Hydrocephalus and Intelligence: The Hollow Men

Some claim the disease hydrocephalus reduces brain size by 95% but often with normal or even above-average intelligence, and thus brains aren’t really necessary. Neither is true.
statistics, psychology, IQ, biology
2015-07-282020-05-13 finished certainty: highly likely importance: 3

Hydro­cephalus is a dam­ag­ing brain dis­or­der where flu­ids com­press the brain, some­times dras­ti­cally decreas­ing its vol­ume. While often extremely harm­ful or life-threat­en­ing when untreat­ed, some peo­ple with severe com­pres­sion nev­er­the­less are rel­a­tively nor­mal, and in one case (Lor­ber) they have been claimed to have IQs as high as 126 with a brain vol­ume 5% of nor­mal brains. A few of these case stud­ies have been used to argue the extra­or­di­nary claim that brain vol­ume has lit­tle or noth­ing to do with intel­li­gence; authors have argued that hydro­cephalus sug­gests enor­mous untapped cog­ni­tive poten­tial which are tapped into rarely for repairs and can boost intel­li­gence on net, or that intelligence/consciousness are non-ma­te­r­ial or tap­ping into ESP.

I point out why this claim is almost cer­tainly untrue because it pre­dicts count­less phe­nom­ena we never observe, and inves­ti­gate the claimed exam­ples in more detail: the cases turn out to be sus­pi­ciously unver­i­fi­able (Lor­ber), likely fraud­u­lent (Oliveira), or actu­ally low intel­li­gence (Feuil­let). It is unclear if high­-func­tion­ing cases of hydro­cephalus even have less brain mass, as opposed to lower proxy mea­sures like brain vol­ume.

I then sum­ma­rize anthro­pol­o­gist John Hawk­s’s crit­i­cisms of the orig­i­nal hydro­cephalus author: his brain imag­ing data could not have been as pre­cise as claimed, he stud­ied a selec­tive sam­ple, the story of the leg­endary IQ 126 hydro­cephalus patient raises ques­tions as to how nor­mal or intel­li­gent he really was, and hydro­cephalus in gen­eral appears to be no more anom­alous or hard-to-ex­plain than many other kinds of brain injuries, and in a com­par­ison, hemi­spherec­tomies, remov­ing or sev­er­ing a hemi­sphere, has pro­duced no anom­alous reports of above-av­er­age intel­li­gence (just deficit­s), though they ought to be just the same in terms of repairs or ESP.

That hydro­cephalus cases can reach roughly nor­mal lev­els of func­tion­ing, var­i­ous deficits aside, can be explained by brain size being one of sev­eral rel­e­vant vari­ables, brain plas­tic­ity enabling cog­ni­tive flex­i­bil­ity & recov­ery from grad­u­al­ly-de­vel­op­ing con­di­tions, and over­pa­ra­me­ter­i­za­tion giv­ing robust­ness to dam­age and poor envi­ron­ments, and learn­ing abil­i­ty. The field of deep learn­ing has observed sim­i­lar phe­nom­e­non in train­ing of arti­fi­cial neural net­works. This is con­sis­tent with Lor­ber’s orig­i­nal con­tention that the brain was more robust, and hydro­cephalus was more treat­able, than com­monly accept­ed, but does not sup­port any of the more exotic inter­pre­ta­tions since put on his find­ings.

In short, there is lit­tle anom­alous to explain, and stan­dard brain-cen­tric accounts appear to account for exist­ing ver­i­fied obser­va­tions with­out much prob­lem or resort to extra­or­di­nary claims.

An argu­ment recently sum­ma­rized by ex-bi­ol­o­gist SF author argues that intel­li­gence of human brains may have lit­tle to do with brain size vari­ables such as neu­ron count, rely­ing on an exam­ple of , a med­ical con­di­tion:

For decades now, I have been haunted by the grainy, black­-and-white x-ray of a human skull. It is alive but emp­ty, with a cav­ernous flu­id-filled space where the brain should be. A thin layer of brain tis­sue lines that cav­ity like an amni­otic sac. The image hails from a , the author, reports that the patient in ques­tion had “vir­tu­ally no brain”. But that’s not what scared me; hydro­cephalus is noth­ing new, and it takes more to creep out this ex-bi­ol­o­gist than a pic­ture of Ven­tri­cles Gone Wild.

The stuff of night­mares. (From )

What scared me was the fact that this vir­tu­ally brain-free patient had an IQ of 126. He had a first-class hon­ors degree in math­e­mat­ics. He pre­sented nor­mally along all social and cog­ni­tive axes. He did­n’t even real­ize there was any­thing wrong with him until he went to the doc­tor for some unre­lated mal­a­dy, only to be referred to a spe­cial­ist because his head seemed a bit too large.

…The authors advo­cate research into “Com­pu­ta­tional mod­els such as the smal­l­-world and scale-free net­work”—net­works whose nodes are clus­tered into high­ly-in­ter­con­nected “cliques”, while the cliques them­selves are more sparsely con­nected one to anoth­er. De Oliveira et al sug­gest that they hold the secret to the resilience of the hydro­cephalic brain. Such net­works result in “higher dynam­i­cal com­plex­i­ty, lower wiring costs, and resilience to tis­sue insults.” This also seems rem­i­nis­cent of those iso­lated hyper­-effi­cient mod­ules of autis­tic savants, which is unlikely to be a coin­ci­dence: net­works from social to genetic to neural have all been described as “smal­l­-world”…The point, though, is that under the right con­di­tions, brain dam­age may para­dox­i­cally result in brain enhance­ment. Smal­l­-world, scale-free net­work­ing—­fo­cused, inten­si­fied, over­clocked—might tur­bocharge a frag­ment of a brain into act­ing like the whole thing. Can you imag­ine what would hap­pen if we applied that trick to a nor­mal brain?

Big if true. This is cer­tainly big if true: were 90%+ of the brain unnec­es­sary, what could it do if the rest of it was used as effi­ciently as the small per­cent­age doing the actual work? And this over­throws the entire under­stand­ing of what the brain is, how it func­tions, and how intel­li­gence works. Not to men­tion it has enor­mous impli­ca­tions for deep learn­ing: why do we throw these vast resources into train­ing ever-larger mod­els in order to increase their capa­bil­i­ties, if all those neu­rons actu­ally aren’t doing any­thing? We could boost AI capa­bil­i­ties dra­mat­i­cally by dis­pens­ing with the redun­dan­cy.

Implausible Implications

These claims how­ever run into a lot of objec­tions:

  1. All the exist­ing evi­dence that Think­ing Is What Neu­rons Do: If neu­ron counts and cor­re­lates like brain vol­umes are so irrel­e­vant, why does point to their close cor­re­la­tion & causal­i­ty, from cross-species com­par­isons to human genetic stud­ies (eg / & )? How do we explain lesion stud­ies or trau­matic brain injuries or…

  2. Evo­lu­tion­ar­ily Absurd: if most of the brain can be dis­pensed with, then, as one of the most, if not the most, meta­bol­i­cal­ly-ex­pen­sive tis­sues in the body, why does evo­lu­tion not extremely aggres­sively prune brains in humans?

    And if it works in humans, then why not every­where in the nat­ural world? (Where are the hydro­cephalus pri­mates, or birds, or any mam­mals? Where are the sib­ling species where one has 20x the neu­rons of the oth­er, because the lat­ter has ‘flipped the switch’ and got­ten the hydro­cephalus ben­e­fit?) Differ­ences in total vol­ume or brain den­sity are one thing, but a neu­ron is a neu­ron and will always be expen­sive to build, main­tain, and oper­ate—if 5% works fine or bet­ter, why an extrav­a­gance of 20x?

  3. Fer­mi’s prob­lem: “Where Are The Aliens”? if the brain can work around such dam­age to the extent of boost­ing intel­li­gence by 2SDs eas­ily while hav­ing only a frac­tion of the brain to work with, why do we not see this workaround trig­gered by any drug, surgery, or envi­ron­men­tal inter­ven­tion, or trig­gered spon­ta­neously at ran­dom, yield­ing poten­tially any­where up to 40S­Ds? We live in a big world with bil­lions of humans, and any­thing that can hap­pen will hap­pen ()—ev­ery­thing from peo­ple to peo­ple . Hydro­cephalus is not that rare, and this mech­a­nism must not be rare either if such extreme cases can be found so appar­ently eas­i­ly. There should not be only a sin­gle for­get­table case.

    For per­spec­tive on what an absolute change of 20x neu­rons might look like, ; why do we not see any humans where the gap between their intel­li­gence and ours is many times larger than that between us and chim­panzees?

  4. Where Are The Geniuses? Sim­i­lar­ly, sup­pos­ing that the repair mech­a­nism can never kick in for healthy humans, result­ing in super­in­tel­li­gences, why does the sup­posed excess reported by Lor­ber not show up any­where? If hydro­cephalus can actu­ally boost intel­li­gence on aver­age, why do we not see gross over-rep­re­sen­ta­tions among high­-IQ cohorts or pro­fes­sions or sta­tus­es?

    It is a basic point of order sta­tis­tics & tail effects that a small increase in a mean or vari­ance com­pared to a ref­er­ence group means that over­rep­re­sen­ta­tion will increase the more extreme the cas­es; even if we do not see super-hu­mans, we should still be struck by all the elite researchers who turn out to be hydro­cephalic (and the more elite they are, the worse their hydro­cephalus should be). But at least as far as Wikipedia is con­cerned, there are few hydro­cephal­ics of note for any rea­son.

Extra­or­di­nary claims require extra­or­di­nary evi­dence. could seri­ously per­suade us that every­thing we know about the brain is wrong and that we need to reboot neu­ro­science, as enthus­es, to study how mem­ory and intel­li­gence are actu­ally stored & con­ducted

in some extremely min­ute, sub­atomic, form, as yet unknown to bio­chemists and phys­i­ol­o­gist­s…out­side the body—ex­tra­cor­po­re­al!…the brain [is] as a receptor/transmitter of some form of elec­tro­mag­netic wave/particle… of course, when speak­ing of extra­cor­po­real mem­ory we enter the domain of “mind” or “spirit” with cor­re­spond­ing meta­phys­i­cal impli­ca­tions.

Problems With The Case Studies

Fake data. For­tu­nately for Watts’s sleep, the case for hydro­cephalus is much worse than it looks. The brain scan he posts is not, in fact, of the IQ 126 case; Oliveira cap­tions it as images from his lab of a nor­mal person/normal brain, a nor­mal person/hydrocephalus, and a hydro­cephalus patient with “deep cog­ni­tive and motor impair­ments”. Fur­ther, Oliveira et al lied about the ori­gin of the images, which were copied from else­where, and the paper has been for­mally retract­ed.1

Sus­pi­ciously unver­i­fi­able anec­dotes. OK, but what about the Lewin ‘review’ (which is really more of a jour­nal­is­tic arti­cle) and the IQ 126 guy? Lewin pro­vides a pair of brain scans as well, but it’s impor­tant to note that he never implies that that scan had any­thing to do with that patient, and is retelling Lor­ber’s anec­dote at third hand. Lor­ber pro­vides no con­crete details about him, includ­ing such basics as what sort of IQ test and when it was done (mak­ing it poten­tially as mis­lead­ing as the peren­nial claim that Richard Feyn­man had a sim­i­lar IQ, based on an anec­dote in his biog­ra­phy about being tested as a child). Sur­pris­ing­ly, as far as I can tell, Lor­ber and asso­ciates (Lor­ber died 16 years lat­er, in 1996) have not pub­lished any­thing at all on their dataset in the 39 years since the Lewin press cov­er­age, and so there are no scans or data on this guy, or any of the oth­ers Lor­ber claimed to have above-nor­mal intel­li­gence.2

Mis­lead­ingly described anec­dotes. How many severe hydro­cephalus cases have even above-av­er­age intel­li­gence? Despite Lor­ber’s claim to have found many cases of nor­mal and a highly above-av­er­age hydro­cephalus patient eas­i­ly, sub­se­quent researchers appear to have failed to do like­wise over the ensu­ing 4 decades. , cited by Watts, him­self only cites 3 instances: Lor­ber’s unver­i­fi­able anec­dotes via Lewin, the retracted & likely fraud­u­lent Oliveira et al, and a third case study of who reports their patient hav­ing an IQ of 75.3 (I am puz­zled how Forsy­dke could pos­si­bly describe Feuil­let’s IQ of 75, nearly retard­ed, as one of “two inde­pen­dent con­fir­ma­tions” of Lor­ber, who was claim­ing a case with an IQ 3.4SDs higher than that; he says the same thing in Forsy­dke 2015 in crit­i­ciz­ing Hawk­s’s crit­i­cism, say­ing that Lor­ber had been vin­di­cated by Oliveira & Feuil­let. Forsy­dke oth­er­wise dodges all of Hawk­s’s points.)

Increas­ingly less than meets the eye. So there appear to be few ver­i­fi­able instances of severe hydro­cephal­ics with above-av­er­age, much less highly above-av­er­age, intel­li­gence. (I also have to won­der what more in-depth test­ing of cas­es, going beyond sum­mary or ful­l-s­cale IQ tests to many other mea­sures of cog­ni­tive func­tion like com­plex reac­tion time, and tests over a lifes­pan, would show; one can only detect deficits in what & when one tests, after all.)

But let us take Lor­ber at face-value and ask, does a brain scan show­ing, say, 95% of vol­ume taken up by fluid imply that they have only 5% of the brain as a nor­mal per­son or only 5% of the neu­rons?

Hawks on Lorber

Anthro­pol­o­gist con­sid­ers what hydro­cephalus means and the Lor­ber anec­dote in 2007. He points out that:

  1. Mea­sure­ment Error: 1970s CT scans were extremely inac­cu­rate; it can­not detect accu­rately at the 1mm as claimed, and can­not show that the cor­tex was uni­formly only 1mm, which is not that far from a nor­mal human’s 2mm thick­ness. There may well be enough white mat­ter total to still con­nect up the grey mat­ter. As Lowin noted in 1980, the inter­pre­ta­tion of brain scans is unre­li­able and may not give a good esti­mate of what exactly is there and how much.

    Fur­ther, as few or no hydro­cephalus cases in the nor­mal range have been sub­ject to detailed scan­ning or post-mortem, it is unclear if a given hydro­cephalus case involves any net loss of neu­rons. It is pos­si­ble that the high pres­sure merely increases neural den­si­ty, which explains much of the cross-species differ­ences already4 (One would not expect this to be a good thing, but then, hydro­cephalus usu­ally isn’t.)

  2. Absurd Intel­li­gence Boosts: Lor­ber’s claimed sta­tis­tics imply increased intel­li­gence, not merely robust­ness to dam­age:

    But the notion that “half” of the patients where ven­tri­cle expan­sion is greater than 95% of the cra­nium have IQ’s greater than 100 is math­e­mat­i­cally implau­si­ble. The defi­n­i­tion of IQ is that the mean is 100. This means that only half of peo­ple with­out ven­tri­cle expan­sion have IQ over 100. Lor­ber seems to have claimed that the most severe cases of hydro­cephalus actu­ally see an increase in the pro­por­tion of high­-IQ indi­vid­u­als, despite “many” being severely dis­abled. I’m not say­ing it’s impos­si­ble, but like “a mil­lime­ter or so,” this is the kind of sta­tis­tic that deserves skep­ti­cism.

  3. Selec­tion Bias: Lor­ber was highly selec­tive in the cases he stud­ied, pre­cisely because he was look­ing for anom­alies; such cases intrin­si­cally will have unknown com­pen­sat­ing fac­tors (such as high genetic poten­tial for intel­li­gence, which can buffer the dam­age), and can yield highly mis­lead­ing results through things like

  4. Miss­ing Caveats: the story of the IQ 126 stu­dent is bizarre, unless Lor­ber is leav­ing some­thing out­—who is referred to a hydro­cephalus spe­cial­ist for expen­sive & unusual brain imag­ing scans (espe­cially back then) just because “his head seemed a bit too large”?

    Nor is it obvi­ous from the reports that the con­di­tion had “no” cog­ni­tive man­i­fes­ta­tions. Much seems to depend on the sin­gle case described above, with an appar­ently nor­mal col­lege stu­dent walk­ing in off the street to dis­cover he had min­i­mal brain mass. But this story is quite obvi­ously incred­i­ble as pre­sent­ed: most neu­rol­o­gists don’t per­form brain scans just because a col­lege stu­dent wears a large hat. It seems rea­son­able to infer that the stu­dent was referred by his doc­tor to Lor­ber, a hydro­cephalus spe­cial­ist, for some rea­son. We can only guess what the rea­son might be, but it hardly gives con­fi­dence in the anec­dote!

  5. Ques­tion­able Dam­age: Hawks con­cludes that anom­alies may not be too use­ful to study here:

    With­out ques­tion, there are many patients who have this out­come—no sig­nifi­cant cog­ni­tive deficit com­pared to non­pa­tients, despite pro­found pathol­o­gy. This is true of almost any pathol­ogy affect­ing the brain, includ­ing tumors, strokes, and devel­op­men­tal abnor­mal­i­ties. The ques­tion is whether this pro­vides a valid model for under­stand­ing the adap­tive impor­tance of brain vol­ume. It seems that later onset hydro­cephalus, where a nor­mal brain is com­pressed within a rel­a­tively nor­mal-sized skull by cere­brospinal fluid pres­sure, does not really apply to the evo­lu­tion­ary ques­tion. The reported cases do not appar­ently involve sig­nifi­cant gray mat­ter tis­sue loss. A “thin” cor­tex does not nec­es­sar­ily imply func­tion­ally small cor­ti­cal vol­ume, even with sub­stan­tial white tis­sue loss.

Coun­ter-ex­am­ple: no high­-IQ hemi­spherec­tomies. Hawks then goes on to exam­ine , an oper­a­tion to treat the most severe epilep­tic cases by sev­er­ing one brain hemi­sphere from another or remov­ing it entire­ly. Unsur­pris­ing­ly, hemi­spherec­tomy patients have many severe cog­ni­tive prob­lems: note, among their many language/reading/behavioral/school prob­lems requir­ing inten­sive inter­ven­tions such as spe­cial school­ing (as well as evo­lu­tion­ar­i­ly-rel­e­vant side-effects like par­tial paral­y­sis), that only “5 (21%) of the 24 patients older than 18 years of age were gain­fully employed.” This is still a suc­cess for hemi­spherec­to­my, since severe epilepsy badly dam­ages nor­mal devel­op­ment and can be fatal (as Hawks points out, since the brain tis­sue being removed or sev­ered is so badly dam­aged already, it’s not nec­es­sar­ily much of a loss), and inci­den­tally demon­strates how / allows com­pen­sa­tion for los­ing access to a whole hemi­sphere, at least when done young enough. Another extreme case is : the true loss of brain, unlike hydro­cephalus, but there are no case reports of nor­mal or above-av­er­age cog­ni­tive func­tion­ing; hydra­nen­cephaly cas­es, while nor­mal-seem­ing for months after birth, are instead pro­foundly dis­abled, barely respond to stim­uli, and are lucky to live for more than a few years (), with 19 years being the record (the case).

What Does Hydrocephalus Mean?

Hawks con­cludes:

So the case of hemi­spherec­tomy does not test the propo­si­tion that nor­mal cog­ni­tive per­for­mance is pos­si­ble after a great reduc­tion in brain size. Instead it pos­si­bly tests the propo­si­tion that a reduc­tion in brain size may be con­sis­tent with nor­mal cog­ni­tive per­for­mance under a spe­cial­ized cul­tural and envi­ron­men­tal regime. That hypoth­e­sis is refuted by the major­ity of cases in the clin­i­cal record, for whom the spe­cial­ized learn­ing envi­ron­ment has not man­aged to elim­i­nate devel­op­men­tal deficits. Still, for many patients some com­bi­na­tion of surgery, ther­apy and learn­ing assis­tance do make a deci­sive differ­ence, and they attain nor­mal cog­ni­tive per­for­mance—even nor­mal for devel­op­men­tal age.

How do these cases apply?

There is no sin­gle con­clu­sion that we can draw from these exam­ples of extreme patho­log­i­cal reduc­tion in brain size in humans. Clear­ly, the brain is capa­ble of remark­able plas­tic­ity in devel­op­ment, includ­ing alter­nate local­iza­tions of some func­tions that are highly local­ized in most adults. But can we apply this plas­tic­ity more gen­er­al­ly, to sug­gest that almost any brain struc­ture might have evolved in ancient human pop­u­la­tions? Even those that involve immense reduc­tions in over­all brain size?

Robust­ness to dam­age & envi­ron­men­tal sub­op­ti­mal­i­ty. It should be men­tioned that these assess­ments build on a rather nar­row view of “cog­ni­tion.” For instance, all hemi­spherec­tomy patients have some paral­y­sis on the oppo­site side from the removed hemi­sphere. The func­tions of the motor and sen­sory cor­tices of the absent side do not appear to have the devel­op­men­tal plas­tic­ity exhib­ited by lan­guage. From the per­spec­tive of fit­ness in pre­his­toric human pop­u­la­tions, the ade­quate con­trol of move­ment and per­cep­tion of sen­sory infor­ma­tion would have a sub­stan­tially greater impor­tance than in today’s cul­tural milieu. So a reduc­tion in brain size that impacts motor and sen­sory func­tion but leaves other aspects of cog­ni­tion intact cer­tainly can­not be said to have no impact. Just because a reduc­tion in per­for­mance can be man­aged within our pop­u­la­tion does not mean that it could have evolved in some past pop­u­la­tion. Also, the attain­ment of “nor­mal” cog­ni­tion, how­ever defined, requires sub­stan­tial invest­ment and teach­ing for the aver­age human. Humans with devel­op­men­tal chal­lenges often can attain nor­mal cog­ni­tive per­for­mance for their age, par­tic­u­larly when sup­ple­men­tary teach­ing and ther­apy is avail­able. All this is to say that human brains are coad­apted with behav­ioral pat­terns that chan­nel devel­op­men­t…Although it may be con­ceiv­able—even if it is far from demon­strat­ed—that a rad­i­cally smaller brain cou­pled with a spe­cial­ized cul­ture might have increased fit­ness, appar­ently there was no avail­able evo­lu­tion­ary path­way to that adap­ta­tion. I would guess that it is sim­ply more diffi­cult to main­tain the nec­es­sary cul­tural spe­cial­iza­tions for such an adap­ta­tion within the con­text of ancient human pop­u­la­tion struc­ture. It is eas­ier to accom­plish devel­op­ment with a large brain that can employ many bot­tom-up strate­gies to build its cog­ni­tive abil­i­ties.

Thus the lost (or just com­pressed) white mat­ter may be no more ‘unnec­es­sary’ than, say, one’s spare lung or kid­ney, most of one’s liv­er, or any of one’s limbs… , describ­ing a mouse hydro­cephalus case, ask:

Sur­vival requires sens­ing the envi­ron­ment, pro­cess­ing the infor­ma­tion, and respond­ing appro­pri­ate­ly. There is no mor­tal threat or com­pe­ti­tion to sur­vive for a rat main­tained in a cage for two years under con­trolled envi­ron­men­tal con­di­tions with food and water ad libi­tum. R222 is a study of the fun­da­men­tal neu­ro­bi­o­log­i­cal and behav­ioral processes that sus­tain a resource ade­quate, safe, ambu­la­tory life. The sever­ity of the hydro­cephalus in R222 in the face of nor­mal body weight and growth, nor­mal motor behav­ior and spa­tial mem­o­ry, and evoked activ­ity to smells, tac­tile stim­u­la­tion and vision would sug­gest neu­road­ap­ta­tion to a life-long abnor­mal­i­ty. This rare case can be viewed as one of nature’s mir­a­cles pro­vid­ing the unique oppor­tu­nity to exam­ine the brain’s capac­ity for neu­ro­plas­tic­ity and reor­ga­ni­za­tion nec­es­sary for sur­vival…Is the cor­tex nec­es­sary? There have been numer­ous stud­ies across a vari­ety of mam­mals look­ing at the devel­op­men­tal con­se­quences of rad­i­cal decor­ti­ca­tion in neonates24,25,26,27,28,29,30. While there are minor deficits par­tic­u­larly in some motor pat­terns and motor coor­di­na­tion, the decor­ti­cate ani­mal can eat, drink, sleep and grow to nor­mal size. They respond to visual and audi­tory stim­uli. They dis­play nor­mal species-spe­cific social, mater­nal, aggres­sive and sex­ual behav­iors. They mate and repro­duce. Nature again has pro­vided sci­ence with an extreme form of decor­ti­ca­tion in human­s–hy­dra­nen­cepha­ly, a rare, inher­ited dis­or­der where by babies are born with­out cere­bral hemi­spheres. There is no treat­ment, yet incred­i­bly, with the proper care and sta­bi­liza­tion these indi­vid­u­als can live for years31, not in a veg­e­ta­tive state and are respon­sive to their sur­round­ings.

This might sound rea­son­able, but how does Fer­ris define ‘nec­es­sary’? Human hydra­nen­cephaly cas­es, as dis­cussed, are lit­tle bet­ter off than coma vic­tims: McAbee describe their first case’s “respon­sive to their sur­round­ings” as con­sist­ing of “turn[ing] his head ipsi­lat­er­ally to sounds, music, and venipunc­ture.” & requir­ing a feed­ing tube, and the sec­ond case “He had no defi­nite aware­ness of his envi­ron­ment. The pupil­lary response to light was min­i­mal; he did not fix or fol­low visu­ally and had no visual response to threat.” How about the mat­ing and repro­duc­ing—­surely com­plex behav­iors, and if decor­ti­ca­tion allows even that, what do we need a cor­tex for? Look­ing at Fer­ris et al 2019’s cita­tions, the only two ref­er­ences for that are less than impres­sive: for females (), “mate and repro­duce” appar­ently means only pas­sively lay­ing in place with one’s rump ele­vated in , and for the males (), when paired with recep­tive (nor­mal) females, they were more infer­tile than nor­mal rats and had to be man­u­ally “stim­u­lated” by the researchers! These cases show that the cor­tex is unnec­es­sary only if one has a low stan­dard for what is nec­es­sary, which is detached from any real-world con­sid­er­a­tions.

I would pre­dict that deeper study of such hydro­cephalus cases would reveal the fol­low­ing:

  1. far less neural loss than expected based on vol­ume changes, con­sis­tent with loss of a large per­cent­age being crip­pling or fatal

  2. a cor­re­la­tion between neu­ron loss & per­for­mance deficits, where high­er-per­form­ing hydro­cephalus is due to greater den­sity for a given vol­ume reduc­tion or the same mass being dis­trib­uted in a weirder and hard­er-to-mea­sure way

  3. sys­temic per­for­mance deficits, where small deficits in each task add up to large total deficits on com­plex tasks—in­creas­ing in harder more-nat­u­ral­is­tic tasks requir­ing more mem­o­ry, learn­ing, and cog­ni­tive flex­i­bil­ity

  4. repro­duc­tive fit­ness penal­ties, with greater fit­ness penal­ties in envi­ron­ments which are more com­pet­i­tive & resem­ble the ances­tral envi­ron­ment (eg the ), as opposed to mod­ern envi­ron­ments which may be unde­mand­ing.

    Con­sid­er­ing cases like Feuil­let’s case, who worked as a French “civil ser­vant”, one might won­der if this indi­cates intel­li­gence is not all that impor­tant: how could some­one with an IQ of 74 so much as feed and dress them­selves, one might say, much less hold a “white-col­lar job” or be mar­ried with chil­dren?

    But one should remem­ber that cases of men­tal retar­da­tion may be caused by dis­or­ders that have many pro­foundly dis­abling side-effects, exag­ger­at­ing the dys­func­tion­al­i­ty; and that oth­er­wise-healthy nor­mal indi­vid­u­als can have many cop­ing strate­gies & sup­port­ive envi­ron­ments even while still hav­ing aston­ish­ing deficits. This is com­pa­ra­ble to cog­ni­tive decline & senil­ity in the elder­ly, or to adult illit­er­a­cy: illit­er­ate adults (often due to dyslex­ia) can be sur­pris­ingly func­tional & fool every­one who knows them by exploit­ing a range of tricks, rang­ing from heavy reliance on oral com­mu­ni­ca­tion like tele­phones and get­ting other peo­ple to read for them to avoid­ing writ­ing-heavy envi­ron­ments in favor of con­crete ones to skilled guess­ing to mem­o­riz­ing spe­cific pieces of writ­ing & pre­tend­ing to read it to sim­ply lying. Human Rights Watch 2001 describes how retarded crim­i­nal defen­dants, even ones being pros­e­cuted , would deny they are retarded or fool their lawyers into sign­ing them up for col­lege cal­cu­lus or lied about fin­ish­ing high school.

    Nev­er­the­less, despite the illu­sion of com­pe­ten­cy, the retarded may still have pro­found deficits, as Human Rights Watch 2001 then goes on to give exam­ples: defen­dants who don’t know how to use stamps or what their stom­ach does, who count on their hands, who will agree to any­thing an author­ity fig­ure says, who don’t know why rape is wrong (sug­gest­ing because they lacked “per­mis­sion” or it was “against her reli­gion”), who under­stand stab­bing but not shoot­ing kills some­one, who smile or sleep through their tri­als, who ask what to wear to their funeral after being exe­cut­ed… In ordi­nary cir­cum­stances, which per­mit extreme lev­els of rep­e­ti­tion & reliance on oth­ers to sup­port a safe envi­ron­ment, they may appear as func­tional as any­one else, yet they will make rare (but extremely impor­tant) errors and can­not han­dle novel prob­lems.5 (To ruin one’s life, one only has to screw up once.) Gre­go­ry’s McNa­ma­ra’s Fol­ly: The Use of Low-IQ Troops in the Viet­nam War describes many exam­ples from of retarded or low-in­tel­li­gence men who could func­tion in a safe civil­ian envi­ron­ment doing sim­pli­fied menial labor, but could not be trained to do any­thing safely in the mil­i­tary or cope with novel envi­ron­ments, suffer­ing extreme casu­alties, often self­-in­flict­ed. While any indi­vid­ual instant may seem unim­por­tan­t—there are many ways to avoid need­ing to under­stand how to use stamps one­self, and most peo­ple would never rape a woman even if they don’t under­stand why rape is bad—they add up over a life­time and push the entire pop­u­la­tion towards bad out­comes. One might note that Feuil­let’s case had a below-av­er­age num­ber of chil­dren and his “white-col­lar” job may mean noth­ing more than stamp­ing the same form for 40 years in the DMV as a kind of covert wel­fare; or con­sider the pro­found con­se­quences for indi­vid­u­als with reduced/no pain sen­si­tiv­ity.

Analogies from Deep Learning

Arti­fi­cial par­al­lels in Deep Learn­ing. Hawk­s’s dis­cus­sion of chan­nel­ing leads to an inter­est­ing par­al­lel with deep learn­ing (DL): it has been repeat­edly estab­lished that all neural net­works are ‘over­pa­ra­me­ter­ized’ in the sense that what­ever per­for­mance a NN reaches at the end of train­ing, one can eas­ily ‘dis­till’ or ‘com­press’ it into a smaller (or faster) NN with slightly worse per­for­mance, by at least a frac­tion and some­times by orders of mag­ni­tudes.

Despite neuron/parameter count clearly caus­ing NN ‘intel­li­gence’. Para­me­ter & neu­ron counts in NNs clearly do cause big­ger NNs are more expres­sive than smaller ones, because when we exper­i­men­tally vary neural net size, we see , even for the largest train­able NNs; but at the same time, we still see that of course size does not explain 100% of vari­ance in per­for­mance—the same archi­tec­ture trained in the same way on the same data will yield NNs with notice­ably differ­ent final per­for­mance, sim­ply due to the sto­chasc­ity of the ini­tial­iza­tion, reg­u­lar­iza­tion mech­a­nisms like dropout, and data selec­tion6, and it would be pos­si­ble to find runs where a smaller NN out­per­forms a larger NN which is oth­er­wise exactly the same, yet, nev­er­the­less, larger is more pow­er­ful. Sim­i­lar­ly, distillation/compression work quite well, and yet can­not be tap­ping into any sort of dual­ism or psi.

DL demon­strates func­tional ben­e­fits of over­pa­ra­me­ter­i­za­tion. Why? The recent work on the strongly indi­cates that most of the work is being done by a rel­a­tively small part of the net­work, and the expres­sive power of the full mod­el, which is enor­mously rich, is largely unnec­es­sary for the task we apply it to. This raises the puz­zling ques­tion of how we can get away with train­ing such large mod­els with­out them over­fit­ting or learn­ing use­less things, when stan­dard sta­tis­ti­cal & machine learn­ing the­ory teaches that this ought to be a car­di­nal sin and guar­an­teed futile. And pecu­liar­ly, know­ing that our NN is far too big & we only need a small NN does not let one train the final smaller NN direct­ly: if one takes a smaller NN and trains a new from-scratch NN with the same size (which one knows is big enough by con­struc­tion), it sim­ply does­n’t work.

Over­pa­ra­me­ter­i­za­tion means always pos­si­ble improve­ments. Why not? The answer seems to be that over­pa­ra­me­ter­i­za­tion is nec­es­sary to enable effi­cient learn­ing, by mak­ing the ‘fit­ness land­scape’ smooth and ensur­ing there is always a para­me­ter which can be tweaked to make some learn­ing pro­gress, while smaller NNs are more likely to get ‘stuck’ in a bad local opti­mum, and have no way out. (An exam­ple of a ‘rough’ fit­ness land­scape would be almost all com­puter pro­gram­ming lan­guages: delet­ing a sin­gle char­ac­ter like a semi­colon may make a pro­gram go from 100% per­for­mance on a task to 0% and per­haps may not even com­pile.) This is par­tially due to hav­ing so many ran­dom sub­-net­works that at least one of them ‘wins the lot­tery’ and is a good fit for the cur­rent task. Thus, one can start off with a big model and then learn effec­tive­ly, and spe­cial­ize down into a much smaller but inflex­i­ble mod­el, but one can­not learn the smaller model to begin with from the raw data (un­less one is exceed­ingly lucky and the small model starts off with a use­ful ini­tial­iza­tion and hap­pens to not run into any trap­s); the small model can only be cre­ated based on heavy guid­ance from the big mod­el.

For a human brain, which must do life­long learn­ing on a con­stantly chang­ing mix of tasks, in a harsh evo­lu­tion­ary con­text filled with com­pet­i­tive zero-sum dynam­ics where small differ­ences can mat­ter & there is no ‘big model’ to copy or learn from, this sug­gests that rely­ing on a small brain would be a ter­ri­ble idea. Since—un­like soft­ware mod­el­s—here is no way to train a human brain to adult­hood and then scoop it out and shrink it to a much more effi­cient ver­sion and copy it every­where else, regard­less of how good it has become, every human brain must start effec­tively from scratch, and can­not ben­e­fit from such com­pres­sion, and human brains must be over­pa­ra­me­ter­ized to ensure it can meet what­ever chal­lenges the world throws at it. (Soft­ware, on the other hand, can be copied, which has inter­est­ing impli­ca­tions for the ‘hard­ware over­hang’ argu­men­t.)

Conclusion: No Evidence

Unlike­ly, poor evi­dence, and eas­ily explained. So to sum up: peo­ple have claimed that hydro­cephalus destroys most of the brain yet hav­ing only a frac­tion of the brain is con­sis­tent with nor­mal or above-av­er­age intel­li­gence, and may even increase it. This is extremely implau­si­ble based on every­thing we know about intel­li­gence and evo­lu­tion and pop­u­la­tion dis­tri­b­u­tions and is a bad descrip­tion of what hydro­cephalus does (con­flat­ing dis­tri­b­u­tion & vol­ume with brain mat­ter). On aver­age, hydro­cephalus and sim­i­lar things like hemi­spherec­tomy do (as expect­ed) induce many deficits rang­ing from mild to sev­ere, which can be accom­mo­dated to some degree by neural plas­tic­i­ty, indi­vid­ual differ­ences, exten­sive envi­ron­men­tal inter­ven­tions, and free­dom from nat­ural selec­tion. Claims to the con­trary, aside from fail­ing to deal with the many objec­tions that this is implau­si­ble, turn out to be based on undoc­u­ment­ed, fraud­u­lent, or mis­lead­ingly described cas­es, and pri­mar­ily pushed by cranks. Ulti­mate­ly, hydro­cephalus does not appear to present any par­tic­u­lar chal­lenge to the stan­dard under­stand­ing of intel­li­gence as being caused by a mate­r­ial brain whose effi­ciency at cog­ni­tive tasks is dri­ven by neu­ron count, wiring pat­terns, neural integri­ty, and gen­eral health: not only is the evi­dence extra­or­di­nar­ily inad­e­quate to jus­tify the extra­or­di­nary claims made by some authors, it is unclear how much, if any, evi­dence there at all.

  1. My notes say the pla­gia­rism was first spot­ted by Ondřej Havlíček on the Neu­roskep­tic blog. Why would any researcher lie about images like that? Who knows. But it demon­strates why one can’t put too much trust in any sin­gle dat­a­point. I have raised my con­cerns about the implau­si­bil­ity of these hydro­cephalus claims & the retrac­tion with Watts on his blog and 3 emails so far.↩︎

  2. I find such lacu­nae to be sus­pi­cious. When you look at the most ques­tion­able find­ings and exper­i­ments in Repro­ducibil­ity Cri­sis & psy­chol­o­gy-re­lated mat­ters, like the , , or , one of the more com­mon hall­marks is a sin­gle extra­or­di­nary find­ing which makes a splash with press atten­tion, and then a strange discli­na­tion by the orig­i­nal author to repli­cate their own find­ing or pub­lish any­thing fur­ther on it which is not a sec­ondary-source-style rehash.↩︎

  3. Feuil­let et al 2007 is some­times mis­quoted as mea­sur­ing an IQ of 84; that refers only to the high­est sub­test score, not the total test score.↩︎

  4. See Her­cu­lano-Houzel’s research, like , Her­cu­lano-Houzel 2012.↩︎

  5. One might say that this is inher­ently true of any kind of abstrac­tion or con­cept or model or long-term mem­o­ry: if one needed them fre­quently, they could be han­dled directly by tri­al-and-er­ror or short­-term mem­o­ry. If eval­u­ated in ordi­nary com­mon cir­cum­stances, their value will be hid­den. They are valu­able because they can extrap­o­late to novel cir­cum­stances, or while used rarely and most uses of lit­tle val­ue, a few uses are extremely impor­tant. A fact may be use­ful just once in a life­time, but that use may save one’s life; and, in all the years up to the dan­ger, such a per­son would look iden­ti­cal to some­one who did­n’t or could­n’t learn that fact. One is reminded of how neural net-gen­er­ated text like can look human-writ­ten save for a sin­gle stray char­ac­ter—a “her” instead of “he”—which shat­ters the illu­sion.↩︎

  6. Or hard­ware non­de­ter­min­ism— and not guar­an­teed to return the same result for a given oper­a­tion! This is just one of the many issues with exact repro­ducibil­ity in deep learn­ing. Deep rein­force­ment learn­ing par­tic­u­larly strug­gles with this lack of repro­ducibil­ity and still higher vari­ance than reg­u­lar DL.↩︎