven­tions’ for which a clin­i­cally as well as sta­tis­ti­cally sig­nifi­cant effect size has been demon­strated and which have proved both trans­fer­able and afford­able in the real world, the ran­dom­ized con­trolled trial of the ‘com­plex in­ter­ven­tion’ (as de­fined, for ex­am­ple, by the UK Med­ical Re­search Coun­cil [3]) should re­main the gold stan­dard when re­search­ing com­plex psy­cho­log­i­cal, so­cial and or­ga­ni­za­tional in­flu­ences on health out­comes.
      • De­spite con­sis­tent and re­peated ev­i­dence that elec­tronic pa­tient record sys­tems can be ex­pen­sive, re­source-hun­gry, fail­ure-prone and un­fit for pur­pose, we need more stud­ies to ‘prove’ what we know to be the case: that re­plac­ing pa­per with tech­nol­ogy will in­evitably save mon­ey, im­prove health out­comes, as­sure safety and em­power staff and pa­tients.

      Last year, Rodger Kessler and Russ Glas­gow pub­lished a pa­per ar­gu­ing for a ten-year mora­to­rium on ran­dom­ized con­trolled tri­als on the grounds that it was time to think smarter about the kind of re­search we need and the kind of study de­signs that are ap­pro­pri­ate for differ­ent kinds of ques­tion.[4]

    • “Build­ing an ev­i­dence base for IVF ‘ad­d-ons’”, Mack­lon et al 2019, like­wise echoes it:

      De­spite these chal­lenges, ma­jor and laud­able RCTs ad­dress­ing clin­i­cal ques­tions in our field reach pub­li­ca­tion in top jour­nals. How­ev­er, in ad­di­tion to shar­ing the nec­es­sary ma­jor fi­nan­cial and man­power in­vest­ment to per­form, their clear ten­dency to pro­duce neg­a­tive find­ings means that they are pri­mar­ily serv­ing to re­move treat­ment op­tions from the clin­i­cian and their pa­tient. This can of course rep­re­sent an im­por­tant con­tri­bu­tion. How­ev­er, when such tri­als test em­pir­i­cal treat­ments (which many IVF ‘ad­d-ons’ are), they risk in­creas­ing con­fu­sion rather than clar­i­ty.

  • Scan­di­na­vian pop­u­la­tion reg­istry stud­ies, which are able to link life­long gov­ern­ment data on medical/tax/school/employment/military/IQ on the en­tire pop­u­la­tion of a coun­try to per­form retroac­tive lon­gi­tu­di­nal stud­ies, are some­times crit­i­cized as not be­ing ap­plic­a­ble to other coun­tries; the irony is that the true Scandinavian/American differ­ence on any re­search ques­tion is likely smaller than the to­tal sys­tem­atic bi­ases + sam­pling er­ror in the non-pop­u­la­tion-reg­istry Amer­i­can stud­ies one would have to use to try to es­ti­mate what that differ­ence is.

  • dual n-back: on IQ and other ‘far trans­fer’ con­sis­tently find that stud­ies with weak method­ol­ogy like ‘pas­sive’ con­trol groups10 get stronger effects than ‘ac­tive’ con­trol groups; does this mean that DNB works?

  • pub­lished an analy­sis of mul­ti­ple ex­per­i­ments fol­low­ing stan­dard psy­chol­ogy pro­ce­dures & sta­tis­tics demon­strat­ing the ex­is­tence of ESP/psi; Bem be­lieves this proves psi, but does this prove psi or in­stead ? (See also Jaynes on ESP.)

    • in­ci­den­tal­ly, a psi en­thu­si­ast states:

      Given para­psy­chol­o­gy’s chronic un­der­fund­ing (it has been cal­cu­lated that ALL the fund­ing that has ever been re­ceived by para­psy­chol­o­gists would fund aca­d­e­mic psy­chol­ogy for ONE mon­th) it is sur­pris­ing the amount that has been learnt so far.

      …S­tu­art said: ‘the ex­per­i­menter’s skep­ti­cal at­ti­tude meant his psi in­ter­fered with the psi of the par­tic­i­pants’—it does­n’t take ex­per­i­menter psi to in­ter­fere with the sub­jects per­for­mance. Sim­ple ex­per­i­menter at­ti­tudes and other sub­tle cues are picked up by the sub­jects and this affects their per­for­mance. Skep­tic Wise­man ran a study with Mar­i­lyn Schlitz in­ves­ti­gat­ing dis­tant in­ten­tion. He got noth­ing, she found sig­nifi­cant ev­i­dence. Ex­actly the same set-up. At­ti­tudes make a differ­ence, and no spe­cial plead­ing to ex­per­i­menter psi is re­quired.

      This has long been a stan­dard psi re­sponse to the ob­ser­va­tion that psi ex­per­i­ments get much smaller or nil effects when con­ducted by non-be­liev­ers: it does­n’t in­di­cate prob­lems with the psi ex­per­i­ments, but rather the skep­tics are un­able to mea­sure psi be­cause their very skep­ti­cism emits an an­ti-psi field de­stroy­ing psi.

  • he­lio­cen­trism vs geo­cen­trism: Greek he­lio­cen­trism fa­mously pre­dicted , which was not ob­serv­able at the time; he­lio­cen­tris­m’s req­ui­site im­plied dis­tance to stars was then used as a modus tol­lens.

  • Be­fore the dis­cov­ery of the tim­ing er­ror, the 2011 was an ex­cel­lent place to ap­ply this ‘I defy (that par­tic­u­lar) data’ rea­son­ing, as are such er­rors in gen­er­al: Steven Kaas puts it nice­ly:

    Ac­cord­ing to [the 2009 blog post] “A New Chal­lenge to Ein­stein”, Gen­eral Rel­a­tiv­ity has been re­futed at 98% con­fi­dence. I won­der if it would­n’t be more ac­cu­rate to say that, ac­tu­al­ly, 98% con­fi­dence has been re­futed at Gen­eral Rel­a­tiv­i­ty.

  • given the fail­ure of per­son­al­ity GWASes and GCTAs in­di­cat­ing near-zero SNP her­i­tabil­i­ty, does a ge­net­ics pa­per claim­ing to iden­tify genes pre­dict­ing al­most all of per­son­al­ity her­i­tabil­ity us­ing only n~4k merely demon­strate that their method ?11

  • does the mean that much of the pop­u­la­tion was re­tarded a cen­tury ago (or does that merely prove that the effect is ‘hol­low’)?

    • more specifi­cal­ly: does the Flynn effect mean that the death penalty has been un­just in ex­e­cut­ing those we would now con­sider men­tally re­tard­ed?12
  • on a pos­si­ble cor­re­la­tion be­tween hor­mone and Obama/Romney vote-shares (highly likely to be spu­ri­ous):

    “There is ab­solutely no rea­son to ex­pect that wom­en’s hor­mones affect how they vote any more than there is a rea­son to sug­gest that vari­a­tions in testos­terone lev­els are re­spon­si­ble for vari­a­tions in the de­bate per­for­mances of Obama and Rom­ney,” said Su­san Car­roll, pro­fes­sor of po­lit­i­cal sci­ence and wom­en’s and gen­der stud­ies.

  • Group differ­ences:

    James H. Bor­land, a pro­fes­sor of ed­u­ca­tion at Teach­ers Col­lege, said that look­ing at the gifted land­scape in New York City sug­gests that one of two things must be true: ei­ther black and His­panic chil­dren are less likely to be gift­ed, or there is some­thing wrong with the way the city se­lects chil­dren for those pro­grams. “It is well known in the ed­u­ca­tion com­mu­nity that stan­dard­ized tests ad­van­tage chil­dren from wealth­ier fam­i­lies and dis­ad­van­tage chil­dren from poorer fam­i­lies,” Dr. Bor­land said…That changed in Sep­tem­ber 2008, when the Bloomberg ad­min­is­tra­tion ush­ered in ad­mis­sion based only on a cut­off score on two high­-s­takes tests given in one sit­ting—the Otis-Len­non School Abil­ity Test, or Ol­sat, and the Bracken School Readi­ness As­sess­ment. The over­haul was meant to stan­dard­ize the ad­mis­sions process and make it fair­er. But the new tests de­creased di­ver­si­ty, with chil­dren from the poor­est dis­tricts offered a smaller share of kinder­garten gifted slots after those were in­tro­duced, while pupils in the wealth­i­est dis­tricts got more.

  • -boost­ing in­ter­ven­tions:

    And in­deed, they’ve got­ten dra­matic re­sults. In one of the best-known stud­ies, low-per­form­ing black mid­dle school stu­dents who com­pleted sev­eral 15-minute class­room writ­ing ex­er­cises raised their GPAs by nearly half a point over two years, com­pared with a con­trol group. Such as­ton­ish­ing re­sults have struck some ob­server­s—­par­tic­u­larly nonpsy­chol­o­gist­s—as nearly mag­i­cal, and pos­si­bly un­be­liev­able. But a grow­ing body of ev­i­dence is show­ing that the in­ter­ven­tions can work, not only among black mid­dle school stu­dents, but also for wom­en, mi­nor­ity col­lege stu­dents and other pop­u­la­tions.

    “When this was first de­scribed to me, I was skep­ti­cal,” says physics pro­fes­sor Michael Dub­son, PhD, of the Uni­ver­sity of Col­orado-Boul­der, who worked with psy­chol­o­gists there on a study with women physics stu­dents. “But now that I think about it, we all know that it’s pos­si­ble to dam­age a stu­dent in 15 min­utes. It’s easy to wreck some­one’s self­-es­teem. So if that’s pos­si­ble, then maybe it’s also pos­si­ble to im­prove it.”

  • if a sur­vey of Mensa self­-re­ported di­ag­noses in­di­cates that high­-IQ in­di­vid­u­als are at rel­a­tive risks of phys­i­cal & psy­cho­log­i­cal dis­or­ders as high as RR = 223, con­tra­dict­ing al­most all pre­vi­ous re­search, does that in­di­cate that re­search was wrong or that Mensa sur­veys are not use­ful?

  • sup­pos­edly was tested as a child by his school with an IQ score in the 130s; given his ac­com­plish­ments, this is highly doubt­ful and a closer look at the source of the anec­dote re­veals many rea­sons why the score is ei­ther false or un­re­li­able

  • if an analy­sis claims that there is only a 1 in ten tril­lion chance that William Shake­speare wrote Shake­speare’s plays…

  • if the FDA uses com­pu­ta­tional mod­el­ing of chem­i­cals to ar­gue that a has “high po­ten­tial for abuse”, a plant which has been used by hun­dreds of thou­sands of Amer­i­cans for decades with­out a known ad­dic­tion epi­demic, does that es­tab­lish an im­pend­ing threat to Amer­i­can pub­lic health, or re­fute their com­pu­ta­tional mod­el­ing?

    And if a mouse model shows some dopamin­er­gic effects of —which has been used by mil­lion of Amer­i­cans over the past 2–3 decades—­does that in­di­cate modafinil is at risk of se­ri­ous drug ad­dict abuse, or ?

  • if Freudian psy­chother­apy works as well as (CBT), does CBT not work? More gen­er­al­ly, if the is true, do any of these psy­cho­log­i­cal treat­ments work or are their the­o­ries true?

  • if and a re­view can be writ­ten propos­ing us­ing it for sports, does that de­bunk re­views or sug­gest that nico­tine might be use­ful?

  • When Seth Roberts ar­gues that one’s sub­jec­tive mem­o­ries about sleep con­flict with the sleep data recorded by one’s EEG sleep­-track­ing de­vice, does that con­sti­tute a dis­proof of the Zeo’s ac­cu­ra­cy? No: es­tab­lish­ing con­tra­dic­tions be­tween one’s memories/subjective im­pres­sions and the Zeo merely tells us that one (or both) are wrong; it does­n’t tell us that the Zeo is wrong un­less you have ad­di­tional data or ar­gu­ments which say that the Zeo is less re­li­able than the mem­o­ries. One could take the Zeo con­tra­dict­ing mem­o­ries as just proof of the fal­li­bil­ity of sleep­-re­lated mem­o­ries (eg Feige et al 2008)13! (The fun­da­men­tal ques­tion of epis­te­mol­o­gy: “What do you be­lieve, and why do you be­lieve it?”)

    For ex­am­ple, if some­one is caught on cam­era sleep­-walk­ing, and de­nies stren­u­ously that he was sleep­-walk­ing, do you take modus po­nens and say his mem­o­ries prove he was not sleep­-walk­ing and re­ject the cam­era footage; or modus tol­lens and say that the claim his sleep mem­o­ries are re­li­able im­ply he could not have been caught on cam­era, but he was, there­fore we can re­ject the claim his sleep mem­o­ries im­ply no walk­ing? But ex­tra­or­di­nary claims re­quire ex­tra­or­di­nary ev­i­dence, so you choose to take modus tol­len­s—be­cause you have pri­ors which say that mem­o­ries are mal­leable and un­trust­wor­thy, while cam­era footage is much harder to fake.

  • model check­ing & un­cer­tainty: if a sta­tis­ti­cal analy­sis as­sumes that hav­ing an ex­tremely high IQ is nor­mally dis­trib­uted (rather than a ) and that the im­pli­ca­tion of a head­count of high IQ types at Har­vard im­plies that >90% of high IQ peo­ple are fail­ures & dis­crim­i­nated against by so­ci­ety, does that show that al­most every smart per­son is doomed, or that the sta­tis­ti­cal model is in­cor­rect?

  • if obe­sity and In­ter­net use are ad­dic­tive in sim­i­lar ways, should we take In­ter­net us­age much more se­ri­ous­ly?

  • Christo­pher Ryan:

    What I’m say­ing is that to ar­gue that our an­ces­tors were sex­ual om­ni­vores is no more a crit­i­cism of monogamy than to ar­gue that our an­ces­tors were di­etary om­ni­vores is a crit­i­cism of veg­e­tar­i­an­ism.

  • if the Civil War wiped out most of the wealth of many rich fam­i­lies, but their chil­dren grad­u­ally re­cov­ered in SES de­spite be­ing raised poor with no in­her­i­tance, does that show the im­por­tance of ge­net­ics to hu­man cap­i­tal, or that ac­tu­al­ly, all in­ter­gen­er­a­tional trans­mis­sion is thanks to “the im­por­tance of so­cial net­works in fa­cil­i­tat­ing em­ploy­ment op­por­tu­ni­ties and ac­cess to credit”?

  • the ex­is­tence of stars, when cal­cu­lated his (earn­ing a No­bel fully 48 years lat­er); ap­par­ently also re­al­ized the im­pli­ca­tion, but “con­cluded that quan­tum laws might be in­valid for stars heav­ier than 1.5 so­lar mass.”, while balked at the im­pli­ca­tions en­tirely:

    Dr. Chan­drasekhar had got this re­sult be­fore, but he has rubbed it in in his last pa­per; and, when dis­cussing it with him, I felt dri­ven to the con­clu­sion that this was al­most a re­duc­tio ad ab­sur­dum of the rel­a­tivis­tic de­gen­er­acy for­mu­la. Var­i­ous ac­ci­dents may in­ter­vene to save the star, but I want more pro­tec­tion than that. I think there should be a law of Na­ture to pre­vent a star from be­hav­ing in this ab­surd way!

    If one takes the math­e­mat­i­cal de­riva­tion of the rel­a­tivis­tic de­gen­er­acy for­mula as given in as­tro­nom­i­cal pa­pers, no fault is to be found. One has to look deeper into its phys­i­cal foun­da­tions, and these are not above sus­pi­cion. The for­mula is based on a com­bi­na­tion of rel­a­tiv­ity me­chan­ics and non-rel­a­tiv­ity quan­tum the­o­ry, and I do not re­gard the off­spring of such a union as born in law­ful wed­lock…

  • is the life span of the old­est per­son to ever live, ,

  • which is more like­ly, that can de­stroy 95% of a hu­man brain with­out nec­es­sar­ily re­duc­ing in­tel­li­gence & pos­si­bly even in­creas­ing in­tel­li­gence in re­ported cas­es—or that this con­flates brain vol­ume with brain mat­ter, and the hy­dro­cephalus cases in ques­tion are ?

  • in de­fend­ing the ex­is­tence of the Pyg­malion effect, Rosen­thal & Ja­cob­son 1968 in­voke an ear­lier an­i­mal ex­per­i­ment of theirs (done to crit­i­cize ): “If an­i­mals be­come ‘brighter’ when ex­pected to by their ex­per­i­menters, then it seemed rea­son­able to think that chil­dren might be­come brighter when ex­pected to by their teach­ers.” How­ev­er, the Pyg­malion effect has been de­bunked; so, if chil­dren do not be­come brighter when ex­pected to by their teach­ers, then surely that casts doubt on their claim that the an­i­mals be­came brighter too…?

  • Bi­o­log­i­cal cells build up waste-prod­ucts like as they age, which they can­not or do not dis­pose of. Chemist Jo­han Bjork­sten noted of such , to quote Mike Dar­win’s de­scrip­tion14:

    He had no­ticed that as or­gan­isms age, they tend to ac­cu­mu­late in­sol­uble, often pig­mented mat­ter in­side their non-di­vid­ing cells. Lipo­fus­cin, which ac­cu­mu­lates most promi­nently in brain and car­diac cells, is one such “age pig­ment.”…B­jork­sten de­ter­mined that this in­sol­u­ble ma­te­ri­al, which could oc­cupy as much as 30% to 40% of the vol­ume of non-di­vid­ing cells in aged an­i­mals, con­sisted largely of cross linked mol­e­cules of lipids and pro­teins. So mol­e­c­u­larly cross linked, com­pact and tough was this ma­te­r­ial that it was com­pletely re­sis­tant to di­ges­tion by trypsin and other com­monly avail­able “di­ges­tive” bi­o­log­i­cal en­zymes.

    This posed a puz­zle for Bjork­sten, be­cause if no liv­ing sys­tems could de­com­pose this ma­te­ri­al, it was so sta­ble that it would nec­es­sar­ily re­main as in­di­gestible de­bris after each or­gan­ism died. Thus, the earth should be cov­ered in such de­bris by now! Clear­ly, this is not case, and so this im­plied to Bjork­sten that there must, in fact, be liv­ing or­gan­isms with spe­cial­ized en­zymes ca­pa­ble of break­ing down this ma­te­ri­al…He set out to find en­zymes in na­ture which could re­verse these cross links and thus, he thought, re­verse ag­ing.

    This ob­ser­va­tion has been broad­ened to the “mi­cro­bial in­fal­li­bil­ity hy­poth­e­sis” of : “mi­croor­gan­isms will be found to de­grade every chem­i­cal sub­stance syn­the­sized by any liv­ing or­gan­ism.” Later geron­tol­o­gist ap­plied the same rea­son­ing in look­ing for en­zymes to break down , specifi­cal­ly, in 2007 (“Up­grad­ing the Bi­o­log­i­cal In­cin­er­a­tors”, pg121):

    …what was needed was a bio­med­ical su­per­fund pro­jec­t…There were ac­tual land sites all over the planet that should be very badly con­t­a­m­i­nated by lipo­fus­cin, be­cause their soil has been seeded with the stuff for gen­er­a­tions. I speak, of course, of grave­yards…there was no ac­cu­mu­la­tion of lipo­fus­cin in ceme­ter­ies—and if there was, we cer­tainly ought to be aware of it, be­cause lipo­fus­cin is flu­o­res­cent. Months lat­er, when I was dis­cussing the is­sue with fel­low Cam­bridge sci­en­tist John Archer, he would put the dis­con­nect suc­cinct­ly: “Why don’t grave­yards glow in the dark?”

    …Sci­en­tists be­came in­ter­ested in this phe­nom­e­non in the 1950s, when it was noted that the lev­els of many hard-to-de­grade pol­lu­tants at con­t­a­m­i­nated sites were present at much lower lev­els than would have been ex­pect­ed. A big part of the ex­pla­na­tion turned out to be the rapid evo­lu­tion of quickly re­pro­duc­ing or­gan­isms like bac­te­ria. Any highly en­er­gy-rich sub­stance rep­re­sents a po­ten­tial feast­—and thus, an eco­log­i­cal niche—­for any or­gan­ism pos­sess­ing the en­zymes needed to di­gest that ma­te­r­ial and lib­er­ate its stored en­er­gy.


  • if the head in­jury rates for cars and bi­cy­cles are sim­i­lar (or var­i­ous sports, or for show­ers, or walk­ing, es­pe­cially in the el­der­ly), and one would­n’t wear a hel­met in­side a car, should one not wear a bi­cy­cle hel­met as well?

  • if a soda has the equiv­a­lent of two Cad­bury choco­late eggs of sugar or 6 donuts, per­haps one should feel less guilty about eat­ing candy

  • if ter­ror­ism is less of a mor­tal­ity risk than bath­tub ac­ci­dents, and we spend much more on fight­ing ter­ror­ism than on house­hold ac­ci­dents, should we spend less on ter­ror­ism or more on ac­ci­dent re­duc­tion?

  • On Swe­den, Tyler Cowen has said:

    Q: “…And then they look at Canada, or Scan­di­navia, which are less cap­i­tal­ist than the US, and that looks pretty good to them. And so they blame cap­i­tal­ism. Do you think that they’re wrong? Which part of that story is wrong?”

    Tyler Cowen: “Peo­ple do look at Swe­den, but they rarely point out that per capita in­come in West Vir­ginia right now is about the same as that in Swe­den. The state that’s sup­posed to be our biggest train wreck is about as rich as Swe­den. I’m not say­ing the qual­ity of pub­lic goods is al­ways as high in West Vir­ginia, but I think that fact is not nearly widely enough known.”

  • : are mass bomb threats to Jews in Amer­ica ev­i­dence of lurk­ing white su­prema­cists em­pow­ered by the elec­tion of Don­ald Trump, or merely hoaxes by a so­cio­pathic Is­raeli Jew?

  • does Eu­ro­pean im­mi­gra­tion into Amer­ica and the Na­tive Amer­i­cans im­ply that Amer­ica should have an Open Bor­ders pol­i­cy?

  • in the ‘so­cial­ist cal­cu­la­tion de­bate’, that cen­tral­ized plan­ning would in­cur in­fea­si­ble com­pu­ta­tional de­mands; now that both com­put­ers & plan­ning al­go­rithms have ad­vanced by many or­ders of mag­ni­tude, does that mean so­cial­ism is pos­si­ble?

  • On Por­tuguese in­vest­ments: “Por­tuguese busi­ness lead­ers say that An­golan in­vest­ments un­fairly at­tract the kind of scrutiny that money from else­where, in­clud­ing Chi­na, does not.”

  • the vi­sual novel re­lies on a “treach­ery of im­ages” twist in or­der to de­ceive the read­er; one reader praises the writ­ing for be­ing so bad that the reader should be able to de­duce this twist from the be­gin­ning:

    Look­ing at this scene now, it’s pretty amaz­ing that I never fig­ured out that Kinzo was dead from the be­gin­ning. Every­thing about not be­ing able to meet him di­rectly and Nan­jo’s hedg­ing is so ridicu­lously sus­pi­cious. Ryuk­ishi was prob­a­bly bank­ing on peo­ple us­ing that to fig­ure out that not every scene could be taken lit­er­al­ly.

  • the log­i­cal in­ver­sion can be used for comedic effect: many Oglaf (NSFW) and SMBC comics in par­tic­u­lar draw on this


Jaynes on ESP

Bayesian , in “Chap­ter 5: Queer uses for prob­a­bil­ity the­ory”, dis­cusses the prob­a­bilis­tic gen­er­al­iza­tion of the rea­son­ing we are en­gaged in when we choose whether to modus po­nens or modus tol­lens, with early ESP ex­per­i­ments as an ex­am­ple, point­ing out that from a Bayesian per­spec­tive, all claims are be­ing eval­u­ated in a larger Bayesian-mod­el-com­par­i­son con­text where is­sues like ex­per­i­menter er­ror or bias are al­ways pos­si­bil­i­ties:

What prob­a­bil­ity would you as­sign to the hy­poth­e­sis that Mr. Smith has per­fect ex­trasen­sory per­cep­tion (ESP)? He can guess right every time which num­ber you have writ­ten down. To say zero is too dog­mat­ic…We take this man who says he has ex­trasen­sory per­cep­tion, and we will write down some num­bers from 1 to 10 on a piece of pa­per and ask him to guess which num­bers we’ve writ­ten down. We’ll take the usual pre­cau­tions to make sure against other ways of find­ing out. If he guesses the first num­ber cor­rect­ly, of course we will all say “you’re a very lucky per­son, but I don’t be­lieve it.” And if he guesses two num­bers cor­rect­ly, we’ll still say “you’re a very lucky per­son, but I don’t be­lieve it.” By the time he’s guessed four num­bers cor­rect­ly—well, I still would­n’t be­lieve it. So my state of be­lief is cer­tainly lower than −40 db. How many num­bers would he have to guess cor­rectly be­fore you would re­ally se­ri­ously con­sider the hy­poth­e­sis that he has ex­trasen­sory per­cep­tion? In my own case, I think some­where around 10. My per­sonal state of be­lief is, there­fore, about −100 db. You could talk me into a ±10 change, and per­haps as much as ±30, but not much more than that.

But on fur­ther thought we see that, al­though this re­sult is cor­rect, it is far from the whole sto­ry. In fact, if he guessed 1000 num­bers cor­rect­ly, I still would not be­lieve that he has ESP, for an ex­ten­sion of the same rea­son that we noted in Chap­ter 4 when we first en­coun­tered the phe­nom­e­non of res­ur­rec­tion of dead hy­pothe­ses. An hy­poth­e­sis A that starts out down at −100 db can hardly ever come to be be­lieved what­ever the data, be­cause there are al­most sure to be al­ter­na­tive hy­pothe­ses above it, per­haps down at −60 db. Then when we get as­ton­ish­ing data that might have res­ur­rected A, the al­ter­na­tives will be res­ur­rected in­stead. Let us il­lus­trate this by two fa­mous ex­am­ples, in­volv­ing telepa­thy and the .

…on the ba­sis of such a re­sult [as Mrs. Stew­art’s ex­per­i­men­tal re­sults in _ Mod­ern Ex­per­i­ments In Telepa­thy_, Soal & Bate­man 1954], ESP re­searchers would pro­claim a vir­tual cer­tainty that ESP is re­al. …it hardly mat­ters what these prior prob­a­bil­i­ties are; in the view of an ESP re­searcher who does not con­sider the prior prob­a­bil­ity par­tic­u­larly small, is so close to unity that its dec­i­mal ex­pres­sion starts with over a hun­dred 9’s. He will then re­act with anger and dis­may when, in spite of what he con­sid­ers this over­whelm­ing ev­i­dence, we per­sist in not be­liev­ing in ESP. Why are we, as he sees it, so per­versely il­log­i­cal and un­sci­en­tific? The trou­ble is that the above cal­cu­la­tions (5-9) and (5-12) rep­re­sent a very naive ap­pli­ca­tion of prob­a­bil­ity the­o­ry, in that they con­sider only Hp and Hf; and no other hy­pothe­ses. If we re­ally knew that Hp and Hf were the only pos­si­ble ways the data (or more pre­cise­ly, the ob­serv­able re­port of the ex­per­i­ment and data) could be gen­er­at­ed, then the con­clu­sions that fol­low from (5-9) and (5-12) would be per­fectly all right. But in the real world, our in­tu­ition is tak­ing into ac­count some ad­di­tional pos­si­bil­i­ties that they ig­nore.

…When we are deal­ing with some ex­tremely im­plau­si­ble hy­poth­e­sis, recog­ni­tion of a seem­ingly triv­ial al­ter­na­tive pos­si­bil­ity can make or­ders of mag­ni­tude differ­ence in the con­clu­sions. Tak­ing note of this, let us show how a more so­phis­ti­cated ap­pli­ca­tion of prob­a­bil­ity the­ory ex­plains and jus­ti­fies our in­tu­itive doubts.

Let Hp, Hf, and Lp, Lf, Pp, Pf be as above; but now we in­tro­duce some new hy­pothe­ses about how this re­port of the ex­per­i­ment and data might have come about, which will surely be en­ter­tained by the read­ers of the re­port even if they are dis­counted by its writ­ers. These new hy­pothe­ses range all the way from in­no­cent pos­si­bil­i­ties such as un­in­ten­tional er­ror in the record keep­ing, through friv­o­lous ones (per­haps Mrs. Stew­art was hav­ing fun with those fool­ish peo­ple, with the aid of a lit­tle mir­ror that they did not no­tice), to less in­no­cent pos­si­bil­i­ties such as se­lec­tion of the data (not re­port­ing the days when Mrs. Stew­art was not at her best), to de­lib­er­ate fal­si­fi­ca­tion of the whole ex­per­i­ment for wholly rep­re­hen­si­ble mo­tives. Let us call them all, sim­ply, “de­cep­tion”. For our pur­poses it does not mat­ter whether it is we or the re­searchers who are be­ing de­ceived, or whether the de­cep­tion was ac­ci­den­tal or de­lib­er­ate. Let the de­cep­tion hy­pothe­ses have like­li­hoods and prior prob­a­bil­i­ties . There are, per­haps, 100 differ­ent de­cep­tion hy­pothe­ses that we could think of and are not too far-fetched to con­sid­er, al­though a sin­gle one would suffice to make our point. In this new log­i­cal en­vi­ron­ment, what is the pos­te­rior prob­a­bil­ity of the hy­poth­e­sis Hf that was sup­ported so over­whelm­ingly be­fore? Prob­a­bil­ity the­ory now tells us: (5-13)

In­tro­duc­tion of the de­cep­tion hy­pothe­ses has changed the cal­cu­la­tion great­ly; in or­der for to come any­where near unity it is now nec­es­sary that: (5-14)

From (5-7), is com­pletely neg­li­gi­ble so (5-14) is not greatly differ­ent from: (5-15)

But each of the de­cep­tion hy­pothe­ses is, in my judg­ment, more likely than Hf, so there is not the re­motest pos­si­bil­ity that in­equal­ity (5-15) could ever be sat­is­fied.

There­fore, this kind of ex­per­i­ment can never con­vince me of the re­al­ity of Mrs. Stew­art’s ESP; not be­cause I as­sert dog­mat­i­cally at the start, but be­cause the ver­i­fi­able facts can be ac­counted for by many al­ter­na­tive hy­pothe­ses.

…In­deed, the very ev­i­dence which the ESPers throw at us to con­vince us, has the op­po­site effect on our state of be­lief; is­su­ing re­ports of sen­sa­tional data de­feats its own pur­pose. For if the prior prob­a­bil­ity of de­cep­tion is greater than that of ESP, then the more im­prob­a­ble the al­leged data are on the null hy­poth­e­sis of no de­cep­tion and no ESP, the more strongly we are led to be­lieve, not in ESP, but in de­cep­tion. For this rea­son, the ad­vo­cates of ESP (or any other mar­vel) will never suc­ceed in per­suad­ing sci­en­tists that their phe­nom­e­non is re­al, un­til they learn how to elim­i­nate the pos­si­bil­ity of de­cep­tion in the mind of the read­er.

It is in­ter­est­ing that per­ceived this phe­nom­e­non long ago. His Es­sai Philosophique sur les prob­a­bil­ités (1819) has a long chap­ter on the , in which he calls at­ten­tion to “the im­mense weight of tes­ti­monies nec­es­sary to ad­mit a sus­pen­sion of nat­ural laws”. He notes that those who make recitals of mir­a­cles, “de­crease rather than aug­ment the be­lief which they wish to in­spire; for then those recitals ren­der very prob­a­ble the er­ror or the false­hood of their au­thors. But that which di­min­ishes the be­lief of ed­u­cated men often in­creases that of the un­e­d­u­cat­ed, al­ways avid for the mar­velous.”

We ob­serve the same phe­nom­e­non at work to­day, not only in the ESP en­thu­si­ast, but in the as­trologer, rein­car­na­tion­ist, ex­or­cist, fun­da­men­tal­ist preacher or cultist of any sort, who at­tracts a loyal fol­low­ing among the un­e­d­u­cated by claim­ing all kinds of mir­a­cles; but has zero suc­cess in con­vert­ing ed­u­cated peo­ple to his teach­ings. Ed­u­cated peo­ple, taught to be­lieve that a cause-effect re­la­tion re­quires a phys­i­cal mech­a­nism to bring it about, are scorn­ful of ar­gu­ments which in­voke mir­a­cles; but the un­e­d­u­cated seem ac­tu­ally to pre­fer them. [see also David Hume’s “Of Mir­a­cles”]

Slavery and Phrenology

(1788–1858) ap­pears to have made a phreno­log­i­cal ar­gu­ment (that African docil­ity im­plies the moral­ity of abo­li­tion rather than main­tain­ing slav­ery) in 3 places—a mar­ginal note on a let­ter from a pro-slav­ery ad­vo­cate in 1839; his Notes on the United States of Amer­ica, 1841; and his Sys­tem of Phrenol­ogy, 1843:

  1. Cald­well’s long let­ter on the ‘an­i­mal or­gans’ reached Combe in Sep­tem­ber 1839; as re­pro­duced & sum­ma­rized by Pos­kett 2016:

    It did not mat­ter, as most whites agreed, that Africans pos­sessed in­fe­rior brains. What mat­tered was whether they could be en­trusted with free­dom. Fill­ing the space to the side of Cald­well’s first sheet, Combe com­pared the char­ac­ter of Africans and Na­tive Amer­i­cans. Draw­ing on Cald­well’s own an­i­mal com­par­ison, he wrote that Na­tive Amer­i­cans were ‘in­domitable, fe­ro­cious sav­ages. They are not tame­able. They are not slaves be­cause they are not tame­able’. With the still rag­ing in Flori­da, this ar­gu­ment ap­pealed di­rectly to south­ern fears of the Na­tive Amer­i­can pop­u­la­tion. In con­trast, Combe ar­gued, ‘Africans are mild, docile & in­tel­li­gent, com­pared with them. They are slaves be­cause they are tame­able’. Here Combe first ex­pressed an idea in note-form which he would later re­turn to in print. Sto­ries of vi­o­lent slave re­bel­lions in Vir­ginia and Ja­maica fu­elled white fears of im­me­di­ate abo­li­tion through­out the 1830s. Cald­well him­self ar­gued that ‘strife and blood­-shed would soon be­come the daily oc­cu­pa­tion’ of the free African. Combe re­sponded to these con­cerns, ar­gu­ing that phrenol­ogy in fact showed African char­ac­ter to be es­sen­tially placid, writ­ing that ‘the qual­i­ties which make them sub­mit to slav­ery are a guar­an­tee that if eman­ci­pated & justly dealt with, they wd [sic] not shed blood’.51 Combe later re­peated this ar­gu­ment in the 1840s, first in his Notes on the United States of North Amer­ica and then in the fifth and ex­panded edi­tion of his Sys­tem of Phrenol­ogy.52 But he first worked it out in the mar­gins of Cald­well’s let­ter.

  2. Notes:

    Mr. Clay re­gards it as cer­tain, that if slav­ery were abol­ished, a war of ex­ter­mi­na­tion would en­sue be­tween the races, which would lead to greater evils than those gen­er­ated by slav­ery. This is the ar­gu­ment of the white man, of the mas­ter, in whose eyes his own losses or suffer­ings are pon­der­ous as gold, and those of three mil­lions of Ne­groes light as a feath­er. Ask the Ne­groes their opin­ion of the mis­eries of the ex­ist­ing sys­tem and weigh this against the evils an­tic­i­pated by the Whites from eman­ci­pa­tion, and then strike the bal­ance. Be­fore I had an op­por­tu­nity of study­ing the Ne­gro char­ac­ter and Ne­gro brain, I en­ter­tained the same opin­ion with Mr. Clay, that a war of ex­ter­mi­na­tion would be the con­se­quence of im­me­di­ate free­dom. More ac­cu­rate and ex­ten­sive in­for­ma­tion has in­duced me to change this view. I may here an­tic­i­pate a state­ment which be­longs, in chrono­log­i­cal or­der, to a more ad­vanced date, name­ly, that I have stud­ied the cra­nia of the North Amer­i­can In­di­ans and of the Ne­groes in var­i­ous parts of the United States, and also ob­served their liv­ing heads, and have ar­rived at the fol­low­ing con­clu­sions. The North Amer­i­can In­di­ans have given bat­tle to the Whites, and per­ished be­fore them, but have never been re­duced ei­ther to na­tional or to per­sonal servi­tude. The de­vel­op­ment of the brains shows large or­gans of De­struc­tive­ness, Se­cre­tive­ness, Cau­tious­ness, Self­-Es­teem, and Firm­ness, with de­fi­cient or­gans of Benev­o­lence, Con­sci­en­tious­ness, and Re­flec­tion. This in­di­cates a nat­ural char­ac­ter that is proud, cau­tious, cun­ning, cru­el, ob­sti­nate, vin­dic­tive, and lit­tle ca­pa­ble of re­flec­tion or com­bi­na­tion. The brain of the Ne­gro, in gen­eral (for there are great va­ri­eties among the African race, and in­di­vid­ual ex­cep­tions are pretty nu­mer­ous), shows pro­por­tion­ately less De­struc­tive­ness, Cau­tious­ness, Self­-Es­teem, and Firm­ness, and greater Benev­o­lence, Con­sci­en­tious­ness, and Re­flec­tion, than the brain of the na­tive Amer­i­can. In short, in the Ne­gro brain the moral and Re­flect­ing or­gans are of larger size, in pro­por­tion to the or­gans of the an­i­mal propen­si­ties now enu­mer­at­ed, than in that of the In­di­an. The Ne­gro is, there­fore, nat­u­rally more sub­mis­sive, docile, in­tel­li­gent, pa­tient, trust­wor­thy, and sus­cep­ti­ble of kindly emo­tions, and less cru­el, cun­ning, and vin­dic­tive, than the other race.

    These differ­ences in their nat­ural dis­po­si­tions throw some light on the differ­ences of their fates. The Amer­i­can In­dian has es­caped the degra­da­tion of slav­ery, be­cause he is a wild, vin­dic­tive, cun­ning, un­tame­able sav­age, too dan­ger­ous to be trusted by the white men in so­cial in­ter­course with them­selves, and more­over, too ob­tuse and in­tractable to be worth co­erc­ing into servi­tude. The African has been de­prived of free­dom and ren­dered “prop­er­ty,” ac­cord­ing to Mr. Clay’s view, be­cause he is by na­ture a tame man, sub­mis­sive, affec­tion­ate, in­tel­li­gent and docile. He is so lit­tle cru­el, cun­ning, fierce, and vin­dic­tive, that the white men can op­press him far be­yond the lim­its of In­dian en­durance, and still trust their lives and prop­erty within his reach; while he is so in­tel­li­gent that his la­bor is worth ac­quir­ing. The na­tive Amer­i­can is free, be­cause he is too dan­ger­ous and too worth­less a be­ing to be valu­able as a slave; the Ne­gro is in bondage, be­cause his na­tive dis­po­si­tions are es­sen­tially ami­able. The one is like the wolf or the fox, the other like the dog. In both, the brain is in­fe­rior in size, par­tic­u­larly in the moral and in­tel­lec­tual re­gions, to that of the An­glo-Saxon race, and hence the foun­da­tion of the nat­ural su­pe­ri­or­ity of the lat­ter over both; but my con­vic­tion is, that the very qual­i­ties which ren­der the Ne­gro in slav­ery a safe com­pan­ion to the White, would make him harm­less when free. If he were by na­ture proud, iras­ci­ble, cun­ning and vin­dic­tive, he would not be a slave; and as he is not so, free­dom will not gen­er­ate these qual­i­ties in his mind; the fears, there­fore, gen­er­ally en­ter­tained of his com­menc­ing, if eman­ci­pat­ed, a war of ex­ter­mi­na­tion, or for su­premacy over the Whites, ap­pear to me to be un­found­ed; un­less, after his eman­ci­pa­tion, the Whites should com­mence a war of ex­ter­mi­na­tion against him. The re­sults of have hith­erto borne out these views, and I an­tic­i­pate that the fu­ture will still far­ther con­firm them.

  3. A Sys­tem of Phrenol­ogy, sim­i­lar­ly:

    I have stud­ied the cra­nia and liv­ing heads of North Amer­i­can In­di­ans and of Ne­groes in var­i­ous parts of the United States, and, after con­sid­er­ing their his­to­ry, I sub­mit the fol­low­ing re­marks. The North Amer­i­can In­di­ans have given bat­tle to the Whites, and per­ished be­fore them, but have never been re­duced ei­ther to na­tional or to per­sonal servi­tude. The de­vel­op­ment of their brains shews large or­gans of De­struc­tive­ness, Se­cre­tive­ness, Cau­tious­ness, Self­-Es­teem, and Firm­ness, with de­fi­cient or­gans of Benev­o­lence, Con­sci­en­tious­ness, and Re­flec­tion. This in­di­cates a nat­ural char­ac­ter that is proud, cau­tious, cun­ning, cru­el, ob­sti­nate, vin­dic­tive, and lit­tle ca­pa­ble of re­flec­tion or com­bi­na­tion. The brain of the Ne­gro, in gen­eral (for there are great va­ri­eties among the African race, and in­di­vid­ual ex­cep­tions are pretty nu­mer­ous), shews pro­por­tion­ately less De­struc­tive­ness, Cau­tious­ness, Self­-Es­teem, and Firm­ness, and greater Benev­o­lence, Con­sci­en­tious­ness, and Re­flec­tion, than the brain of the na­tive Amer­i­can. In short, in the Ne­gro brain the moral and re­flect­ing or­gans are of larger size, in pro­por­tion to the or­gans of the an­i­mal propen­si­ties now enu­mer­at­ed, than in that of the In­di­an. The Ne­gro is, there­fore, nat­u­rally more sub­mis­sive, docile, in­tel­li­gent, pa­tient, trust­wor­thy, and sus­cep­ti­ble of kindly emo­tions, and less cru­el, cun­ning, and vin­dic­tive, than the other race.

    These differ­ences in their nat­ural dis­po­si­tions throw some light on the differ­ences of their fates. The North Amer­i­can In­dian has es­caped the degra­da­tion of slav­ery, be­cause he is a wild, vin­dic­tive, cun­ning, un­tame­able sav­age, too dan­ger­ous to be trusted by the white men in so­cial in­ter­course with them­selves, and, more­over, too ob­tuse and in­tractable to be worth co­erc­ing into servi­tude. The African has been de­prived of free­dom and ren­dered “prop­erty”, be­cause he is by na­ture a tame man, sub­mis­sive, affec­tion­ate, in­tel­li­gent, and docile. He is so lit­tle cru­el, cun­ning, fierce, and vin­dic­tive, that the white men can op­press him far be­yond the lim­its of In­dian en­durance, and still trust their lives and prop­erty within his reach; while he is so in­tel­li­gent, that his labour is worth ac­quir­ing. The na­tive Amer­i­can is free, be­cause he is too dan­ger­ous and too worth­less a be­ing to be valu­able as a slave: the Ne­gro is in bondage, be­cause his na­tive dis­po­si­tions are es­sen­tially ami­able. The one is like the wolf or the fox, the other like the dog. In both, the brain is in­fe­rior in size, par­tic­u­larly in the moral and in­tel­lec­tual re­gions, to that of the An­glo-Saxon race, and hence the foun­da­tion of the nat­ural su­pe­ri­or­ity of the lat­ter over both; but my con­vic­tion is, that the very qual­i­ties which ren­der the Ne­gro in slav­ery a safe com­pan­ion to the White, will make him harm­less when free. If he were by na­ture proud, iras­ci­ble, cun­ning, and vin­dic­tive, he would not be a slave; and as he is not so, free­dom will not gen­er­ate these qual­i­ties in his mind; the fears, there­fore, gen­er­ally en­ter­tained of his com­menc­ing, if eman­ci­pat­ed, a war of ex­ter­mi­na­tion, or for su­premacy over the Whites, ap­pear to me to be un­found­ed; un­less after his eman­ci­pa­tion, the Whites should com­mence a war of ex­ter­mi­na­tion against him. The re­sults of eman­ci­pa­tion in the British West In­dia Is­lands have hith­erto borne out these views, and I an­tic­i­pate that the fu­ture will still far­ther con­firm them.

  1. If I may be per­mit­ted to try to im­prove on David Hume’s eval­u­a­tion of Berke­ley’s ide­al­ism (An En­quiry Con­cern­ing Hu­man Un­der­stand­ing, part 1): “This ar­gu­ment is drawn from Dr. Berke­ley; and in­deed most of the writ­ings of that very in­ge­nious au­thor form the best lessons of skep­ti­cism, which are to be found ei­ther among the an­cient or mod­ern philoso­phers, Bayle not ex­cept­ed. He pro­fess­es, how­ev­er, in his ti­tle-page (and un­doubt­edly with great truth) to have com­posed his book against the skep­tics as well as against the athe­ists and free-thinkers. But that all his ar­gu­ments, though oth­er­wise in­tend­ed, are, in re­al­i­ty, merely skep­ti­cal, ap­pears from this, that they ad­mit of no an­swer and pro­duce no con­vic­tion. Their only effect is to cause that mo­men­tary amaze­ment and ir­res­o­lu­tion and con­fu­sion, which is the re­sult of skep­ti­cism.” Or as Mencken puts it: “Meta­physics is al­most al­ways an at­tempt to prove the in­cred­i­ble by an ap­peal to the un­in­tel­li­gi­ble.” (p238, Mi­nor­ity Re­port, H. L. Menck­en’s Note­books, 1956) Imag­ine be­ing Neo in The Ma­trix and hear­ing that the ma­chines use hu­mans as en­ergy sources and ob­ject­ing, along with so many of the view­ers, that this is ab­surd given ther­mo­dy­nam­ics; but where did Neo learn about ther­mo­dy­nam­ics?↩︎

  2. Al­though in math­e­mat­ics, many proofs have turned out to be in­cor­rect not be­cause of the premises (which are typ­i­cally ax­iomatic and diffi­cult to say whether or not they are ‘true’), but be­cause of in­cor­rect in­fer­ence; this poses a to the to­tal re­li­a­bil­ity of any claims based on math­e­mat­i­cal mod­els in ad­di­tion to the un­re­li­a­bil­ity of their em­pir­i­cal in­puts. As says, “One can’t pro­ceed from the in­for­mal to the for­mal by for­mal means.”.↩︎

  3. Not that all skep­tics are nec­es­sar­ily up­set by this. For ex­am­ple, aims at un­der­min­ing ‘dog­ma­tism’ & faith in ab­stract rea­son­ing, elim­i­nat­ing ab­surd be­liefs to pro­duce .↩︎

  4. Al­though be­fore con­nect­ing this to the “back­fire effect” in psy­chol­o­gy, one should be aware that the back­fire effect has failed to repli­cate. There are, how­ev­er, sit­u­a­tions in which Bayesian rea­son­ers can draw di­a­met­ri­cally differ­ent con­clu­sions from the same ob­ser­va­tion based on their models/priors which can look like a ‘back­fire effect’: eg Bul­lock 2007, Jern et al 2009.↩︎

  5. Ap­par­ently coined by in his 1979 “The Prob­lem of Evil and Some Va­ri­eties of Athe­ism”.↩︎

  6. What then would be the point of show­ing a con­tra­dic­tion in arith­metic? Well, it would clar­ify the rea­sons why we use it, per Kripke (pg 66–67, , 1982):

    A skep­ti­cal so­lu­tion of a philo­soph­i­cal prob­lem be­gins… by con­ced­ing that the skep­tic’s neg­a­tive as­ser­tions are unan­swer­able. Nev­er­the­less our or­di­nary prac­tice or be­lief is jus­ti­fied be­cause—­con­trary ap­pear­ances notwith­stand­ing—it need not re­quire the jus­ti­fi­ca­tion the scep­tic has shown to be un­ten­able. And much of the value of the scep­ti­cal ar­gu­ment con­sists pre­cisely in the fact that he has shown that an or­di­nary prac­tice, if it is to be de­fended at all, can­not be de­fended in a cer­tain way.

  7. , episode 1↩︎

  8. For a sim­i­lar null re­sult from In­ter­net-wide tests of fu­ture-knowl­edge, see .↩︎

  9. Rossi 1987 (em­pha­sis added):

    …A pos­si­bil­ity that de­serves very se­ri­ous con­sid­er­a­tion is that there is some­thing rad­i­cally wrong with the ways in which we go about con­duct­ing eval­u­a­tions. In­deed, this ar­gu­ment is the foun­da­tion of a re­vi­sion­ist school of eval­u­a­tion, com­posed of eval­u­a­tors who are in­tent on call­ing into ques­tion the main body of method­olog­i­cal pro­ce­dures used in eval­u­a­tion re­search, es­pe­cially those that em­pha­size quan­ti­ta­tive and par­tic­u­larly ex­per­i­men­tal ap­proaches to the es­ti­ma­tion of net im­pacts. The re­vi­sion­ists in­clude such per­sons as Michael Pat­ton (1980) and Ego Guba (1981). Some of the re­vi­sion­ists are re­formed num­ber crunch­ers who have seen the er­rors of their ways and have been re­born as qual­i­ta­tive re­searchers. Oth­ers have come from so­cial sci­ence dis­ci­plines in which qual­i­ta­tive ethno­graphic field meth­ods have been dom­i­nant. Al­though the is­sue of the ap­pro­pri­ate­ness of so­cial sci­ence method­ol­ogy is an im­por­tant one, so far the re­vi­sion­ist ar­gu­ments fall far short of be­ing fully con­vinc­ing. At the root of the re­vi­sion­ist ar­gu­ment ap­pears to be that the re­vi­sion­ists find it diffi­cult to ac­cept the find­ings that most so­cial pro­grams, when eval­u­ate for im­pact as­sess­ment by rig­or­ous quan­ti­ta­tive eval­u­a­tion pro­ce­dures, fail to reg­is­ter main effects: hence the de­fects must be in the method of mak­ing the es­ti­mates. This ar­gu­ment per se is an in­ter­est­ing one, and de­serves at­ten­tion: all pro­ce­dures need to be con­tin­u­ally re-e­val­u­at­ed. There are some ob­vi­ous de­fi­cien­cies in most eval­u­a­tions, some of which are in­her­ent in the pro­ce­dures em­ployed. For ex­am­ple, a pro­gram that is con­stantly chang­ing and evolv­ing can­not or­di­nar­ily be rig­or­ously eval­u­ated since the treat­ment to be eval­u­ate can­not be clearly de­fined. Such pro­grams ei­ther re­quire new eval­u­a­tion pro­ce­dures or should not be eval­u­ated at all…

  10. Con­trol groups who merely do pre/post-tests and don’t oth­er­wise do any­thing; an ‘ac­tive’ con­trol group is given an ac­tiv­ity to do which is be­lieved to have no par­tic­u­lar ben­e­fit. Ac­tive con­trol groups help re­duce ‘ex­pectancy’ effects, wherein mem­bers of a pas­sive con­trol don’t try hard as they could on the post-test be­cause they “should­n’t” im­prove.↩︎

  11. This is a par­tic­u­larly com­mon use of modus tol­lens in statistics/genetics/psychology/machine learn­ing: many re­sults are ‘too good to be true’.↩︎

  12. Bryan Ca­plan, Book Re­view: Are We Get­ting Smarter?:

    Other sec­tion­s—­most no­tably Mr. Fly­n­n’s at­tack on the death penal­ty—are also tainted by se­ri­ous left­-wing bias. He is ea­ger to ar­gue that the “com­pe­tent” mur­der­ers of 1960 were “men­tally re­tarded” by mod­ern stan­dards. You could just as eas­ily con­clude, how­ev­er, that the “men­tally re­tarded” mur­der­ers of to­day are “com­pe­tent” by the stan­dards of 1960. Mr. Flynn briefly con­sid­ers this ar­gu­ment, and ob­jects that while IQ has risen, “prac­ti­cal in­tel­li­gence”—the “abil­ity to live au­tonomous lives”—has­n’t. If so, the “Flynn effect” has no effect on this de­bate: Can’t we sim­ply use un­ad­justed IQ scores as a proxy for prac­ti­cal in­tel­li­gence?’

  13. That sleep affects con­scious­ness & mem­ory is an un­con­tro­ver­sial claim; eg in the WSJ:

    One lit­tle known as­pect of in­som­nia is that the seem­ingly sleep­-de­prived often un­der­es­ti­mate (or over­es­ti­mate) how much shut-eye they’re get­ting, says Matt Bianchi, di­rec­tor of the sleep di­vi­sion at Mass­a­chu­setts Gen­eral Hos­pi­tal in Boston. “They could sleep seven hours in the sleep lab and they would say they did­n’t sleep one min­ute,” says Bianchi, adding that many pa­tients also wake up mul­ti­ple times with­out re­mem­ber­ing it. That dis­con­nect has sparked nu­mer­ous apps and gad­gets that offer to help peo­ple gauge how much sleep they’re get­ting.

  14. A par­tial Bjork­sten bib­li­og­ra­phy (see also Dean 2014):